Climate Risk Management in Agricultural Extension for Ethiopia 2nd Edition REFERENCE GUIDE Recommended citation Dinku T, Hansen J, Trzaska S, Grossi A, Nsengiyumva G, Siebert A, Han E, Mangen O, Lemma E, Dendoba T, Solomon D, Demissie T. Climate Risk Management in Agricultural Extension in Ethiopia Reference Guide. 2nd ed. Palisades, NY, USA: International Research Institute for Climate and Society (IRI), Columbia Climate School; 2024. Acknowledgements This collaborative effort was made possible by the sustained partnership, thoughtful inputs, and critical feedback from the Ethiopia Ministry of Agriculture, Ethiopian Institute for Agricultural Research (EIAR), and Ethiopian Meteorology Institute (EMI) and International Livestock Research Institute (ILRI). The contributions from experts across these institutions were invaluable in the preparation of this Reference Guide. The development of this guide was supported by the Accelerating Impact of CGIAR Climate Research for Africa (AICCRA) project; and the Columbia World Project, ACToday, Columbia University in the City of New York. Accelerating Impacts of CGIAR Climate Research for Africa (AICCRA) is a project that helps deliver a climate-smart African future driven by science and innovation in agriculture. It is led by the Alliance of Bioversity International and CIAT and supported by a grant from the International Development Association (IDA) of the World Bank. Explore its work at aiccra.cgiar.org. The Adapting Agriculture to Climate Today (ACToday) project worked in six countries (Bangladesh, Colombia, Ethiopia, Guatemala, Senegal and Vietnam) to enable partners to incorporate climate services solutions in their efforts to end hunger, achieve food security, improve nutrition, and promote sustainable agriculture. Photos Cover image: Jacquelyn Turner, IRI; pp. 1, 39: Petterik Wiggers, IWMI; pp. 2, 62, 63, 97: Georgina Smith, CIAT; p. 38: Maheder Haileselassie, IWMI; p. 98: Apollo Habtamu, IWMI Design and layout: Solomon Kilungu Disclaimer This guide has not been peer reviewed. Any opinions stated herein are those of the author(s) and do not necessarily reflect the policies or opinions of AICCRA, donors, or partners. Licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. ©2024 CIAT, IRI, and ILRI. The copyright over this report is jointly owned by Centro Internacional de Agricultura Tropical (CIAT), the International Research Institute for Climate and Society (IRI) of the Columbia Climate School, and the International Livestock Research Institute (ILRI). Partners Module 1: Climate Basics provides Module 2: Climate Information a foundation of knowledge about Products and Tools Available for climate concepts, processes, data, Agriculture provides an overview forecasts, and probability. A basic of relevant weather and climate understanding of climate will provide information products and services necessary background and context that are [or will soon be] available for subsequent learning about through the Ethiopian Meteorology the use of climate information Institute (EMI). to improve agricultural risk management. Climate Risk Management in Agricultural Extension REFERENCE GUIDE Module 3: Climate-Sensitive Module 4: Integrating Climate Agricultural Decisions aims to Services into Agricultural Extension strengthen understanding of equips extension staff to bring the interaction between climate climate services into the services that and farm decision-making. they provide to their client farmers. Understanding what climate- sensitive decisions farmers make and the factors that influence their decisions is essential for understanding how climate services can support farmer decision making. Foreword Climate change and year-to-year climate variability have resulted in widespread, pervasive impacts to ecosystems and people in Ethiopia, including increases in the intensity of weather extremes such as droughts and floods. While these shifting and varying temperature and precipitation patterns have affected the productivity of many climate-sensitive sectors, the agricultural sector has been the most impacted, resulting in reduced food availability and increased food prices, ultimately jeopardizing food security, nutrition, and livelihoods of millions of people. Climate information has an enormous role to play in supporting the resilience of agricultural and food systems. Whether it is a farmer needing to make key agricultural decisions for a season or within a season based on temperature or rainfall patterns, a humanitarian aid worker who needs to anticipate and respond to climate extremes such as a drought or flood, or even a policymaker who must plan and arrange for agricultural inputs and investments annually, climate information can and should pragmatically inform and be integrated within specific decisions and decision-making processes for adaptation. The Ethiopian Meteorological Institute (EMI), in its mission and mandate to deliver a range of weather and climate services to the user community, has for more than 40 years strived to contribute to the advancement of socio-economic activities through climate-informed planning and practice. In 2021, the EMI led the development and launch of the country’s first National Framework for Climate Service to coordinate, facilitate, and strengthen collaboration amongst national institutions to improve the co-production, tailoring, delivery, and use of science- based climate information and services. Despite these efforts and advances, however, the effective use of climate information in decision-making processes to manage risks associated with climate variability and change has remained limited, especially at the most grassroots levels. This is in spite of one of the continent’s strongest agricultural extension systems, with more than 72,000 agricultural agents serving over 16 million farmers. Part of the problem stems from a lack of foundational capacity and common understanding of climate basics, including a shared vocabulary and knowledge of basic terminology and dynamics to both articulate and engage in meaningful collaborations and design of climate solutions. Another part stems from the reality that even when available, climate information may not be easily accessible or transformed to decision-relevant formats for those working in the agricultural sector to act upon, or that there is poor awareness of products that do so. Still, another major issue with the exploitation of climate information for agricultural decision-making lies with poor understanding of how climate impacts agriculture and the pragmatic climate-sensitive decisions that farmers and the network of actors that support them must make as a result of these impacts. Acknowledging these gaps and the resultant enormous opportunity to capacitate the agricultural extension system with the knowledge and skills that can support climate resilience, the Climate Risk Management in Agricultural Extension (CRMAE) Reference Guide and Handbook represent a seminal effort to equip Ethiopia’s agricultural actors with foundational knowledge and, most critically, the concrete skills to support climate adaptation and climate-informed agricultural decision-making on the ground. In particular, the heart of the curriculum addresses a longstanding bottleneck identified by Ethiopia’s Climate Smart Agriculture Roadmap for 2020-2030 and the National Strategy for Ethiopia’s Agricultural Extension System of support for location-specific climate-smart adaptation interventions and practices, by capacitating extension and development agents on the use of high-resolution historical, monitoring, and forecast climate information products freely and readily available from the Ethiopian Meteorological Institute ( EMI). These products are vital for enabling the extension system to tailor recommendations most appropriately according to location and context, and for empowering the country to maximize the use of its own locally-owned and high-quality data for sustainable and resilient agricultural development. Moreover, this long-awaited curriculum is instrumental for raising awareness of such products and enabling users to better understand and articulate their climate i Ethiopia information needs to ensure climate information is meeting the real needs of those in the agricultural sector. Systems-level problems demand systems-level solutions, and I am therefore pleased to have seen persistent and enthusiastic collaboration amongst national institutions and actors such as the EMI, the Ministry of Agriculture and Livestock, the Ethiopian Institute of Agricultural Research, and Agricultural Technical Vocational Education and Training (ATVET) colleges with the International Research Institute for Climate and Society (IRI) of the Columbia Climate School towards achieving a climate-smart and resilient agricultural sector through the development of the CRMAE curriculum. This effort has involved invaluable contributions from across these institutions and across projects including the Adapting Agriculture to Climate Today, for Tomorrow (ACToday) Columbia World Project and the Accelerating Impacts of CGIAR Climate Research for Africa (AICCRA) project. Foundational, competency-based capacity development initiatives such as this have been, and will continue to be a crucial means of supporting Ethiopia to deliver on its development and climate commitments. I fully expect the CRMAE curriculum and its associated teaching and learning materials to be transformative towards these ends and guide Ethiopia’s agricultural extension system towards a sustainable, climate-informed future. Fetene Teshome Director General Ethiopian Meteorological Institute Climate Risk Management in Agricultural Extension | REFERENCE GUIDE ii Foreword By the Ministry of Agriculture The agricultural sector is the backbone of the economy and millions of livelihoods in Ethiopia, including that of smallholder farmers whose hands produce more than 90% of the agricultural output. Yet, heavy reliance on rain-fed agriculture and increasingly erratic rainfall resulting from climatic changes have resulted in unprecedented challenges for Ethiopia’s farmers and the network of actors who support them. Following the 2015 El Niño, for example, more than one-third of the country experienced loss of crop and livestock. While efforts have been made to adapt to these changes including the development of early warning systems for climate extremes with the Ethiopian Meteorological Institute (EMI), the severity of food insecurity in Ethiopia is still among the worst globally, with regular yearly multibillion-dollar humanitarian appeals and increasingly frequent and severe droughts and floods exacerbating the challenges already faced by vulnerable households. As such, changing temperature and precipitation patterns have continued and will continue to play a central role in causing or contributing to all the main multi-hazard shocks affecting Ethiopia across all sectors, but most especially the agricultural sector, and we must adapt and seek solutions that build resilience at a systems level. Strengthening the capacity of agricultural extension staff at the most local levels to help farmers adapt is foremost amongst these solutions. And for this, the Climate Risk Management in Agricultural Extension (CRMAE) curriculum has an enormous role to play. This Handbook, which moves beyond knowledge to equip Ethiopia’s extension staff and development agents with practical and foundational skills to access and use climate information from the EMI at local scales, communicate such often-complex information in understandable terms, and ultimately help farmers to inform decisions take adaptive action, meets a longstanding need in the agricultural sector for actionable guidance towards a climate-smart future. While both Ethiopia’s Climate Smart Agriculture Roadmap for 2020-2030 and the National Strategy for Ethiopia’s Agricultural Extension System have long-identified location-specific agro-ecology based interventions and climate-smart adaptation practices as one of the main systemic bottlenecks for effective adaptation, for example, Ethiopia’s more than 72,000 agricultural agents serving over 16 million farmers have not until this point been supported with practical training materials to on how to access and use historical, monitoring, or forecast climate information products available through the EMI that will enable them to tailor their recommendations more appropriately. In this way, having the capacities to both understand and actionably use such information will better equip our extension actors plan for, manage, and respond to climate extremes and emergencies, as well as to support routine agricultural planning in a changing climate. Moreover, the competencies targeted through the CRMAE curriculum underpin Ethiopia’s ability to transform the agricultural sector to meet its growth and development goals, including those outlined in the Climate Resilience Green Economy (CRGE) Strategy and 2030 Digital Agricultural Extension and Advisory Services (DAEAS) Roadmap. By strengthening capacity of the agricultural extension at local levels to through use best-available climate information and products available through EMI and promoting digital skills that support robust decision- making in an uncertain climate, the CRMAE curriculum reflects the ambitions of the Ministry of Agriculture towards cultivating a productive, inclusive, and sustainable agri-food system through the collaborative delivery of customized services to all farmers in Ethiopia. I am pleased therefore to recognize this guide as a collaborative effort not only between projects, including the Adapting Agriculture to Climate Today, for Tomorrow (ACToday) Columbia World Project and the Accelerating Impacts of CGIAR Climate Research for Africa (AICCRA) project, but also with the International Research Institute ii Ethiopia for Climate and Society (IRI) of the Columbia Climate School alongside the Government of Ethiopia through various institutions including the Ministry of Agriculture, Ethiopian Institute of Agricultural Research (EIAR), the Ministry of Labor and Skills (MoLS), and the EMI, in the spirit of Ethiopia’s recently launched National Framework for Climate Services to strengthen collaboration amongst national institutions for improved climate services at all levels. It is our vision that this guide will be used in capacity development of agricultural experts, meteorologists, and decision-makers, including supporting Ethiopia’s Agricultural Technical Vocational Education and Training (ATVET) colleges, for which it has been incorporated as a course to be delivered to new development agents, as well as universities as reference material in courses and tailored training. Such capacity development innovations are fundamental for advancing food security and sustained capacity development beyond the life of any project in the long-term, and supporting Ethiopia’s agricultural sector to adapt to changing conditions, as these issues continue to evolve. I am therefore pleased to present to educators, trainers, development actors, and all of those working to support the resilience of Ethiopia’s farmers with this practical guide for managing climate risk in the agricultural sector. Dr. Meles Mekonen State Minister Agriculture and Horticulture Development Sector Ethiopian Ministry of Agriculture Climate Risk Management in Agricultural Extension | REFERENCE GUIDE iv Table of Contents Foreword i Acknowledgements viii About this guide ix Introduction x Module 1: Climate basics Section 1.1: Basic Climate concepts Definitions 3 Dimensions of climate 3 Main components of weather and climate 5 Section 1.2: The Climate of Ethiopia Main features of Ethiopian climate 10 Temporal characteristics of climates in Ethiopia 12 Main factors determining climate of Ethiopia 13 Climate Variability in Ethiopia 16 Section 1.3: Climate Data and Information Various types of Climate Data 19 Climate analyses 23 Section 1.4: Seasonal Climate Forecasting Forecastingl methods 28 Formats of seasonal climate forecast 31 Use of seasonal forecast 34 Understanding and interpreting most common climate maps and charts 36 Section 1.5: Climate Change The mechanism of the current climate change 39 The causes of current climate change 41 Global warming or climate change? 42 The impacts of current climate change 43 Climate change skepticism 46 Climate Information for adaptation 48 Module 2: Climate Information Products and Tools available for Agriculture Section 2.1: Overview of Products Available Through Ethiopian Meteorological Institute (EMI) Section 2.2: The ENACTS Maprooms Climate maproom 55 What is the climate maproom used for 55 Inside the climate maproom 55 Daily climate analysis 57 Monthly climate analysis 58 Seasonal climate analysis 59 Extreme rainfall analysis 60 Extreme temperature analysis 62 Probability of Seasonal Rainfall Conditioned on ENSO 62 Section 2.3: Specialized Climate Analyses Available for Agriculture Climate and agriculture maproom 64 What is the climate and agriculture maproom Used for? 64 v Ethiopia Inside the climate and agriculture maproom 64 Daily precipitation analysis 64 Historical onset and cessation dates 67 Seasonal totals 68 Section 2.4: Climate Monitoring Maproom What is the Climate Monitoring Maproom Used for? 70 Inside the Climate Monitoring Maproom 70 Section 2.5: Climate Forecast Maproom Flexible Presentation of Seasonal Rainfall Forecast Maproom 73 Module 3: Climate-Sensitive Agricultural decisions Section 3.1: How Climate Risk Affects Farmers Risk concepts 78 Biological productivity impacts 78 Impacts on the welfare of rural households 80 Section 3.2: Understanding Climate-Sensitive Agricultural Decisions Farmer characteristics and decision making 82 Time and climate-sensitive agricultural decisions 83 Describing risky decision problems with decision trees 85 Section 3.3: Decision Making under Uncertainty Risk aversion and related concepts 88 Decision analysis when preferences are unknown 90 Section 3.4: Tools to Analyze Climate-Sensitive Decisions Analyzing climate-sensitive agricultural production decisions 92 Estimating productivity: Crop simulation models 92 Estimating profitability: Enterprise budget analysis 95 Combining crop simulation and enterprise budget analysis 97 Section 3.5: Farm Level Option for Managing Climate Risk Production technologies 100 Diversification 101 Forecast-Based Seasonal Farm Planning 104 Section 3.6: Index-Based Agricutural Insurance Index insurance basics 108 Is index insurance worth the cost? 110 Index insurance good practice 111 Module 4: Communicating Climate information and Supporting its use Section 4.1: Rural Climate Service Communication Satrategies Communication to support gender and social equity 115 Differing needs for weather and climate information 117 Participatory climate communication processes 118 Digital and media climate communication channels 120 Combining communication channels to support climate services 122 Section 4.2: Farmer Participatory Seasonal Forecast Training and Planning Workshop Rationale and overview 123 Workshop process 124 Practical considerations 128 Climate Risk Management in Agricultural Extension | REFERENCE GUIDE vi References Cited 130 Glossary 133 Apendix I: Seasonal Forecast Participatory Workshop Overview 142 Materials needed: 142 Step 1: Introduce the workshop purpose and key concepts 145 Step 2: From memory to variability 147 Step 3: From variability to probability 148 Step 4: Understand a seasonal forecast as a shift in the probability distribution 150 Step 5: Present the current seasonal forecast 152 Step 6: Plan how to adjust farm management based on the forecast 153 Appendix A: Self-Check Answers 154 Appendix B: Activity Worksheets Activity 1.5. Understanding and Interpreting Common Maps and Graphs 162 Interpreting Graphs (Primarily Histograms and Time Series) 167 Activity 3.1. Agricultural calendar 171 Activity 3.2. Use a decision tree to represent a farmers cultivar selection and fertilizer rate decision 177 Activity 3.3. Simulate crop management options with SIMAGRI 178 Activity 3.4. Use an enterprise budget to analyze crop management options 179 Activity 3.5. Use risk efficiency analysis to identify prefered fertilizer strategies 182 Activity 3.6. Analyze a rainfall insurance contract 183 Activity 4.1. Seasonal forecast training and planning workshop 184 Step 1: Introduce the workshop purpose and key concepts Workshop purpose 186 Step 2: Understand past variability Relate time series graphs to farmers’ collective memory 187 Step 3: Introduce !the probability of exceedance graph Develop a probability graph 189 Step 4: Understand a seasonal forecast as a shift in the historical probability distribution 190 Step 5: Present the current seasonal forecast 192 Step 6: Plan how to adjust farm management based on the forecast 193 vii Ethiopia Acknowledgements The Climate Risk Management in Agricultural Extension Reference Guide represents a collaborative effort, made possible by the sustained partnership, consistent input, and critical feedback from the Ethiopian Meteorological Institute (EMI), the Ethiopian Institute of Agricultural Research (EIAR), Ethiopia’s Ministry of Agriculture and Livestock (MoAL), and the Agricultural Technical Vocational Education and Training (ATVET) program. This work was jointly prepared through the Adapting Agriculture to Climate Today, for Tomorrow (ACToday) Columbia World Project, the CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS) in East Africa, and the Accelerating Impacts of CGIAR Climate Research for Africa (AICCRA) project. The contributions from experts across these projects and institutions were invaluable in the preparation of this Reference Guide, which is an accompaniment for the Climate Risk Management in Agricultural Extension course and its Handbook aiming to equip extension staff to access, understand, and incorporate climate information into their professional work. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE viii About this guide This Guide is an accompaniment to the Climate Risk Management in Agricultural Extension course and Handbook. What is this Guide? This comprehensive Reference Guide is an accompaniment to the abridged Climate Risk Management in Agricultural Extension Handbook. Both the Reference Guide and Handbook are training and reference materials intended to be used during implementation of the Climate Risk Management in Agricultural Extension course. Who is this Guide for? This Reference Guide was designed for Ethiopia’s subject matter specialists (SMS) and extension staff, including development agents (DAs). It may also be used by other actors, such as non-governmental organizations (NGOs) or community-based organizations (CBOs), who work closely with farmers and those who support them. Why was it developed? This Reference Guide was designed to provide foundational knowledge on climate and agricultural decision making; and practical tools to analyze climate- related risks, use appropriate weather and climate information to support agricultural decisions, communicate complex climate information effectively with farmers, and integrate climate services into agricultural extension activities. How do I use this Guide? This Reference Guide is organized into four sections corresponding to the four modules of the Climate Risk Management in Agricultural Extension course: Climate Basics, Climate Information Products and Tools Available for Agriculture, Climate- Sensitive Agricultural Decisions, and Integrating Climate Services into Agricultural Extension. ix Ethiopia Introduction Improvements in the weather and climate information available in Ethiopia provide new opportunities for agricultural extension personnel to help farmers better manage the risks that they face, and to adapt recommended technology packages to local climatic conditions. This 2-week course aims to equip Development Agents (DAs) and Extension Officers to access, understand and incorporate climate information into their professional work. It is designed to provide foundational knowledge on climate and agricultural decision making; and practical tools to analyze climate-related risks, use appropriate weather and climate information to support agricultural decisions, communicate complex climate information effectively with farmers, and integrate Module 1 Climate Basics provides foundational knowledge about climate concepts, data and common data analyses, and forecasts. A basic understanding of climate will provide necessary background and context for subsequent learning about the types and use of climate information to improve agricultural risk management. It also prepares participants to address client farmers’ questions and concerns about weather and climate. The module includes probability concepts that are foundational for understanding and managing risk, and for interpreting and using climate information to support risk management. Module 2 Climate Information Products and Tools Available for Agriculture provides an overview of relevant weather and climate information products and services that are [or will soon be] available through the Ethiopian Meteorology Institute (EMI). It teaches participants how to navigate and use relevant historical, monitored and forecast information products available through EMI’s web page and online Maprooms. Module 3 Climate-Sensitive Agricultural Decisions strengthens participants’ understanding of the interaction between climate and farm decision-making. It enables them to perform basic analyses of climate-sensitive farm management decisions under uncertainty. A number of factors, in addition to crop and livestock productivity, influence farm management decisions, particularly at seasonal and longer time scales. To equip participants to provide appropriate support and guidance to their farmers, the module builds understanding of factors that lead to different management decisions by different farmers and under different climate conditions. Module 4 Integrating Climate Services into Agricultural Extension equips participants to bring climate services into the services that they provide their client farmers. Building on learning from the other three modules, Module 4 equips them to lead farmers in a participatory seasonal planning workshop, informed by historical and seasonal forecast information. The course concludes with development and presentation of plans to integrate climate services into extension activities with participants’ client farmers. These plans will address: information and support needed for key climate-sensitive management decisions; differing needs of different types of farmers; communication, training and support strategies; annual calendar of climate service activities; and monitoring and responding to feedback. This course is designed for Development Agents and Extension Officers who actively support farmers with information, advice and other services. Participants are expected to have completed at least a 3-year DA training program at an Agricultural Technical and Vocational Education and Training (ATVET) College, and have at least three years of experience in agricultural extension. A basic level of knowledge is expected in: probability, descriptive, statistics, and economics. Course activities require working knowledge of web browsers, Microsoft Word, Excel and PowerPoint. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE x Climate Basics MODULE 1 ABOUT THIS MODULE Module 1: Climate Basics provides a foundation of knowledge about climate concepts, processes, data, forecasts, and probability. A basic understanding of climate will provide necessary background and context for subsequent learning about the use of climate information to improve agricultural risk management. It also prepares participants to address client farmers’ questions and concerns about climate. The module includes probability concepts that are foundational for understanding and managing risk, and for interpreting and using climate information to support risk management. Competencies • Understand the concepts of weather, climate and climate change • Characterize the climate of the country • Understand climate data, climate information and their limitations • Apply statistical concepts in climate • Interpret seasonal climate forecasts SECTION 1.1 Basic Climate Concepts This section helps to identify whether the problem at stake or information provided relates to weather, climate or climate change. Definitions In climate science, it is customary to distinguish between three concepts: weather, climate and climate change: Weather is the state of the The term weather… atmosphere at a given time and • is generally restricted to conditions over short periods of time place with regard to important (two weeks or less) atmospheric variables including • includes whatever is happening outdoors at a given place and temperature, precipitation, humidity, time air pressure, wind, cloudiness, and • can change a lot within a very short time sunshine. • is what we learn about on TV/Radio and other news daily Climate is the statistics of weather at The term climate... any place over some specific, usually • usually involves relatively long periods of time (i.e., years and long, period of time. months, instead of days or weeks) • includes the cycle of seasons (e.g. dry vs rainy seasons) • includes the characterization of extreme weather events (floods, cyclones) • informs on the typical conditions in a specified location and timeof year • is often reflected in the vegetation or agricultural/other economic activities Climate change refers to changes More information about climate change will be provided later in this in the characteristics of climate section. over long periods of time. These changes occur not only in the mean conditions but also in the variability. Section 1.1.1: Dimensions of Climate Climate and its characteristics vary from one location to another as well as from one period to another. It is customary to conceptualize climate as having spatial and temporal dimensions and present climate characteristics in both dimensions. Figure 1.1-1 shows different representations of climate in its spatial and temporal dimensions: Figure 1.1-1a shows a map - spatial dimension - of rainfall amounts received on average in one year in different locations; Figure 1.1-1b shows the amount of rainfall that fell each month in Addis Ababa between January 1984 and December 2014, thus the temporal dimension of climate; Figure 1.1-1c shows the rainfall that falls on average in Addis Ababa in each month of the year – another representation of temporal dimension of climate; Figure 1.1.-1d shows how many times the annual rainfall recorded in Addis Ababa between 1984 and 2014 fell in each category of less than 800mm/year, between 800 and 900mm/year, 900-1000mm/year etc. This is yet another way to represent the temporal dimension of climate. 3 Ethiopia MODULE 1 Figure 1.1-1: Spatial and temporal distribution of climate characteristics. a) map of annual rainfall, expressed in mm/month; b) time series of monthly rainfall in Addis Ababa, January 1983- December 2014 (in mm/month); c) average seasonal cycle of rainfall in Addis Ababa (in mm/month); d) number of years annual rainfall was in a given range (in mm/year) Figure 1.1-1: Spatial and temporal distribution of climate characteristics. a) map of annual rainfall, expressed in mm/month; b) time series of monthly rainfall in Addis Ababa, January 1983- December 2014 (in mm/month); c) average seasonal cycle of rainfall in Addis Ababa (in mm/month); d) number of years annual rainfall was in a given range (in mm/year) Think Deeper 1. Identify the areas of highest and lowest rainfall on the map (Figure.1.1-1 a). What type of vegetation or crop do you think grows in these areas? 2. Can you identify a systematic pattern in the time series of monthly rainfall (Figure 1.1-1 b)? What changes the most from one year to another? 3. In the seasonal cycle of rainfall (Figure 1.1-1 c), identify the month of highest rainfall. Identify the months with no rainfall. What are the months of the rainy season? What are the moths of the dry season? 4. Based on Figure 1.1-1 b and c, do you expect the rainfall in the wettest month to always be 250 mm? 5. Figure 1.1-1 d shows the distribution of the total amounts of rainfall that fell in different years of the period 1983-2014. a. Identify the amount that fell most frequently; in how many years was such amount recorded? b. What was the smallest range recorded? In how many years? c. What was the highest range? In how many years? Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 4 Section 1.1.2: Main Components of Weather and Climate Main Components of Weather and Climate Both weather and climate are measured through a number of variables. Some of the main ones are described in the remainder of this section, along with the typical units and methods of measurement. Temperature measures how hot or cold a substance is. Routine measurements for weather and climate focus on air temperature, although water temperature, especially that of ocean or large lakes can also be of interest. For agronomic applications, soil temperature is sometimes needed. Temperature captures how much energy the surface of the Earth receives and is an important limiting factor for plants’ growth. Temperature is measured using a thermometer, in most of the countries using the Celsius scale (°C) where 0°C and 100°C correspond to the freezing and boiling of water respectively.1 Meteorological Services record the hottest and the coolest temperature of the day – Tmax and Tmin respectively. Average temperature of that day (Tmean) is the average of these two numbers. Temperature can vary between two locations as the function of elevation (cooler at higher elevation) or latitude (cooler in high latitudes), between seasons (winter vs summer, dry season vs rainy season) as well as between day and night. Think Deeper 1. Based on your experience, when is the hottest temperature of the day, when is the coolest? 2. In which months do you experience the hottest temperatures, in which months the coolest? Which of the seasons is hotter: rainy season or dry season? Why? 3. Where are the coolest places in Ethiopia? Why? Do you know where the hottest place is? Precipitation is the water that falls to the ground as rain, snow or hail. Water is critical for many economic activities, from domestic use to hydropower generation; a critical limiting factor for plant growth in the tropics; and is closely monitored by meteorological and hydrological services. Rainfall is measured using a rain gauge which is typically a plastic or metal receptacle with graduated markings denoting the depth of accumulated precipitation, while the amount of water contained in snow will be estimated from the depth of the snow layer. The units used most often are mm. Unlike temperature, water can accumulate in the soil or lake, thus rainfall can be measured as cumulative amount over different lengths of time such as daily, weekly, dekadly (10-day), monthly or annually.2 On the weather scales, precipitation strongly varies in time and space. It can start and stop raining within minutes, and it can also rain on one side of a town or valley, and not on the other. 1 Occasionally, the Fahrenheit scale is used, mostly in the USA and associated territories. In that scale water freezes at 32°F and boils at 212°F. To convert to degrees Celsius T°C=5/9(T°F-32). In some scientific applications degree Kelvin may be used. Water freezes at 273.15K and boils at 373.13K. T°C=TK-273.15. A simple rule of thumb to recognize that non-Celsius scale is used: if air temperature values are systematically somewhere between 20 and 90 degrees, Fahrenheit degrees are being used. If they are above 200 degrees, Kelvin is used. 2 It will most often be expressed in mm/day, mm/week, mm/dekad, mm/month or mm/year. However, occasionally it can be expressed differently: for example, average annual rainfall in mm/month means that total annual rainfall was divided by 12. ‘Average daily rainfall’ indicates the ‘average’ amount of rain that can be expected during a rainy day and was obtained by dividing total annual rainfall by the number of days when rainfall was observed. It is very important to pay attention to the units in which rainfall is expressed. Note that, 10mm of rainfall measured by a rain gauge is equivalent to 1 liter of water over a surface of 1m2. In some scientific applications rainfall is expressed in units of volume, e.g. rainfall intensity can be expressed in m3/s. 5 Ethiopia MODULE 1 Temperature, on the other hand, tends to vary less on those scales. Unless there is a change in elevation (e.g. from the bottom of the valley to the top of the mountain), in vegetation (e.g. from prairie to forest) or in proximity to a water body, temperature is unlikely to change within a kilometer or so (spatial dimension) or within minutes (temporal dimension). On climate scales, seasonal cycles or year-to-year differences (temporal dimensions) for variables such as temperature and rainfall behave very differently between tropical and temperate regions (spatial dimensions). Figure 1.1-2: Mean seasonal cycle of precipitation (blue) and temperature (orange) in four locations, a) Addis Ababa [8.98° N, 38.76° E], b) Niamey [13.51° N, 2.12°E], c) Cape Town [33.92° S, 18.42° E] and d) Moscow [55.75° N, 37.62° E]. Tropical climates, with the exception of equatorial regions, often show large contrasts in precipitation with temperature showing relatively small variations during the year. Tropical climates are characterized by monthly average temperatures of 18°C or higher year-round, with little variations between months. Conversely, with the exception of equatorial regions, precipitations can show large contrasts between the dry and the rainy seasons. In mid-latitudes, precipitations vary relatively little, while temperature shows large contrasts. A class of mid- latitude climates called ‘temperate’ or ‘oceanic’ encompasses areas where the monthly average temperatures is above 0°C but below 18°C. In ‘continental’ mid-latitude climates temperature in cooler months can drop below 0°C. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 6 Think Deeper Use Figure 1.1-2 to consider the following questions: 1. What is the maximum average monthly precipitation in Addis, in which month does it occur? What is the minimum average monthly precipitation in Addis? In which month? What is the annual amplitude (the difference between the maximum and minimum, the seasonal contrast) of the average monthly precipitation in Addis? 2. What is the maximum average monthly temperature in Addis, in which month does it occur? What is the minimum average monthly temperature in Addis? In which month? What is the annual amplitude of the average monthly temperature in Addis? 3. Locate all the four cities on a map of the world. a) What are the value and the month of maximum average monthly precipitation in each city? What are the value and the month of the minimum monthly precipitation? What is the annual amplitude of rainfall? Which city receives the most rainfall in one month? Which city has the longest dry season? Which city has the shortest or none? b) What are the value and the month of maximum and minimum average monthly temperature in each city? What is the amplitude of the seasonal cycle of temperature? Which city has the coldest months? Which city is the warmest year-round? c) How do maxima and minima in precipitation and temperature relate to each other in a given city? Thus, in the tropics availability of water is the main limiting factor for agricultural activities while in mid-latitudes low temperatures are the main barrier to crops. Perennial vegetation in mid-latitudes goes dormant during the cold season. Cape Town, with its marked seasonal cycle in precipitations and to some degree in temperature, represents a class of transition climates called ‘Mediterranean’, with warm and dry summers. Think Deeper Total annual precipitation in Niamey is around 500mm/year while in Moscow it is around 600mm/ year. The difference is in the order of 20%. Figure 1.1-3 shows natural vegetation around Niamey and Moscow in early summer. 1. Would you expect these differences? 2. How can climatic and other factors explain such differences? Figure 1.1-3: Natural vegetation around Moscow (panel) and Niamey (right) in early summer. 7 Ethiopia MODULE 1 Humidity measures how much water vapor is in the air at a given time. There are two standard measurements of humidity: Absolute humidity is a measure of the amount of water vapor in a given parcel of air and is typically measured in mass/volume or volume/volume units. Relative humidity is a measure of the amount of water vapor in the air at a given time relative to the maximum amount of water vapor a parcel of air can retain, called saturation level. It is typically expressed in percentages (of saturation). Saturation level strongly depends on temperature – the warmer the air, the more water vapor it can contain. If saturated air cools (e.g. at night), water will condensate and create fog or precipitation. Note that the same level of absolute humidity in two locations in Ethiopia may produce very different results. In the hot lowlands, the amount of water contained as vapor may not be enough to produce a high relative humidity and precipitation. However, in the cool highlands, that same level of absolute humidity may be close to or over the saturation point and may be accompanied by fog or rain. Both absolute and relative humidity tend to be correlated with precipitation in drier regions and seasons experiencing lower humidity than in wetter regions and seasons. Solar radiation/cloud cover: The existence of nearly all life on Earth is fueled by light from the Sun. Most plants use the energy of sunlight, combined with carbon dioxide and water through the process of photosynthesis, to grow. Animals, including humans, consume plants and other animals to survive and grow. Humans further use fossil fuels, the remnants of ancient plant and animal matter formed using solar energy, to support various activities. Solar radiation is also the source of energy for movements within the climate system such as winds and ocean currents. The amount of solar radiation that is received in a given location will depend mostly on the length of the day and cloud cover. The latter have different effects on the amount of energy received locally, details beyond the scope of this course. Solar radiation received in a given location can be measured in different ways: (i) solar irradiance – a direct measure of the instant energy received per unit of surface, in kW/m2, it can further be transformed into insolation i.e. energy received over a period of time; (ii) number of hours of sunshine, which is then transformed in solar energy received in a day; (iii) estimate of cloud cover, allowing estimation of total insolation in a day. Winds are an important component of the climate system as they move around air masses with different properties (temperature, humidity). They can bring moist or dry air to an area or, conversely, blow away humid air generated locally through evaporation and evapotranspiration. It is not uncommon that the arrival of the rainy season is preceded by a change in prevailing winds and the new winds bring moisture that fuels rainfall from nearby oceanic or rainforest areas. The withdrawal of the rainy season often corresponds to the new winds not bringing moisture to the area any more. Wind are also important in estimation of evapotranspiration of plants. Atmospheric pressure is a measure of the force or weight of the overlying air per unit area on the surface. Winds can be driven by differences in atmospheric pressure, usually blowing, in the tropics, from areas with higher pressure towards areas with lower pressure. Lower atmospheric pressure tends to occur, with a few exceptions, over areas experiencing warmer temperatures than their surroundings and high pressure is often located over cooler areas. Warmer air in low pressure areas will rise and, if it contains enough humidity, the rising motion may lead to rainfall as the humid air cools at higher altitude and cannot hold the same amount of water vapor. In high pressure areas, air often descends and will be relatively dry, as it comes from upper, cooler atmospheric layers, containing less moisture. Deserts, however, are an exception to this rule as, usually, higher pressures overlay hot surfaces. This is related to moist processes in the air as well as general circulation patterns and diurnal cycle of temperature. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 8 Think Deeper The elements above are the most important variables, necessary to describe current atmospheric conditions as well as predict their evolution in the near future, from 24 hours to a few weeks. They impact plants and crops in different ways. 1. What was the last or the most impactful weather or climate event that you remember having an adverse effect on crops? 2. Was it weather or climate related? Self-Check: Basic Climate Concepts 1. Recall the differences between weather, climate, and climate change. 2. What are the two dimensions of climate. How do weather and climate differ in those two dimensions 3. Recall three main variables used to describe climate in a given location *Answers can be found in Appendix A 9 Ethiopia MODULE 1 SECTION 1.2 The Climate of Ethiopia In this section, we focus on the description of the climate through the long-term characteristics of rainfall and temperature. Section 1.2.1: Main Features of Ethiopian Climate Spatial Distribution of Climates Ethiopia has a complex mosaic of natural vegetation closely following the distribution of climates (Figure 1.2.-1). Global climate classification such as the one shown in Figure 1.2.-1 are based on annual total rainfall and temperature since they capture the most important parameters for plants. Note the presence of cooler temperate climates, mostly found in extratropical regions of Europe, Asia and the Americas. Their presence in Ethiopia is linked with cooler temperatures in the highlands. Figure 1.2-1: Spatial distribution of (a) climates and (b) potential vegetation. Sources: Koppen- Geiger climate classification for Ethiopia, 2018, available from https://commons.wikimedia.org/wiki/ File:Koppen-Geiger_Map_ ETH_present.svg, retrieved April 22, 2021; Map of the potential vegetation types of Ethiopia from Friis et al. (2010 copied it from Asefa et al, 2020) Think Deeper 1. What are the main types of climate in Ethiopia? Describe their approximate location. 2. Find on the vegetation map (fig. 1.2-1 b) areas covered by Moist Evergreen Afromontane forest and by Transitional Rain forest. To what climates do they correspond approximately? 3. Climates in several areas are classified as ‘temperate’. Did you expect temperate climates in Ethiopia? Why are there temperate climates in Ethiopia? 4. There are many variants of the temperate climates in Ethiopia, what are the differences? Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 10 Figure 1.2-2 allows to explain some of the features of the climate mosaic in Ethiopia. It is easy to see how temperature follows the topography (figur 1.2-1 a and b) and defines temperate climates. However, not all the areas experiencing high maximum temperatures (above 30°C) are arid or semi- arid. Rainfall also plays an important role in defining different types of tropical climates. In Ethiopia, rainfall distribution is also closely related to the topography, but interestingly the highest amounts of annual rainfall are observed to the west of the mountain peaks and lowest temperatures, farther from intuitive sources of moisture such as the Red Sea and the Indian Ocean. This rainfall distribution explains why the lowlands to the west, experiencing similar temperatures would be covered with wooded grassland while the lowlands to the east would support acacia and bushland only. The reasons for this rainfall distribution will be explained further in this chapter. Figure 1.2-2: Spatial distribution of (a) annual average maximum temperature, (b) topography, and (c) average annual rainfall in Ethiopia., Sources: topography - Wikipedia, https://en.wikipedia.org/ wiki/Geography_ of_Ethiopia#/media/File:Ethiopia_Topography.png, accessed on April 22, 2021; climatologies of maximum temperature and rainfall are based on Worldclim dataset and were downloaded from Colorado State University, http://ibis.colostate.edu/nr505/ethiopia2009/nr505_fall09_group11_Rubin_Barker/BaseMap.html, accessed on April 22, 02021 Figure 1.2-2 explains some of the features of the climate mosaic in Ethiopia. It is easy to see how temperature follows the topography (1.2-1 a and b) and defines temperate climates. However, not all the areas experiencing high maximum temperatures (above 30°C) are arid or semi-arid. Rainfall also plays an important role in defining different types of tropical climates. In Ethiopia, rainfall distribution is also closely related to the topography, but interestingly the highest amounts of annual rainfall are observed to the west of the mountain peaks and lowest temperatures, farther from intuitive sources of moisture such as the Red Sea and the Indian Ocean. This rainfall distribution explains why the lowlands to the west, experiencing similar temperatures would be covered with wooded grassland while the lowlands to the east would support acacia and bushland only. The reasons for this rainfall distribution will be explained further in this chapter. 11 Ethiopia MODULE 1 Think Deeper 1. What is approximatively the altitude above which the annual average maximum temperature is below 25°C? What is the average maximum temperature experienced by the arid areas? 2. How much rainfall (approximately) falls annually in the areas potentially covered by Moist Evergreen Afromontane forest and by Transitional Rain forest? How much rain falls annually in the arid areas? 3. What do the maps in Figure 1.2-2 tell you about the climate of the Danakil Depression? Temporal Characteristics of Climates in Ethiopia In addition to spatial distribution, rainfall is also unevenly distributed during the year, leading to rainy and dry seasons. The length of these seasons will define the type of vegetation and crops that can be grown in a given location. Note that the length of the rainy season is not necessarily related with the total amount rainfall, especially in the regions experiencing two rainy seasons. Figure 1.2-3 shows the regions in Ethiopia experiencing different seasonal cycles of rainfall. Figure 1.2-3: Different types of seasonal cycles of rainfall in Ethiopia. (after Tesfaye Haile) Areas in region A are characterized by three distinct seasons that are locally known as Bega (October to January), Belg (February to May) and Kiremt (June to September). The rainfall in region A has two distinct peaks during the year, with Kiremt receiving significantly more rainfall than Belg, and Bega being essentially dry Rainfall in region C, which includes the southern and south-eastern parts of the country, is also characterized by two distinct rainfall peaks with a longer dry season in between. In region C, the first wet season is from March to May and the second is from October to January. Region B receives rainfall in a single rainy season and has distinct wet and dry seasons. However, their respective lengths differ: the rainy season is longest in the Southwest where it lasts from February/ March to October/November; it is a little shorter in the western art of the country, lasting from April/ May to October/ November and is the shortest in the northwest, from June to September. Conversely, the dry season is the longest in the northwest and the shortest in the southwest. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 12 Section 1.2.2: Main factors determining climate of Ethiopia The climate at a given location results from an interaction between global and local factors. Global Factors The latitude, which defines how much solar energy per m2 a given location receives; simply put this amount decreases from the equator to the poles, due to the curvature of the Earth. These differences fuel the ocean and atmospheric circulations in the climate system. The rotation of the Earth around the sun. Because Earth’s axis is tilted polar regions receive very different amounts of energy, depending on the season. During the winter of the northern hemisphere, northern polar regions don’t receive any sunlight and mid-latitudes receive less solar radiation within shorter days than their southern counterparts. During the southern winter, the North Pole and northern mid-latitudes receive more solar radiation than their southern counterparts. The variation in the energy received in the tropics due to seasonal cycle is much smaller. The distribution of ocean- and land-masses. Water, land surface and atmosphere have different capacities to store and distribute energy. Those differences define atmospheric circulations between land masses and oceans. Figure 1.2-4 schematically presents those global factors and the resulting circulations, in the absence of local factors. Figure 1.2-4: Schematic representation of the main global factors influencing local climate. a) differences in energy received by latitude; b) resulting atmospheric circulations, in the absence of land- masses; c) the seasonal cycle and d) the distribution of land-masses and oceans. Sources: http:// www.ux1.eiu.edu/~cfjps/1400/ circulation.html, https://www.climate.gov/news-features/blogs/enso/ walker-circulation-ensos-atmospheric- buddy 13 Ethiopia MODULE 1 Think Deeper 1. Note the circulations in the vertical dimension on Figure 1.2-4 b. Where is the air rising, where is it sinking? Note the complex circulation pattern in the vertical dimension the discussion of the multiple cells is beyond the scope of this course. Identify the equator. The cells immediately north and south of the equator are called Hadley Cells and define the extent of the tropics. Note the winds on the surface of the Earth in the tropics. In which direction are they blowing? They are called Trade Winds. 2. Under the effect of the seasonal cycle depicted in Figure 1.2-4 c the system rom Figure 1.2-4 b is not symmetrical and shift towards the south and towards the north respectively. So does the zone of rising air near the equator. Do you expect the zone of rising air to be north of the equator or south of the equator during June-September i.e. during the Northern Hemisphere summer? 3. Which of the global factors above can best explain the differences in the seasonal cycle of temperature noted in Figure 1.1-2? 4. In Figure 1.2-4 d identify the land-masses and oceans. Name the main continents. In this figure, vertical arrows represent rising and sinking motions. Rising moist air is likely to generate rainfall. Where is the region of strongest rising motion (called ‘convection’)? Is the air rising or sinking over East Africa? Which way blow the winds over East Africa? Local Factors The theoretical circulations linked with global factors interact with local factors such as mountain ranges, inland water bodies, types of vegetation and soil, etc. The effects of local factors are schematically represented on Figure 1.2-5. Figure 1.2-5: Schematic representation of local factors influencing local climate. Think Deeper 1. Which local factors do you think are the most important in defining the climates in Ethiopia? Seasonal Circulations in Ethiopia Figure 1.2-6 shows the seasonal cycle of winds and humidity near the surface. Winds are shown as arrows pointing in the direction that the wind blows and their length is proportional to the strength of the wind – the longer the arrows, the stronger the wind. Humidity is shown in blue colors with white indicating lowest humidity and dark blue highest levels of humidity. The months of December and January – when the southern hemisphere receives more solar radiation are shown at the bottom of the figure and the months of June-July, when Northern Hemisphere receives more solar radiation are shown at the top of the figure. Take a moment to observe: Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 14 1. In each month, there are areas with virtually no wind. They often correspond to areas of high humidity. Those areas are regions where winds from different directions meet, or converge. Winds blowing from moist areas bring moisture to areas of convergence (areas of low wind) where it accumulates. High moisture levels and wind convergence are conducive to rainfall. Note that areas with strongest winds often correspond to areas with low humidity (e.g Sudan and Egypt and Arabic peninsula, and to a certain degree Indian Ocean). 2. Over the course of the year, the areas of convergence and high humidity (and rainfall) migrate. They are located to the south in December-February and reach their northern location in June-September. In December-February, highest moisture levels are located over the Congo, dry winds from the Sahara extend all the way to the southern border of South Sudan. Over the Indian Ocean the winds blow towards the Southwest. There is a little wind blowing from east to west over the horn of Africa but the humidity levels remain low. In March, winds blowing northward from the Congo basin intensify, winds from the Indian Ocean change direction and start blowing towards northwest over the horn of Africa and winds from the Sahara start retreating towards the North. This situation intensifies in April and May and high humidity levels extend all the way to the southern part of Sudan and Ethiopia. In June to September the winds from the Congo basin bringing humidity are well established over the southern half of Sudan and western Ethiopia which experience the highest levels of humidity and rainfall of the year. The dry winds from the Sahara have retreated to the north of Sudan. The winds over the Indian Ocean and the horn of Africa are now blowing towards North and Northeast. In October to December the winds from the Sahara bringing dry air intensify, and the winds from the Congo basin progressively retreat southwards, slowly bringing the zone of high moisture to the south. Winds from the Indian Ocean also change directions. In December the moist air is back to its southernmost location and the cycle has been completed. Think Deeper 1. In light of the description above, are you able explain the seasonal cycle of rainfall observed in different regions of Ethiopia and depicted in Figure 1.2-3? 2. Can you explain why is the maximum of moisture and rainfall experienced to the west of the main mountain range in Ethiopia? Figure 1.2-6: Climatology of low level wind and specific humidity, computed over the period 1956- 2016. Data source: NCEP-NCAR Reanalysis. 15 Ethiopia MODULE 1 Section 1.2.3: Climate Variability in Ethiopia The succession of the seasons occurs on a regular basis but with small deviations from one year to another: the rainy season can be more or less rainy, more or less long, the temperature can be a little warmer or cooler can the year before etc. Those variations are usually small and for the most part socio- economic systems are well adapted to withstand them, within a certain range. They aren’t however without consequences on crop productivity and animal and human wellbeing. Part of climate science focuses on understanding and predicting those variations that occur from one to another. Based on the elements seen so far it is easy to understand that a small change in the seasonal wind pattern, such as stronger winds from the Sahara or the change in the timing of direction switch in the winds from the Indian Ocean can delay the rainy season and make it shorter. Similarly, less humidity brought by the winds can make the season less rainy. It is easy to imagine numerous combinations of those variations. In addition to those factors in the direct viscinity of Ethiopia, variations in places as remote as the tropical Pacific Ocean can also affect climate in Ethiopia, by disrupting the cells established between the land- masses and the oceans and depicted in the Figure 1.2-4 d. Two main remote phenomena have profound impact on the interannual variability of climate in Ethiopia: El Nino-Sothern Oscillation and Indian Ocean Dipole. El Niño-Southern Oscillation El Niño-Southern Oscillation (ENSO) is a phenomenon in the ocean and in the atmosphere in the equatorial Pacific region. In neutral conditions, due to trade winds, there is an accumulation of warm waters and a lot of evaporation and rainfall in the eastern equatorial Pacific, and in general wet climatic conditions in southeastern Asia. On the western side of equatorial Pacific, surface waters are cooler and the atmosphere is drier, leading to dry conditions on the west coast of tropical South America (Figure 1.2-7 a). Periodically, these conditions are relaxed, warmer waters and rain-producing systems spread westward and can cause torrential rains on the coast of South America (Figure 1.2-7a b). This is an El Niño episode. During the opposite episode, La Niña, the atmospheric circulations and the contrast in oceanic temperatures strengthen (Figure 1.2-7a). El Niño/La Niña mostly occur around December-January ad draw their name from their co-occurrence with Christmas (El Niño=the Child). Figure 1.2-7: Schematic representation of (a) Neutral, (b) El Niño and (c) La Niña conditions in Equatorial Pacific Think Deeper 1. Locate the circulation in figure 1.2-4d that is disrupted by the El Niño/La Niña phenomenon. Which other circulation it can disrupt? Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 16 The anomalies in the atmospheric circulation and ranfall in the equatorial Pacific propagate around the world through connections in the atmosphere and atmospheric waves, and can generate climate anomalies in various regions of the globe. Figure 1.2-8 shows the main climate anomalies in El Niño and La Niña phases respectively. While these figures are global, in a nutshell, El Niño tends to maintain moist air in the southern parts of East Africa, thus enhance rains in the South of Ethiopia and decrease rains in the North of Ethiopia. La Niña tends to have opposite effects. You can see more precisely the impacts of ENSO on climate of Ethiopia in the dedicated maproom in the Climate Analysis section of the maprooms, described in Module 2. Figure 1.2-8: Impacts of (left) El Niño and (right) La Niña on temperature and rainfall around the globe in the December-February season. Source: NOAA, https://www.climate.gov/news-features/featured- images/global- impacts-el-niño-and-la-niña, accessed February 8, 2021. Figure 1.2-9 shows the time evolution of sea surface temperature anomalies in the equatorial Pacific, capturing El Niño/La Niña events, with positive /negative anomalies respectively. Figure 1.2-9: Time series of SST anomalies in equatorial Pacific over the period 1950-2020. Source: https://fews. net/el-niño-and-precipitation, accessed Feb 8, 2021. Indian Ocean Dipole Indian Ocean Dipole (IOD) is another ocean-atmosphere phenomenon that influences rainfall and temperatures in East Africa. Figure 1.2-10 displays a schematic view of the IOD and figure 1.2-11 displays the time evolution of associated sea surface temperatures anomalies. More precise impacts of IOD on your country can be found in dedicated maprooms in the Climate Analysis section of the maprooms. 17 Ethiopia MODULE 1 Figure 1.2-10: Indian Ocean Dipole. Neutral, positive, negative phases and their impacts on left, middle and right panels respectively. Source: NOAA, https://www.climate.gov/news-features/blogs/enso/meet-enso’s- neighbor-indian-ocean-dipole, accessed February 8, 2021. Figure 1.2-11: Yearly changes in the Dipole Mode Index (DMI) - DMI is defined as the sea surface temperature anomaly difference between tropical western Indian Ocean and the southeastern Indian Ocean. Positive DMI values (red) point to a positive IOD year, and negative values (blue) to negative IOD years. Image credit: Dr. Saji N Hameed. Retrieved from: https://watchers.news/2019/09/17/indian-ocean-dipole-iod-2019-strongest-on-record/, on February 8, 2021. Several maprooms, introduced in module 2 provide information on how ENSO and IOD impact Ethiopian climate. Think Deeper Looking at the time series of ENSO (Figure 1.2-9) and IOD (Figure 1.2-11) 1. How often do El Niño/La Niña events occur? 2. How often do positive/negative IOD events occur? 3. In what years did the two strongest El Niño and two strongest La Niña events occur? 4. Do you recall their impact on crop production? 5. For the strongest El Niño and La Niña events after 1999 what was the status of the IOD? Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 18 Self-Check: Climate Variability in Ethiopia 1. What are the two main classes of factors that define the characteristics of climate in Ethiopia? Cite three elements in each class. 2. Which local factor is the most important in Ethiopia. Cite two-three ways it impacts climate in Ethiopia? 3. What are the three main regions of Ethiopia with respect to the seasonal cycle of rainfall? Which class of factors is the most important for the seasonal cycle of rainfall? 4. El Nino/La Nina is a phenomenon in tropical Pacific. Explain the mechanism by which it affects climate in Ethiopia (hint: which factor is important in this mechanism?) *Answers can be found in Appendix A SECTION 1.3 Climate Data and Climate Information By the end of this section, you will be able to: • Identify different types of meteorological and climate data, their strength and weaknesses • Identify some statistical aspects of climate • Describe and interpret climate information provided • Explain different climate dimensions and scales An important distinction needs to be made between weather and climate data on the one hand and weather and climate information on the other hand. Data refers to arrays of measurements of weather and climate variables from which information that is more useful in everyday life is distilled or that are ingested in applications such as crop or hydrological models. Weather and climate information are based on various analyses of data made with specific objectives in mind. For example, rainfall data is a collection of amounts of rainfall recorded every day at a given location over a certain period (e.g. 30 years); average annual rainfall at that location or the record amounts of rainfall that have only been observed once every 10 years (or three times over the entire record) is information extracted from these values. Most people need weather and climate information, not data, for their every day and professional activities. However, there are different types of data and they influence the type and the quality of climate information that is possible to derive from them. It is important to always check what type of data underpins given information and keep in mind related limitations. Section 1.3.1: Various Types of Climate Data Our understanding of weather and climate, their variability and long term evolution under climate change relies on the analysis of data pertaining to the variables listed in section 1.1 and collected on a routine basis by various devices and organizations, from Meteorological stations managed by the Meteorological Services, to data collected high up in the atmosphere by radars and planes or in the middle of the ocean by ships and buoys. For the past 40 years, meteorological conditions have also been observed by satellites orbiting around the earth. Figure 1.3-1 portrays various data sources contributing to our knowledge. 19 Ethiopia MODULE 1 Figure 1.3.1: Illustration of various sources of weather and climate data. a) schematic representation of various devices measuring weather variables. b) a picture of typical meteorological station; c) ship and buoy – main sources of measurements over the oceans; d) the constellation of various satellites orbiting around the Earth Think Deeper Take a moment to identify different ways of measuring weather and climate variables in figure 1.3- 1 a. Which platforms are used to collect data at the surface of the earth, which in the atmosphere and which above earth’s atmosphere? These measurements lead to three types of data: 1. Station data are exact measurements in a given location, mostly at the surface of the Earth, although radar and stationary balloons also record information about the atmosphere above. However, there are regions of the globe with very few stations making the measurements. Figure 1.3-2 shows the distribution of meteorological stations that reported collecting data on June 1st, 1997. Figure 1.3.2: Location of the meteorological stations that reported collecting data on June 1st, 1997. Each red circle represents one station. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 20 2. Gridded data partition Earth’s surface in small squares and attributes a value of a weather/climate variable to each square. In this way Earth’s surface is evenly covered with values, even in places where there is no observation. The way in which the values in places without observations are derived depends on the dataset and its specifics and ranges from a simple mean of closest observed values to more complex methods that include, for example, the elevation. The advantage of such datasets is that there are no areas without values, which helps with a lot of analyses. Most of such datasets are also global. However, there are limitations on how well data in places with no observations represent the reality. Currently, the resolution of gridded datasets – the size of the squares – ranges between a few km2 to a few hundreds km2. Figure 1.3.3: Map of winds close to the oceanic surface (arrows) and of a measure of their convergence (colors) based on gridded data. 3. Satellite data is a special class of gridded data. Satellites can observe vast swaths of Earth and atmosphere, even in remote, hard to access oceanic areas. However, they too have some issues: There are different types of satellites – some orbit around the Earth and pass over the same place approximately every 15 days. This means that they measure weather/climate variables every 15 days. Some are stationary over a given place, which means that they observe every minute the conditions in that place and a circle of several kilometers around but there is not enough of them to have measurements covering the entire globe.Satellites make observations from a very long distance and the measurements can be affected by what is going on between the satellite and Earth’s surface. For example, on a cloudy day they don’t ‘see’ Earth’s surface thus cannot measure surface temperature. Similarly, they will have difficulties having exact values of temperature if there is a lot of pollution or smoke. Finally, some variables are not measured directly but values are calculated from measurements that reflect the variable. Rainfall, for example, is such an indirectly measured variable. As mentioned above, satellites cannot measure what is hidden from them by clouds. Clouds and rain are indissociable; thus satellites cannot measure the rain that hits the ground. They make estimates of how much it rains below the clouds based on how tall the clouds are – the taller the clouds the more intense the rainfall. The height of the clouds is estimated from the temperature of their top – the taller the cloud the coldest the temperature. Measurements of rain from stations on the ground are then used to associate cloud’s temperature with the rainfall that is effectively observed. This method is not without flaws as some local characteristics, such as mountains, lakes or even vegetation can influence how much rain will be produced by a cloud of the same height. Therefore, we talk about ‘satellite rainfall estimates.’ In summary, satellites add a great deal of information, particularly for remote areas, but this information is not the ‘ground truth’ and satellite data often need to be corrected using observations, where these are available, in a process called ‘merging’. The merged datasest have the advantage of even spatial coverage and are closer to the truth on the ground than satellite-only dataset. An important merged dataset in Ethiopia is the rainfall and temperature high resolution dataset developed under the Enhancing National Climate Services (ENACTS) initiative. Its current resolution is 4km x 4km and it spans the period 1982-present. It is the basis of numerous maprooms which will be discussed in Module 2. 21 Ethiopia MODULE 1 4. Model Data refers to data generated by Numerical models. Such models — General or Regional Circulation Models (GCM or RCM) — are a collection of equations capturing physical and chemical laws governing the weather and climate system. By ingesting available data describing current conditions and solving these equations using powerful computers they are able to predict weather and climate conditions in the future as well as in places with no observations. The GCMs will be explained more in detail in the section about forecasting. At this stage, it is important to understand that they can generate gridded data, by filling missing data, on Earth’s surface as well as in the atmosphere, through very complex procedures based on physical equations. They create more physically coherent gridded datasets and are particularly in favor of climate scientists, especially the dataset called ‘Reanalysis’, but may suffer from systematic biases, similar to those of gridded datasets. Think Deeper 1. Take a moment to identify different ways of measuring weather and climate variables in Fiigure 1.3-1a. Which platforms are used to collect data at the surface of the earth, which in the atmosphere and which above earth’s atmosphere? 2. Consider Figure 1.3.2. a) Identify regions with a lot of stations and those with few stations. b) What do think are the reasons for having fewer stations? c) What consequences it can have on the quality of weather and climate information in the areas with few stations? 3. Compare Figures 1.3-2 and 1.3-3 in terms of data over oceans. Suppose a gridded dataset fills missing observations with the average of neighboring observed values. There are stations measuring rainfall and temperature in two valleys separated by a mountain range with no observations. The dataset derived the values corresponding to the mountain range as an average between the observations in the valleys. a) How well do you think the gridded values represent the conditions in the mountains? What aspects will be well represented? What aspects will be less well represented? b) What error do you expect for temperature? What error do you expect for rainfall? 4. List the advantages and disadvantages of the different types of data. 5. What precautions does one needsto take while using station data? 6. What precautions does one need to take when using gridded or model data? Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 22 Section 1.3.2: Climate Analyses As mentioned before, most of the time climate information, i.e. data that have been analyzed, is the most useful for many applications. In this section we will discuss the most common data analyses. Some of them apply to both spatial and temporal dimensions of climate, some to only one. Mean This is probably the simplest and the most common analysis. It consists of the sum of the values divided by the number of values. It gives us an overall idea of the values in the dataset, knowing that they are not all equal to the mean. It is convenient for comparisons of climate characteristics between locations. For example, average or mean annual rainfall in location 1 is 600mm/year while in location 2 it is 800mm/year. It can be computed in space, e.g. average rainfall over a district or a river catchment, or in time, e.g. average rainfall over the 1991- 2020 period. When computed over time, it is common practice to use at least 30 years of data for the mean to be representative of the climate. The mean of the past 3 or 5 years, while still giving some information, may not be really representative of the climate since those years could have been exceptionally wet or dry or cold or warm. Meteorologists often refer to ‘climatic normals’, these usually span 30- year periods starting at the beginning of the decade and are updated every 10 years. The current normal is the normal of the most recent fully finished decade. Current climatological normal is 1991-2020. Anomaly The mean is a useful way of describing climate but it does not capture its variability, both in space and in time. It might be useful sometimes to know that part of the catchment or of the district receives more rainfall than another part or is cooler than another part. It might also be useful sometimes to tell that one year was cooler than the average or warmer than the average. Anomalies are calculated by subtracting the mean for each value in the spatial domain or in the time series. Certain values of the anomalies will be positive and certain negative. While the variability can be seen from the raw values on a map or in the time- series, expressing it as anomalies describes it in terms of values ‘lower than average’ or ‘higher than average’, and allows to see more quickly where or when there were deficits in rainfall or heatwaves. It may sometimes be useful to estimate what is the total range of variability or anomalies around the average value. This We can also say: annual rainfall total in a given location ranges between 950 and 1250mm/year, or: annual rainfall total in that location is 1100 +/- 150mm/year. The latter gives us a better overall sense of the amount of rainfall and adds the range of variability to it. If the values to estimate the range are the maximum and the minimum of the values on the record, the range is called amplitude. Standard Deviation The standard deviation is another measure of variability around the mean. Instead of taking the maximum and the minimum values in the data, which, most of the time only occurred once on the record thus are exceptional, it may be interesting to figure out what is the ‘average’ range of anomalies. To do so a sum of squared anomalies is computed first to take into account the departures from the mean without their sign (so that they do not cancel each other), a square root of this sum is taken (to get back to the same units as the initial variable) and the value obtained is divided by the number of observations, to obtain the ‘aver-age’ anomaly. The temperature r rainfall in a location will then be presented as mean +/- standard deviation. Note that the standard deviation will be smaller than the amplitude discussed earlier. Providing a measure of variability around the mean is important for understanding local climate and risks. Think Deeper 1. Compare two locations in terms of risks to the crops: a) Location A receiving on average 1250mm or rainfall per year with std=100mm and b) Location B receiving on average 1250mm/year with std=400mm 2. Compare two locations in terms of risks to the crop: a) Location B receiving on average 1250mm of rainfall per year with std=400mm and b) Location C receiving 600mm/year with a std=400mm 23 Ethiopia MODULE 1 Coefficient of variation and standardized anomalies Mean and standard deviation are compact ways of presenting the climate as the capture the ‘total’ experience of climate – what is it on average and how much it can vary. While comparing averages is straightforward, the comparisons become more complicated when variability is also included in the comparisons. Two additional manipulations make comparisons, and understanding climate easier: the coefficient of variation and standardized anomalies. Coefficient of variation is the standard deviation divided by the mean. By expressing the ‘average’ anomaly as fraction of the average of the variable it gives us a sense of ‘big’ the variability is and a sense of its potential impact. Standardized anomalies are the anomalies on the record divided by the standard deviation. On some occasions looking at relative anomaly, rather than its value might be more important. This is the case when we want to know whether two locations co-vary (have anomalies of the same sign at the same time and their larger and smaller anomalies coincide) – we are less interested in whether the larger anomalies are 100 or 200mm but that they are large compared to the entire range of anomalies. Standardizing anomalies can also be useful when comparing anomalies between different variables for the same location or different variables for different locations. For example, to estimate the impacts of El Nino on rainfall in Addis one could look at the time series of standardized anomalies of sea surface temperatures in the key ENSO area and standardized anomalies of rainfall in Addis. A statistical procedure called ‘correlation analysis’ makes use of standardized anomalies to estimate how well variables co-vary, but is beyond the scope of this course. Think Deeper 1. Compute the coefficients of variation for locations A, B, and C from the previous sub- section. In which location would you prefer to farm? Why? Frequencies and Probabilities, Extremes The measures of variability above, while useful, still do not allow to describe climate with enough detail for a lot of applications. For example: how often do values close to the maximum or minimum occur? How often do values close to the mean occur vs close to the extremes? What are the values that occur with a certain frequency, of interest to the application (e.g. once every 10 years, once every 50 years)? How often is a certain threshold crossed (e.g. seasonal rainfall amount necessary to grow certain crop, temperature or rainfall threshold linked with epidemic malaria outbreaks)? Frequency analysis and construction of probability density functions and probability of exceedance curves allow to answer those and similar questions (figure 1.3-4). In those frameworks, long term changes in climate properties are represented as shifts in the distributions and allow to capture changes in the mean climate as well as in the variability. A hands-on activity will allow you to discover and practice how similar plots are done and how to interpret them. Figure 1.3.4: Examples of probability density function (left) and cumulative probability (right) for annual and seasonal rainfall values. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 24 Standardized Precipitation Index Drought is one of the most impactful climate anomalies as it can affect crop production and water availability and quality, thus food security and sanitation. Drought can be roughly defined as resulting from lower levels of precipitation than what is considered normal. Drought is therefore a phenomenon relative to local conditions, in space and time. The same rainfall anomaly will have different impacts in the Sahel and in the Amazon. Similarly, the same rainfall anomaly will have different impacts in the heart of the rainy season and outside of the rainy season. The same anomaly of monthly rainfall will have different impact if it is embedded in a multi-month deficit or is one of a kind. Rainfall deficits persisting over 1 to 6 months will affect soil moisture and agriculture while deficits persisting over 6 to 24 months will impact streamflow, reservoir and groundwater. Thus, different sectors will define drought differently and it is possible to experience drought on one time scale but not on another. The Standardized Precipitation Index (SPI) was designed to quantify precipitation deficits at multiple scales and relate them to normal conditions prevailing locally. The possibility of computing the index at multiple scales allows us to detect different types of droughts that impact different sectors. The standardization makes it comparable between regions. The SPI can compute drought intensity over any desired interval, e.g., one month, five months or 200 days. Technically, the SPI is the number of standard deviations that the observed value would deviate from the long-term mean, for a normally distributed random variable. Since precipitation is not normally distributed, a transformation is first applied so that the transformed precipitation values follow a normal distribution. Positive SPI values indicate greater than median precipitation and negative values indicate less than median precipitation. Because the SPI is normalized, wetter and drier climates can be represented in the same way; thus, wet periods can also be monitored using the SPI. Key strengths: • Designed to quantify the precipitation deficit for multiple timescales which correspond to the time availability of different water resources (e.g. soil moisture, snowpack, groundwater, river discharge and reservoir storage) • Uses precipitation only; • Comparable across regions with different climates • Relatively simple to calculate Key limitations: • As a measure of water supply only, the SPI does not account for evapotranspiration, and this limits its ability to capture the effect of increased temperatures (associated with climate change) on moisture demand and availability • Sensitive to the quantity and reliability of the data used to fit the distribution; 30-50 years recommended • Does not consider the intensity of precipitation and its potential impacts on runoff, streamflow, and water availability within the system of interest. 25 Ethiopia MODULE 1 Figure 1.3.5: Example of a 3-months SPI for August-October 1984. Source: EMI Data Library, http://213.55.84.78:8082/maproom/Climatology/Climate_Analysis/spi.html?T=Aug-Oct%20 1984&plotrangeT1=1984&plotrangeT2=2014 accessed 08/24/2021 Think Deeper 1. What is the magnitude of the three-month SPI in Figure 1.3-5? 2. Which areas were the most affected? 3. Do you know what caused the extreme SPI? 4. Do you know what was the impact on crop production? Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 26 Self-Check: Climate Data and Information 1. What is the difference between climate data and climate information. Give one example of each. 2. List the advantages and disadvantages of the different types of data. What precautions need to be taken when using information based on each type of data? 3. Read carefully the caption in Figure 1.3-6 below. What is plotted on the figure? Is it spatial or time dimension? Over what period of time? What data have been used? What analysis have the data undergone? What are the main types of variations presented? What is the main caveat of the result presented? *Answers can be found in Appendix A Figure 1.3.6: Anomalies of global mean surface temperature based on station data. Thin red line is the anomaly of annual mean, thick red line is the 5-year running average (NASA figure adapted from Goddard Institute for Space Studies https://earthobservatory.nasa.gov/features/GlobalWarming/ 27 Ethiopia MODULE 1 SECTION 1.4 Seasonal Climate Forecasting This section only presents a few basic concepts in seasonal forecasting and is not intended to provide in-depth information on specific methods that are being used operationally. The basis for seasonal forecasting lies in the slower evolution of temperatures of the oceans and phenomena such as ENSO or IOD, reviewed in Section 1.3. There are two main classes of methods to forecast the evolution of climate in the next season. By the end of this section, you will be able to: • Identify different types of forecasts • Explain forecast uncertainty • Interpret probabilistic seasonal forecasts Section 1.4.1: Forecasting Methods Statistical Methods Statistical methods take advantage of historical climate records and establish statistical relationships between the variable of interest (predictand), like rainfall, number of rainy days or temperature anomalies during an upcoming season, and variable(s) that drive their variations (predictor), such as SST anomalies in key oceanic regions. The relationships, sometimes called “statistical model”, can be as simple as anomalies conditioned on SST anomalies or a linear regression between two indices, or can involve more complex spatio-temporal analyses, such as Canonical Correlation Analysis. Statistical methods are simple to understand and to implement. However, they rely heavily on relationships between variables and selection of predictors that may not hold under all circumstances. For example, a strong IOD can interfere with the impacts of El Niño in East Africa. Statistical methods take advantage of historical climate records and establish statistical relationships between the variable of interest (predictand), like rainfall, number of rainy days or temperature anomalies during an upcoming season, and variable(s) that drive their variations (predictor), such as SST anomalies in key oceanic regions. The relationships, sometimes called “statistical model”, can be as simple as anomalies conditioned on SST anomalies or a linear regression between two indices, or can involve more complex spatio-temporal analyses, such as Canonical Correlation Analysis. Statistical methods are simple to understand and to implement. However, they rely heavily on relationships between variables and selection of predictors that may not hold under all circumstances. For example, a strong IOD can interfere with the impacts of El Niño in East Africa. Special care needs to be taken to capture all the potential influences. However, it is easy to select too many predictors that are either redundant or show correlations that cannot be supported by plausible mechanisms explaining their influence on the region of interest. Therefore, it is recommended that predictors be always checked for co-variability and (at least hypothetical) mechanisms be proposed to explain the impacts of selected predictors on seasonal rainfall and temperature in your region, to avoid overfitting. Measures of uncertainty of the forecasts should also be provided (see section on Uncertainty, below). Dynamical Methods Dynamical methods refer to the use of models based on physical equations that govern interactions between various components of the climatic system, called General Circulation Models, and similar to the models used in Numerical Weather Prediction. These models describe the evolution of each meteorological variable using physical equations. Figure 1.4-1 shows schematically (top the main processes included in numerical weather and climate models and (bottom) the main processes and interactions described by the equations Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 28 Figure 1.4.1: Schematic representation of the main physical processes in the climate system (top). Main processes and interactions captured by the equations in a GCM (bottom). To describe the state of the atmosphere in different locations on the globe and through its height, they partition Earth’s surface and the atmosphere above into small building blocks and computing changes in each weather and climate variable with time. Figure 1.4.2 schematically represents the partition of the atmosphere and the main variables computed at each step. Such models capture our current knowledge of climate systems and complex interactions within and take advantage of advances on computing technologies. They are used to predict weather and they are very good at predicting ENSO several months ahead as well as how ENSO will disrupt the circulations around the globe and impact climate in Ethiopia. However, they too have several caveats. 29 Ethiopia MODULE 1 Figure 1.4.2: Schematic representation of a GCM, the partition into grid-cells and the main processes variables computed at the surface of the Earth and in the atmosphere. The main caveat is the model resolution (horizontal size of the grid cell). Typical grid cell size ranges from about 50 km x 50 km to 100 km x 100 km.The conditions are assumed uniform within the grid. Thus, the models have difficulty generating different values where there are inhomogeneities within a grid such as mountain ranges or water bodies. However, increasing the resolution (decreasing grid cell size) is not trivial as solving the equations requires a lot of computing resources which limits the number of grids into which we can partition the atmosphere. Another major caveat is in our imperfect knowledge of the atmospheric conditions at the start of the forecast. Think Deeper 1. Assume that in a GCM the surface of the Earth is partitioned into 10,000 grid points and that the height of the atmosphere is partitioned into 20 levels. To predict the future state of the atmosphere the model needs to solve 100 equations in each cube. And it can only advance by steps of 15 minutes i.e. in order to predict the state of atmosphere in one hour it needs to solve the equations 4 times, advancing in time 15 minutes each time. Assume that the computer can solve 1 million equations per second. How long will it take for the model to compute the state of the atmosphere in 24 hours from now? In one month? In 3 months? In one year? 2. Assume you want your grid point to be smaller and cover half the distance in each direction. How long will it take the model to compute the state of the atmosphere in 24 hours from now? One year from now? Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 30 Mixed Methods To take advantage of the objective forecasts allowed by the GCMs but correct some of their systematic biases, statistical treatments are sometimes applied to GCM (or RCM) outputs. Such treatments are often called Model Output Statistics (MOS). Climate Predictability Tool developed by IRI allows for such MOS corrections. Statistical treatments can also be applied to further downscale model outputs. Section 1.4.2: Formats of Seasonal Climate Forecast Uncertainty Uncertainty is an inherent feature of predictions of future climate, from weather to climate change scales. The main categories of sources of uncertainty in the predictions are: • Uncertainty on the initial conditions of the forecasting process, due to spatial gaps in the weather observing system (e.g. in oceanic and sparsely populated areas, such as deserts) • Uncertainty on our representation of the system, e.g., a regression line is only an approximate representation of a scatter plot; GCMs only imperfectly capture local conditions due to their coarse resolution • Uncertainty on the evolution of the system and of the main drivers, e.g. changes in SST over the period of seasonal forecast, changes in the emission rates for climate projections. The relative importance of these factors depends on the horizon of the forecast, with the uncertainty on the initial conditions being most important for short term, weather forecast and the uncertainty on the evolution of the system and external drivers, for longer horizons such as climate change. It is important to note that uncertainty can and should be evaluated and communicated to convey forecasters confidence in the forecast. Similarly, evaluation of the uncertainty should be an integrant part of forecast evaluation. Probabilistic formats are a convenient way to convey the uncertainty of the predictions. Flexible Seasonal Forecast As discussed in previous sections, climate and its variability can be expressed as a probability distribution of historical values of the variable of interest, e.g. annual or seasonal rainfall, number of rainy days, monthly average temperature, etc. Historical distribution gives us a first guess of the likelihood of a given value to occur or the be exceeded/not exceeded. Probabilistic forecast indicates potential shifts in those probabilities, as shown on the Figure 1.4-1. This format of seasonal forecast allows the user to access shift in the probabilities for a value that is of interest to his decisions. Maprooms allow the visualization of historical and predicted probabilities for a given location, like in Figure1.4-1, as well as plot maps of probability of exceedance (or non- exceedance) of a user-defined value. In addition, maprooms allow to plot maps of probability of exceeding/not exceeding thresholds of interest Figure 1.4-3: Predicted shifts in probabilities in the seasonal rainfall totals, predicted in September 2020 for Oct-Dec 2020 season, for the location 31.5E, 12.25S. a) shifts in the distribution of probabilities, b) shifts in the probabilities of exceeding rainfall. 31 Ethiopia MODULE 1 Think Deeper 1. Looking at the Figure 1.4-3a, what does the forecast suggest for next season’s rainfall? a) What is the most probable forecasted rainfall amount? What is the associated probability? b) What is the most probable climatological rainfall? What is the climatological probability associated? c) What is the forecasted probability for the amount from question b? is the shift in probability with respect to climatology large? 2. Looking at the figure 1.4-3b, what is the meaning of the median? a) What is the climatological rainfall related to median? b) What is the forecasted rainfall amount related to 50% probability of exceedance? Is the shift in the predicted median large? Tercile Forecast Tercile format is an older format of the probabilistic forecast where the historical distribution was first divided into three equiprobable segments – the terciles – that had a nominal probability of occurrence of 33.3%. The forecast indicated predicted shifts from the expected 33.3% probabilities, as shown in the Figure 1.4-4. Figure 1.4-3: Predicted shifts in probabilities in the seasonal rainfall totals, predicted in September 2020 for Oct-Dec 2020 season, for the location 31.5E, 12.25S. a) shifts in the distribution of probabilities, b) shifts in the probabilities of exceeding rainfall. Maprooms allow the display spatial distribution of predicted probabilities of individual terciles as well as ‘summary’ maps, where the most probable tercile is displayed. The Tercile forecast format has the inconvenience of pre-partitioning the full distribution into predefined segments that do not always correspond to thresholds relevant to decisions. However, due to historical reasons, a lot of forecasts are still presented in this format. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 32 Figure 1.4-4 Probabilistic forecast of seasonal rainfall for the July-September season in Ethiopia showing most probable terciles. Source: EMI http://213.55.84.78:8082/maproom/Climatology/ Climate_Forecast/DomTerProb. html, accessed 08/25/2021 Think Deeper Looking at the Figure 1.4-4: 1. What do the green colors correspond to? What do orange/yellow colors correspond to? What do grey colors correspond to? 2. Based on your recollection from the section 1.1, which parts of Ethiopia experience rainfall in July-September season? 3. What does the forecast suggest for the JAS 2018 season’s rainfall? 33 Ethiopia MODULE 1 Section 1.4.3: Use of Seasonal Forecasts User Interpretation While it may be tempting for users of climate information to seek climate forecast information that seems very precise (either in spatial resolution) or in the nature of the forecast itself (a high confidence of a particular outcome), it is important for users to understand the inherently uncertain nature of climate forecasting and know how to interpret and act on probabilistic information. Historically, many of the regional climate outlook forums (RCOFs) have produced seasonal forecast consensus maps based on forecasting efforts using an array of software and engagement from the meteorologists from many nations within the region. Prior to the RCOF, the technical experts convene on the findings of their different forecast efforts and a qualitative consensus map is drawn delineating different sub-regions with respect to tercile probabilities for precipitation and temperature for the upcoming season (the probability of the season being “above normal”, “near normal” or “below normal”). This format has often made the meteorological community feel comfortable but has created challenges for the user-community. Interpreting a tercile-based consensus probabilistic forecast into an actionable decision requires specific knowledge of what “above normal”, “near normal” and “below normal” mean for a given location and season - this is knowledge that the users may not have directly. Furthermore, critical decisions for agriculture, health or water management may depend on thresholds other than the 33rd or 67th percentiles of temperature and precipitation. In response to these challenges, there has been an effort in recent years (particularly in the Greater Horn of Africa Climate Outlook Forum GHACOF) to present forecast information with the full forecast distribution and to use a fully objective approach to forecast generation. This has been mainstreamed into the forecasting efforts of the IGAD Climate Prediction and Application Center (ICPAC).4 On these web pages, users can select a region or point location of interest and look at the probability of exceedance graph for rainfall (in mm), the rainfall anomaly forecast, the percent of median forecast, and the probability of exceeding or not exceeding a given threshold. By providing the users with the ability to select their own threshold and focus on a region of specific interest, the hope is that this newer format will be better tailored to the needs of the users. Other Considerations on the Use of Forecasts In reality, crop yield depends on a number of factors beyond the seasonal rainfall total: the timing of the onset and cessation dates, the timing of planting and harvest dates, the intra-seasonal distribution of rainfall, the number and length of dry spells, the frequency and severity of heat waves, and a host of non-climatic factors (soil fertility, management practices, land use pressures, etc.). Navigating these decisions successfully requires that the meteorological community be forthright about matters of forecast skill and uncertainty and adequately verify forecasts against historical observations. Further, the NMHS also needs to be sensitive to the needs of users and to try to present information in a usable way for the different user sectors. However, the users also need to appreciate that climate information is inherently probabilistic and uncertain and need to frame their decision making around the probabilistic information that can be produced by the technical experts. Perfectly skillful forecasts many months in advance of a target season are generally not possible. But many forecast outputs can still be useful at guiding decisions. Professionals in various sectors that constitute the user community need to be clear enough in their own understanding of what level of skill and lead time are necessary from a forecast in order to effectively inform a decision. That understanding should guide the users’ engagement with the NMHS. There should also be an understanding on the users’ side that because of the uncertain, probabilistic nature of climate forecasting, some forecasts will appear “wrong” (i.e., the forecast leaned strongly towards wet conditions and dry conditions actually occurred). Such events may imply a need to refine and reevaluate the forecast methodology, but do not necessarily mean that the forecasting enterprise (led by regional organizations and NMHS) is inherently untrustworthy or lacking in value. Both the user community and the meteorological community should bring a historical perspective to this engagement and strive to keep in mind the mixture of successes and failures over time to try to continuously improve the process of forecasting, communication and engagement. 4 http://digilib.icpac.net/maproom/Climatology/Climate_Forecast/SeasRainFcst.html The more general ICPAC climate maproom page can be found at http://digilib.icpac.net/maproom/Climatology/ Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 34 Section 1.4.4: The Forecasting Continuum While the main focus of this module is the seasonal forecast it is important to understand that forecasts are made at different tome scales and horizons, from the familiar weather forecasts to climate change projections. Different forecasts are based on different premises. For example: Weather forecast predicts the short-term evolution of atmospheric conditions over a period of a few days and is mostly based on our knowledge of current atmospheric conditions Seasonal forecasts predict the climatic conditions several months ahead and are based on the fact that warmer or cooler sea surface temperatures, such as in the case of ENSO or IOD, impact large scale circulations that bring moisture into the continents and impact rainfall. Climate Change projections attempt to predict the evolution of climate several decades ahead and are based on the change of atmospheric composition thus the change in the energy that our climatic system receives and redistributes via global ocean and atmosphere circulations and uses to melt the ice and increase the temperature of the oceans and continents. It is important to understand that the longer the horizon of the forecast, the greater the uncertainty and the less precise the forecast in terms of spatial and temporal resolution as well as amounts. Figure 1.4-6 is a schematic representation of different types of forecasts, associated time and space scales and underpinning drivers. Forecasts at the scales in between weather, seasonal and climate change, such as sub- seasonal (characteristics of rainfall or temperature within the season) or decadal (evolution of rainfall) Figure 1.4-7: Probabilistic Schematic representation of different types of forecasts as a functionof temoral and spatial scales and uncertainties. The main factors underpinning different types f forecast are listed at the top of the figure. Source: adapted from CLIVAR. 35 Ethiopia MODULE 1 Self-Check: Use of Seasonal Forecasts 1. What are the two main types of forecast? Cite advantages and drawbacks of each. 2. What are the main sources of uncertainty? Can uncertainty be avoided? How do forecasts deal with the uncertainty? 3. Would you base your on-farm decisions on a deterministic (i.e. giving precise values), high resolution seasonal forecast (e.g. at the level of the farm)? Explain why or why not. *Answers can be found in Appendix A Understanding and interpreting most common climate maps and charts This section only presents a few basic concepts on how to interpret maps and graphs. All maps shown in this section are accessible from Ethiopia’s Meteorological Institute maproom and are based on Ethiopia’s ENACTS climate data or digital elevation data. By the end of this section, you will be able to: • Identify the basic parts of a map and graph • Interpret maps and graphs Section 1.5.1: Interpreting Maps In climate science, a lot of information is conveyed by maps that display the value of different variables over space. They allow us to see the spatial dimension of climate. The key elements of a map are as follows: Compass: The map’s compass or direction arrow generally shows which direction on the map is North. The directions for East, South and West are implied from the direction for North, because East is always 90 degrees to the right of north, South is always opposite north and West is always 90 degrees to the west of north (or opposite East). Most maps have the convention that North is directly up on the page, where East is to the right, West is to the left and South is down. If there is no compass or direction arrow given, this is the assumption that should be made. Scale: The scale of a map indicates how the distance on the map corresponds to distances in real life. Most maps are “linear” in that a fixed distance on the page corresponds to a fixed horizontal distance in real life. For point of reference, near the equator, one degree of latitude or longitude is approximately equal to 110 km. If a scale is not given in a map, but the geographic entity depicted (hemisphere, continent, country, state, city, etc.), is known, the scale of the map can be estimated by looking up the length or coordinate dimensions of the geographic entity and measuring the distance between boundaries on the page or the screen image. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 36 Key/Legend: The key or legend is usually presented as an insert in a corner of the map or a feature below or above the map that provides insight into what the symbols on the map mean. Coastlines, national boundaries and sometimes state or region boundaries are often demarcated by lines (usually black solid lines) and often, the name of the geographic entity is written as text inside those boundaries. But maps in general tend to plot other features either through a colorscale, contour lines or a pattern of shading or hatching. The key should give an indication of the meaning of the colorscale, contour lines or shading/hatching pattern. The key/legend will also generally indicate the units of the variable featured in the map (eg. mm/year or mm/month for annual or monthly rainfall, degrees C for temperature, and meters above sea level for elevation, etc.). Colorscale: In theory, a colorscale can be any arrangement of colors in any sequence and can be scaled in any mathematical context. However, there are some conventions in mapping geophysical variables that are useful to know. Ocean areas are often shown in blue. Relief maps often show low elevation regions in green, moderate elevation regions in yellow, high elevation regions in brown and very high elevation regions in white. Temperature maps often show cooler areas in cooler colors (blue, green, purple) and warmer areas in warmer colors (yellow, orange, red, pink or brown). Maps concerning precipitation can be a bit more variable. Some may use cool colors (blue, green, purple) to denote high rainfall areas and warm colors (yellow, orange, red) to denote dry areas. In other instances, the colorscale may be the other way around with cool colors depicting low precipitation and warm colors depicting high precipitation. Some precipitation maps also show different levels of precipitation as different degrees of saturation of the same color. So it is always important to look carefully at the legend and colorscale in order to accurately interpret the information shown. Isolines: The lines drawn on a map connecting locations with equal values of a particular variable (the prefix “iso” means equal). Isolines may be drawn explicitly and labeled, may be drawn explicitly with the spacing of the isolines specified in the key or may be implied as the demarcations between areas of different color in a colorscale map. Examples of isolines may be contour lines on a relief map indicating locations at the same elevation, isotherms (lines connecting locations of equal temperature), isohyets (lines connecting locations of equal precipitation) or isobars (lines connecting locations of equal atmospheric pressure). In most cases, colorscales and isolines are spaced at equal intervals (eg. every 100 meters elevation or 50 mm annual rainfall or 2 degrees C of average temperature). However, some colorscales are not equally spaced (for example, a precipitation map with different colors at 50 mm, 100 mm, 150 mm, 200 mm, 300 mm, 400 mm, 600 mm and 1000 mm). For practice reading the most common graphs and self-check exercises for Section 1.5, please turn to Activity 1.5 in Appendix B. Figure 1.5-1: Example of a colroscale. The cooler colors (blue, green) represent low rainfall, and the warmer colors (red, orange) represent high rainfall. Self-Check: Reading Maps and Charts For practice reading the most common types of graphs and maps, please turn to the Activity 1.5 worksheet in Appendix B. 37 Ethiopia MODULE 1 SECTION 1.5 1.1 Climate Change While climate varies on many time scales and sections 1.1 and 1.2 have discussed some of those scales and section 1.4 dealt with forecasting across time scales, the focus of this section will be on long-term climate change. There is a great deal that can be written on this topic, but the objective of this section is only to provide some of the most essential highlights. The earlier sections of this chapter have discussed climate variability on time scales from a few weeks to several decades and our working definition of climate generally refers to a collection of aggregate characteristics established across a period of multiple decades. Climate change happens at an even longer time scales and refers to changes in the climate characteristics over centuries or even longer. The Earth’s climate has changed many times and quite considerably over the course of its 4.6-billion-year history – sometimes quite quickly - due to internal processes within the climate systems and sometimes, due to external events such a large meteorite strike (see Think Deeper box). Some of those changes led to massive animal and plant extinctions, some to slow evolution of ecosystems and their inhabitants. Most of long-term variations in climate occurred long time ago, before humans appeared and were affected by them. The current concern with the climate change lies in how it will impact human societies, either directly (e.g. through flooding, unbearable temperatures, drinkable water shortages, affecting places where we live, etc.) or indirectly through affecting the environment in which we live or depend upon for our economies and wellbeing (e.g through droughts, crop and livestock loss, water and energy shortages, etc.). The concern is also due to the fact that climate change is happening quite fast – the effects are quite noticeable from one generation to another – much faster than our societies are able to adapt. Ultimately, it will have large socio-economic costs. On the other hand, measures can be taken to slow the climate change (climate change mitigation) and anticipate and adapt to new climatic conditions (climate change adaptation). This section will review the causes of current climate change, put it into perspective relatively to past changes in climate and review the impacts of climate change globally and in Ethiopia. Think Deeper Impact of meteorites on the Earth’s climate: In simplest terms, a meteorite is a rock that falls to Earth from space. The vast majority of meteorites are small, no larger than a pebble, and have very little effect on the Earth system. Very large meteorite impacts are rare, but an impact 66 million years ago in what is now the Yucatan Peninsula in Mexico is hypothesized to have caused the extinction of many plant and animal species, including all dinosaurs, except the ancestors of the living birds. While the crater of the Yucatan meteorite is no longer visible (eroded and partly underwater), figure 1.30 shows the crater of another meteorite, in South Africa. The impact of the Yucatan meteorite affected the Earth system in many ways, including1 • Earth to several hundred degrees Celsius for a few minutes due to the friction of the rock passing the atmosphere at high speed; • releasing a lot of dust particles to the atmosphere that obscured the sunlight and caused drop in Earth temperature for several months, reduced the photosynthesis of plants; • causing a massive earthquake and related tsunami, and changing the landscape; • melting the rocks and releasing vast amounts of carbon dioxide to the atmosphere; causing massive fires, ultimately leading to the extinction of many organisms and evolution/adaptation of others Figure 1.30: Vredefort Crater (South Africa) This is the largest verified impact crater on Earth, with an estimated diameter of about 300 kilometers. It was formed over 2 billion years ago and is one of the oldest known impact structures. 1 adapted from ‘ Meteorite impact’ https://ugc.berkeley.edu/background-content/meteorite-impact/#:~:text=Melting%20of%20 rocks%20at%20the,the%20atmosphere%20by%20the%20impact. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 38 1. Based on the description of Yucatan meteorite impacts, which ones lasted a short of amount of time and which ones had longer term consequences? 2. Which of the impacts were local and which were global? 3. Based on what you learned in Section 1.2.2 think about the different ways or mechanisms the meteorite strike affected the climate globally. 4. One of the impacts was temporary dimming of the solar radiation reaching the Earth. Can you think of any other phenomena that could have similar effects? 1.1.1 The mechanism of the current climate change We have learned in section 1.2.2 that solar radiation is the primary source of energy for Earth and climate and that ocean and atmosphere play a critical role in redistributing the energy between equatorial zones and higher latitudes, between oceans and land masses etc. The atmosphere plays also a critical role in keeping the Earth warmer than it would be without it. Indeed, while the atmosphere is transparent to the visible light from the Sun – the reason we can see the Sun and its radiation reaches Earth’s surface – it does not let all types of radiation easily pass through. In particular, it absorbs the so called infra-red radiation, i.e. the molecules of some of the gases in the atmosphere will absorb this specific type of energy and their temperature will rise. To explain the infrared radiation in very simple terms: any object whose temperature is in the range of +/- 80°C emits thermal longwave infrared radiation2. We cannot see it like the radiation emitted by the Sun, which is much hotter, but we can feel it in the form of heat. Some nocturnal animals and some special cameras can see it and capture as image (see fig. 1.31) and it is sometimes referred to as night vision. Figure 1.31: Examples of Visible and Infrared images. Left: a picture of a man in visible light; Middle: a picture of the same man in Infrared light; Right: an infrared picture of an elephant, taken at night. Sources: http://nhm. ac.uk, https://www.elephantlisteningproject.org/thermal-imaging/ Think Deeper Looking at the IR images in figure 1.31: if warmer objects are brighter yellow or red and cooler are purple or blue (the images are not in the same color-scale): 1. what are the warmer and the cooler part of the man (fig. 1.31, middle)? Why there are such differences? 2. Looking at the scale in the right figure (in °C) , what is the approximate temperature of the warmest parts of the elephant? What is the approximate temperature of the coldest parts? What is the approximate temperature of the water? Figure 1.32 shows the schematic paths of the visible (emitted by the Sun) and infrared (IR, emitted by the Earth) radiations. Solar energy mostly passes through the atmosphere and, upon reaching the surface of the Earth, is either reflected or absorbed by the surface, increasing its temperature. Some of the solar energy can be reflected or absorbed by the clouds and aerosol partices in the atmosphere. The infrared radiation, or terrestrial 2 In reality any object whose temperature is above absolute zero or -273.15 °C emits electromagnetic radiation whose properties depend on object’s temperature. For example, the Sun, whose surface temperature is about 5,600°C emits radiation that is visible to human eye while human body, with temperature around 36°C emits longwave infrared radiation, invisible to human eye. The full theory behind radiative properties of objects is beyond the scope of this course. 39 Ethiopia MODULE 1 energy, is mostly reflected and absorbed by the molecules in the atmosphere and only a fraction crosses to outer space. In other terms, the atmosphere acts like a blanket and keeps the Earth warmer than it would be without it. The average global temperature of the Earth is around 15°C and would be around -18°C, without the atmospheric ‘blanket’, known also as greenhouse effect . This effect3 occurs naturally in the atmosphere and helps keeping the Earth habitable. Figure 1.32: Left: paths of solar (visible) radiation through the atmosphere. Right: paths of the IR radiation emitted by Earth surface; source https://forces.si.edu/atmosphere/02_04_07.html Not all the gases in Earth’s atmosphere contribute to the greenhouse effect. The main gases in the atmosphere are: Nitrogen constituting around 78% of the atmospheric gases, oxygen around 20.9% and Argon around 0.9%. They do not contribute to the greenhouse effect. The remainder, less than 1%, is composed of approximately 12 different gases, among which carbon dioxide, water vapor, methane, nitrous oxide and water vapor and fluorinated gases are the most active in trapping IR. In other words, a very small fraction of atmospheric gases, measured in parts per million (ppm) is responsible for keeping our planet habitable. It is easy to see that even a small perturbation in the amount of these ‘Greenhouse Gases’ (GHG) can have large effects on the amount of energy trapped by the atmosphere. For the last 150 years or so, human activities linked to industrialization, have added GHG to the atmosphere, in amounts sufficient to cause additional trapping of energy in the system. Figure 1.33 illustrates the natural vs human-induced greenhouse effects. 3 The term greenhouse effect comes from an analogy to greenhouses – structures with glass walls and roofs, often used to grow vegetables or flowers in cold climates, or to control growing conditions, like humidity etc. Both greenhouses and the greenhouse effect work by retaining heat from sunlight, but the way they retain heat differs. Greenhouses retain heat mainly by preventing the air heated by the solar radiation from escaping the greenhouse. In contrast, the greenhouse effect retains heat by restricting the rate at which heat, not air, escapes to space. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 40 Figure 1.33. Schematic Diagram of the natural greenhouse effect and the human enhanced greenhouse effect. 1.1.2 The causes of current climate change The main greenhouse gases that are causing climate change include carbon dioxide and methane. The current increase in GHG amounts in the atmosphere are unequivocally linked with human activities. Since the 1800s, human activities have injected vast amounts of carbon dioxide into the atmosphere, primarily due to the burning of fossil fuels like coal, oil and gas for heating and transportation. Clearing land and cutting down forests also release carbon dioxide. Agriculture, oil and gas operations are major sources of methane emissions. Energy, industry, transport, buildings, agriculture and land use are among the main sectors releasing greenhouse gases (fig. 1.34). Figure 1.34: Contribution of various sectors to GHG emissions. Source: IPCC (2014) Due to different of development and industrialization levels, different countries do not contribute equally to such emissions, with the contribution of developing countries per capita (per person) being smaller (figure 1.35). At the same, their economies being dominated by climate-sensitive sectors such as agriculture, they will be the most affected by climate change. 41 Ethiopia MODULE 1 Figure 1.35: GHG emissions per capita in 2021. Source https://ourworldindata.org/grapher/per-capita-ghg- emissions?time=latest 1.1.3 Global Warming or Climate Change? Although people tend to use these terms interchangeably, global warming and climate change are not exact synonyms. “Global warming” refers to the rise in global temperatures due mainly to the increasing concentrations of greenhouse gases in the atmosphere. “Climate change” refers to the changes in climate over a long period of time linked with the increase in the energy of the climate system due to increase in concentrations of greenhouse gases in the atmosphere. These changes will affect precipitation, temperature, and wind patterns amd more. As we have seen in section 1.2.2 the energy in the ocean-atmosphere system and the differences herein are responsible for many processes within the climate system, such as ocean currents, movements of air masses, evaporation etc. It is easy to imagine that all these aspects will be affected by the increased greenhouse effect trapping more energy within the system. In fact, the additional energy in the system will not all be used to increase the global temperature, but will contribute to various climatic processes such as melting polar ice sheets, increase evaporation, increase the intensity of storms, and some of it can be stored e.g. in the ocean. Figure 1.35 illustrates the split of additional energy between various types of energy and processes within the climate system. If these different mechanisms were not at play and all the additional energy was transformed into temperature increase, our Earth would be much warming even faster. Figure 1.35: Illustrative partition within the climate system of the additional energy linked with the increase in greenhouse gases in the atmosphere Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 42 Think Deeper 1. The current climate change is caused by the increase of the amount of GHG in the atmosphere. Thinking about the Yucatan Meteorite: which of the meteorite impacts had similar effect? What is the difference between that event and the current climate change mechanism? 2. Looking at the map of average temperatures in Ethiopia (figure 1.2-2) what are the minima and the maxima? Section 1.5.1 stated that the natural greenhouse effect maintains the global temperature at 15°C. Does this temperature match what you see in Ethiopia? If not, how can you explain the differences? 3. Which of the terms ‘Global Warming’ or ‘Climate Change’ applies to what will happen in Ethiopia in the coming decades? 1.1.4 The impacts of current climate change How is climate change predicted? To predict climate over next 100 years or so., scientists use General Circulation Models, like the ones described in section 1.4, figure 1.4-2. The models are essentially like mini-laboratories, where different scenarios can be tested. Because they will simulate climate over a very long period of time, these models are first tested by simulating climate observed over past 100 years. If they successfully simulate the main characteristics of past climate, such as average global temperature and the amplitude of its variation, like the ones in the figure 1.3-6, the models are instructed to increase the GHG concentration in atmosphere according to anticipated emission rates. Considering the exact future GHG emission rates are not known and highly depend on the way various societies develop in the future, five possible scenarios of GHG increase are tested, describing broad socioeconomic trends that could shape future society, intended to span the range of plausible futures. They include: a world of sustainability-focused growth and equality (SSP1); a “middle of the road” world where trends broadly follow their historical patterns (SSP2); a fragmented world of “resurgent nationalism” (SSP3); a world of ever-increasing inequality (SSP4); and a world of rapid and unconstrained growth in economic output and energy use (SSP5). These shared socioeconomic pathways, in turn, lead to different scenarios of climate evolution and a range of possible outcomes. Note that: • As in the case of seasonal forecasts, climate change projections are not about predicting a sequence of weather events but about their statistics and how they can vary over time. • The models are imperfect and only capture the reality to a certain degree. Thus, climate change projections should not be taken at face value at a given location but as indications of probable outcomes. • In addition, different models may have different strengths and weaknesses capturing various climate phenomena, as for seasonal forecasts, it is best to consider outcomes provided by a range of models • Therefore, probabilistic approaches and ranges of outcomes should always be considered. However, despite the caveats above, with appropriate approaches we can have confidence in climate change projections and models because of their physical basis, and their skill in representing observed climate and past climate changes. Models have proven to be extremely important tools for simulating and understanding climate, and there is considerable confidence that they are able to provide credible quantitative estimates of future climate change, particularly at larger scales. Figure 1.36 provides the latest projections of global mean surface temperatures, according to various development scenarios. Note that the extent of future climate change depends on what we do now to reduce greenhouse gas emissions. The more we emit, the larger future changes will be and some scenarios project an increase on global temperature on the order of 5 °C, much greater than the increase observed over past 100 years, and represented in fig, 1.3-6, leading to uncharted territories for local climate. 43 Ethiopia MODULE 1 Figure 1.36. Observed and simulated global surface temperature change relative to the 1850-1900 mean temperature for the different SSP emissions scenarios, IPCC, AR6 Global impacts of Climate Change The changing climate impacts society and ecosystems in a broad variety of ways. For example, climate change can alter rainfall, influence crop yields, affect human health, cause changes to forests and other ecosystems, and even impact our energy supply. Climate-related impacts are occurring across the countries and over many sectors of our economy. Many efforts are made to assess and anticipate potential changes in various sectors in various regions of the world and reviewing them all is beyond the scope of this course. However, we can focus on the impacts of climate change on one resource– water – as approximately 75% of the Earth’s surface is covered by liquid water and ice and it is essential to all life on Earth. Water is constantly moving between the atmosphere, land, and ocean, shaping our planet’s climate and ecosystems. Water also stores and exchanges heat among different parts of the Earth system. The location and balance of that water between different reservoirs: the oceans, glacial ice, ground water, surface fresh water, and the atmosphere depends very much on the temperature. In colder global climate periods, more of the water is locked in glacial ice, sea levels fall, evaporation rates fall and less water is in vapor form in the atmosphere. In warmer global climate periods, the opposite is true – glaciers shrink, sea levels rise, evaporation rates increase and more water vapor is held in the atmosphere. The current climate change will have various adverse effects on the water cycle and water resources. Warmer air can hold more moisture than cool air. As a result, in a warmer world, the air will suck up more water from oceans, lakes, soil and plants. The drier conditions this air leaves behind could negatively affect drinking water supplies and agriculture. The warmer, wetter air could also endanger human lives as higher humidity will make future higher temperatures unbearable in some places, by blocking the cooling effects of our sweat. When all that extra warm, extra wet air cools down, it drops extra rain or snow to the ground. Thus, a warmer world means we get hit with heavier rain and snowstorms. By changing air temperatures and circulation patterns, climate change will also change where precipitation falls. Some areas are expected to get drier, some wetter. The heavier bursts of precipitation caused by warmer, wetter air can lead to flooding, which can endanger human lives, damage homes, kill crops, and hurt the economy. Heavier rainstorms will also increase surface runoff — the water that flows over the ground after a storm. This moving water may strip nutrients from the soil and pick up pollutants, dirt, and other undesirables, flushing them into nearby bodies of water. Those contaminants may muck up our water supplies and make it more expensive to clean the water to drinking standards. In addition, as runoff dumps sediments and other contaminants into lakes and streams, it could harm fish and other wildlife. Fertilizer runoff can cause algae blooms that ultimately end up suffocating aquatic critters and causing a stinky mess. The problem is compounded by warming water, which can’t hold as much of the dissolved oxygen that fish need to survive. These conditions could harm fisheries. In the oceans, temperature changes also have the potential to alter major ocean currents. Because ocean temperatures also drive atmospheric circulation patterns, this could change weather patterns all over the world as well as make rainfall more variable and less predictable. Warmer temperatures and increased acidity (due to chemical reactions between carbon dioxide and seawater) are making life difficult for sea creatures. These changes are transforming food chains from the bottom-up. In addition, many fish are moving poleward in search of cooler waters, which has implications for the fishing industry and people who like to eat fish. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 44 Increased temperature will also have impact on the snowpack and ice sheets. Ordinarily, as winter snowpack melts in the springtime, it slowly adds fresh water to rivers and streams and helps to replenish drinking water supplies. However, as the air warms, many areas are receiving more of their precipitation as rain rather than snow. This means less water is being stored for later as snowpack. The lack of snowpack can lead to drier conditions later in the year, which can be bad news for regions that rely on snowmelt to refill their drinking water supplies. And of course, as ice sheets and mountaintop glaciers melt, they’re dumping extra water into the oceans; the resulting sea level rise jeopardizes coastal regions around the world. In addition to changing the water cycle, climate change could change how we use water and how much we need. Higher temperatures and evaporation rates could increase the demand for water in many areas. Given the very many impacts of climate change on water and temperature it is easy to imagine the multitude of further impacts on societies and ecosystems but these will highly depend on specific location Climate Change Impacts in Ethiopia While there are many findings and analyses for many regions of the globe within the IPCC, the key messages with relevance for Ethiopia are: temperatures are likely to become hotter, heat waves are likely to become more frequent, cool spells are likely to happen less frequently and both droughts and floods are likely to become more frequent. The reason behind the increase in temperatures and heat extremes is fairly self-evident. The reason for the anticipated increase of both droughts and floods is part of a broader narrative of “hydrological intensification” described in the previous section. As the atmosphere warms and can retain more water vapor, stronger, wetter storms are possible – and so flooding risk and intense precipitation risk is likely to increase. In addition, heat extremes can also exacerbate and intensify (and be exacerbated by) drought conditions by increasing the rate of evapotranspiration from the land surfaces and vegetation. When there is less soil moisture and moisture in plants, the Earth’s radiated energy does more to directly heat the surface. This interaction between heat and drought can enhance the likelihood of soil desiccation, crop damage and wild-fires. The spatial distribution of rainfall and drought extremes over time depends on the dynamics of individual storm systems and it is impossible to predict how individual extreme events are likely to unfold, but Ethiopia, like many regions of the world is expected to see increases in the frequency of both flooding and drought under climate change. These changes in the climate system are likely to have significant impacts on ecosystems and the water, agriculture, energy and health sectors. Climate change may also amplify risks to urban dwellers or lead to migration. A range of recommended approaches can help managing climate risks; including developing early warning systems, community engagement-based projects, ecology-based adaptation measures, integration of climate adaptation into social protection plans and livelihood diversification. Think Deeper • What are the main climate risks in your region? • In the light of what you learned about climate change and its effects, how do you think these risks may change in the future? 45 Ethiopia MODULE 1 1.1.5 Climate change skepticism A “climate skeptic” is someone who challenges or downplays the prevailing understanding of global warming, especially the notion that it’s driven primarily by human actions. These individuals often believe that global warming is a natural and cyclical event. Some even argue that climate change might eventually yield positive outcomes or that technological advancements will eventually address the issue. More often than not, this climate skepticism stems from a lack of understanding, an unease about confronting unsettling realities, or a resistance to implementing stringent measures. It’s worth highlighting: Misconceptions and falsehoods about the climate are rampant on social media platforms. It’s crucial to be vigilant about the sources from which we glean information. Many of the climate-skeptic arguments rely on the fact that Earth’s climate, including temperature and carbon dioxide content in the atmosphere, have varied in the past. While it’s true that our planet has experienced several climate changes over its long history, the one we currently live in is faster and more violent than ever. In addition, it is directly caused by human activity - that was not the case before. Previous climatic variations had two main causes: • Volcanic eruptions, which are responsible for cooling (by emitting particles into the atmosphere, which return sunlight); • Alternation between glacial and warm periods (approximately every 10,000 years). A change caused by the variation of the Earth’s orbit relative to the Sun. Current changes are not related to any of these phenomena. Climate scientists are certain that the primary reason for the increases in the atmospheric concentrations of carbon dioxide, methane and other greenhouse gases is because of human activity. There is no serious debate about this point in scientific circles, even though there is (unfortunately) a political and social debate about it in some regions of the world. Think Deeper Figure 1.37 represents the variations of global surface temperature over past 20,000 years, together with projections for the next 100 year or so. Note that the temperature on the vertical axis is presented as anomalies with respect to late 19th century – beginning of the industrialization. Note also unequal intervals for time the time, on the horizontal axis, whereby the past and the next 100 years have the same spacing as 1000 or even 10,000 years earlier – this is important for reading the graphs. 1. Have a look at the period going back approximately 10,000 from now. This is a period of relatively stable climate in recent Earth’s history that allowed emergence of agriculture and establishment of human societies as we know them now. There were, however variations in Earth’s surface temperature within this period. Identify the main ‘warm’ and cold periods on the figure. What were approximate global surface temperature anomalies during these periods? 2. Have a look at the future 100 years. The dotted line represents projections. a. What is the meaning of the three arrows? b. Taking the middle global temperature increase, how much warmer is the global temperature projected to be with respect to today’s temperature? 3. Have a look at the end of the last Ice age that happen about 20,000-15,000 years ago. a. What was the range of global temperature anomalies then? b. How long did it take for the temperature to reach the zero line? c. What would be the temperature increase over the same length of time if the temperature continues to increase at the rate projected for the next 100years 4. Can you use the elements above to explain how unusual the current climate change is? Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 46 Figure 1.37. Reconstruction of the global Earth’s surface temperature over past 20,000 years. Note the stable period over the last 6-8,000 years and the 4-5 degree C global mean temperature difference between the Last Glacial Maximum and the present. Below are a couple of climate-skeptic’s arguments and their rebuttals. • The climate was warmer in the past, like in the Middle Ages Drawing parallels with today is misleading. The warmth during this era was regional (Europe), not global. Furthermore, its impact pales in comparison to the climate changes we observe currently. • There is no global warming, it’s still cold! It’s crucial not to confuse weather with climate. Weather pertains to short-term atmospheric phenomena and aims to forecast conditions in specific locations over brief periods. In contrast, climate refers to statistics of weather and one cold event will not affect observed warming trends. Moreover, isolated instances of record low temperatures in a specific location over a short duration aren’t indicative of overall climate trends. To gain a comprehensive understanding, you need to consider average temperatures globally. • 1°C is nothing, scientists are too alarmist. While a single degree might seem inconsequential to us, the ecosystems certainly feel the impact. Consider the minor shift required to change rain to snow, and vice versa. In addition, we are talking about 1°C global average which can translate to much more in certain locations, as well as to other consequences such as more frequent extreme rainfall and winds leading to flooding and other destruction, or prolonged droughts impacting food production etc. • It doesn’t matter, humans, fauna and flora are able to adapt! Throughout history, climate variations have played a pivotal role in mass extinctions. It’s an undeniable fact. Countless animal and plant species are already feeling the harsh impacts of the rapid climate shifts. Such swift changes don’t afford them sufficient time to adapt, which in many cases would involve migration. 47 Ethiopia MODULE 1 Humans aren’t exempt from these challenges. While our adaptability may be superior, massive human migrations could become necessary. However, in today’s world, this isn’t a straightforward proposition. Habitats that remain suitable for habitation are tightly regulated by well-defined borders. This means that cross-border migrations have become far more complex than they might have been in ages past. If we fail to anticipate and prepare for these shifts, the inevitable large-scale migrations could lead to heightened tensions at borders, potentially even escalating to armed confrontations. • You can’t even get weather forecasts right for a few days ahead, how can you pretend to predict climate? With climate projections we do not predict individual patterns but general trends. Suppose we heat a pan of water; we know the water will get hotter, but we don’t know where the bubbles will appear when it starts boiling. So, we can predict general trends even if sometimes we can’t predict the exact detail. • We’re heading into an ice age Ice age cycles are on the other of tens of thousands of years. Global warming will have consequences in the next couple of decades. We need to worry about the next 100 years, and not hope that an ice age in over 10,000 years will solve our loomimg problems. • Solar cycles cause global warming While it is true that Sun’s activity and emitted radiation vary over time, in recent decades, the sun has been slightly cooling & is irrelevant to recent global warming. 1.1.6 Climate Information for adaptation In general, climate information needed to design adaptation plans will depend on the scope of the adaptation planning/analysis, ranging from initial risk screening and detailed risk analysis to assessing risk management options (Lu, 2007). While the issue of climate change brings about the notion of modifications of what is usually deemed to be steady and immutable, climate varies on a number of scales. Simply put, human activities as well as natural systems develop in response to such variations and can cope with a range of climatic conditions without serious damage. Figure 1.37 conceptualizes current climate variations and the coping range. Occasionally, climate variations exceed the coping range and the system is put under higher stress or severely damaged. If the frequency of such events is low, the system will progressively recover or reach a new state. Within such framework climate change means that severe impact events may happen more frequently, and/or events of unprecedented magnitude can occur, putting the system under more stress, potentially not recoverable (Figure 1.37, right side). Figure 1.38. Conceptual illustration of historical and future climate, coping range and adaptation. Source: Lu, 2007, adapted from Carter et al. 2007. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 48 Adaptation is aimed to alleviate or support the recovery from such damaging events. Note that high impact events in the future can occur more frequently due to (Fig. 1.39): • Change in the average, keeping the amplitude of the variability the same (Fig. 1.39a) • Change in the amplitude of the variability, keeping the average the same (Fig. 1.39b) • Change in the frequency of rare events on one side of the distribution (Fig. 1.39c) • Any combination of the above at the same time. Figure 1.39. The effect of changes in temperature distribution on extremes. Different changes in temperature distributions between present and future climate and their effects on extreme values of the distributions: (a) effects of a simple shift of the entire distribution toward a warmer climate; (b) effects of an increase in temperature variability with no shift in the mean; (c) effects of an altered shape of the distribution, in this example a change in asymmetry toward the hotter part of the distribution. Source: IPCC (2012) While designing adaptation plans and seeking climate information it is important to keep in mind the following elements: Climate Information vs. Impacts Information Diverse past and projected statistics of meteorological data, as cited above and applied to rainfall, temperature, atmospheric moisture, solar radiation, wind etc. are climate information and should be available from Meteorological Services. Changes in crop yields due to variations in temperature, forecasts of areas flooded in a flash flood after heavy rain or changes in vector-borne disease incidence due to variations in temperature 49 Ethiopia MODULE 1 and rainfall, are usually beyond the mandate of climate information producers. Such information is usually co- developed with a given sector or decision maker, using impact models (e.g., crop models, hydrological or flood models, vector and disease models). Uncertainty The practitioners need to accept that information about future climate is provided with a certain level of uncertainty, inherent to projections of the future. This needs to be factored in the decision process. However, uncertainty does not mean that future climate is totally unknown or that projections are false. Moreover, uncertainty can be quantified and decisions can still be made4. There are four main sources of uncertainty in climate projections: 1. Future levels of anthropogenic emissions and occurrence of natural phenomena (e.g., volcanic eruptions) 2. Imperfections of the climate models used to project changes in climate 3. Imperfect knowledge of current climate that serves as starting point for the projections 4. Difficulty in representing, thus reliably projecting, interannual and decadal variations in climate. Time Horizon The type of information needed will depend on the type of adaptation intervention or investment and, among others, on how far into the future the return on investment is expected. Figure 1.40 presents examples of different decision types as function of their time horizon. Larger infrastructure investments (irrigation, transportation network, dams, etc.) have usually longer time horizons than individual decisions/investments such as cropping portfolio or farm planning. Time horizon of the intervention or investments will impact the precision with which the information can be provided. While the amounts of rainfall or the probability of long dry spell in the coming rainy season can be predicted with a certain degree of accuracy, only general tendencies in rainfall, with a wide uncertainty, can be provided for a horizon of 30 to 50 years. Information with such a range of uncertainty might not be suitable for farm planning. In addition, farming systems 50 years from now will most probably result from successive adaptation stages rather than from changes planned 50 years ahead of time. Furthermore, not all of these changes will be driven by climate, with global and local markets and economy, customer preferences etc. playing a significant role. Figure 1.40. Adaptation decision contexts and their associated time horizons. Source: Lu, 2007 4 Decisions are routinely made in the context of military operations and financial investments where uncertainty is greater than that of climate projections. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 50 Thus, in a lot of cases interventions other than large scale infrastructures will focus on much shorter time horizons, more compatible with decisions at individual and community levels and project cycles. At such horizons, changes in climate might be less important than interannual variability, and even not significant. In addition, IPCC (2012) recommends focusing on measures that provide benefits under future climate scenarios but also under current climate, so called no-regret options, as valuable starting points for addressing potential future trends in climate. They have the potential to offer benefits now and lay foundation for addressing changes in the future. With respect to the conceptual Figure 1.33, such approach expands the current coping range, irrespective of the direction that climate change is going to take. Figure 1.40 presents a more complete continuum of horizons, including shorter time scales, and related adaptation applications. Note that the ability to provide climate information ahead of time for horizons shorter than several decades derives from drivers that are different from anthropogenic emissions, and that different time-scales have different drivers and different data needs. Thus, the issue of adaptation to climate change requires focusing on data collection and climate information not only for the distant future. It is useful to differentiate between time-frames below (Figure 1.41). Figure 1.41. Schematic diagram showing the relationship between climate timescales, from weather to climate change, and emergency and adaptation mechanisms, from relief operations to climate change planning. Source: Mason et al. 2015, adapted from WMO (n.d) 51 Ethiopia Climate Information and Tools Available for Agriculture MODULE 2 ABOUT THIS MODULE Module 2: Climate Information Products and Tools Available for Agriculture provides an overview of relevant weather and climate information products and services that are [or will soon be] available through the Ethiopian Meteorology Institute (EMI). It teaches participants how to navigate and use relevant historical, monitored and forecast information products available through EMI’s web page and the interactive online climate information products (Maprooms). Learning Objectives • Understand agriculture-relevant products and services available through EMI’s web page. • Navigate EMI Maproom products. • Use EMI Maproom products. SECTION 2.1 Overview of Products Available through the Ethiopia Meteorological Institute As a mandated government institution, Ethiopian Meteorological Institute (EMI) regularly provides weather and climate information for different users and sectors, the main one being that of agriculture. This information is provided through different channels that includes newspapers, television, radio, weather bulletins, emails, social media, word of mouth, and through its web page. This section provides a brief overview of the EMI’s web page. The official web page of the EMI1 offers a wealth of information about the EMI (history, mission, strategies, reports, etc.) and climate data and information products. These products include information on how to request data; daily, three-day,2 and ten-day sub-seasonal and seasonal climate forecasts; satellite images; different types of bulletins3; and the ENACTS maproom discussed in Section 2.2 of this module. Some of these products, such as all the bulletins and sub-seasonal and seasonal forecasts, can also be downloaded in PDF format allowing them to be shared offline with users such as farmers that do not have constant internet access. Users (stakeholders, users, farmers, research/sector professionals) are encouraged to explore this webpage to be used for agricultural, policy, and other decision-making. SECTION 2.2 The ENACTS Maproom The ENACTS Maproom is a collection of interactive maps and other graphs that enable users to generate information on past, present, and future climate. The Maproom is dynamic in that the maps and figures are generated by the user instantly, as the maprooms are linked to the original climate data. Users can focus on specific areas of interest with information extraction and summarization capabilities. The current version of the ENACTS Maproom includes three “generic” Climate Maprooms as well as three application-specific Maprooms. The diagram below, Figure 2.2-1, shows the first page of EMI’s Maproom.4 However, only maproom products that are very relevant to the current topic are presented here. Figure 2.2-1 EMI’s ENACTS Maproom showing the generic “Climate” maproom and other maprooms for use by users from the agriculture, health, and water sectors. 1 www.ethiomet.gov.et 2 http://www.ethiometmaprooms.gov.et/forecasts/three_day_forecast 3 http://www.ethiometmaprooms.gov.et/bulletins/bulletins 4 http://213.55.84.78:8082/maproom/ Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 54 Figure 2.2-1 Circular diagram showing the maprooms, potential uses, and the related products in concentric circles. Different sub (branch) maproom are observed as one goes out from the center of the circle. (Grossi, IRI). The Climate Maproom The Climate Maproom5 provides information about past climate, such as rainfall and temperature amounts at any point, at national or subnational levels. It has three sub-maprooms: Climate Analysis (past); Climate Monitoring(present); and Climate Forecast (future). What is the Climate Maproom Used For? Historical climate data could be used in many different ways that help us understand the drivers of observed climate variability and change. For example, it could help in determining whether observed changes in agricultural productivity are linked to variations in climate or some other factor, such as soil degradation. It can also support informed decision-making in agricultural practices. For instance, historical climate information can help in deciding on appropriate times for land preparation, selection of seed or animal breeds, planting, weeding, application of fertilizer, and control of pests and diseases. ENACTS maprooms provide instant and adjustable analyses and visualizations in the form of interactive maps and graphs that can depict climate characteristics such as seasonality, trends, extremes and others. Inside the Climate Maproom (The Climate Maproom has three different tabs. Here, we focus on the “Climate Analysis” tab (Figure 2.2-1). The “Climate Analysis” tab in the Climate Maproom consists of several sub-maprooms (or products) that allow analysis at different timescales (daily, dekadal [10-day], monthly and seasonal). Analysis of historical climate data can be conducted in this maproom, including the following (Figure 2.2-3): 5 http://213.55.84.78:8082/maproom/Climatology/index.html 55 Ethiopia MODULE 2 • Daily rainfall analysis (e.g., mean intensity of rainfall, number of wet/dry days, probability of dry/wet spells, and more) • Dekadal rainfall analysis • Monthly rainfall analysis • Seasonal mean, trends, and probability of extremes • Extreme rainfall and temperature analysis • ENSO/IOD analysis (in some country maprooms this may appear under Climate Forecast) • Standardized Precipitation Index (SPI) long-term climatic information of an area (in most cases, is best presented in the form of graphs or maps. This is because graphs and maps can present data recorded over a long period in a simple form to reveal important information such as trends, variations and other changes over a specific time or space.) Figure 2.2-3: The Climate Analysis Maproom Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 56 Daily Climate Analysis This sub-maproom allows the user to explore historical daily precipitation by an array of different statistics. Many options are available for producing annual time series of a chosen seasonal diagnostic of the daily precipitation, such as the number of dry/wet days and dry/wet spells. The user can then choose to map (Figure 2.2-4) the mean, standard deviation or probability of exceeding the chosen threshold, over years. Clicking on the map will generate a location-specific graph (Figure 2.3-5) for the chosen diagnostic. In this and other maprooms, users can extract and plot information over different administrative boundaries (country areas) by clicking on the map or picking an administrative name from the drop-down list, once having chosen which level of administration they want. Among other analyses, this component may be used to determine: • The increase/decrease of rainfall in a certain area over time; • How dry/wet spells (consecutive x number of dry/wet days) affect crop growth; • Which areas experience heavy rainfall and have the potential to be flooded; • Which areas may experience drought? Figure 2.2-4: Average number wet days during Jul - Sep season over Ethiopia. The map shows that the southern and southwestern part of the country (brown color) has less than five rainy days during the whole season (because July to September is not their rainy season), while areas in red may have over 80 rainy days during the Kiremt season. Figure 2.2-5: Probability of having wet days above a given threshold during the July- September season for a specific location over Ethiopia. The X (horizontal) axis shows the number of wet days while the Y(vertical ) axis shows the probability(chance) of getting above a given number of rainy days(exceedance probability). For example, the probability of having above 60 rainy days for the selected location is over 90%, while the chance of having above 70 rainy days is about 50% 57 Ethiopia MODULE 2 Note that there are tabs, common to all maprooms, that provide description of the product you are exploring, some information about the datasets used to generate the products, explanations of the common tabs used to navigate the products, and contact information of developers (Figure 2.2-6). Figure 2.2-6: Useful tabs available for all maprooms. The “Description” tab provides some information about the maproom products one is looking at; “Data Documentation” is meant to describe that data used to create the maprooms; “Instructions” provide explanations of some useful buttons; and “Contact Us” provides contact info for the institution or person managing the maproom (in this case EMI). Figure 2.2-7: Useful buttons described under the “Instructions” Tab. Monthly Climate Analysis The monthly analysis allows users to view rainfall; maximum, minimum and mean temperature; monthly climatology, seasonality, anomalies, and trends. The monthly analysis, which is based on 30-years of historical observation, may be used to characterize the climate of a locations including, spatial distribution of rainfall or temperature during a specific month (Figure 2.2-8), seasonality (Figure 2.2-9), and which years experienced unusual increase or decrease in rainfall or temperature (Figure 2.2-10). Figure 2.2-4: Climatology (normal) of Minimum temperature during the month of January. Bluish colors show colder temperatures while orange and red colors show warmer temperatures. It is clear that the spatial patter of temperature follows that of elevation. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 58 Figure 2.2-9: The Climate Analysis Maproom Figure 2.2-10: Climatology (normal) of Minimum temperature during the month of January. Blueish colors show colder temperatures while orange and red colors show warmer temperatures. It is clear that the spatial pattern of temperature follows that of elevation. Seasonal Climate Analysis This sub-maproom allows users to view rainfall; maximum, minimum and mean temperature climatology; anomalies; and probability of exceedance at seasonal time scale. The user defines the season (in terms of constituent months). The analysis is presented as a mean in the map (Figure 2.2-11), anomalies (Figure 2.2-12), trends, and probability of exceeding a chosen threshold. 59 Ethiopia MODULE 2 Figure 2.2-11: Spatial distribution of rainfall climatology(normal) during the Kiremt( July-August- September) season. This figure shows seasonal rainfall totals of above 1200mm (orange and red colors) over western parts of the country, and mostly dry conditions (light blue colors) over southern and southeastern part of the country Figure 2.2-12: Seasonal rainfall anomalies during the Jul-Sep season for a selected location. For each year, the bars show whether seasonal total rainfall for a given year was above normal (brown colors) or below normal (green colors). The length of the bars shows how much below or above normal the seasonal rainfall for a given year was. Extreme Rainfall Analysis This interface facilitates the exploration of extreme monthly and seasonal rainfall characteristics (these could be total rainfall, number of wet/dry days, rainfall intensity, and number of wet/dry spells). The user can define the season as well as the daily statistics of interest and then map the probability of having values above (exceeding) or below(non-exceeding) a user-defined threshold. The map (Figure 2.2-13) shows how likely or unlikely the selected threshold will be crossed. The user can also look at the variance or coefficient of variation for a quantity of interest to get a sense of the range of variability across years. Clicking on the map will then produce the probability of exceedance as well as the probability density function (PDF) graphs for the selected location (Figure 2.2-14). Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 60 Figure 2.2-13: Probability of having over 45 dry days during the Jul-Sep season. Figure 2.2-13: Probability of exceedance (top panel) and probability density function (bottom panel) of number of dry days during Jul-Sep season for a selected woreda in Ethiopia. 61 Ethiopia MODULE 2 The top graph shows the probablity (Y-axis) of having above(exceeding) a given number of dry days (X-axis) for the selected location. For example, the probability of having above 20 dry days is about 70%, while the chance of having above 30 dry days is about 20%. The bottom graph (PDF) shows the probability of getting exactly given number of dry days. For example, the most likely (with highest probability) number of dry days for this woreda is about 24 . On the other hand, the probabilities of having below 12 dry days or above 40 dry days are very low. This is different from the top graph (probability of exceedance) which shows the chance of having above a given threshold. Extreme Temperature Analysis This maproom is similar to Extreme Rainfall, except that it analyzes minimum and maximum temperature. Probability of Seasonal Rainfall Conditioned on ENSO This maproom allows users to explore the probability that seasonal average rainfall will fall within the upper (wet), middle (normal), or bottom (dry) category (tercile) of the historical rainfall distribution during El Niño, La Nina, or Neutral ENSO (El Niño -Southern Oscillation) conditions. This information is not a forecast. It is based just on historical observations of rainfall and sea surface tempera- ture conditions. However, it is a good tool for exploring the effect of different ENSO phases (El Niño, La Nina, or Neutral) on seasonal rainfall over an area of interest. For instance, if the coming season is expected to be an El Niño season, then this maproom can help to assess the expected impact of El Niño over an area of interest. This maproom may be used to analyze: • Maps showing probabilities of a location being drier, wetter, or normal for a given season and ENSO phase (Figure 2.2-15); and • Time series of seasonal rainfall total at a given location for a particular season and the ENSO phase of each year (Fig 2.2-16). Figure 2.2-15: Probability of a dry Jul-Aug-Sep season during El Nino. This map shows the chance of the season being dry during El Niño years. Areas with red color have a high chance of being dry during El Niño, while areas with green or light blue colors have very low chance of being dry during El Niño. Similar analyses could be performed for La Nina and Neutral years. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 62 Figure 2.2-16: Time series of the seasonal rainfall total during Jul-Aug-Sep season and an ENSO phase for a selected woreda in Ethiopia. Rainfall values(Y-axis) below the black horizontal line represent below normal values, values between the black and green horizontal lines represent normal values, while values above the green line indicate above normal values. The red bars represent El Niño years, blue lines La Nina years, gray lines neutral years. Almost all El Niño years show below normal rainfall for this woreda. Self-Check: Use of Seasonal Forecasts 1. Which maproom is best for identifying the driest and wettest months of a given location? 2. Increase in temperature is expected to decrease coffee production. How would you check this risk over a given coffee growing aera? 3. You have been asked to check if a given wheat variety, which requires about 450 mm of seasonal total rainfall, can be grown in a given woreda. Which maproom, which “Seasonal daily statistics”, and which “Yearly seasonal Statistics” would you use to check this? 4. The above wheat variety cannot tolerate more than two consecutive dry spells in the growing season. Which maproom, which “Seasonal daily statistics”, and which “Yearly seasonal Statistics” would you use to check this? 5. The EMI has announced that the coming season would be under neutral ENSO condition. How would you check the potential impact of this condition over your woreda? *Answers can be found in Appendix A 63 Ethiopia MODULE 2 SECTION 2.3 Specialized Climate Analyses Available for Agriculture: The Climate and Agriculture Maproom There are a number of benefits of climate information to agriculture and food security, including developing sustainable and economically viable agricultural systems, improving production and quality, reducing losses and risks, reducing costs, increasing efficiency in the use of water, labor, and energy, and conserving natural resources. The Climate and Agriculture Maproom offers users an array of climate information products that can aid different agricultural decision making. What is The Climate and Agriculture Maproom Used For? The Climate and Agriculture Maproom enables users to analyze different variables of historical climate information to help them make critical decisions in their agriculture planning and practice. These decisions may include: when to plant, when to apply fertilizer or pesticides, and selecting other suitable practices to make a crop grow under different climate conditions. In addition to this agriculture-specific maproom, the climate maprooms presented in the previous chapter can also provide valuable information for agriculture-related applications. Inside the Climate and Agriculture Maproom The Climate and Agriculture Maproom has sub-maprooms (products) that include daily precipitation analysis, historical onset and cessation dates as well as length of the season. Onset date is the start of the rainy season as identified by agronomic criterion. These often include passing a certain threshold in cumulative rain over a few days and not being followed by a long break in rains (dry spells) that could damage germinating crops. The cessation date is the date when the season is assumed to end and is based on the available soil moisture. Having the onset and cessation dates also allows us to estimate the length of that specific season (by calculating the difference between the cessation and the onset dates). Extracting and presenting information at any administrative level enables focus on a specific area of interest. Figure 2.3-1 is the Climate and Agriculture Maproom for Ethiopia. Figure 2.3-1: ProbabilOverview of the climate and agriculture maproom Daily Precipitation Analysis This maproom, which was also described earlier under the Climate Maproom, explores historical daily precipitation using some seasonal statistics. Many options can be specified to produce yearly time series of a chosen seasonal diagnostic of the daily precipitation. The user can then choose to map the mean, standard deviation or probability of exceeding a chosen threshold (Figure 31). Clicking on a map will then produce a local time series (Figure 32) of the chosen diagnostic. Among other analyses, this maproom may be used to determine: the increase/ decrease of rainfall in a certain area over time; how dry/wet spells (consecutive x number of dry/wet days) compare; which areas might have flooded or water- logged due to high rainfall intensity; or which areas have experienced drought. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 64 Figure 2.3-2: Probability of having 4 or more dry spells during Jul 1 to Sep 30 over Ethiopia. 65 Ethiopia MODULE 2 Figure 2.3-3: Time series of number of wet days (top) and probability of exceeding a given number of wet days (bottom) during Jul 1 to Sep 30 seasons for selected location in Ethiopia. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 66 Historical Onset and Cessation Dates This maproom enables users to explore historical rainy season onset and cessation dates based on user-defined criteria. The date when the rainy season starts is critical for deciding on the optimum planting date. By enabling the exploration of the history of onset dates, the Maproom also allows users to understand the spatial and temporal variability of onset dates and therefore characterize the risk for a successful agricultural campaign associated with it. The definition of the onset used here is a significantly wet event (e.g., 20mm in 3 days) that is not followed by a dry spell (e.g., a 7-day dry spell within the following 21 days). The actual date is the first day of the wet event. The onset date is computed instantly for each year according to the definition, and is expressed in days since an early start date (e.g., June 1). The search for the onset date is made from that early start date and for a certain number of following days (e.g., 60 days). The default definition of the cessation date is the first date after a given date (e.g., September 1) in 90 days when the soil water balance falls below 5mm for a period of 3 days. Then the maproom shows yearly statistics of the onset or cessation date, including the mean (by default) and the standard deviation or probability of exceeding a chosen starting or end date. Clicking on the map (Figure 2.3-4) will then produce a local yearly time series (Figure 2.3-5) of onset and cessation dates, as well as a table with the actual dates (as opposed to days since early start); and a probability of exceedance graph. Note that if the criteria to define the onset or cessation date are not met within the search period, the analysis will return a missing value. And if the analysis returns 0 (days since the early start), it is likely that the early start date is picked within the rainy season. Figure 2.3-4: Mean onset dates for the different parts of the country counted from the earliest start date (Jun 1). Thus, 10 represents the onset date of June 10th, while 60 represents onset date of July 30th. White areas means that either the criteria is not met or this is not a rainy season for that area. 67 Ethiopia MODULE 2 Figure 2.3-5: Example time series of onset dates for a selected location. Seasonal Totals This Maproom explores historical rainy season length and total rainfall amount based on user-defined definition of onset and cessation dates (seen in point B above). Clicking on the map (see map example in Figure 2.3-6) will then produce a local yearly time series (Figure 2.3-7) of rainy season length values. Figure 2.3-6: Average length of the rainy season (in days) over Ethiopia for areas receiving rains between June and September. Season length is of course zero over area that do not receive rainfall during this season, while areas over the western part of the country could have season length of over 150 days (about five months) Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 68 Figure 2.3-7: Example time series of seasonal length for a selected location in Ethiopia. In some case there could be break points showing when either the onset or the cessation criteria are not met. Self-Check: Climate and Agriculture Maproom 1. How do you find the probability of having onset before or after a give date (say July 1st)? 2. What are the white areas on an onset or cessation map? 3. What is the difference between seasonal total rainfall in the “Seasonal Analysis” maproom and the seasonal total in the Agriculture maproom? *Answers can be found in Appendix A 69 Ethiopia MODULE 2 SECTION 2.4 The Climate Monitoring Maproom Tab The Climate Monitoring Maproom tab under the Climate Maproom enables monitoring of the current season at dekadal, monthly and seasonal time scales. Different maps and graphs are used to compare the current season with the climatology or with recent years. What the Climate Monitoring Maproom Tab is Used for? The Climate Monitoring Maproom Tab provides routinely updated dekadal (10- day), monthly and seasonal rainfall for the whole country along with analysis tools. It enables monitoring of the current season, allowing for the comparison of the current season to either the climatology or recent years. Users can monitor the season for any specific location by selecting that region (either point, district, provinces, or any administrative boundary) on the map and choosing the type of analysis desired Inside the Climate Monitoring Maproom Tab The Climate Monitoring Maproom Tab has sub-components (products) that allow users to analyze the data at different temporal and spatial (point, box, or administrative boundaries) scales. Available analyses include dekadal, monthly and seasonal anomalies as well as analyses of extreme rainfall events. Figure 2.4-1 shows the different sub-maprooms in the Climate Monitoring Maproom Tab for Ethiopia. Figure 2.4-1 The Climate Monitoring Maproom showing the available analyses options. This analysis allows users to view different displays of the information about the most recent dekad, month, or season. The default map shows rainfall totals for the most recently available dekad (Figure 2.4-2), but totals for previous dekads can also be displayed. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 70 Figure 2.4-2: Total Rainfall amount for 2nd Dekad (11-20) of March 2021. The Climate Monitoring Maproom can also be used to display maps of rainfall anomalies, cumulative anomalies (over a user-defined time period) and the likelihood of extreme events presented as Standardized Precipitation Index (SPI) for the current year. This is done by selecting from the dropdown menu under “Analysis”. Anomalies show the difference between the most recent dekadal, month, or season rainfall and the expected (long-term average) expressed in mm or as a percentage. These maps offer an array of options to assess the performance of the current season and take action accordingly. Clicking a location on the map will generate four different time series graphs (Figure 2.4-3), representing the following: a) Dekadal rainfall totals for the selected region over the last 3 years. This can be used to visually compare the performance of the current year with the previous years. b) Dekadal rainfall anomalies for the selected region over the last 3 years, which can be used to assess the performance of the current season relative to the climatology for that location. c) Dekadal rainfall for the current year (thick black line) compared to the previous three years. d) Comparison of the cumulative dekadal rainfall (solid blue line) to the cumulative long-term average rainfall (solid black line) from the start of the year until the latest dekad. The grey plume indicates the range of the 5th and 95th percentile cumulative historical rainfall percentiles. The 5th percentile represents the dry extreme while the 95th percentile represents the wet extreme. Thus, this graph can be used to estimate how extreme the current season is compared to what is normally expected (climatology) for that location. 71 Ethiopia MODULE 2 Figure 2.4-3: Climate monitoring graphs (see explanation in the text) Self-Check: Climate and Agriculture Maproom 1. Which product would you use to check the severity of a dry dekad or month? 2. Which one more useful between “Cumulative anomalies” and “Cumulative anomalies in percentage” and why? From the cumulative graph, where would the blue line be relative the dark line for a very 3. wet year? Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 72 SECTION 2.5 The Climate Forecast Maproom Tab The Climate Forecast Maproom presents seasonal forecasts in a user-friendly format. Here it should be noted that this Maproom does not actually generate forecasts; it just presents forecasts generated by the NMA in a more decision- relevant format. Figure 2.5-1: Frontpage of the Climate Forecast Maproom Tab for Ethiopia. Flexible Presentation of Seasonal Rainfall Forecast Maproom This maproom has transformed how seasonal rainfall forecasts are presented. It deviates from the usual tercile format and presents seasonal forecasts in a flexible format that can be easily understood and used. Instead of the usual terciles (below normal, normal, above normal) presentation, the new forecast Maproom allows users to choose a threshold they are interested in either as percentiles or rainfall amounts. For instance, one can explore the probability that the total rainfall for the coming season is would be above or below 1200mm, or 75th percentile of the historical distribution. The forecast maps are also provided at higher spatial resolutions, making the forecast locally more relevant. The default map shows the most likely seasonal rainfall total amount for the forecasted season. The forecast can be expressed in different ways: • Rainfall: most likely seasonal total rainfall • Anomaly: deviation in mm of the most likely seasonal total rainfall from yearly average of the most likely seasonal total rainfall • Percent of Median: deviation in percentages of the most likely seasonal total rainfall from yearly median of the most likely seasonal total rainfall • Probability of exceeding or not exceeding a user-defined threshold: forecast probability of seasonal total rainfall to be below or above the threshold (Figure 2.5-2) 73 Ethiopia MODULE 2 Figure 2.5-2: Probability of the Kiremt (Jun-Sep) 2020 seasonal rainfall total forecasted in May 2020 exceeding 600 mm. The probability in the color bar is given as fraction, but can be converted into percentage by multiplying by 100. Clicking on the map will display specific forecast information for the selected location. This includes the probability of exceedance as well as the full forecast distribution (Figure 2.5-3) at that given location, compared with the historical distribution Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 74 Figure 2.5-3: Top: Probability of Exceedance comparing the forecast (red) with the expected (climatological in blue). The black line is actual data while the blue line is smoothed versions of the data(observation). Comparing the forecast and expected lines enables the user to assess whether the predicted seasonal rainfall will be above or below the expected for each rainfall threshold. For instance, the climatological probability of exceeding 300 mm is about 70%, while the forecasted probability of exceeding 300 mm is about is over 90%. Thus, rainfall for this location during Kiremt 2020 was expected to be above normal Bottom: The graphics compares the full probability distribution of the forecasted(red) and climatological (black line) for the selected location. The shift of the forecast probability to right shows that the season was expected to be above normal. Self-Check: Climate and Agriculture Maproom 1. What is the difference between the tercile presentation of seasonal forecast and the presentation in the Flexible Forecast Maproom? 2. Using the Probability exceedance graph, how would you tell if the forecast is above or below normal ? *Answers can be found in Appendix A 75 Ethiopia MODULE 2 Climate-Sensitive Agriculture Decisions MODULE 3 ABOUT THIS MODULE Module 3: Climate-Sensitive Agricultural Decisions aims to strengthen understanding of the interaction between climate and farm decision-making. Understanding what climate-sensitive decisions farmers make and the factors that influence their decisions is essential for understanding how climate services can support farmer decision making. In this module, you will learn how climate variability interacts with farmers’ environments, goals and constraints to influence their management decisions. You will gain awareness of some of the tools and methods used to characterize and analyze climate-sensitive farm management decisions. You will also about options that are available to farmers to manage climate-related risk, including production technologies, diversification, seasonal climate forecasts, and index-based agricultural insurance. By the end of this module, you will be able to: • Understand factors that lead to different management decisions by different farmers and in different years. • Perform basic analyses of climate-sensitive farm management decisions. • Understand basics of index-based agricultural insurance. SECTION 3.1 Section 3.1. How Climate Risk Impacts Farmers Climate is among the most important driver of agricultural risk, particularly in rainfed sub-humid to arid environments. Climate-related risk complicates farm decision making, and contributes to poverty and food insecurity. This section covers mechanisms by which climate-related risk impacts agricultural productivity and profitability, and the wellbeing of rural households who depend on agriculture. Risk concepts Farmers operate in an environment that is affected by stochastic variability, meaning that they are affected by randomness. Sources of stochastic variability, such as climate and market prices, are important drivers of agricultural risk. Risk refers to impacts of stochastic variability on productivity, profitability or some aspect of wellbeing. The term risk is used in two contrasting ways. The fields of agricultural economics and decision analysis view risk as the full range of variability of impacts, and considers the impact of the resulting uncertainty about future conditions. This perspective uses dispersion statistics (e.g., standard deviation, coefficient of variation) to quantify risk. This course generally uses this definition of risk. A hazard perspective, used widely in disaster risk management, treats risk as the likelihood that a disaster or other damaging event will occur. This perspective uses discrete probabilities of experiencing adverse events to quantify risk. Several concepts are relevant to risk and its impacts on agriculture. In the context of climate, hazard, shock and extreme event refer to a climate-related physical phenomenon that has the potential to trigger adverse impacts on human wellbeing – directly or through the natural resource base. A stress refers to a condition that can lead to adverse impacts, but that occurs over a longer period than a shock. Compound risk refers to instances when more than multiple types of simultaneous hazards interact. Exposure refers to the frequency and severity of environmental shocks or stresses that a community experiences. Vulnerability refers to characteristics of human communities or social systems that cause them to be susceptible to adverse impacts, such as increased food insecurity, poverty or mortality, when exposed to an external shock or stress. This perspective treats the likelihood of experiencing adverse impacts as a function of internal characteristics that determine vulnerability, and exposure to external shocks or hazard events. Note that some disciplines and communities use the term vulnerability to refer to the combined effects of internal characteristics and exposure on the likelihood of adverse impacts. In the context of development, resilience refers to the capacity of an individual, community or society to resist moving to a less desirable state (e.g., poverty), or to recover if it is already in an undesirable state (relative to a development goal), in the face of stochastic shocks or long-term stresses. From a development perspective, vulnerability and resilience can be viewed loosely as opposite concepts, although resilience considers wellbeing over long time scales while vulnerability may focus also on immediate impacts of shocks. The concept of resilience is used differently by different disciplines and communities, and has evolved considerably from its initial focus on ecological systems (Barrett et al., 2021) Coping refers to the actions that individuals or communities take to withstand a shock and limit its immediate adverse impacts. Although the coping strategies of rural households reduce the immediate impacts, some coping strategies reduce livelihood opportunities or increase vulnerability in the long term. Adaptation involves adjustment in natural or human systems in response to actual or expected climatic stimuli or their effects, which moderates harm or exploits beneficial opportunities. Although adaptation is used in several ways, it typically refers to changes over longer time periods than coping. Biological productivity impacts Climate conditions impact crop and livestock production systems through a range of mechanisms (Table 3.1-1). Solar radiation drives conversion of atmospheric CO2 into plant biomass through photosynthesis. Temperature controls the rate of plant phenological development, and most plant physiological processes. Precipitation determines the water that is available to plants from the soil, and influences availability and movement of soil nutrients. Potential evapotranspiration drives plants’ uptake of water and nutrients from the soil. Climate impacts on livestock production tend to be more indirect, since animals do not directly use resources, such as solar radiation, CO2 or precipitation from the atmosphere. Episodes of severe drought leads to widespread livestock mortality in pastoralism and other extensive livestock production systems in semi-arid to arid lands, by reducing rangeland, forage and fodder plant productivity. Heat stress, from a combination of high temperatures Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 78 and low potential evaporation, suppresses meat and milk production. Hydrometeorological hazards, such as flooding, wind storms and lightening, can kill unprotected animals. Climate, particularly temperature, humidity and wind, has an important but indirect impact on crops and livestock by influencing the population dynamics and movement of insect pests and microbial pathogens. Table 3.1-1. Examples of weather and climate impacts on biological productivity. Parameter Impacts Crops Solar radiation CO2 conversion into plant biomass through photosynthesis Temperature • Rate of phenological development • Most plant physiological processes Precipitation • Water available from the soil • Movement and availability of soil nutrients Potential evapotranspiration Drives uptake of water and nutrients from the soil Temperature, humidity, wind Population dynamics and movement of pests, pathogens Livestock Drought • Reduced food availability from rangeland, forage, fodder • Widespread mortality from severe or prolonged drought High temperature and humidity Reduced weight gain and milk production Hydrometeorological hazards Unprotected animal mortality (flood, wind storms, lightening) Temperature, humidity, wind Population dynamics and movement of insect pests, vector-borne diseases Variations in the amount, timing and distribution of precipitation is often the dominant source of year-to-year variability of production, income and consumption in dryland (i.e., rainfed production in sub-humid to arid environments) agricultural systems in the topics, and in flood prone locations. Temperature variability can also be an important driver of production variability in temperate climates, and in tropical environments where temperatures fluctuate near upper or lower thresholds that crops or animals can tolerate. Biological response to climate conditions is generally nonlinear and sometimes nonmonotonic (i.e., increasing and decreasing over different parts of the range of variability), and climate variability interacts with this nonlinearity to reduce average productivity. This interaction can be illustrated graphically with a nonlinear yield response curve to growing season rainfall (Figure 3.1-1). Although both panels have the same mean rainfall, the higher rainfall variability in the right panel results in lower mean yield. Note that the decreased mean productivity does not capture the damaging impact of a year with very low production, which can impact household wellbeing in subsequent years. Figure 3.1-1. Nonlinear grain yield response curve to growing season rainfall, illustrating reduced mean yield in response to increased rainfall variability. 79 Ethiopia MODULE 3 Impacts on the welfare of rural households Climate-related disasters impact poor countries, and the relatively poor within countries, disproportionately. Shocks associated with extreme climate events trigger cascading productivity and market responses that lead to acute food insecurity. The uncertainty associated with climate variability suppresses agricultural production and livelihoods. Climate impacts on food accessibility propagate through the economy. The adverse impacts of climate risk on food security and its precursors can persist long after a period of climatic stress – a phenomenon known is a “poverty trap.” The discussion of each of these mechanisms below draws from reviews by Hansen et al. (2022, 2019), and is summarized in Figure 3.1-2. Climate risk Extreme Uncertainty events Ex-ante risk “Moving Ex-post aversion target” coping effect strategies Reduced Reduced Depleted investment average productive adoption production assets income “Poverty tray” Figure 3.1-2. Main pathways by which climate risk contributes to long-term rural poverty. Adapted from Hansen et al. (2019). Near-term impacts of climate shocks Extreme weather and climate events, such as droughts, flooding, or high or low temperature extremes, can lead to reduced food consumption, and resulting health impacts expressed as wasting (i.e., low weight-for-height), stunting (i.e., low height-for-age), underweight (i.e., low weight-for-age) and mid-upper arm circumference (MUAC). While a shock might impact an entire community, the severity of food security and health impacts differs among individuals due to differences in characteristics such as gender, age, asset ownership and social capital. There are important differences between the sequence of impacts of slow onset (e.g., drought) and rapid onset shocks (e.g., flooding, storms) (Hill et al., 2019) (Figure 3.1-3). A slow onset shock typically triggers a cascade of impacts discussed below: reduced crop production, increased staple food prices, reduced terms of trade for labor and capital goods, declining incomes, and reduced food consumption. a: Slow onset shock (e.g, drought): Drought Food crop Staple food Terms of Income production price trade Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 80 b: Rapid onset shock (e.g flood or storm): Food crop Staple food Terms of Income production price trade Flood, storm Infrastructure Food damage Water Food contamination utilization Figure 3.1-3. Typical cascade of impacts from a (a) slow-onset and a (b) rapid-onset climate shock. The most direct impact of a climate shock is often reduced crop production. For smallholder farm households, a failed harvest directly reduces the availability of food from subsistence production, and often leads households to ration food consumption earlier in the hunger season, prior to the next year’s harvest. A climate-driven reduction in availability of a staple crop can increase its price through market equilibrium. A change in production of an agricultural commodity tends to lead to an opposite change in its price. This can be a particular challenge for horticultural crops that have short shelf lives and high transportation costs. For many agricultural commodities, trade and storage reduce price response to variations in productivity. Rural households who are net buyers of food face the combined impact of reduced availability through subsistence production and reduced accessibility through higher food prices, whereas increased prices received can partially compensate for the impact of a negative productivity shock on the incomes of farmers who are net sellers. As farm households deplete their food stocks and savings, they increasingly turn to off-farm casual employment to meet the shortfall, which can flood a local labor market. The value of durable assets decreases as affected households seek to exchange assets for food, through distress sales or barter, at the same time demand for these assets is decreasing in response to falling incomes and rising food costs. If a drought reduces staple crop production or reduces grazing resources, it could lead livestock keepers to sell animals, which could result in over-supply and drop in market price. The impacts of rapid onset shocks can be more complex and more context-specific. In addition to damaging crops, immediate impacts of storms or floods can include loss of livestock and other capital assets, diseases triggered by contaminating drinking water, and infrastructure damage that leads to loss of access to markets and vital services. Climatic uncertainty impacts farmer decision making While the impacts of extreme climate conditions (e.g., droughts) on rural communities are more visible, the uncertainty due to climate variability also contributes to agricultural productivity and livelihoods by reducing the efficiency of input use, and acting as a disincentive to adopting improved agricultural practices and investing in agriculture. Uncertainty is a fundamental characteristic of the climate, and a key challenge for climate-sensitive management decisions. If farmers knew what daily weather and seasonal climate conditions would be like in the future, they would be able to select the best management options available for those future conditions. However, farmers must make decisions before they experience those conditions. The uncertainty associated with climate variability creates a “moving target” for management that reduces efficiency of land and production inputs and hence profitability, because management (e.g., cultivar choice, fertilizer rates, planting density) that is optimal for average climatic conditions can be far from optimal for weather conditions in most growing seasons. Furthermore, most farmers are risk averse, meaning that they prefer options that reduce risk. In the face of climate variability, risk-averse farmers do not optimize their management for average conditions, but for adverse conditions. They employ precautionary strategies such as: selecting less risky but less profitable crops and cultivars, generally avoiding investing their resources in production assets and technologies, under-use of fertilizers, using livestock for precautionary savings rather than income, and shifting household labor to less profitable off-farm activities. Risk aversion and its effect on climate-sensitive decisions are discussed in more detail in Section 3.2. 81 Ethiopia MODULE 3 Impacts persist through coping responses and health consequences The food security impacts of a climate shock often persist long after climate conditions return to normal. This is due both to long-term consequences of early childhood health impacts, and to household coping strategies that deplete productive assets. A severe or prolonged crisis that leads to malnutrition in utero or during the critical first 1000 days of life can adversely impact the individual’s physical and mental health, educational achievement, and income into adulthood. For example, early childhood stunting reduced income later in life by 5-7%, averaged across 34 developing countries that account for 90% of the world’s stunted children (Galasso and Wagstaff, 2018). A study of 106,330 women in 19 sub- Saharan African countries showed that drought experienced during early childhood reduced educational attainment and wealth as adults, adversely affected empowerment, and increased the likelihood that their children would have low birth weight – for rural but not for urban populations (Hyland and Russ, 2019). When a severe climate shock, such as a drought, flood or heat wave, reduces the availability and accessibility of food, vulnerable households typically employ a sequence of coping strategies to endure the immediate crisis. The hardship from initial coping strategies, such as rationing meals, typically lasts only until the next harvest. If the stress is severe and persists long enough, they may implement increasingly drastic coping strategies that reduce their future capacity to provide for their sustenance and livelihood, for example: defaulting on loans, liquidating productive assets, withdrawing children from school and over-exploiting natural resources. When a household or community is unable to accumulate the necessary resources to escape poverty, the condition is often described as a “poverty trap” (Barrett, 2005). In rural settings, a poverty trap is characterized by subsistence food production, poor adoption of improved agricultural production practices, and chronic poverty and food insecurity. Climate risk contributes to poverty traps through the mechanisms described in this section and summarized in Figure 3.1-2. SECTION 3.2 Understanding Climate-Sensitive Agricultural Decisions A management decision is climate sensitive if the decision maker would select different options if they observed or anticipated different weather or climate conditions. For example: • In seasons with good rainfall, applying nitrogen fertilizer can increase cereal yields and farm income. However, the same fertilizer application may have little impact on yields, and the fertilizer cost may reduce income, in a season when rainfall is inadequate. • An unexpected rainfall event at harvest maturity can damage the standing crop, but early harvest and post- harvest management may avoid that damage. Understanding what climate-sensitive decisions farmers make, and the factors that influence their decisions, is essential for understanding how climate services can support farmer decision making. Farmers, and the environments in which they operate, can be quite variable. Management practices developed through agricultural research are more likely to be adopted and to benefit farmers if they take these differences into account, and if agricultural extension supports farmers to adapt those practices to their farms, goals, constraints and expected climatic conditions. By the end of this section, you will be able to: • Explain how differences in household goals and resource constraints can lead different farmers to select different management options. • Identify the time scales of important agricultural decisions, and climate information that would support those decisions. • Use decision trees to describe simple decision problems under uncertainty. Farmer characteristics and decision making Farm management recommendations are sometimes based on simple criteria such as maximizing the yield of a particular crop. However, adoption rates of recommended practices are often low, particularly for smallholder farmers. This is because smallholder farmers consider several different goals beyond maximizing productivity, and face a range of resource constraints. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 82 Goals that can influence farm management decisions include: • sustenance (household food production); • income; • accumulation of land, livestock and other assets; • avoidance of drudgery; • social status; and • providing better livelihoods for the next generation. The relative importance of these and other goals can vary considerably across cultures and from farmer to farmer, and influenced by a range of factors including local culture, religion and education. Farm decisions are also influenced by a range of resource constraints, including: • land available for cultivation; • availability and timing of household labor; • capital (i.e., savings, credit, equipment); and • social capital (i.e., ability to access a range of help through social and institutional contacts). These constraints limit farmers’ ability to achieve their goals. Smallholder farmers; and disadvantaged social groups such as women, relatively poor farmers, and those lacking education or literacy within smallholder communities; tend to be impacted more by resource constraints than larger-scale commercial farmers. When we deal with climate-related risk and the role of information in managing risk, one of the most important factors to consider is the farmer’s attitude towards risk. Most individuals dislike risk. Although commercial and smallholder farmers might have similar attitudes toward risk, commercial farmers are often better able to manage climate-related risks through financial instruments such as savings, credit, and insurance. As discussed earlier (Section 3.1), climate variability complicates farm decision making in two main ways. First, climate variability creates a moving target for management that reduces efficiency of input use and hence profitability. Second, climate variability and risk aversion lead farmers to select less risky but less profitable crops, use less than the optimal amount of fertilizer, shift household labor to less profitable off-farm activities, and avoid investing in productive assets and improved technologies. Time and climate-sensitive agricultural decisions As farmers are aware, timing is very important for agricultural management. We consider three aspects of timing: seasonality, time horizon and lead time. Seasonality refers to the times of the year when agricultural activities occur and key reoccurring management decisions are made. For example, rainfed annual crop farmers make important decisions about land allocation, and choice of crops and production inputs just before the start of the rainfall season. The annual climate cycle has a large influence on the seasonality of agricultural activities and management decisions. Government financial and policy processes may also influence the annual cycle of agricultural extension activities. Decision time horizon refers to how far into the future a decision maker considers when evaluating the consequences of a decision option. For example, the choice of what annual crop or cultivar to plant typically has a time horizon of 3-4 months, because the productivity and profitability of that decision can be affected by conditions from the 3-4 months between planting and harvest. Once a crop is growing, decisions about field operations such as pest management, supplemental irrigation or when to harvest have much shorter time horizons. On the other hand, the decision to plant or remove perennial crops such as coffee can have a planning horizon on the order of 5-10 years. Lead time refers to the time that is needed between making a decision and implementing the decision. Some farm decisions require a period to prepare, for example to secure credit, purchase inputs, contract labor or draught animals, before the decision can be implemented. These three aspects of timing are important for climate services. For climate information to be useful for agricultural decision making, the time scale of the information should match the time horizon plus the lead time of the decision. Furthermore, climate services must consider seasonality, as appropriate climate information and support should be available at the times of the year when farmers make key climate-sensitive decisions. The lead time and time horizon of a decision determine the time scale of weather or climate information that might support the decision. For a typical annual rainfed crop, Table 3.2-1 gives a few examples of when decisions must be made relative to the growing season, and types and time scales of weather and climate information that may be relevant to those decisions. 83 Ethiopia MODULE 3 Table 3.2-1. Examples of climate-sensitive decisions for a typical annual rainfed crop. Timing Decision Relevant Info Infrequent, before major capital Choice of farming and cropping Historical rainfall and temperature investments system climatology Farm land allocation Seasonal rainfall forecast Before growing season Crop cultivar selection Seasonal rainfall forecast, onset date forecast Land preparation, basal fertilizer Seasonal rainfall forecast, onset date application forecast Time of sowing Onset date forecast During growing season Fertilizer side dressing Seasonal rainfall forecast updates, sub- seasonal rainfall forecast Irrigation Management Weather monitoring and forecast, soil water balance Pest and disease management Weather monitoring and forecast (rainfall, wind, humidity) Frequency and intensity of Seasonal rainfall forecast updates, sub- weeding seasonal rainfall forecast Time of harvest Weather forecast (rainfall, wind) End of growing season Post-harvest handling Weather forecast, rainfall humidity An agricultural calendar is a planning tool that summarizes the timing of agricultural activities for a particular location. In the case of annual crop farming, an agricultural calendar identifies normal time windows for sowing, crop growth and harvesting. It could be expanded to also indicate times for phenological stages such as: vegetative growth, flowering, grain fill, and physiological maturity; and farm activities such as: land preparation, fertilizer application, and weeding. Because crop growth stages and key agricultural activities are sensitive to weather conditions, it is useful to overlay the calendar with graphs of the seasonal climatology of rainfall and temperature. Figure 3.2-1 shows an example calendar for maize at one location (Melkasa), overlaid with seasonal climatology graphs for precipitation and temperature. Figure 3.2-1. Example of an agricultural calendar for Melkasa, Ethiopia, overlaid with seasonal climatology graphs for precipitation and temperature. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 84 Describing risky decision problems with decision trees The outcome of a climate-sensitive agricultural management decision is a function of the option that a farmer chooses, and climate and other external factors that are outside of the farmer’s control. Climate-sensitive management decisions typically require a farmer to choose among a set of options while climate conditions and other states of nature that affect the outcome are still uncertain. A decision tree provides a useful way to describe management decisions under uncertainty, when the decision has a small set of options, and when the risks that influence the decision outcomes can be described by a small set of states of nature. A decision tree captures these three aspects of a decision problem (Figure 3.2-2). First, the farmer selects one of the decision options. A square represents a decision node. Each option available for the decision is represented by a line coming from the right of the decision node. Second, conditions, or “states of nature,” that are uncertain and beyond the farmer’s control may affect the outcome of the decision. For example, the range of possible seasonal total rainfall can be expressed in terms of three equally probable states of nature: the “below-normal,” “near-normal” and “above- normal” terciles of the historical distribution A circle represents a chance node, which is followed by two or more lines representing uncertain states of nature and their probabilities. Third, the farmer either experiences the outcome of the decision, or faces another decision that may affect the final outcome. An end node, shown as a triangle, represents the outcome of a given sequence of decision options and states of nature. For a crop farmer, the outcome might refer to the yield of the harvested crop, expressed as kg/ha; or the income from selling the crop, expressed as birr/ha. Figure 3.2-2. Components of a decision tree. A decision tree provides a useful way to describe management decisions under uncertainty, when the decision has a small set of options, and when the risks that influence the decision outcomes can be described by a small set of states of nature. The branches that follow each decision node and each chance node give this type of diagram its name. The decision tree represents the flow of time: from decision, to uncertain climate conditions, to the final outcome. Recall that the time between a decision and its outcome is called the decision time horizon. Decision trees can also represent sequences of decisions, by adding decision nodes after each state of nature following a chance node. The example below (Figure 3.2-2) represents a sequence of two simple decisions, each with only two options, followed by only two uncertain states of nature. As this example suggests, realistic sequences of farm decisions can easily lead to too many nodes and branches to represent in a diagram. If the consequence (e.g., birr of income or kg/ha of yield) of each end node is known, and if the probability of each state of nature is known, the expected value of each decision option can be calculated by multiplying the values of each end node by the probability of its state of nature, and then adding the result across all states of nature. 85 Ethiopia MODULE 3 Figure 3.2-2. A decision tree that represents a sequence of two decisions. A Decision Tree Example As an example, suppose a farmer faced the choice of whether to purchase crop insurance. Assume that income from crop production would be $150 if seasonal rainfall were to fall in the below-normal tercile, $500 if seasonal rainfall were near normal, and $650 if seasonal rainfall were above normal. The insurance premium costs $75, and pays out $200 if seasonal rainfall falls in the below-normal tercile. The premium reduces income by $75 for any state of nature, and increases income by $200 if rainfall is below normal. The probability of rainfall falling in any given tercile is 1/3, or approximately 0.333. This can be represented by the decision tree below (Figure 3.2- 3). Figure 3.2-3. Decision tree representation of an insurance decision. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 86 Expected income can be calculated from the information in the decision tree: • With no insurance, expected return is: 0.333 × $150 + 0.333 × $500 + 0.333 × $650 = $433 • With insurance, expected return is: 0.333 × $275 + 0.333 × $425 + 0.333 × $575 = $425 For a farmer who wants to select the management option with the highest expected return, the decision tree can be simplified by replacing the chance nodes for each option with end nodes showing the expected value (Figure 3.2-4). In this case, the decision to purchase insurance is less profitable, on average. Figure 3.2-4. Decision tree representation of an insurance decision, with uncertain consequences replaced with expected returns. Forecasts can change the probabilities that different states of nature will occur. In the insurance example, suppose there is a forecast of 25% probability of above normal, 30% probability of near normal, and 45% probability of below normal seasonal rainfall (Figure 3.2-5). In this example: • With no insurance, expected return is: 0.45 × $150 + 0.30 × $500 + 0.20 × $650 = $380. • With insurance, expected return is: 0.45 × $275 + 0.30 × $425 + 0.20 × $575 = $395. The expected return from both decision options is reduced, and the expected return is higher with insurance than without, because of the forecast. Figure 3.2-5 Decision tree representation of an insurance decision, with probabilities modified by a seasonal forecast. 87 Ethiopia MODULE 3 Limitations of decision trees Although decision trees can be a helpful way to describe decisions under uncertainty, they have several important limitations when they are applied to real decision problems. First, decision trees can only describe decisions with a small set of distinct options. Decisions such as what crop or cultivar to plant can be described with decision trees. However, decisions such as how much fertilizer to use or what percent of available land should be planted with a particular crop have an infinite number of possible options. Second, decision trees can only describe risk with a small set of distinct states of nature. Climate variables such as rainfall amount or average temperature during a particular period can take on an infinite number of possible values. Although it is sometimes convenient to use terms like above-normal, near-normal and below-normal – as national meteorological services often do when presenting seasonal forecasts – this may over-simplify aspects of climate variability that are important for a management decision. Third, decision trees easily become too complex to draw or analyze for realistic sequences of decisions, with their corresponding options, even if risk is represented by a small number of states of nature. SECTION 3.3 Decision Making Under Uncertainity This section will expand on the idea that individuals, including farmers, tend to be risk averse. It will present concepts associated with risk aversion, and discuss how to identify management options that could be preferred by a range of farmers who have differing degrees of risk aversion. By the end of this section, you will be able to: • Understand the role of risk and risk aversion in farm management decisions • Analyze production management options using risk analysis Risk aversion and related concepts Section 3.1 introduced the concept of risk aversion. While farmers generally prefer options that increase average production or income, risk aversion means that a farmer also prefers options that reduce risk expressed as the year-to-year variability of production or income. Risk management involves managing tradeoffs between the goal of maximizing expected production or income, and the goal of minimizing its variability. The relative importance of these two goals is a function of an individual’s degree of risk aversion. To explain risk aversion and closely related concepts, we will imagine a game in which you, as an agricultural professional, must choose between receiving an uncertain income, and receiving a lower but certain income. The “coin toss game” that follows is an unrealistic scenario, but a useful way to understand these concepts. Imagine that your employer offers you a choice: Instead of getting your annual salary, in the form of a monthly paycheck, you could play a game. Your employer will flip a coin and, if it lands on heads, you will get twice as much in each paycheck for the next year. If it lands on tails, you will not receive any pay for the next year. If you do not play, you are certain you will receive your annual income (y). If you do play, there is a 0.5 probability of receiving two times your normal annual income, or 2y, and a 0.5 probability of receiving an annual income of 0. Therefore, the expected value of playing the game is 0.5 × 2y + 0.5 × 0 = y. Figure 3.3-1 represents this decision problem with a decision tree. Figure 3.3-1. Decision tree representation of the “coin toss problem” (1 of 4). Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 88 Would you play the game? Why or why not? Most people would choose not to, because the negative consequence of having zero income for the year is much greater than the added benefit of having double their annual income. What if your employer would only give you 90% of your normal annual income if you decide not to play? In this case (Figure 3.3-2), the expected value of playing the game (y) is greater than the expected value of not playing (0.9 y). Would you play? Would you play instead of receiving only 80% of your annual income with certainty (Figure 3.3-3)? …or 70%, or 60% (Figures 3.3-2, 3.3-3)? Figure 3.3-2. Decision tree representation of the “coin toss problem” (2 of 4). Figure 3.3-3. Decision tree representation of the “coin toss problem” (3 of 4). At some point, you would be indifferent between playing the game, which gives you expected return equal to your normal annual income (y) with considerable risk, and obtaining a lower but certain return (Figure 3.3-4). Figure 3.3-4. Decision tree representation of the “coin toss problem” (4 of 4). 89 Ethiopia MODULE 3 This value is known as the certainty equivalent, or CE. The CE is a guaranteed return that someone would accept, rather than taking a chance on a higher but uncertain return in the future. The CE expresses the subjective value an individual places on an uncertain economic return, adjusted for their subjective cost of risk. If the CE is less than the expected value of a risky prospect, this indicates that the decision maker is risk averse. For a risk averse individual, uncertainty reduces the value of an option. On the other hand, if a decision maker is risk-neutral, meaning indifferent to risk, then for a given decision option their CE would be equal to the expected return of that option. According to decision theory, a decision maker who follows a particular set of assumptions will select the available management option with the highest CE. The difference between the expected return and CE is known as the risk premium, or RP, defined as the amount of average return that a decision maker would be willing to sacrifice or, equivalently, the maximum amount that the decision maker would be willing to pay, to eliminate risk. RP is a measure of the cost of the combined effects of risk and risk aversion. Increasing either risk aversion or the risk (i.e., the variability) of returns will increase RP. Limitations It would be possible to calculate the CE of each option, and hence identify the best management strategy if complete information were available about: • the individual’s level of risk aversion; • the probability distribution of the consequences of each management option; and • the farm household’s assets, income sources and expenses. This type of analysis is generally not used routinely for agricultural extension for three reasons. First, the math used for the analysis is complicated, and requires economic information about the entire farm household. Second, methods to measure individual risk preferences are complicated and time consuming. Third, management recommendations are made for many farmers who may have differing economic circumstances and levels of risk aversion. Maximizing CE is a useful principle for guiding risk-averse decision makers to select options that are most consistent with their actual risks and with their goals and risk preferences. However, there is considerable evidence that people often chose options that are inconsistent with their goals and risk preferences. Widespread challenges to making rational risk management decisions are related to: • incomplete knowledge about the probabilistic consequences of decision options; • inconsistent preferences, which involves changing preferences at different times, or depending on how a risky prospect is presented; and • cognitive errors such as loss aversion, which involves putting greater weight on a decision consequence framed as a loss than a consequence framed as the equivalent final wealth. Decision analysis when preferences are unknown It is generally not feasible base to identify agricultural management practices that maximize CE for a target set of farmers. However, simple risk efficiency analysis methods can be used to evaluate options for a group of farmers who may have different risk preferences, based on a few general assumptions. The simplest type of risk efficiency analysis uses graphs to show the tradeoff between expected return and risk. Expected return is expressed as the mean, and risk can be expressed as variance or standard deviation. It is sometimes referred to as E-V (expected return – variance) or E-S (expected return – standard deviation) analysis. E-S and E-V analysis assume that any decision maker prefers higher over lower expected return, and lower risk over higher risk. The E-S graph in Figure 3.3-6 shows the mean and standard deviation of returns for six management options. A decision maker who wishes to maximize expected return would choose option E. A decision maker who wishes to minimize risk would choose option A. A decision maker who is somewhat risk averse might select option C, and a decision who is more risk averse might select option B. No decision maker who follows these assumptions would select option F, because options are available with higher expected return and lower risk (options B, C, and D). What about option D? This option has a higher return than options A and B, and lower risk than options C and E. However, the decision maker could obtain a better combination of risk and expected return by choosing a combination of options B and C. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 90 Figure 3.3-6. Risks and expected returns for a set of six management options, represented in a mean - standard deviation (E-S) diagram. To find the set of options that decision makers who meet the assumptions might choose, draw straight lines to connect the decision options that lie above and to the right of other options, as in Figure 3.3-7. These options are known as the efficient set. In this example, the efficient set consists of options A, B, C and E. The options that are not part of the efficient set in our example, options D and F, would not be selected by any decision maker, and therefore should not be included in recommendations. Figure 3.3-7. E-S diagram showing the efficient set of management options. In addition to assuming that decision makers prefer higher over lower expected return, and lower risk over higher risk, E-V and E-S analysis requires two additional assumptions: • Decision makers’ expectations are consistent with measured mean return and its variability (variance or standard deviation). In other words, the decision maker knows what the mean and variability are for each option. • The shape of the probability distribution should approximate a normal distribution. Since real-world decisions often do not completely meet these assumptions, the results of E-V and E-S should be treated as an approximation. 91 Ethiopia MODULE 3 SECTION 3.4 Tools to analyze Climate-Sensitive Decisions This section introduces tools that are useful for translating information about climate, soils, genetic characteristics and management into estimates of crop productivity and profitability. By the end of this section, you will be able to: • Describe the role and appropriate use of crop simulation models to estimate productivity. • Use enterprise budget calculations to compare the profitability of crop and livestock management options. Analyzing climate-sensitive agricultural production decisions When enough information is available, decision analysis can be used to identify which management options are best adapted to farmers’ goals, attitudes towards risk, expected climate conditions and local environment. A rigorous analysis of agricultural production typically includes the following steps: 1. Understand goals and constraints that influence farmers’ decisions. 2. Describe the decision problem and management options. 3. Identify relevant climate uncertainties and how they influence production. 4. Estimate probability distributions of crop or livestock production for each management option, based for example on field experiments, crop simulation models, statistical models or expert knowledge. 5. Develop a realistic enterprise budget, and use it to estimate probability distributions of financial returns for each management option. 6. Using climate forecasts, if they are relevant, to modify probability distributions of production and returns for each management option. 7. Use risk analysis to compare management options and identify preferred options, based on based on: • CE, if risk preferences can be estimated; or • E-S or other risk efficiency analysis when preferences are unknown. Rigorous agricultural decision analysis requires access to good data about the production system, the environment, and farmers’ goals and constraints. As an agricultural extension worker, you may not perform these analyses routinely in your work with farmers. However, understanding decision analysis concepts and tools will help you better understand the challenges that farmers face as they make risky climate-sensitive decisions, and enable you to more effectively support them with climate information, training and technical guidance. Estimating productivity: Crop simulation models A model is a simplified representation of a system. Crop simulation models are computer programs that use mathematical functions to mimic the growth and development of a crop. The models are based on existing knowledge of crop physiology and the ecology of crop response to management practices and the environment. The early development of crop simulation models emphasized their ability to integrate knowledge of different processes, and to simulate interactions between genetics, environment and management, sometimes referred to as “G×E×M” interactions. As their ability to simulate these interactions advanced, crop models were increasingly seen as a tool to transfer results of traditional agronomic research results to locations other than where they were developed and tested, and to tailor production technologies to specific environments without having to replicate field research at every location. Their ability to simulate crop yield response to weather, soil and management makes crop simulation models a useful tool for understanding and supporting climate- sensitive crop management decisions, when they are used appropriately, with an understanding of their capabilities and limitations. Levels of analysis The suitability of a crop model for a given environment depends in part on whether it represents the key determinants of production in that environment. Rabbinge (1993) classified levels of crop production (more appropriately, “levels of analysis”) by the factors that limit production (Figure 3.3-2). The evolution of crop models paralleled these levels of production. As crop models incorporate additional processes, they tend to grow in complexity and data requirements. In the first level, potential production, crop production is limited only by crop genetic characteristics, solar irradiance, temperature, daylength and CO2. Early crop models incorporated basic physiological processes that determine potential production, including phenology, photosynthesis, respiration, and growth of the various plant parts. A second level of detail, attainable Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 92 production, accounts for soil water and nitrogen availability. Soil water, N and organic C dynamics and uptake were added to crop models to simulate attainable production. This level is sometimes divided between water- limited and N-limited production. Many widely-used crop models are able to describe crop growth and yield with these assumptions. The third level, actual production, may be limited by a range of additional constraints such as pests, diseases, and various soil nutrient limitations and toxicities. Although no model simulates the full range of determinants of actual production, some crop models have attempted to address some determinants at this level, including P dynamics; and response to pests, diseases and various soil physical and chemical. Figure 3.4-1. Levels of crop production. Model processes • Most widely-used crop models simulate how a community of plants, with specific genetic traits, respond to their environment and management over time. They typically simulate the following processes (Figure 3.4-2): • soil water and nitrogen balance, • phenology, or crop growth stage, • increase in biomass through photosynthesis, • decrease in biomass through respiration, • partitioning of crop biomass among parts of the plant (roots, shoots, leaves, grain or other harvested products), and • loss of biomass through senescence. Each day, or sometimes at shorter intervals, a typical crop model calculates rates of change in the state of the soil and plant in response to the environment, and calculates the new state of the soil and plant based on the rate of change. The process is repeated each day until the crop reaches physiological or harvest maturity. 93 Ethiopia MODULE 3 Initial inputs Management Water and N balance Daily weather and Phenology management inputs Photosynthesis Cultivar or genetic Daily outputs parameters Respiration No Partitioning Senescence Maturity ? Yes Final outputs Figure 3.4-2 Simplified flow diagram of processing control for a crop production model operating on a daily time step. Blue and brown arrows indicate key points for flow of information. Arrows with dashed lines indicate inputs for cultivar-specific or genetic parameters. Possible feedbacks (e.g., of partitioning on water use and photosynthesis) are not shown. Adapted from Plant, Cell & Environment 36(9): 1658-1672, DOI: 10.1111/ pce.12119 Most of these processes are influenced by daily weather data (Table 3.4-1). Rainfall drives the soil water and nitrogen balance. Potential evapotranspiration is a component of the soil water balance and influences photosynthesis and crop water stress. Potential evapotranspiration is modeled as a function of daily minimum and maximum temperature, solar radiation, and sometimes humidity and wind data. Phenology is controlled by temperature, and for photoperiod- sensitive crops, by daylength. Solar radiation drives photosynthesis. The main physiological processes (phenology, photosynthesis, respiration, partitioning, senescence) are influenced by temperature, and indirectly by rainfall through the soil water and nitrogen balance. Crop simulation models typically require daily precipitation, minimum and maximum temperature, and solar radiation data. Some also use wind and either relative humidity or dewpoint temperature to provide more accurate estimates of potential evapotranspiration. Table 3.4-1. Daily weather data that influence crop growth and development in crop simulation models (indicated by red shading). Process Rainfall Temperature Daylength Solar Radiation Potential evapo- transpiration Soil water and N balance Phenology Photosynthesis Respiration Partitioning By simulating the complex interactions between daily weather, soil water and nitrogen dynamic, crop phenology and crop physiology, crop simulation models generally capture crop response to seasonal climate conditions much better than statistical relationships (Figure 3.4-4). Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 94 Figure 3.4-3. Observed soybean yields (Georgia, USA, variety trials) vs. seasonal rainfall, temperature, and simulated yields. Source: James W. Jones, University of Florida. Appropriate use and limitations Crop simulation models have important limitations. They do not simulate many of the processes that can limit crop production, for example, soil micronutrients, soil toxicity, pests and diseases. As a result, they are likely to overestimate the yields that smallholder farmers experience on their fields. The accuracy of crop simulation results depends on the quality of the input data. Obtaining accurate input data is particularly challenging for soil properties, which can vary a great deal over short distances; and for initial soil moisture and nitrate content, which vary in space and in time. If crop simulation models have been tested in the environment in which they will be used, and if good input data and expertise is available, crop simulation models can contribute to extension recommendations in at least three ways. First, crop models can be used to screen potential crop management strategies for a particular environment, and identify a small set of promising candidates to test through field experiments or on-farm trials. Second, crop models can test how sensitive recommended crop management practices are across a greater range of range of soil and weather conditions than can be sampled through field experiments. This can indicate whether a particular cultivar or management practice can be recommended with confidence across a wide range of soils and climatic conditions, or if it should be adjusted for specific conditions. Finally, and perhaps most important, decision support tools that employ crop models provide a useful way for farmers and their advisors to learn principles of climate risk management. Such tools support learning through accelerated experience and experimentation with decision scenarios. When these tools are used in a group setting to experiment with management options under different climate scenarios or other environmental conditions, they sometimes challenge the existing beliefs of farmers, advisors and even researchers, leading to new ways to think about their risk management challenges. For this reason, these crop model-based tools are sometimes referred to as “discussion support tools.” Estimating profitability: Enterprise budget analysis The profitability of farming depends not only on the income the farmer receives from selling their products, but also on the cost of production. A farmer will be unlikely to adopt a technology that increases costs more than income. Enterprise budget analysis is a useful tool for estimating the profitability of a farm enterprise, and for comparing the profitability of alternative ways of managing farm enterprises. It can also be used to translate probabilistic information about expected crop yield response to particular management options into economic terms that are more relevant to farm decision making. This session will present enterprise budget concepts and terms, how to derive a simple crop enterprise budget, basic calculations, and how to derive probability distribution of economic returns for a set of crop management options. An enterprise budget is a useful tool for estimating the profitability of a farm enterprise, and for comparing the profitability of alternative enterprise management options. An enterprise is defined as the production and marketing of a single crop or livestock commodity that actually produces a marketable product. The enterprise is the lowest managed level in the hierarchy of agricultural systems, and the simplest to analyze. A diversified farm can include many production enterprises, and a farm household may be supported by a combination of farm and non-farm enterprises (Figure 3.4-5). 95 Ethiopia MODULE 3 Rainfed Cereal 1 Poultry Rainfed Cereal 2 Goats Irrigated Off-farm jobs vegetables Farming system Figure 3.4-4. Illustration of a farming system with several crop, livestock and non-farm enterprises. An enterprise budget describes the sources and amounts of income, costs and resulting profit associated with a farm enterprise, including: • variable and fixed costs of production, • expected yield or productivity, • expected price of the farm commodity, • income, or gross receipts, calculated as yield × price, • gross margin: net income after production costs, • breakeven price: minimum harvest price needed to cover costs, and • breakeven yield: minimum yield needed to cover costs. Enterprise budgeting involves two simplifications that make analyses simpler and more generalizable across farmers. First, it looks at the smallest unit for which farmers make management decisions that impact their livelihoods. Second, enterprise budgets are expressed on the basis of a unit of land (i.e., one hectare) or livestock (i.e., one animal or tropical livestock unit (TLU)) basis, so they can be scaled easily to different farm sizes. A partial budget further simplifies the enterprise budget by ignoring costs that do not change among the management options that are under consideration. Enterprise partial budgets can be used to compare different management options, as production management does not affect fixed costs. Because of these simplifications results can be interpreted relatively easily across farmers employing similar technology under similar environmental (i.e., soil, weather, economic) environmental conditions. Enterprise budget calculations Gross margin, or GM, refers to the net return, or profit, that a farmer would earn per hectare or per animal unit from a single enterprise in a given year. Profitability GM is the difference between income generated by the production enterprise, usually from the sale of a crop or livestock product; and the costs incurred in producing the product. For a crop enterprise gross margin can be calculated as: GM=yield× crop price-production cost where GM is in birr/ha, yield in kg/ha, farmgate price is in kg/ha, and production cost is in birr/ha. For a simple crop enterprise that includes expenses for seed, fertilizer, labor, harvest and post-harvest processing, gross margin is: GM=yield× crop price -planting density×seed price -fertilizer rate × fertilizer price -labor used × labor price-yield ×harvest and postharvest costs -fixed costs For a livestock enterprise, gross margin can be calculated as: GM=animal weight×price-production cost Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 96 where animal weight is in kg/animal, farmgate price is in birr/kg, and production cost is in birr/animal. Yields and prices are typically the most important sources of uncertainty – in part because of climate variability – that affect farm decision making. When a farmer plants a crop, she does not know what the yield and price will be at the time of harvest. Breakeven analysis uses the results of an enterprise budget to identify the minimum yield needed to break even (i.e., not experience a net income loss) for a given farmgate price: breakeven price=(production cost)⁄(yield,) or the minimum farmgate price needed to break even for a given yield: breakeven yield = (production cost)⁄(price.) Practical considerations The actual formula for an enterprise budget can vary depending, for example, on what aspects of management incur costs, which costs are assumed to vary with management and which are assumed constant, whether the production enterprise generates a single commodity or multiple sources of income, and whether the analysis accounts for the value of household inputs (land, labor) and consumption. Fixed costs, such as rental or tax on land, must be paid every year or season regardless of whether the enterprise is implemented or how it is management. Fixed costs are sometimes included and sometimes omitted from calculations. Tables are often used to list costs; show units, unit prices and quantities of inputs; and compare costs, gross receipts and gross margins for multiple management options. Institutions such as national agricultural research and extension organizations often have their own templates or guidelines for enterprise budged calculations and presentations. Combining crop simulation and enterprise budget analysis Crop simulation models can be run with historical weather data to generate probability distributions of productivity. Enterprise budget analysis is often combined with crop simulation to translate production risk into income risk. The resulting stochastic enterprise budget analysis offers a powerful tool for estimating how climate risk and management options impact the profitability of farming enterprises, as long as the input data and simulation results are properly validated. SIMAGRI Running crop simulation models, collecting and organizing all of the required input data, analyzing model outputs and performing requires considerable time and expertise. A number of software tools have been developed that combine crop simulation models with enterprise budge analysis, and package these tools with data sets and a user-friendly interface to make them more accessible to agricultural advisors. SIMAGRI is a web-based decision support tool that was developed to translate climate information into agricultural and economic terms (Han et al., 2019). SIMAGRI provides a simplified graphical user interface (Figure 3.4-6) to simplify use of the crop simulation models that are part of the Decision Support System for Agrotechnology Transfer (DSSAT). For a given location, crop and set of management options, SIMAGRI runs the crop models either based on historical weather data or probabilistic seasonal forecasts. Users have the option to enter prices and calculate production costs and gross margins for management scenarios. The output is expressed in the form of time series graphs or as probability of exceedance graphs or box plots that represent the probability distribution. Although SIMAGRI’s simplified user interface makes it easier for extension personnel to run complex crop models, it also limits their flexibility as users can only access the limited set of locations and management options that have been incorporated into the menus. SIMAGRI has been tested and adapted for use in Ethiopia, in collaboration with Ethiopian Institute for Agricultural Research (EIAR). 97 Ethiopia MODULE 3 Figure 3.4-5. SIMAGRI-Ethiopia user interface screen capture. Example: Maize fertilizer management To illustrate the usefulness of combining crop simulation and enterprise budget analysis, the example below examines the influence of fertilizer rate and seasonal rainfall on the productivity and profitability of maize. SIMAGRI was used to simulate maize response to varying nitrogen (N) fertilizer rates. The simulation used weather (1982-2018) and soil data from Awassa, the ‘BH540’ maize cultivar, a 20 April planting date, and five fertilizer rates (0, 30, 60, 90, 120 kg N/ha). Yields and gross margins showed high year-to-year variability, largely in response to rainfall variability. Although yields are consistently higher with higher fertilizer rates, this is not the case for gross margins. The graph in Figure 3.4-6 shows how both average yield and average gross margin response to fertilizer rate. In this example, yields show a consistent positive response to fertilizer rate at up to 120 kg N/ha (Figure 3.4-6a). On the other hand, gross margin increases with fertilizer rates up to 60 kg N/ha, but decreases with higher fertilizer rates (Figure 3.4-6b). In general, the level of input use that maximizes profitability (expressed as gross margin) is lower than the level that maximizes productivity (expressed as yield). This suggests that fertilizer recommendations should consider cost and profitability, and not be based on maximizing yields. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 98 Figure 3.4-6. Maize yield (a) and gross margin (b) time series simulated with daily weather from Awassa, Ethiopia. Simulations are based on weather (1982-2018) and soil data from Awassa, ‘BH540’ cultivar, and 20 April planting date. Figure 3.4-7. Simulated maize yield and gross margin as a function of N fertilizer rate, average of all years of weather data. Simulations are based on weather (1982-2018) and soil data from Awassa, ‘BH540’ cultivar, and 20 April planting date. To illustrate the interaction between fertilizer rate and seasonal rainfall, Figure 3.4-8 compares yield and gross margin fertilizer response curves averaged across all years with response curves in a relatively high (1996) and a relatively low rainfall (2004) year. July-September rainfall was 585 mm 1996 and 429 mm in 2004, compared to a 459 mm average for 1981-2018. Both the yield and the gross margin response curves were higher than average in the wet year, and lower than average in the dry year selected. The profit-maximizing fertilizer rate was higher (90 kg N/ha) in the high-rainfall year than on average (60 kg N/ha). In general, the levels of input use that maximize profitability are higher in years with favorable seasonal climate conditions and lower in years with unfavorable climate conditions. 99 Ethiopia MODULE 3 Figure 3.4-8. Simulated maize yield (a) and gross margin (b) as a function of N fertilizer rate, for all years, a wet year (1996) and a dry year (2004). Simulations are based on weather (1982-2018) and soil data from Awassa, ‘BH540’ cultivar, and 20 April planting date. SECTION 3.5 Farm Level Options for Managing Climate Risk This section addresses three types of strategies that are available to farmers to manage climate-related risk: crop production technologies, farm diversification, and adaptive management based on seasonal forecasts. By the end of this section, you will be able to: • Explain how different types of agricultural production technologies can reduce climate-related production risks. • Identify opportunities to use diversification to reduce risk for a given farming system. • Describe opportunities for farmers to take advantage of seasonal climate forecasts for their planning decisions. • Recognize challenges and identify solutions to common challenges to using seasonal forecasts to improve agricultural decision making. Production technologies Crop and livestock production technologies vary widely in their productivity, profitability and sensitivity to climate conditions. In their role of developing, testing and recommending production technologies, international and national agricultural research organizations have paid increasing attention to risk characteristics, and in particular to interactions with varying weather and climate conditions. Because the risk benefits of particular production technologies are often quite context specific, it is important to draw on local expertise when selecting or recommending available technologies, and to do adequate field testing across a range of climate conditions when developing new technologies. The discussion below, which draws from Hansen et al. (2019) and references therein, touches on three examples of production technologies that aim to reduce climate-related production risk. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 100 Crop genetics The Green Revolution, which dramatically increased production of staple cereals and reduced rates of hunger and poverty over roughly the last third of the 20th century, was largely driven by new high-yielding crop varieties. These early improvements in crop genetics targeted high-potential environments, and were supported by resource-intensive water, soil fertility and pest management packages. However, the productivity and welfare gains from Green Revolution technologies largely bypassed large, marginal rainfed agricultural regions, where the impacts of climate variability are greater, and where high rural poverty and hunger rates have been more persistent. International research and breeding programs responded by increasing attention to tolerance to climate-related stresses, which led to advances in drought-tolerant maize varieties in sub-Saharan Africa, and rice varieties with increased tolerance to drought, flooding and salinity across regions. In sub-Saharan Africa, drought-tolerant maize cultivars consistently outperform commercial varieties under drought conditions in field trials, but their performance relative to standard commercial varieties under non-stressed conditions is mixed. In field experiments in Asia, drought and flood tolerant rice experienced no significant yield penalty under non- stress conditions. Conservation agriculture Conservation agriculture (CA) aims to sustainably improve and stabilize production through a combination of reducing soil disturbance from tillage, maintaining soil cover with organic material, and crop diversification through intercropping or rotations. CA is not a fixed technology, but a varying bundle of practices that sometimes include additional soil fertility, water or weed management practices. The yield and risk reduction benefits of CA are associated with a combination of improved water infiltration and retention, accumulation of soil C, and reduced heat stress due to the presence of organic residues. Where these benefits have been observed, they generally increase over time, and may not be realized in the first few years. The effectiveness of CA varies with context. Some studies have concluded that CA shows greater risk management and productivity benefits in more drought-prone environments. Other studies show more complex interactions between CA practices and environmental conditions. The complexity of CA, competing demand for organic residues and the time lag before yield benefits are experienced have contributed to highly variable adoption rates and cases of dis-adoption in sub-Saharan Africa. Agricultural water management Irrigation has the potential to dramatically increase productivity and reduce the risks associated with rainfed agriculture in dryland (i.e., dry sub-humid and semi-arid) environments. The storage capacity of an irrigation system largely determines its capacity to buffer the agricultural impacts of climate variability. Large reservoir-based irrigation systems that can store enough water to meet demand for multiple agricultural seasons can protect crops from major droughts. However, large reservoir-based systems are expensive to build, operate and maintain. Existing systems are typically the result of major government or donor infrastructure investments, and often serve multiple water demands such as industrial, municipal and hydroelectric power generation. Small-scale agricultural water management alternatives have lower capacity to buffer the impacts of prolonged or severe droughts, but are generally more accessible to small-scale farmers due to their lower initial cost. Small-scale agricultural water management options include on-farm water harvesting and groundwater irrigation. Groundwater irrigation, using electric, gasoline or diesel pumps, often an option where the water table is sufficiently high and reliable. Efforts to reduce greenhouse gas emissions from agriculture and the diminishing cost of solar panels have resulted in increasing attention to solar-powered groundwater irrigation. Rainwater harvesting is an alternative where groundwater irrigation is infeasible or impractical. Diversification Diversification is a well-known strategy for managing risk in fields as diverse as financial investments and agricultural production. Farmers can often use diversification to reduce their income and consumption risk, even if resource constraints limit their access to other risk management options. Education in agricultural sciences, on the other hand, might not emphasize diversification, but focus instead on how to use technologies such as crop and animal genetics, agronomic practices, phytosanitary practices and veterinary care to manage risks for individual production enterprises. 101 Ethiopia MODULE 3 What can farmers diversify? Farmers and agropastoralists have a range of options for diversifying their livelihoods, including farm and non- farm income sources, livestock species and breeds, crop species and cultivars, field locations and topography, and timing of production and marketing activities. Increasing the number of crop species does little to mitigate climate risk if all of the crops respond similarly to climate stresses, as their yields and returns are likely to be strongly correlated. Correlations of yields among crops or cultivars can be reduced, and risk reduction benefits increased, if timing of sensitivity to climatic stresses is different due to differing phenology or planting dates. This is because the impact of short-term drought or heat stress within the growing season is reduced if only a portion of the plants cultivated are at a critical stage (e.g., anthesis or grain set) at the time. Growing crops in different fields with differing soil and hydrological environments can also reduce correlation and overall variability. A traditional strategy for managing risk in extensive dryland farming systems, particularly in the West African Sahel, is to divide crop cultivation between depressional areas, which collect and retain water during short drought periods; and upland areas, which are less susceptible to waterlogging and anoxic conditions during periods of heavy rain (Brouwer et al., 1993)which reduce crop yield and complicate interpretation of agronomic field experiments, are traditionally seen as problems. Data are presented that suggest that, in subsistence farming systems in the semi-arid tropics of West Africa, where nutrient and water availability alternate in limiting agricultural production, soil and crop growth micro-variability may be an asset to farmers. These farmers are more interested in relatively good yields in poor rainfall years (a satisfactory level of ‘assured’ production. Livestock, like crops, are a source of income and subsistence consumption. However, in addition to providing a source of income and household consumption, livestock are often treated as a form of savings, and slaughtered or sold at strategic times. When farmers sell animals to fill a gap in income from crop production, the income from crop and livestock sales is negatively correlated. The flexibility to use livestock as a buffer during periods of stress may therefore make diversification between crop- and animal-based farm enterprises an effective risk management strategy. Diversified farming systems are not always more resilient to climate risk, as environmental and market constraints may limit the range of viable farm livelihood options or the benefits of diversification. In an Africa- wide study, Waha et al. (2018) used data from 28,000 households in 18 countries to study the relationship between climate, diversification and food security. They found that households with more diversified production systems were generally better able to meet their consumption needs. However, the benefit from farm diversification was influence by factors such as household market orientation, livestock ownership, non- agricultural employment opportunities, and available land resources. The authors concluded that the greatest opportunities for diversification of food crops, cash crops, and livestock are located in areas where annual rainfall has a mean in the range of 500-1000 mm, and a coefficient of variability of annual rainfall in the range of 17-22%. Relationship between diversification and risk The risk benefits of diversification depend on the number of activities or income streams, and on how correlated they are in time. Increasing the number of activities or income sources reduces risk. Figure 3.5-1 illustrates this with a time series of incomes from portfolios that combine different income sources. The income from the average of five income sources (Figure 3.5-1b) is less variable than the income from the average of two income sources (Figure 3.5-1a). Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 102 Figure 3.5-1. Time series of portfolios of mixtures of enterprises and their mean (bold line), in equal proportion, with hypothetical returns of mean 0 and standard deviation 1, with (a) 2, and (b) 5 enterprises. Decreasing the correlation between different income sources also reduces risk. Figure 3.5-2 illustrates this with a time series of incomes from portfolios that combine two income sources. The income from income sources with weak negative correlation (Figure 3.5-2b) is less variable than the income from the average of income sources with weak positive correlation (Figure 3.5-2a). Risk reduction is greatest when the different income streams are weakly or negatively correlated. Combining income from two sources that are perfectly correlated has no effect on risk. Figure 3.5-2. Time series of portfolios of two enterprises and their mean (bold line), in equal proportion, with hypothetical returns of mean 0 and standard deviation 1, and correlations of (a) +0.5 and (b) -0.5. 103 Ethiopia MODULE 3 Recall from section 3.3 that a decision maker who follows a set of rationality assumptions would select options that fall on the frontier to the left (i.e., lower risk) and top (i.e., higher returns) of an E-S graph. Figure 3.5-3 shows a mean-standard deviation (or E-S) graph of returns from two hypothetical enterprises, A and B, with the specified mean, standard deviation and cross-correlation, and for mixes of the two at 10% increments. The parabolic shape of the curve in Figure 3.5-3 is characteristic of E-S graphs of diversified portfolios. Under the assumptions in this example, a rational decision maker would never select 100% enterprise A because combining A with up to 40% enterprise B would reduce risk and increase expected return. Whether a diverse mixture exists that has lower risk than the least risky individual enterprise depends on the mean, standard deviation and correlation of the individual options. Figure 3.5-3. Mean-standard deviation (E-S) graph of returns from combinations of two hypothetical enterprises with specified means, standard deviations and correlation. Guidelines for using diversification to manage risk Understanding how diversification influences risk suggests a few guidelines for using farm diversification as a climate risk management strategy. First, a key goal for farm diversification is to combine income or food sources that have low or negative correlation. For crops, correlation may be reduced by selecting crops or cultivars with differing phenology, staggering planting dates, and selecting different fields with different topographic characteristics. Diversifying crop commodities can reduce income risk if farmgate prices are weakly or negatively correlated, even if yields are strongly correlated. Second, in mixed farming systems, the complementary roles of crops and livestock can be exploited. Animals serve as a form of savings. Crops must be harvested and generally sold at a particular time of year. Animals, on the other hand, can be sold for income to compensate for loss of income or staple food production when a harvest fails. Third, consider all potential sources of production and income, not just on-farm production. Farm households often have other sources of income than farm production, such as off-farm salaried employment and day labor, remittances from extended family, social protection programs, and emergency humanitarian assistance. If a farm household’s production and income are already diversified, look for opportunities to adjust portfolio of activities and incomes to better meet the particular farmers’ goals and risk preferences. Consider whether the portfolio should be adjusted as new climate and market information become available. Forecast-Based Seasonal Farm Planning Role of seasonal forecasts in farm management Although farmers make management decisions many times throughout the year, the period shortly before the main agricultural season is a particularly crucial time when farmers must make many strategic decisions. Seasonal forecasts are potentially relevant to a wide range of agricultural production systems and decisions. However, seasonal forecast use has been most studied, and evidence suggests is most widespread, in the case of rainfed annual crop production. This is because the seasonal time scale matches annual crop production cycles. Table 3.7-1 lists a few common agricultural management decisions that farmers adjust based on seasonal forecasts. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 104 Table 3.7-1. Decisions that farmers frequently report adjusting in response to seasonal climate forecasts. Decision Timing Allocation of farm land, labor, capital Before growing season Land preparation Crop, cultivar selection, planting density Basal fertilizer application Time of planting Fertilizer side dressing Early season Weed management Herd stocking, de-stocking Varies Herd migration, transhumance Feed, forage management The benefit from seasonal forecast use for crop production management is a function of interactions among seasonal climate conditions, management options, prices of inputs and farm commodities, productivity and profitability. In general, optimal management intensity (e.g., investment in production inputs, fertilizer application rates, weeding intensity) increases when favorable climate conditions are anticipated, and reduces when rainfall deficits or other adverse conditions are anticipated. Figure 3.7-1 illustrates this interaction for a modeling study of maize nitrogen fertilizer management at a semi-arid location in Kenya. The graphs show simulated grain yields for individual fertilizer rates and fit to a nonlinear function, and gross margin calculated from the fitted yield response curve. Although yields increase over the range of fertilizer levels, the increased production cost causes gross margin to decrease above an optimum. When averaged among all years of weather data, the profit-maximizing fertilizer rate is 90 kg ha-1. However, optimal fertilizer rate increases to 146 kg N/ha for a wet year (1994), and decreases to 28 kg N/ha in a dry year (1995). Figure 3.7-1. Simulated maize yields and gross margin as a function of N fertilizer rate, and profit-maximizing fertilizer rate, for all years (1968-2002), (b) a dry year (1995) and (c) a wet year (1984). Based on Hansen et al., 2009. Challenges and solutions Farmers may encounter several common challenges when using seasonal climate forecasts. A good understanding of these challenges suggests ways that agricultural extension workers can help farmers overcome them and make better use of seasonal forecasts. The discussion below draws heavily on Hansen et al. (2011; 2022)with a view to understanding and exploiting opportunities to realize more of its potential benefits. Interaction between the atmosphere and underlying oceans provides the basis for probabilistic forecasts of climate conditions at a seasonal lead-time, including during cropping seasons in parts of SSA. Regional climate outlook forums (RCOF. 105 Ethiopia MODULE 3 The economic, technical, policy and social constraints keep many smallholder farmers trapped in poverty can also limit their ability to adjust their management to take advantage of seasonal forecast information. For example, a high-yielding hybrid might perform much better than the usual cultivar in a season that is predicted to have longer growing season with higher rainfall than normal. Yet smallholder farmers cannot change cultivars if seed is not available, or if they cannot access credit to purchase the seed. Although such constraints can reduce the options available, particularly for relatively poor smallholder farmers, available evidence from across Africa shows that the majority of crop farmers who access seasonal forecasts change some management decisions in response (Hansen et al., 2011; Vaughan et al., 2019)with a view to understanding and exploiting opportunities to realize more of its potential benefits. Interaction between the atmosphere and underlying oceans provides the basis for probabilistic forecasts of climate conditions at a seasonal lead-time, including during cropping seasons in parts of SSA. Regional climate outlook forums (RCOF. Although severe resource constraints can exclude relatively poor smallholder farmers from some forecast-based management options, adjustments to allocation of land and labor among farm activities, and to the timing of farm operations, are less likely to be affected by resource constraints. When availability of key inputs, such as seeds or credit, limit farmers’ ability to benefit from seasonal forecasts, agricultural institutions and value chain actors might be in a position to take actions that alleviate farmers’ constraints to acting on forecasts. Some practitioners and literature on seasonal forecasting express concern that smallholder farmers cannot cope with the consequences of a “wrong” forecast. Although this concern reflects an important challenge, it also reveals a problem with how seasonal forecasts are sometimes communicated and interpreted. In the absence of seasonal forecasts, farmers must account for climate variability when they make seasonal planning decisions if they are to survive as farmers. Seasonal forecasts are probabilistic. They are not fundamentally different from the climatological distribution that describes farmers’ routine experience with climate variability, but merely shift the distribution for the upcoming season. A few plausible situations might lead a farmer to make decisions differently with and without seasonal forecast information. The first is the very real danger that probabilistic information about the forecast distribution could be lost or distorted somewhere in the forecast generation, dissemination, interpretation and application process. Underestimating uncertainty can lead to excessive responses that are inconsistent with decision makers’ risk tolerance, and can damage the credibility of the forecast provider. Second, the process of learning to use the new information in new ways could increase risk. Well-designed training can accelerate this learning process and reduce this risk. Finally, agricultural extension advisories or policy interventions that promote particular forecast responses could coerce farmers to make seasonal management decisions that are inconsistent with their goals or risk tolerance, and thereby increase their risk exposure. If farmers are to make decisions that are consistent with their level of risk aversion, and consistent with the way they factor climate variability into their planning in the absence of forecasts, they must understand the uncertainty of the forecasts. Furthermore, they should understand that the use of seasonal forecasts is expected to improve their wellbeing average, but not necessarily in every year. Section 4.3 presents a participatory seasonal forecast training and planning workshop that aims to help farmers understand forecasts as probability shifts in the context of their local climate, and not deterministic statements of what will happen. Most Regional Climate Outlook Forums and National Meteorological Services (NMS) in Africa have adopted a subjective process to arrive at a consensus among different sources of prediction, and express the seasonal forecast as probabilities that rainfall in the upcoming season will fall in “below-normal,” “normal” or “above- normal” historical tercile categories. In many cases, the probabilities are replaced with deterministic statements, e.g., “Rainfall will be normal to above normal.” There is a significant mismatch between these forecast conventions and farmer needs. Experience with farmers across many countries and contexts has revealed several weaknesses that limited the usability of these forecasts for local agricultural decision-making. The most widely reported criticism is that they do not directly provide information about expected climate conditions at the local scale at which most climate-sensitive decisions are made. Second, forecast probabilities are associated with thresholds that are defined by tercile boundaries (i.e., 67th and 33rd percentiles of the distribution) and not by needs of decision makers. In agriculture, relevant thresholds are context-specific, defined by factors such as crop water and growing season length requirements. Third, forecast categories are difficult to interpret. The need to process both directional shifts from the “normal” tercile, and probability shifts from the 33% climatological tercile probability (e.g., “above-normal probability of below-normal rainfall”) imposes a level of complexity that often leads to misinterpretation. Furthermore, these forecast categories are often misinterpreted as something other than historic terciles. Fourth, acting appropriately on forecast information requires understanding the degree of accuracy, or conversely the uncertainty, of the forecast. Although probabilistic tercile forecasts account for forecast uncertainty, the format leaves ambiguity about the accuracy of the forecast system, which can in turn lead to inappropriate management responses. Finally, farmers need information beyond average conditions during the growing season, such as timing of the season start and end, and risk of damaging dry spells or other extremes. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 106 These constraints reflect conventional practice and not inherent characteristics of seasonal forecasts. Most of these weaknesses can be eliminated or reduced by switching to objective forecast methods, using local historical climate data to calibrate and downscale forecasts, and presenting probability distributions of the forecast along with the climatological distribution (Table 3.7-2). Downscaling the forecast onto local climate data, and presenting the climatological distribution alongside the forecast distribution, in physical units (e.g., mm rainfall), provides all of the information needed to anticipate upcoming seasonal climate conditions at the local scale of decision-making. Presenting the full forecast distribution provides forecast quantities or probabilities associated with any decision-relevant threshold, and avoids misinterpretations that are common to categorical probabilistic forecasts. The ability to compare the shapes and position of the forecast and climatological distributions conveys the degree of uncertainty of the forecast relative to climatology, and hence the skill of the forecast, more clearly than tercile probability shifts. Assuming hindcast residuals are used to calibrate a forecast distribution, the narrower the forecast distribution, the greater the accuracy of the forecast system. The EMI Flexible Forecast maproom incorporates these improvements. ] Table 3.7-2. Tercile seasonal forecast usability challenges, and solutions available through NextGen forecast methods and Flexible Forecast presentation. Adapted from Hansen et al., 2019. Challenge Solution Lack information about local climate Downscale forecasts Bundle forecasts with local climatology Categories are difficult to understand Provide full forecast probability distribution Categories don’t match decisionmaker thresholds Limited relevance of seasonal averages Forecast additional variables Translate into impacts, options Different farmers are likely to respond quite differently to different seasonal forecasts. In general, seasonal planning decisions tend to be more sensitive to differences in individual farmer characteristics and their physical and market environments than are the decisions that farmers make at shorter time horizons once they are committed to a crop or livestock enterprise. In rainfed annual cropping systems, farmers must decide before the start of each growing season how they will allocate their scarce land and household labor, and how much cash from savings or borrowing they will invest, among a number of alternative farm and non-farm enterprises. These resource allocation decisions are quite sensitive to differences in resource availability, to changes in expected commodity and input prices, and to farmer goals and level of risk aversion. Results of a model-based study of the value of ENSO information for a simple crop mix problem in Argentina (Messina et al., 1999) illustrate the sensitivity of seasonal farmland allocation decisions to risk preferences and prices (Figure 3.7-2). Figure 3.7-2. Modeled farmland allocation responses to ENSO-based seasonal forecast information, as influenced by risk preferences and prices, Pergamino, Argentina. Based on Messina et al., 1999. 107 Ethiopia MODULE 3 The sensitivity of many seasonal farm management decisions to individual farmer characteristics and contexts has important implications for how extension and advisory service providers support farmers’ use of seasonal forecasts. Seasonal farm management decisions that are very sensitive to individual farmer characteristics and contexts do not align well with top-down seasonal forecast-based advisories. For such decisions, advisories should be tailored to varying farmer characteristics and contexts would likely not be widely adopted or appropriate. An effective approach is to support farmers’ understanding and use of seasonal forecasts with participatory communication and planning approaches (Section 4.3). SECTION 3.6 Index-Based Agricultural Insurance Index-based agricultural insurance, also known as “index insurance,” is a relatively new innovation that has overcome major challenges to providing insurance to smallholder farmers and pastoralists. Interest in agricultural insurance is growing in Ethiopia, building on the achievements of a few successful insurance pilot projects. This section presents an overview of index insurance, considers the circumstances under which its benefits outweigh the costs, and provides several guidelines for addressing challenges and strengthening the effectiveness of index insurance. By the end of this section, you will be able to: • Summarize main roles of insurance for agriculture. • Explain main differences between indemnity-based and index-based insurance. • Interpret a basic index insurance contract. • Identify circumstances in which farmers are likely to benefit from index insurance. Index insurance basics Role of agricultural insurance Like other forms of insurance, agricultural insurance protects policyholders from specific types of risks by providing a monetary payout when a covered loss occurs. In the case of agriculture, insurance can contribute to the wellbeing of farmers in two ways. First, insurance protects farmers from losses. An extreme climate event, such as a drought or flood, can have long-term impacts on livelihoods. For example, a farmer who does not harvest enough to repay loans, due to a drought or flood, may lose the ability to borrow money for farm inputs in the future. Likewise, it can be difficult for a pastoralist to recover their herd if a drought depletes grazing resources to a point where many animals die. Insurance payouts can help farmers avoid coping strategies that deplete their productive assets, or replace lost assets, resulting from an extreme climate event. Second, insurance can promote farmers’ livelihoods by improving the adoption of improved production practices and access to credit. By providing payouts after an extreme event, insurance not only protects farmers from resorting to coping strategies that threaten their livelihoods by depleting productive assets, but it can also unlock investments that are more risky but more profitable on average. For smallholder farmers, the risk of an infrequent but severe event, such as a flood or a drought, is a disincentive to investing in improved seeds, fertilizer and other costly agricultural inputs. Risk also has a negative impact on the development of rural financial services and the availability of credit to smallholder farmers. Knowing that insurance will provide payouts in the event of extreme climate conditions increases farmers’ confidence to invest in improved agricultural practices, and increases the confidence of credit providers to lend to smallholder farmers. In this way, farmers can benefit from insurance even in years when they do not receive a payout. What is index-based insurance? Understanding index-based insurance requires recognizing how it differs from traditional indemnity insurance. With any insurance, the farmer or other client pays an agreed amount, known as a premium, on a regular basis. If a specified hazard is experienced, the client receives a monetary payout. The amount and timing of the premium, the conditions under which a payout is made, and the amount of the payout are all specified clearly in a contract, well in advance of experiencing the particular hazard. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 108 Indemnity insurance insures against a specified loss, and requires farm visits to verify that the loss occurred. This type of insurance has sometimes worked for large-scale farms. However, a few challenges have made this type of insurance generally infeasible for smallholder farmers. Moral hazard refers to the incentive for farmers to neglect good risk management in order to receive payouts. Adverse selection refers to the tendency for insurance to be purchased preferentially by farmers who experience more frequent losses. Moral hazard and adverse selection increase the cost of insurance. Third, the need for farm visits to verify claims results in high processing costs and long payout delays. Index-based agricultural insurance is a recent innovation that triggers payouts based on an index that is correlated with agricultural losses, rather than actual losses. Indexes include rainfall or temperature during a defined period, yields sampled over a larger region, and remote sensing of vegetation conditions or flood extent. Index insurance seeks to cover specific threats, and for a number of farmers, which can be captured by the selected index. Basing payouts on an objectively measured index, rather than the performance of individual farmers, reduces the incentive for farmers to neglect good risk management (moral hazard), or to base their decision to purchase insurance on the losses they expect to experience (adverse selection). Without the need to process claims and verify losses, payouts can be made quickly and with low administrative cost. The main limitation of index-based insurance is basis risk, which represents the imperfect relationship between farmers’ losses and the index that triggers payouts. As a result, farmers who pay an insurance premium might not receive a payout when they experience significant crop or livestock loss, or they might receive payouts when they did not suffer actual losses. Both forms of basis risk can reduce the benefits farmers receive from insurance. There are three main types of index-based agricultural insurance programs: 1. Weather index insurance is based on meteorological variables such as rainfall or temperature, or sometimes the result of a soil water balance model driven by meteorological data. 2. Area yield insurance is based on yields estimated over a specified region, using statistical sampling from many farmers’ fields. 3. Index-based livestock insurance is either based on livestock mortality data estimated over a specific region, or on vegetation remote sensing data. Vegetation remote sensing data is a useful index of livestock mortality where drought impacts on grazing resources are a major cause of livestock mortality. Table 3.6-1. Contrasts between indemnity-based and index-based insurance. Characteristic Indemnity insurance Index insurance Payouts based on: Measured loss An index that is correlated with loss Farm visits needed? Yes No Farmer incentive to No: poor yield leads to payout Yes: yield not directly related to payout manage crop well? Advantages • Payouts based on actual loss • Quicker payout • Lower operating cost allows lower premium or higher payouts • More accessible to smallholder farmers Limitations • High operating costs lead to high • Basis risk: some losses may not have payouts premium or low payouts • Slow payouts Index insurance contract structure Figure 3.5-1 illustrates how a simple weather index insurance contract is structured. This example index is based on rainfall, and is intended for use in drought index insurance schemes. A contract window is the specific range of dates when the insurance covers a particular risk. A well-designed weather insurance contract provides protection during critical phases when the crop is particularly vulnerable to the particular weather hazards. Timing is perhaps one of the most important factors to consider when designing a rainfall-based index for agricultural index insurance. The trigger is a pre-identified threshold value of the index that triggers the beginning of a payout, and the exit is a pre-identified threshold that triggers the maximum payout amount. Weather index insurance contracts often increase payouts linearly when the index value falls somewhere between the trigger and the exit. 109 Ethiopia MODULE 3 Figure 3.6-1. Components of a simple rainfall index insurance contract. Is index insurance worth the cost? Insurance is both an effective way to manage some risks, and an expensive way to manage risk. For commercial insurance, the premiums that insured farmers pay must cover several costs. Part of the premium is saved towards payouts that the company owes insurance holders when the covered hazard occurs. Part of the premium is used to cover operating costs and provide profit to the insurance company that is assuming that risk. Since insurance premiums must cover the average cost of payouts, operating costs, and insurance company profits, farmers can expect to pay more for premiums than they receive back in the form of payouts, when averaged over many years. Insurance is likely to be successful in the long term only if the benefits outweigh the costs to the farmer. For this to happen, CE of farm income must be higher with insurance than without insurance (Section 3.3). It is useful to consider three cases in which this might happen: 1. The positive effect of reduced variability outweighs the negative effect of reduced average income on CE. 2. Insurance leads to management changes that increase net income. 3. Premium subsidies reduce the cost of insurance. Case 1: Risk reduction effect outweighs reduced income effect If we assume that (a) insurance premiums are not subsidized, and (b) insurance does not change agricultural management practices, then insurance has two opposite effects on CE. First, insurance reduces average net income, which tends to reduce CE. This is because the average payout is less than the annual premium. Second, insurance reduces variability of income, which tends to increase CE. Recall that combining sources of income that are negatively correlated reduces variability of income (section 3.5). Because insurance payouts occur in years when production is very low, income from payouts is designed to have a strong negative correlation with agricultural income and hence reduce variability of farm net income. Since insurance has two opposite effects on CE, it is possible that insurance will benefit some farmers but not others when the two assumptions (no premium subsidy and no effect on agricultural management) hold. Insurance is most likely to increase CE of farm income when: • risk is very high; • risk aversion is very high; • insurance greatly reduces an important risk or prevents a shock from reducing future income through loss of productive assets; and • insurers’ operating cost plus profit margin is a small percentage of the premium cost. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 110 One example where these conditions is likely to be held is index-based livestock insurance (IBLI) for pastoralists. Pastoralists who depend on rangelands in semi-arid to arid lands are among the most vulnerable population to drought risk in Ethiopia. Index-based livestock insurance (IBLI) was developed by ILRI in northern Kenya in 2010, and extended into the Borana region of Ethiopia in 2012. The Normalized Difference Vegetation Index (NDVI), based on satellite remote sensing, is a good predictor of rangeland conditions, and is strongly correlated with livestock mortality in drought conditions. IBLI is designed to protect pastoralists against herd loss and resulting poverty traps during severe drought, by triggering payouts when NDVI in traditional grazing areas falls below a threshold. Case 2: Insurance leads to management changes that increase net income The second case, in which insurance increases expected farm income, is possible if insurance leads to change in farm management, which increase average farm income. Several index-based agricultural insurance initiatives were designed to overcome risk as a barrier to adoption of improved agricultural management or market opportunities. Insurance can lead to improved management if it either enhances farmers’ willingness to adopt improved practices, invest scarce household capital or borrow; or enhances the willingness of lenders or value chain actors to provide services to farmers. In the case of crop-based farming systems, experience suggests that the benefits from enabling improved management practices are generally more important than the benefits from reducing income variability. Factors that determine cost Since insurance is effective in the long term only if the benefits outweigh the costs, it is important to design insurance programs that are cost-effective. Several aspects of index insurance contract and program design influence the cost to farmers: • Payout size and frequency: For a given premium, farmers face a tradeoff between the frequency and the magnitude of payouts. The frequency of payouts in an index insurance contract should closely reflect the frequency of extreme events that lead to serious losses that exceed a household’s coping capacity or that may lead farmers to default on loans. Index-based crop insurance often covers the cost of production inputs to protect farmers’ access to credit or to allow replanting if an early-season dry spell kills the germinating crop. Insuring the full harvest value of a crop would be much more expensive. • Operating costs: The higher the operating cost per farmer, the higher the premium. Low transaction cost is one of the main advantages of index-based agricultural insurance relative to indemnity insurance. Operating costs per farmer typically decrease as the number of farmers who adopt a particular insurance product increase increases and fixed costs can be spread among more client farmers. Once an insurance program has been designed and implemented, the operational fees and profit margin set by insurance companies are usually not negotiable. • Uncertainty: If an insurance company does not have enough data to estimate the insured risk accurately, companies will increase the cost of insurance to protect themselves from that uncertainty. • Subsidy: In some countries, governments subsidize insurance premiums to reduce the cost to farmers and thereby stimulate adoption. However, subsidies that are not implemented carefully have the potential to discourage other means of managing risk, create dependence on future subsidized assistance, and distort incentives for insurers and banks. Index insurance good practice the high cost of insurance, basis risk, heterogeneity of farmers’ needs, and trust in insurance are key challenges to achieving the potential benefits of index-based agricultural insurance. Recognizing these challenges suggests some practices that address those challenges. To improve the cost-effectiveness of insurance: • Insure only low-frequency, high impact risks that cannot be covered through other risk management options, exceed coping capacity, or lead to future loss of livelihood potential. • Design insurance to provide benefits in good climatic years when payouts do not occur, for example by bundling with credit or production technology packages that aim to improve farmers’ productivity and income. • If insurance is subsidized to reduce cost to farmers and stimulate adoption, use only smart subsidies. A smart insurance subsidy should: (a) serve a well-defined policy objective, (b) target a well-defined set of beneficiaries, (c) be informed by monitoring and evaluation, and (d) have either a clear exit strategy or a viable long-term financing strategy. 111 Ethiopia MODULE 3 To reduce basis risk: • Use high quality, long-term, complete data to design and validate indexes and calculate risk. • Use agrometeorology tools and expertise to design indexes that are strongly correlated with agricultural losses. However, bear in mind that index design must also consider issues such as understanding and trust in the product, vulnerability to manipulation, and availability of data. • Validate index performance against objective data (e.g., historical production statistics, experiment station records or remote sensing data) and farmers’ experience. To respond to farmers’ varying needs: • Design insurance products to mee the needs of specific farmer groups, recognizing that the benefits might not outweigh the costs for some types of farmers. • Involve farmers in the design process to ensure that products align with their demand and constraints, and cover the risks that they consider important. • Verify that the index captures the risks that are most important to the farmers. To foster trust: • Support insurance with a strong legal and regulatory environment that enforces contracts that both buyer and seller can trust. • Invest in transparent communication and farmer education to enable farmers to understand contract terms, the basis for payouts, and the reason why it is possible to experience a loss without a payout. • Market insurance through existing distribution channels that farmers use and trust. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 112 Integrating Climate Services into Agricultural Extension MODULE 4 ABOUT THIS MODULE Module 4: Integrating Climate Services into Agricultural Extension equips extension staff to bring climate services into the services that they provide to their client farmers. Integrating climate services into agricultural extension activities and responsibilities can improve the effectiveness of both types of services. Delivering climate services to farmers involves much more than disseminating information. It requires understanding farmers’ diverse needs, strengthening their capacity to interpret and act appropriately on new types of information, and communicating farmers’ demands and feedback to those who produce climate information and agricultural advisories. Communication processes are therefore crucial to enabling farmers to benefit from climate services. By the end of this module, you will be able to: • Formulate a climate service communication strategy for a given context. • Lead a group of farmers through a basic seasonal forecast communication and planning workshop process. SECTION 4.1 Section 4.1. Rural Climate Service Communication Strategies Delivering climate services to farmers involves much more than disseminating information. Effective services must begin with understanding farmers’ diverse communication and capacity needs. The different channels by which rural communities access information have differing roles, strengths and limitations. This section discusses different communication needs and communication channels that are relevant to rural climate services. It guides learners to consider how communication channels can be combined strategically to support the climate service needs of rural communities. By the end of this section, you will be able to: • Recognize the factors that lead to inequitable benefits from climate services, and potential solutions. • Understand why weather and climate information time scales require different communication strategies. • Explain how group participatory processes help farmers understand and use probabilistic climate information. • Identify and use appropriate communication processes and channels to deliver climate services to farming communities in a given context. Communication to support gender and social equity Climate services have potential for strengthening the wellbeing of rural women and other disadvantaged groups. However, they risk reinforcing existing inequalities if service providers fail to understand and effectively address their differing needs and constraints, and include them in the development of services. Climate service social equity challenges Sociocultural norms, institutional biases, and differing resource endowment and capacities among individuals reinforce inequalities within rural populations. These factors are crucial challenges to agricultural development efforts generally, and to equitably assess, use and benefit from climate services (Table 4.1-1). The discussion below draws on a review by (Gumucio et al., 2020). Table 4.1-1. Examples of how sources of gender and social inequality in rural communities lead to climate service equity challenges. Sources of inequality Climate service equity challenge Social norms about work and decision • Differing climate-sensitive decisions could lead to differing making, participation in group processes, climate information needs. based on gender, age, ethnicity, etc. • Social norms reduce women’s access to ICT-based services, group processes, institutions. Bias of agricultural institutions towards Women face obstacles accessing information or support to act on men, or crops that are considered men’s information, through agricultural extension or farmer associations. responsibility Individual differences in wealth Relatively poor farmers are excluded from fee-based services, some ICT channels. Individual differences in education Farmers with education or literacy constraints are excluded from text- opportunities based resources (e.g., SMS, bulletins). Sociocultural norms that differentiate farming and decision-making roles based on gender, age, ethnicity or caste can lead to differences in participation in farm and household decision making, and hence differences in the types and time scales of information that individuals can act on. In some cultures, gender roles and interactions within households, which are influenced by sociocultural norms, constrain women’s access to mobile phones or the times when they are available to access information through broadcast media programs or group meetings. Historically, agricultural extension and related service providers across Africa have often favored men’s participation or target agricultural commodities that are considered men’s responsibility. Farmers’ associations and cooperatives sometimes exclude or underserve women due to membership criteria based on land ownership and other capital requirements. In many cases, education and policy efforts are working to help agricultural institutions overcome gender biases and better serve rural women. Furthermore, sociocultural norms or local power dynamics that restrict participation of women in group meetings can lead to differences in the institutional channels available to access climate-related information, training and support. 115 Ethiopia MODULE 4 Differing individual resource endowments and educational opportunities also contribute to differing ability to benefit from climate services. Differing local language present a barrier to accessing information or advisories that are available only in official national languages, and present a challenge to resource-constrained information providers. Information that is distributed in text form (e.g., SMS) can reinforce inequalities due to differing access to education and literacy levels. Because all of these factors are context- and culture-specific, efforts to develop equitable agricultural climate services should be informed by knowledge of how aspects of identity interact with sociocultural norms and institutional biases in a given country or local context. A growing number of African countries are promoting commercial, fee-based weather and climate information as a way to supplement inadequate NMS budgets. However, available willingness-to-pay studies in Africa show that a substantial proportion of farmers are unwilling or unable to pay for climate information (Table 4.1-3). Depending solely on fee-based services risks reinforcing wealth-based inequalities, as those lack the willingness or ability to pay are disproportionally relatively poor subsistence-oriented farmers, and in strongly patriarchal cultures women who lack control over household finances. Table 4.1-2. Published studies of farmers’ willingness to pay for weather and climate information. Study Country Product Willing to pay >0 Mean WTP* N Amegnaglo et al., 2017 Benin Seasonal forecasts 81% $23.42 354 Antwi-Agyei et al., 2021 Ghana various 43% n/a 193 Seasonal forecast 21% Daily forecast 24% Donkoh (2019)in Ghana Weather forecast 54% $28.37 Hoyos (2010) Burkina Faso Seasonal forecast 53% $7.51 169 10-daily information 33% $2.29 Daily forecast 53% $4.27 Agro-advisories 33% $3.50 Zongo et al., 2015 Burkina Faso Seasonal forecasts 64% $1.19 629 *Annual amount per farmer, averaged among entire sample, adjusted to 2022 USD. There is more limited evidence that the sociocultural norms that often disadvantage women and influence the decisions that are under their control also lead them to need different types and time scales of information, or at different times of the year. Where this is the case, intentional effort is needed to identify and respond to these differing information needs. Improving distribution of climate service benefits Once disadvantaged groups are identified and the main constraints that they encounter to accessing, using and benefiting from services are understood, several options are available that can foster more equitable climate services (Table 4.1-4). Available evidence suggests that the most widespread challenges and hence the most promising opportunities to improve the distribution of benefits from rural climate services involve communication processes and channels. One key area relates to ICT channels based on mobile phones. Rural women in Sub-Saharan Africa tend to be less able than men to access and use weather and climate information via digital channels, and less willing to pay for commercial service. Factors that tend to work against rural women accessing climate and other information through mobile phones – including limited control over household financial resources, spouse disapproval, lower rates of formal education and literacy, and gender norms that limit control over climate- sensitive decisions and hence demand climate information. On the other hand, a few experiences demonstrate that group participatory communication processes reduce or eliminate this gender gap in climate service access and use. A second key area relates to group communication processes. While group participatory communication processes are effective at raising the capacity of rural women to understand, access and act on climate information, women may face obstacles participating due to institutional biases or cultural norms. Where this is the case, partnering with women’s organizations, or inclusive community or religious gatherings, might improve Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 116 women’s participation. In any group communication process, it is important that facilitators know how to recognize when gender, age or social status adversely affects some individuals’ group participation. Within agricultural development, including climate services that support agriculture, there is growing interest in empowering rural women. Empowerment involves “the expansion of the capacity to make strategic and meaningful choices by those who have previously been denied this capacity, but in ways that do not merely reproduce, and may indeed actively challenge, the structures of inequality in their society” (Kabeer, 2017: 651). It is not realistic for climate services alone to drive the changes in the cultural and institutional drivers of gender inequality that woman’s empowerment entails. Yet a few instances in which participatory climate service activities have contributed to improvements in women’s confidence, status within the home and community, and participation in decision making suggest that climate services could play an incremental role in broader efforts to empower rural women. Climate services are likely to contribute more to women’s livelihoods and empowerment if service providers collaborate with civil society organizations that are working to address the underlying socio-cultural institutional drivers of inequality. Table 4.1-3. Potential solutions to common gender and social equity challenges in rural climate services. Equity challenge Potential solutions Gender differences in climate Separately assess information needs of women and men, and of female- information needs headed households. Obstacles to accessing ICT, Combine communication channels suited to differing needs. broadcast media, group communication channels Schedule meetings, broadcasts to accommodate women’s time and mobility constraints. Partner with women’s groups to deliver information, build capacity. Train extension, other intermediaries to recognize when gender, age or status adversely affects participation. Inability to pay for services Balance private fee-based services and pro-poor public-good services. Education and literacy constraints Build capacity through group participatory processes. Provide information and training in local languages, and in audio and video formats. Differing needs for weather and climate information To review: Weather refers to the state of the atmosphere at a particular time and place. Examples include forecasts of daily or hourly weather conditions for up to about ten days into the future, and records of recent daily or hourly weather conditions. Climate refers to statistical information about weather. Examples include seasonal rainfall or temperature forecasts for the coming 3-5 months, and summaries of historical data showing seasonal cycles, variability and trends. Weather is relatively simple straightforward to understand and factor into decision-making because people experience weather every day. In contrast, climate is a more abstract statistical concept and climate information is inherently probabilistic. Since climate information is more challenging to understand and act upon, different communication processes and more support are required to use it appropriately. Practical considerations Several differences between weather and climate information have practical implications for how the information would be best communicated to a given community of farmers or pastoralists. Weather information is used frequently throughout the agricultural calendar, particularly the growing season and, because of its short lead time, must be communicated quickly. On the other hand, climate information tends to be used infrequently, at particular times during the year. For example, seasonal forecasts are consulted shortly before the rainfed growing season, to plan crop production strategies and obtain inputs. Farmers would likely consult historical climate information less frequently, to assess whether the farming system is adapted to the risks, trends and seasonal cycles of their local climate. Weather information is relatively simple to understand. Frequent repetition helps farmers learn to understand weather information, assess its accuracy, and act appropriately on the information. Climate information is more challenging to understand because the information is more complex and inherently probabilistic. 117 Ethiopia Because climate information is consulted and at most a few times each year, decision makers must depend on statistical descriptions instead of personal experience to assess its accuracy, and be trained to interpret and act appropriately on the information. Table 4.1-4. Differences between weather and climate information that have implications for communication needs. Weather Information Climate Information Used frequently, needed quickly Used at particular strategic times Relatively simple information Complex, probabalistic information Users learn to interpret and act on information through Users need training and support to interpret and act repetition on information Simple messages, frequency of information fit Timing, training and support needed fit mobile phone, broadcast media channels group participatory processes Psychological considerations Psychological research describes two different systems that the human brain uses to process decisions that are based on uncertain information: • Analytical processing refers to the system our minds use to process information that is in the form of a statistical description, such as a probability distribution, or the probability of experiencing an outcome within a particular category such as above-normal seasonal rainfall. • Experiential processing refers to the system our minds use to process information that is obtained through repeated experience. Experiential processing relates current situations to memories of one’s personal experience, or the stories of other people’s experience. The brain’s experiential processing system is likely to play a major role when people interpret weather information, because they experience weather conditions on a daily basis and have many memories of positive and negative experiences with weather. Farmers, in particular, are likely to have strong memories, including some that are quite negative, from their experience with weather and its impact on their farming operations. Since climate refers to statistics (i.e., averages, variability, seasonal cycles and trends) that are typically described with graphs or statistical summaries, and not to events that people experience at particular times, analytical processing is likely to play a major role when people interpret climate information. When people are able to relate abstract new information to their own experience, and hence integrate their analytical and experiential processing systems, they are more likely to act on that information and more able to avoid some of the errors that are commonly associated with making decisions with probabilistic information. This understanding of how the human mind processes complex information suggests that communication strategies that help farmers relate climate information to their own experience is a key to enabling them to use new climate information effectively (Marx et al., 2007). Based on these practical and psychological considerations, weather and climate information require different communication strategies. Mobile phone and broadcast media channels, for example, are a good match for the simple content, and the frequency and time-sensitive nature that characterize weather information. Structured training and group participatory processes are a good match for the degree of learning and support that are needed to understand and act appropriately on climate information. The next section will discuss communication options and strategies to support rural communities with climate services. Participatory climate communication processes Participatory communication processes with groups of farmers can be effective at increasing their understanding and willingness to act on complex, probabilistic climate information because they provide opportunity for formal training and social learning, and because they help overcome the cognitive difficulties that people experience when processing and using uncertain information. People generally have difficulty interpreting probabilistic information and tend to make several common errors when using that information for decision making; particularly when this information is presented and processed Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 118 as statistical summaries. However, those difficulties are reduced when people combine analytical and experiential processing when they make decisions based on climate information (Marx et al., 2007). Furthermore, experiential processing tends to have greater influence than analytical processing on decision- making because it is connected to a person’s memories and emotions. Participatory communication processes that incorporate group activities can help farmers relate statistical climate information to their personal or collective experience. Examples include: • Games that provide accelerated experience with simulated decisions; • Activities that relate climate information graphs to farmers’ memory of recent agricultural seasons; • Group discussion of experience with traditional climate indicators; and • Group discussions of hazards, opportunities and decision options associated with seasonal climatic conditions (e.g., above- or below-normal rainfall) they have experienced in the past. Several participatory climate communication processes include useful training and guidance materials. Section 4.3 presents a participatory workshop process that the IRI developed to help farmers relate downscaled seasonal forecasts in probability-of-exceedance format to their own memory of historic climate variability. It specifically supports the use of the Flexible Forecast Maproom presentation, although it was developed before National Meteorological Services started adopting this format. While the goal of this approach is to equip farmers to understand and use seasonal forecasts for their planning, most of the time is spent using local historical climate data to understand the concept of probability and the risks that seasonal forecasts modify. Participatory Integrated Climate Services for Agriculture (PICSA), is a structured participatory climate communication, training and planning process, developed by the University of Reading. PICSA combines location-specific climate information with participatory activities including resource mapping, activity calendars and budgeting activities to empower farmers to make better-informed crop, livestock and livelihood decisions. PICSA has a strong focus on using historical climate information to understand risks that are important for agriculture. PICSA includes the use of seasonal forecasts, but this component uses the older tercile forecast convention, and is not as developed. Like the seasonal planning workshop covered in the next section, PICSA makes extensive use of graphs of the local climate. The Participatory Scenario Planning (PSP) process, developed by CARE, brings together local governments and a range of stakeholders to incorporate seasonal forecasts into their planning. PSP accounts for the uncertainty of forecasts by discussing options that would be appropriate for each of the forecast tercile categories. The Farmer Field School approach, which FAO developed initially to support integrated pest management, brings convenes farmers frequently through an agricultural season for training and shared learning. Climate services, and management of climate-related agricultural risk, has been incorporated into farmer field school programs and curriculum in Indonesia, The Philippines, and recently East Africa (Climate Resilient Farmer Field School). Because they go into detail about a range of topics and specific agricultural management issues, these farmer field schools are the most detailed and time-consuming of the participatory communication approaches for climate services. Additional information about these participator processes is available from these publications: • p’Rajom MO, Oroma GW, Osumba J, Recha J. 2020. Climate resilient farmer field schools handbook. CGIAR Research Program on Climate Change, Agri- culture and Food Security. https://ccafs.cgiar.org/resources/ publications/cli- mate-resilient-farmer-field-schools-handbook • Osumba, J.J.L., Recha, J.W., Oroma, G.W. (2021). Transforming Agricultural Extension Service Delivery through Innovative Bottom–Up Climate-Resilient Agribusiness Farmer Field Schools. Sustainability 13, 3938. https:// doi. org/10.3390/su13073938 • Ambani, M., Shikuku, P., Maina, J. W., & Percy, F. (2018). Practical guide to PSP: Participatory Scenario Planning using seasonal forecasts. https://carecli- matechange.org/publications/practical-guide-to- participatory-scenario-planning-seasonal-climate-information-for-resilient-decision-making/ • Dorward, P., Clarkson, G., and Stern, R. (2015). Participatory Integrated Climate Services for Agriculture (PICSA): Field manual. University of Reading. https://hdl.handle.net/10568/68687 119 Ethiopia MODULE 4 Digital and media climate communication channels Agricultural extension services and climate services often use a range of communication channels to reach their clients. Agricultural extension services traditionally emphasized face-to-face communication processes (e.g., farm visits, training visits and workshops, farmer field schools) and broadcast media. In response to expanding mobile phone coverage and cost considerations, agricultural extension services increasingly use digital communication channels. Broadcast media In much of rural Africa, rural radio remains the communication channel that offers the greatest reach at the lowest cost. It is used widely and effectively for weather information, and for awareness and education campaigns around agricultural innovations that are widely applicable for farmers in a given region. Television programming can be more engaging and informative, but is generally less available to smallholder farmers. Broadcast media allow information to be disseminated without local presence of extension agents or meteorologists, but offer limited opportunities for responding to demand or incorporating feedback from users. For reasons that were already discussed, this is an advantage for time-sensitive weather forecasts and extreme weather alerts, but may be a disadvantage for information at a climate time scale. Although broadcast media (radio, and in some regions television) is generally a one-way dissemination channel, several innovations can make communication more interactive, and hence better able to engage farmers and meet their context-specific needs. For example, creative programming that involves farmers and experts in village dialogs, radio drama and reality television programs such as “Shamba Shapeup” in Kenya, helps farmers relate new information and concepts to their experience. Community listening groups combine the reach of radio with the benefits of group interaction and social learning. Radio listen groups can be effective at fostering change in farm practices, and have been part of several efforts to support rural climate services in Africa. Mobile phone channels New digital services that use mobile phones have potential to address both the limitations of broadcast media and the resource and scaling challenges of traditional face-to-face communication channels. If information is available about farmers’ locations and characteristics, it is possible to tailor climate and extension messages to their needs. Interactive mobile phone channels exist that enable on-demand access to content and two-way information flow. Mobile phones are used in several ways. Broadcasting messages to all subscribing farmers, in the form of short text (SMS) or audio messages, is an efficient but supply- driven communication channel. This can be an efficient way to disseminate time-sensitive information that is relevant to many individuals, such as daily weather forecasts or emergency warnings of extreme weather conditions. More demand-responsive mobile phone strategies allow users to request SMS or audio messages from a menu of options, through interactive voice response (IVR) or Unstructured Supplementary Services Data (USSD) menu applications. Call centers that allow a user to talk with an expert are the most demand-driven but also the most expensive mobile phone communication strategy. Internet tools and digital applications For farmers who have access to data services, and either smartphones or computers, digital apps and web-based tools offer powerful and flexible options to access climate- related information. Digital apps and web-based tools can provide information that is tailored to particular decision makers and locations, and in multiple formats (text, visual, audio). In Ethiopia, a growing set of climate-related internet tools and apps target farmers and agricultural extension providers. Relevant examples include: • EMI’s Climate Maprooms (discussed in Module 2); Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 120 • Ethiopian Digital AgroClimate Advisory Platform (EDACaP) 1 is an information and decision support platform that aims to strengthen farmers’resilience through agro-climate advisories that digitally integrate climate, soil, crop and agronomic data. Advisories are delivered to Development Agents and subscribing farmers through SMS, IVRS and radio in local languages. • Ethiopia’s national Ag-Data Hub 2 is a system that brings together data from multiple sources in one place to support agricultural decision making. The system combines data from previously independent systems from the Ministry of Agriculture, national agricultural research and extension systems, CGIAR centers, and other organizations. • Ethiopia’s Agricultural Stress Index System (“ASIS”), also known as the NextGen Agricultural Drought Monitoring and Warning System (NADMWS), enables users to monitor agricultural areas or “hotspots” with a high likelihood of water stress at the national, regional, zonal and woreda (district) levels. The system aims to automate the analyses that an expert in remote sensing would undertake, and simplifies the interpretation and use of the data for users who are not remote sensing experts. Extension videos Short training videos by lead farmers or extension workers can be an effective way to share concepts and information with farmers. Short videos can be shared with groups of farmers, using for example a portable projector; made available on demand on computer monitors or projectors at local government offices, training centers or other community facilities; or played on smartphones, tablets or laptop computers. Institutional networks Institutions represent a third type of communication channel, in addition to group participatory processes and digital communication channels. In Ethiopia, as in most African countries, farmers and other local stakeholders access climate other types information through many different institutions. These institutions play differing roles in the generation, translation, transfer and use of climate information. It might not be obvious to an end user whether an institution that provides them with information generated that information, or whether it is transferring information that they obtained from another institution. In Ethiopia, a complex network of governmental, non-governmental, civic society, faith-based organizations, and international institutions are involved in some aspect of agricultural climate services ((Tesfaye et al., 2020)). More organizations are involved in translation and communication than in generation of climate information. Among government organizations, the different levels of government (federal, regional, zonal, woreda, kebele) add to the complexity of climate service networks. Information that is produced at the federal or regional level often doesn’t reach farming communities or local government. Although some of the institutions that are involved in agricultural climate services in Ethiopia provide mandated services on a sustained basis, many are involved in climate service activities through time-bound projects. Projects can increase farmers’ access to key institutions, during the project lifecycle. As a result, the institutions that interact with farmers vary in space and time. Farmers tend to have greater awareness, access and capacity to act on climate information in locations where climate service projects have been active. The diversity of institutions involved in climate services presents both opportunities and challenges. In terms of challenges, when information passes through a network of institutions, there is a risk that the information will be distorted. Loss of information about the probabilistic 1 https://dev.edacap.scio.services/#/Forecast 2 https://datahub.moa.gov.et/ 121 Ethiopia MODULE 4 nature and inherent uncertainty of the information is a particular concern in the case of climate information. Furthermore, the diversity of institutions raises the risk that farmers will receive conflicting information from different institutions. This is particularly challenging when the source and the quality of the information is not communicated, or when farmers are given conflicting management recommendations. The diversity of institutions also presents opportunities. First, institutions that farmers already know and trust may be very effective at introducing new types of information, and building farmers’ confidence and capacity to act on that information. Trust in a messenger is an important requirement for acceptance and use of information. Second, women’s’ organizations and other non-governmental organizations that serve disadvantaged groups may be well positioned to reach those groups of farmers with climate services and strengthen their capacity to understand and act effectively on the information. Third, agricultural extension and advisory service providers, and agribusiness organizations have the potential to bundle climate information with other types of information or services in a manner that exploits their synergies and increases uptake. Finally, institutions who work with farmers have the potential to amplify farmers’ influence on providers of information and services. Combining communication channels to support climate services The various communication channels that are available in a given region play different roles in support of rural climate service. Examples of climate service activities include: • raising awareness of available services; • “climate literacy:” educating farmers about climate risks and their management; • adapting farming systems to the local climate based on historical data; • forecast-based planning for the next agricultural season; • accessing daily weather monitoring and forecasts; • disseminating weather warnings; • providing customized information, advisories and decision support tools; • bringing farmers’ feedback into the co-production of climate services: and • supporting extension workers’ access to climate information, advisories and decision support tools. Some communication channels are better suited than others to support these particular activities. For example, videos, and radio and television programs are well-suited to building awareness, reinforcing concepts introduced through participatory processes, and delivering frequent forecasts and advisories at a weather time scale. Mobile phones can push location-specific weather forecasts and alerts of extreme events as SMS or voice messages; and provide targeted information on demand. Group processes are effective for initial learning and seasonal planning around complex information at a climate time scale, and provide an opportunity for the learning and feedback that are needed to design effective services. For a given context, table can be used to represent climate service communication functions (rows), and available communication channels (columns). The cells can be used to identify which communication channels are best suited to each climate service function. The example in Table 4.2 suggests how suitable a particular set of communication channels might be to a particular set of climate service functions. However, the climate service activities and the communication channels available to support them may be different in different locations. Table 4.2-1. Suitability of different communication channels for climate service communication functions (1 = not useful, 2 = somewhat useful, 3 = very useful). Adapted from Hansen et al., (2019). Climate services can support African farmers’ context-specific adaptation needs at scale. Frontiers in Sustainable Food Systems 3:21. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 122 Function Communication Channel Group Meetings Broadcast Media Mobile Phone Internet, Smart Phone Awareness of services 2 3 1 2 Climate literacy 3 2 1 2 Adapting to local climatology 3 1 1 2 Forecast-based seasonal planning 3 2 1 2 Weather monitoring and forecasts 1 3 3 2 Extreme weather alerts 1 2 3 2 Customized information, advisories 2 2 3 3 Farmer feedback 3 1 2 3 Provide information, support to DAs 1 1 1 3 SECTION 4.2 Farmer Participatory Seasonal Forecast Training and Planning Workshop Farmers can benefit more from seasonal climate forecasts if they understand what they mean for their own local climate. In this session, we will learn how to lead a group of farmers in a seasonal forecast training, presentation and planning workshop. This process uses historical climate graphs, and the probability-of- exceedance format for seasonal forecasts, which is available through the EMI Flexible Forecast Maproom. This section draws on Hansen et al., 2022 and provides a summary of the workshop process. Appendix 1 provides detailed step-by-step instructions for facilitating the workshop. By the end of this section, you will be able to: • Understand the steps involved in a participatory seasonal forecast communication and planning workshop. • Lead farmers in a basic participatory seasonal forecast communication and planning workshop. Rationale and overview The workshop is designed to equip farmers to interpret seasonal forecasts in the context of the variability of their local climate, and incorporate this information into their seasonal planning decisions. It uses probabilistic seasonal forecasts expressed as probability-of-exceedance graphs, which are available in the EMI Flexible Forecast Maproom. The rationale for using this format with farmers is discussed in Section 3.7. The presentation below draws from Hansen et al. (2022). The workshop strategy is based on a few assumptions about participating farmers. First, farmers understand seasonal climate variability and resulting risk, and factor this information into their agricultural and livelihood management decisions. Their understanding of climate risk is based on their memory of past seasonal climate conditions and their impacts on their own farms, and not necessarily on statistical or graphical descriptions of variability or risk. Second, although farmers understand climate risk, they are susceptible to cognitive biases and decision errors that research shows affect most people when they make decisions based on probabilistic information. Third, with simple training, farmers can understand information presented visually in graphs. Fourth, farmers have a basic level of numeracy (i.e., understanding of numbers, counting, proportions). The workshop does not require literacy, but the process would need to be adapted when participants cannot read. The workshop uses several approaches to strengthen farmers’ capacity to understand probabilistic seasonal climate forecasts and incorporate it into their decisions. First, relating climate information to farmers’ 123 Ethiopia MODULE 4 experience helps them connect their analytical and experiential cognitive processing modes. As discussed in Section 4.2, people are more likely to use new, probabilistic climate information, and less likely to make significant errors in interpreting and acting on this information, if they use both their analytical and experiential cognitive processing systems. Second, the workshop presents climate in terms of natural frequencies associated with historical variability, before it introduces the probability of future seasonal climate conditions. Research shows that some of the widespread errors that people make interpreting and acting on probabilistic information are reduced or eliminated if the information is expressed in equivalent natural frequencies, even though historical frequencies are used to estimate future probabilities (Gigerenzer and Hoffrage, 1995). Fourth, like other participatory climate communication processes, the workshop incorporates group discussion to take advantage of farmer-to-farmer social learning. Within a rural community, a few farmers typically understand new information or adopt new innovations more quickly than others. These lead farmers who quickly understand the new climate information products and their implication for farm management decisions can help their peers learn. Workshop process The six steps in the seasonal forecast workshop are summarized in Table 4.3-1 and detailed in the rest of this section. The workshop process follows a logical progression that starts with farmers’ memory of climate variability, and leads them through a step-by-step process to interpret the forecast in the context of their local climate, and apply it to their routine seasonal planning decisions. It seeks to introduce complex new information and concepts in small incremental steps (Figure 4.3-1), including participatory graphing activities for farmer groups who have never been exposed to graphs. Table 4.3-1. Summary of steps in farmer seasonal forecast training and planning workshop. Adapted from Hansen et al. (2022). Step Purpose Process Options 1. Purpose, Shape participants’ Present workshop purpose Discuss indigenous climate concepts expectations. Define key terms and concepts indicators to build trust. 2. From memory Relate time series data Elicit recent growing season Participatory activity drawing to variability and graphs to participants’ conditions. time series graph, to scale, for experience. Introduce time series graph, elicited years. validate against collective memory. Calculate probability of exceeding thresholds from time series graph. 3. From variability Understand relationship Sort time series into frequency of Participatory activity sorting to probability between variability and exceedance graph. recent time series into frequency probability, and interpret POE Re-define y-axis from relative of exceedance, to scale. graph. frequency to probability, and   discuss relationship between past frequency and future probability. Practice reading probability of experiencing above or below a threshold. 4. Forecasts shift Understand forecast Show POE curve shifted to the Highlight El Niño years in time probabilities as a shifted probability right and/or to the left of the series graph, and show POE distribution. climatological distribution, and graph for El Niño years alongside discuss its implication for climate all years. and for farming. Identify a familiar location with Practice reading shifts in a wetter or drier climate, and probability of exceeding a discuss how that climate would threshold. affect farm performance and management. 5. Current Present forecast for upcoming Present current forecast and other Enlist NMS staff to discuss forecast growing season. information relevant to planning. the forecast and local climate, Review forecast interpretation. but only if trained in Flexible Respond to any questions. Forecast communication. 6. Farm planning Facilitate discussion of farm Present framing questions   management plans for Discuss management options in upcoming season breakout groups. Present and discuss group plans to plenary. Address farmer questions and needs for additional support. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 124 Memory of past Time series Frequency of Probability Forecasts Shift Interpret and use conditions graph past conditions graph probabilities forecast Figure 4.3-1. Sequence used to train farmers to interpret and use Flexible Forecast graph. Step 1: Introduce the workshop purpose and key concepts Step 1 introduces the purpose and roadmap of the workshop, and explains key concepts including: weather, climate, variability, frequency, uncertainty, probability, and forecast. Several concepts are crucial to understanding and using probabilistic climate information appropriately, it is a good idea to discuss the concepts with farmers at the beginning, and perhaps leave a short definition or description available as a reminder (for participants who can read) during training and planning meetings. First, it is important to explain the time scale of seasonal forecasts. In many places, farmers and the general public think about long-term climate change when they encounter the concept of “climate.” In other places, farmers may have weather forecasts in mind when seasonal forecasting is introduced. If your farmers do not already have a good understanding of climate variability and seasonal climate forecasts, then discuss these concepts: weather (atmospheric conditions at a particular time and place), climate (statistics such as average, seasonal cycles, variability, trends over longer time periods), climate variability (how seasons differ from year to year), and climate change (changes in things like average temperature and average rainfall over at least several decades driven in part by changes in the atmosphere due to human activity). Second, define and explain these concepts: variability (how climate conditions differed in past years), frequency (expresses variability with numbers), uncertainty (Because the climate has been variable in the past, I am uncertain about what the weather will be like in next growing season.), probability (expresses uncertainty with numbers), and forecast (or prediction). A key message is that a forecast is new information that changes the probabilities about the future. A forecast reduces uncertainty, but doesn’t eliminate it completely. Since the workshop will use these concepts in ways that might be new to the community, either because their language does not have an equivalent word, or the closest word has a different meaning within the farming community, it is important to agree on what words best represent each concept, and on the meaning in the context of climate information. When translating into the closest word or phrase in the local language, are farmers likely to have a different meaning in mind for that word? Does each word fully convey the concept, or do you need to define how we (climate service providers, communicators or facilitators) are using the words? In climate services, we often use words that have a particular technical meaning that is different from how the general public might use that word. To avoid misunderstanding, it is important to explain and agree with participating farms what we mean when we use key terms and concepts. It is important to discuss key concepts, especially if the local language does not have words for some of the key concepts. Two additional topics may help prepare participants for the remainder of the workshop, if time permits. Traditional climate knowledge. Recognizing and discussing the community’s traditional climate indicators can help demonstrate respect for farmers’ knowledge, build trust, and foster two-way discussion. In most cultures, farmers, or other members of the community such as traditional forecasters, have identified methods to predict what the future season might be like or what the weather will be. Since most farming communities make some use of traditional climate knowledge and indicators, it can be useful to discuss how those indicators influence their farming decisions, and whether they are 100% reliable or have some uncertainty. Forecast and decision analogies. You could discuss examples of ways that participants use uncertain information outside of farming, and how new information that shifts the probabilities is a form of forecast. Use examples that are relevant to the farmers that you work with. One illustration that might work in some contexts is betting which team will win a sporting event, such as football. Past records of wins and losses against a particular team (or similar teams) give an idea of the probability that your favorite team will win the next game. Suppose you learn that the star forward on your team (or the opposing team) is injured and can’t play. This new information provides a forecast; it changes the probability that the team will win the next game. 125 Ethiopia MODULE 4 Step 2: Understand past variability Step 2 is designed to help farmers relate time series graphs to their collective experience. The process starts with eliciting participants’ collective memory of growing season rainfall and resulting agricultural performance for the five most recent years. Five years is suggested because the period is recent enough for at least some farmers to remember, and because it is easy to convert number of years out of five into percent. If participants are not already familiar with graphs, involving them in a participatory activity to constructing a time series graph, to scale, for the set of years they discussed fosters understanding and confidence. Once they are comfortable with the format, a computer-generated graph with the full set of available historical data is presented and discussed. The discussion of the time series graph should highlight three key issues. First, the graph should represent their experience with seasonal rainfall for the specified growing season. Does the graph seem to be consistent with the way they classified each of the past five years? Does it show years that they remember as unusually wet or unusually dry? Second, the time series graph is a way to summarize information about past climate variability. Does the graph show that seasonal rainfall is more variable or less variable than they expected? Does it seem to show any pattern (e.g., increasing or decreasing trend), or just random variability? Third, the time series graph can be used to find what percent of years (or how many out of ten) had more than or less than a particular threshold amount. If at least some of the participants are familiar with division and percentage, it is useful to lead them to calculate the percent of years in which seasonal rainfall exceeded a meaningful threshold amount. This exercise will help them interpret the probability-of-exceedance graph in Step 3, and recognize its usefulness. Step 3: Introduce the probability of exceedance graph Step 3 introduces the probability-of-exceedance (POE) graph and helps participants to interpret it. However, it is first presented in frequency terms, as a way to describe historical variability. Similar to Step 2, involving farmers in a participatory graphing activity can help farmers understand the POE graph, and how it is related to the time series graph. This involves participants in sorting recent growing season rainfall totals onto a blank POE graph, with frequency rather than probability on the vertical axis. The vertical axis is then changed from number of years to percentage of years (or number of years out of ten) with at least a given amount of seasonal rainfall. Once they are comfortable with the format, a computer-generated graph with the full set of available historical data is presented and discussed. At this point, the facilitator discusses how relative frequency of past variations relates to probabilities in the future. Even through relative frequency and probability are equivalent mathematically, the use of past relative frequency to estimate probability for the next growing season might be a new concept for participating farmers. Participants will learn to find the probability of experiencing rainfall above, and then below, some threshold that they agree would be relevant to farm decisions. If they calculated the percent of years when seasonal rainfall exceeded a particular threshold in Step 2, they should recognize how much easier it is to find these probabilities from the POE graph. It is important to emphasize that the POE graph contains the same information as the time series graph, just arranged in a way that makes it easier to find probabilities. Step 4: Understand a seasonal forecast as a shift of the historical probability distribution Step 4 presents a forecast as new information that shifts the historical probability distribution, and equips participants to interpret a forecast in POE format. Two activities are used to reinforce this idea. First, where farmers are familiar with El Niña and/or La Niña, the shifted distribution can be obtained by showing a POE graph for El Niño alongside the climatological distribution, and the corresponding time series graph with El Niño years highlighted. A simplified explanation of what meteorologists mean when they refer to El Niño (i.e., warmer than normal eastern equatorial Pacific Ocean temperatures), and how it can influence climate conditions in other parts of the world, can help build confidence. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 126 The second activity involves identifying familiar location with a wetter (or dryer) climate, and discussing how that climate differs from the climate at the farmers’ location. If participants interpret the probability of exceedance graph correctly, they should expect the POE curve for the wetter location will be shifted to the right of their location. If they use a dryer location, it will be shifted to the left. Optionally, the time series and POE graphs both locations (participating farmers’ location, and the wetter or dryer location. Discussing what it would be like to have their farm, but with the climate of the selected wetter (or dryer) location, will help them relate a hypothetical forecast to their farming experience. To help participants interpret a shift in the POE towards the right (i.e., probability shift wards wetter conditions) or left (i.e., probability shift towards dryer conditions), they repeat the activity of finding the probability of experiencing rainfall above or below the same threshold. This can be done using either the POE curve for El Niña (or La Niña) years for their location, or for the historical POE curve for a specified location with a wetter (or dryer) climate. Step 5: Present the seasonal forecast Steps 1–4 equip farmers to understand and trust the seasonal forecast. The way Steps 5 and 6 are implemented depends on whether the workshop occurs just before the start of the rainfed growing season, or as a training activity during a fallow period long before the growing season: • If the workshop occurs just before the season, the focus in steps 5 and 6 shift from training to planning for the upcoming season based on the current forecast. • If the workshop occurs long before the start of the growing season, steps 5 and 6 use a forecast from a previous year or a hypothetical forecast to prepare participants to understand the forecast and use it for planning for the next growing season when it becomes available. In this case, farmers would participate in a short workshop that covers Steps 5-6 once the forecast for the next season becomes available. When the current or prior-year forecast is presented, participants should discuss its interpretation. Explanation of the seasonal forecast should include the brief explanation about the context and implications of the forecast that EMI includes in its seasonal forecast bullets. Participants should understand: • what institution (i.e., EMI) produced the forecast; • how seasonal forecasts are produced; • what variable(s) and time period the forecast covers; and • how to interpret the forecast as a shift in probabilities. If the forecast includes additional seasonal variables (e.g., season onset and cessation dates, probabilities of dry spells, growing degree-days), these forecasts can be presented in the same format and discussed. To reinforce interpretation, participants will identify the probabilities of experiencing more or less than a threshold seasonal rainfall based on the historical and on the forecast POE curves, using the same threshold value that was used in Steps 2-4. 127 Ethiopia MODULE 4 Step 6: Plan how to adjust farm management based on the forecast The forecast presentation leads into Step 6, farm planning, when farmers discuss and decide what seasonal management decisions, if any, they will change based on the seasonal forecast. If the workshop occurs as a training activity long before the season, then they discuss what seasonal management decisions, if any, they would change if the forecast for the next season were to look like the forecast presented in Step 5. Group discussion allows participants to learn from other farmers, and get feedback on their ideas. This can be done either in the large group, or in breakout groups that then summarize their plans to the larger group. A given seasonal forecast may or may not lead farmers to change their management plans. Whether they choose to adjust their plans for the upcoming season, and the kinds of adjustments that they make, may depend on how much the forecast shifts probabilities that are important for decisions, and whether options are available that would be better suited than existing plans to the forecast conditions. They may agree as a group to change some of their plans, while some decisions might be different for different farmers. While Development Agents and any other professionals may provide information about management options and answer questions, the process aims to support farmers’ decision making. Refer to Appendix 1 for step-by-step guidelines for facilitating the workshop. Practical considerations Timing The full forecast training, communication and planning workshop is needed the first time farmers are exposed to the downscaled forecast in the Flexible Forecast format. The entire process would typically take about 8 to 12 hours, spread over two days. Farmers and extension workers tend to be quite busy shortly before the start of the growing season. Furthermore, there may be a short time between when EMI releases a seasonal forecast, and when farmers must purchase seed and other inputs and prepare their fields for planting. Because of these timing constraints, it might be desirable to conduct the training (steps 1-4) during a slow part of the agricultural calendar, and hold a shorter workshop to present the actual forecast and adjust plans for the upcoming season (steps 5-6). If the process is separated into a training workshop and a forecast presentation and planning workshop, then a forecast from a previous year should be presented, and once farmers have been trained are familiar with the format of the forecast, and understand how it relates to historical variability, they will only require a short workshop to: • briefly review what they have learned in steps 1-4; • present the current seasonal forecast (step 5); and • plan how to adjust farm management based on the forecast (step 6). Without the training (steps 1-4), a seasonal forecast presentation and planning workshop would typically take about 2-3 hours. Figure 4.3-2 summarizes these options. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 128 Figure 4.3-2. Options for scheduling seasonal forecast training, presentation and planning workshops. Group size and composition The size of a workshop should generally be limited to about 25-30 farmers. If the group is too large, it will be difficult to involve all participants in discussion and participatory activities, and to assess whether all participants understand the concepts. If farmers are already organized in groups, those existing groups should be used as much as possible. In some cultures, and contexts, women farmers might be reluctant to participate fully in a mixed group. Consider local culture, your experience with mixed vs. single sex groups, and any relevant guidelines when deciding whether to hold separate workshops for men and women, and whether female facilitators are needed to ensure that women participate freely and equitably. Adapting to different farmer backgrounds The workshop process can be adapted to farmers with differing education and literacy levels. First, if literacy rates are low, then more discussion and repetition may be required. Pictures could be used instead of text to identify graph components (e.g., axis labels) and to characterize farmers’ memory of recent years. Second, if participants have been exposed to graphs in school or other training, then the participatory activities deriving time series and POE graphs could probably be eliminated and overall time reduced. While past pilot training workshops took 12–14 hours over two days, it may be feasible to reduce them to a single day if the majority of farmers are literate and already comfortable with graphs. Third, if participants are unfamiliar with percentages, then probabilities could be expressed as number of years out of ten. 129 Ethiopia MODULE 4 References Cited References Cited Amegnaglo, C.J., Anaman, K.A., Mensah-Bonsu, A., Onumah, E.E., Amoussouga Gero, F., 2017. Contingent valuation study of the benefits of seasonal climate forecasts for maize farmers in the Republic of Benin, West Africa. Climate Services 6, 1–11. https://doi.org/10.1016/j.cliser.2017.06.007 Antwi-Agyei, P., Amanor, K., Hogarh, J.N., Dougill, A.J., 2021. Predictors of access to and willingness to pay for climate information services in north-eastern Ghana: A gendered perspective. Environmental Development 37, 100580. https://doi.org/10.1016/j.envdev.2020.100580 Barrett, C.B., 2005. Rural poverty dynamics: development policy implications. Agricultural Economics 32, 45–60. https://doi.org/10.1111/j.0169-5150.2004.00013.x Barrett, C.B., Ghezzi-Kopel, K., Hoddinott, J., Homami, N., Tennant, E., Upton, J., Wu, T., 2021. A scoping review of the development resilience literature: Theory, methods and evidence. World Development 146, 105612. https:// doi.org/10.1016/j.worlddev.2021.105612 Brouwer, J., Fussell, L.K., Herrmann, L., 1993. Soil and crop growth micro-variability in the West African semi- arid tropics: a possible risk-reducing factor for subsistence farmers. Agriculture, Ecosystems & Environment 45, 229–238. https://doi.org/10.1016/0167-8809(93)90073-X Donkoh, S.A., 2019. Farmers’ willingness-to-pay for weather information through mobile phones in northern Ghana. Ghana Journal of Science, Technology and Development 6, 19–36. https://doi.org/10.47881/166.967x Galasso, E., Wagstaff, A., 2018. The Aggregate Income Losses from Childhood Stunting and the Returns to a Nutrition Intervention Aimed at Reducing Stunting. The World Bank 38. Gigerenzer, G., Hoffrage, U., 1995. How to improve Bayesian reasoning without instruction: Frequency formats. Psychological Review 102, 684–704. https://doi.org/10.1037/0033-295X.102.4.684 Gumucio, T., Hansen, J.W., Huyer, S., van Huysen, T., 2020. Gender-responsive rural climate services: a review of the literature. Climate and Development 12, 241–254. https://doi.org/10.1080/17565529.2019.1613216 Han, E., Baethgen, W.E., Ines, A.V.M., Mer, F., Souza, J.S., Berterretche, M., Atunez, G., Barreira, C., 2019. SIMAGRI: An agro-climate decision support tool. Computers and Electronics in Agriculture, BigData and DSS in Agriculture 161, 241–251. https://doi.org/10.1016/j.compag.2018.06.034 Hansen, J., Hellin, J., Rosenstock, T., Fisher, E., Cairns, J., Stirling, C., Lamanna, C., van Etten, J., Rose, A., Campbell, B., 2019. Climate risk management and rural poverty reduction. Agricultural Systems 172, 28–46. https://doi. org/10.1016/j.agsy.2018.01.019 Hansen, J., List, G., Downs, S., Carr, E.R., Diro, R., Baethgen, W., Kruczkiewicz, A., Braun, M., Furlow, J., Walsh, K., Magima, N., 2022. Impact pathways from climate services to SDG2 (“zero hunger”): A synthesis of evidence. Climate Risk Management 35, 100399. https://doi.org/10.1016/j.crm.2022.100399 Hansen, J.W., Dinku, T., Robertson, A.W., Cousin, R., Trzaska, S., Mason, S.J., 2022. Flexible Forecast Presentation Overcomes Longstanding Obstacles to Using Probabilistic Seasonal Forecasts. Frontiers in Climate. https://doi. org/10.3389/fclim.2022.908661 Hansen, J.W., Mason, S.J., Sun, L., Tall, A., 2011. Review of seasonal climate forecasting for agriculture in sub- Saharan Africa. Ex. Agric. 47, 205–240. https://doi.org/10.1017/S0014479710000876 Hansen, J.W., Mishra, A., Rao, K., Indeje, M., Ngugi, K., 2009. Potential value of GCM-based seasonal rainfall forecasts for maize management in semi-arid Kenya. Agricultural Systems 101, 80–90. https://doi.org/10.1016/j. agsy.2009.03.005 Hansen, J.W., Vaughan, C., Kagabo, D.M., Dinku, T., Carr, E.R., Körner, J., Zougmoré, R.B., 2019. Climate Services Can Support African Farmers’ Context-Specific Adaptation Needs at Scale. Frontiers in Sustainable Food Systems 3, 21. https://doi.org/10.3389/fsufs.2019.00021 Hill, R., Skoufias, E., Maher, B., 2019. The Chronology of a Disaster. World Bank, Washington, DC. https://doi. org/10.1596/31721 131 Ethiopia Hyland, M., Russ, J., 2019. Water as destiny – The long-term impacts of drought in sub-Saharan Africa. World Development 115, 30–45. https://doi.org/10.1016/j.worlddev.2018.11.002 Kabeer, N., 2017. Economic Pathways to Women’s Empowerment and Active Citizenship: What Does The Evidence From Bangladesh Tell Us? The Journal of Development Studies 53, 649–663. https://doi.org/10.1080/00 220388.2016.1205730 Marx, S.M., Weber, E.U., Orlove, B.S., Leiserowitz, A., Krantz, D.H., Roncoli, C., Phillips, J., 2007. Communication and mental processes: Experiential and analytic processing of uncertain climate information. Global Environmental Change 17, 47–58. https://doi.org/10.1016/j.gloenvcha.2006.10.004 Messina, C.D., Hansen, J.W., Hall, A.J., 1999. Land allocation conditioned on El Niño-Southern Oscillation phases in the Pampas of Argentina☆☆Florida Agricultural Experiment Station, Journal Series No. R-06795. Agricultural Systems 60, 197–212. https://doi.org/10.1016/S0308-521X(99)00032-3 Ouédraogo, M., Barry, S., Zougmoré, R.B., Partey, S.T., Somé, L., Baki, G., 2018. Farmers’ Willingness to Pay for Climate Information Services: Evidence from Cowpea and Sesame Producers in Northern Burkina Faso. Sustainability 10, 611. https://doi.org/10.3390/su10030611 Rabbinge, R., 1993. The ecological background of food production, in: Crop Protection and Sustainable Agriculture. John Wiley & Sons, Chichester, UK, pp. 2–29. Tesfaye, A., Hansen, J., Radeny, M., Belay, S., Solomon, D., 2020. Actor roles and networks in agricultural climate services in Ethiopia: a social network analysis. Climate and Development 12, 769–780. https://doi.org/10.1080/17 565529.2019.1691485 Vaughan, C., Hansen, J.W., Roudier, P., Watkiss, P., Carr, E., 2019. Evaluating agricultural weather and climate services in Africa: Evidence, methods, and a learning agenda. WIREs Clim Change 10. https://doi.org/10.1002/ wcc.586 Waha, K., van Wijk, M.T., Fritz, S., See, L., Thornton, P.K., Wichern, J., Herrero, M., 2018. Agricultural diversification as an important strategy for achieving food security in Africa. Global Change Biology 24, 3390–3400. https://doi. org/10.1111/gcb.14158 Zongo, B., Diarra, A., Barbier, B., Zorom, M., Yacouba, H., Dogot, T., 2015. Farmers’ Perception and Willingness to Pay for Climate Information in Burkina Faso. JAS 8, 175. https://doi.org/10.5539/jas.v8n1p175 Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 132 Glossary GLOSSARY absolute humidity A measure of the amount of water vapor in a given parcel of air that is typically measured in mass/volume or volume/volume units. adverse selection The tendency for insurance to be purchased preferentially by farmers who experience more frequent losses. agricultural calendar A planning tool that summarizes the timing of agricultural activities for a particular location. amplitude The difference between the maximum and minimum, the seasonal contrast. analytical processing System the human mind use to process information that is in the form of a statistical description. anemometer Measures wind speed and direction. anomaly The difference between the average or baseline area yield insurance Index insurance based on yields estimated over a specified region, using statistical sampling from many farmers’ fields. atmospheric pressure A measure of the force or weight of the overlying air per unit area on the surface. barometer Measures atmospheric pressure basis risk The risk that insurance payouts will not match losses due to the imperfect relationship between factual losses and the index that triggers payouts. breakeven price Minimum farmgate price needed to break even for a given yield. breakeven yield Minimum yield needed to break even for a given farmgate price. certainty equivalent A guaranteed return that someone would accept, rather than taking a chance on a higher, but uncertain, return in the future. chance node Decision tree symbol representing uncertainty due to multiple probabilistic states of nature. climate The statistics of weather at a given location, over a specified period of time. climate change Changes in the characteristics of climate over long periods of time. It is important that these changes occur not only in the mean conditions but also in the variability. climate data arrays of measurements of weather and climate variables from which information that is more useful in everyday life is distilled or that are ingested in applications such as crop or hydrological models Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 134 climate information Structured data describing past, current, or future climate climate-sensitive A decision is climate sensitive if a decision maker would select different options decision under different weather or climate conditions. co-production A process that involves climate information generators and users engaging throughout the production of climate services coefficient of A statistical measure of the relative dispersion of data points in a data series around variation the mean communication channel Mediums through which you can send a message to its intended audience constraint A shortage of some resource that prevents a decision maker from implementing a desired option. contract window The specific range of dates when insurance covers a particular risk. crop simulation moddel A simulation model that describes processes of crop growth and development as a function of weather conditions, soil conditions, and crop management. decision node Decision tree symbol representing a decision among multiple options. decision support tool A tool or information system that supports organizational decision- making activities decision tree A decision support tool that uses a tree-like model of decisions and their possible consequences, including chance event outcomes, resource costs, and utility. dynamical methods The use of models based on physical equations that govern interactions between various components of the climatic system, called General Circulation Models, and similar to the models used in Numerical Weather Prediction efficient set the set of optimal portfolios that offer the highest expected return for a defined level of risk or the lowest risk for a given level of expected return ENACTS The Enhancing National Climate Services initiative, or ENACTS, is led by the International Research Institute for Climate and Society and is a unique, multi- faceted initiative designed to bring climate knowledge into national decision making by improving availability, access to, and use of climate information. end node Decision tree symbol representing the consequence of a decision option, typically in monetary terms. enterprise the production and marketing of a single crop or livestock commodity that actually produces a marketable product enterprise budget is a useful tool for estimating the profitability of a farm enterprise, and for comparing the profitability of alternative ways of managing farm enterprises exit A pre-identified threshold of an insurance index that triggers the maximum payout amount. 135 Ethiopia expected return the anticipated amount of returns that a portfolio may generate, making it the mean (average) of the portfolio’s possible return distribution expected value A predicted value of a variable, calculated as the sum of all possible values each multiplied by the probability of its occurrence. experiential processing System the human mind uses to process information that is obtained through repeated experience. fixed cost Also known as indirect costs or overhead costs, fixed costs are business expenses that are not dependent on the level of goods or services produced by the business. forecast / prediction The process of making predictions based on past and present data and most commonly by analysis of trends frequency The rate at which something occurs or is repeated over a particular period of time or in a given sample. global warming Also known as climate change, a gradual increase in the overall temperature of the earth’s atmosphere generally attributed to the greenhouse effect caused by increased levels of carbon dioxide, chlorofluorocarbons, and other pollutants. gridded data Partitions Earth’s Surface in small squares and attributes a value of a weather/ climate variable to each square. gross margin Refers to the net return, or profit, that a farmer would make on an enterprise in a given year, ignoring fixed costs. It is calculated as gross receipts from the sale of the crop or livestock enterprise, minus the cost of production. gross receipts The total amounts received from all sources during a year humidity Measures how much water vapor is in the air at a given time. indemnity insurance Insurance that pays out based on a verified loss. index-based insurance Insurance that pays out based on the value of data that are used as a proxy for a particular loss, rather than actual losses. index-based livestock Insurance for livestock based on livestock mortality data estimated over a specific insurance region, or on vegetation remote sensing data as a proxy for livestock mortality. insurance payout The amount of money that insurance pays a client when contractual conditions (i.e., a verified loss or an index exceeding a specified threshold) ocurrs. insurance premium Contractual amount of money that a client pays periodically for insurance. lead time The amount of time that is needed between making a decision and implementing the decision. Maproom Interactive online climate information products Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 136 mean Average median The median number is the number with the middle value of the set when arranged in ascending order. merged data data are constructed datasets based on a blend of satellite and station information. model data A collection of equations capturing physical and chemical laws governing the weather and climate system moral hazard The incentive for farmers to neglect good risk management in order to receive payouts. normal distribution a function that represents the distribution of many random variables as a symmetrical bell-shaped graph. oceanic data data are based on observations of oceanic properties (temperature and velocity are of particular climate/weather relevance). participatory process A facilitated communication or planning process in which all participants play an active role. Partitioning Allocation of new dry matter, produced by photosynthesis, among different plant parts. Phenology The study of cyclic and seasonal natural phenomena, especially in relation to climate and plant and animal life. Photosynthesis The process by which green plants and some other organisms use sunlight to synthesize foods from carbon dioxide and water. Photosynthesis in plants generally involves the green pigment chlorophyll and generates oxygen as a byproduct. precipitation Water that falls to the ground as rain, snow or hail. Water is critical for many economic activities, from domestic use to hydropower generation; a critical limiting factor for plants growth in the tropics; probability Probability of an event occurring is the chance that that event will occur relative to all possible outcomes. probability density A function whose value at any given sample in the sample space can be interpreted function (PDF) as providing a relative likelihood that the value of the random variable would equal that sample. probability of A graph that allows one to see the probability that seasonal climate conditions will exceedance graph fall above or below any threshold probability of The probability that a certain value will be exceeded or not exceeded exceedance/non- exceedance rain gauge A plastic or metal receptacle with graduated markings denoting the depth of accumulated precipitation to measure rainfall. 137 Ethiopia reanalysis data data are based on a mix of observational inputs from the above forms of data and dynamical models of how the atmosphere is expected to behave. relative humidity A measure of the amount of water vapor in the air at a given time relative to the saturation level (the maximum amount of water vapor a parcel of air can retain). It is typically expressed in percentages (of saturation). respiration The process of respiration in plants involves using the sugars produced during photosynthesis plus oxygen to produce energy for plant growth return/payoff Profit risk A situation involving exposure to danger or loss risk analysis Inolves analyzing management options under uncertainty risk efficiency Analysis used to evaluate management options based on a few general analysis/criteria assumptions, for a group of decision makers who may have different risk preferences. risk aversion the tendency of people to prefer outcomes with low uncertainty to those outcomes with high uncertainty risk premium A measure of excess return that is required by an individual to compensate them for being subjected to an increased level of risk risky prospect An outcome associated with high risk satellite data special class of gridded data. Satellites can observe vast swaths of Earth and atmosphere, even in remote, hard to access oceanic areas. satellite data data are based on reflected radar signals sent from and received by satellites. seasonality Times of the year when agricultural activities occur and key reoccurring management decisions are made. senescence biological aging thermometer Tool/instrument for measuring temperature. time horizon The amount of time into the future a decision maker considers when evaluating the consequences of a decision option. time series graph Time series graphs are created by plotting an aggregated value (either a count or a statistic, such as sum or average) on a time line. trend A pattern in a set of results displayed on a graph. trigger A pre-identified threshold value of an insurance index that triggers the beginning of a payout. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 138 tropical livestock unit Livestock numbers converted to a common unit. An increased number of animals (TLU) per adult available to support the household, indicates improved food security and household resilience uncertainty A state arising from imperfect or unknown information upper air data data are recorded by dropsondes, radiosondes, weather balloons or aircraft mounted sensors. These data offer important insight into the vertical profile of the atmosphere with respect to temperature, precipitation, pressure, wind and other variables, but the spatial and temporal coverage is quite limited to the flight path of the aircraft, weather balloon or sonde. utility A unitless measure of subjective value placed on different economic returns or levels of wealth variability Liability to vary or change variable cost A cost that varies with the level of output vertical measurements Measurements to assess atmospheric properties at different levels in the atmosphere weather The state of the atmosphere at a given time and place with regard to variables such as temperature, precipitation, humidity, air pressure, wind, cloudiness and sunshine. weather index Insurance based on a meteorological index such as rainfall or temperature data. insurance weather stations A facility, either on land or sea, with instruments and equipment for measuring atmospheric conditions to provide information for weather forecasts and to study the weather and climate. wet-bulb thermometer/ used to infer ambient humidity. hygrometer 139 Ethiopia Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 140 Appendix 1 Seasonal Forecast Participatory Workshop In this activity, you will learn how to lead a group of farmers in a seasonal forecast training, presentation and planning workshop. This process uses historical climate graphs, and the probability-of-exceedance format for seasonal forecasts, which is available through the EMI Flexible Forecast Maproom. You will divide into groups of 5-6roup members. Group members will take turns playing the role of farmers. Make sure that all group members are actively involved in leading parts of the workshop. Overview The workshop is designed to equip farmers to interpret seasonal forecasts in the context of the variability of their local climate, and incorporate this information into their seasonal planning decisions. It follows a logical progression that starts with farmers’ memory of climate variability, and leads them through a step-by-step process to interpret the forecast in the context of their local climate, and apply it to their routine seasonal planning decisions. The six steps in the seasonal forecast workshop are summarized in Table 1 and detailed in the rest of this section. See Hansen et al. (2022)1 for more information about the rationale for the Flexible Forecast and workshop process. Table 1. Summary of the six steps in the seasonal forecast workshop. Step Purpose Process 1. Purpose, Shape participants’ expectations. • Present workshop purpose concepts • Define key terms and concepts 2. From memory Relate time series data and graphs to • Elicit memory of recent years to variability participants’ experience. • Participatory time series graph activity • Interpret time series graph 3. From variability Understand relationship between • Participatory probability graph activity to probability variability and probability, and use • Interpret probability graph probability graph to interpret forecast categories. 4. Forecasts shift Understand forecast as a shifted • Show how El Niño shifts the probability distribution probabilities probability distribution. • Use a familiar location to discuss what a wetter or drier climate might look like 5. Current Present forecast for upcoming • Present current forecast and other information relevant forecast growing season. to planning. • Review forecast interpretation. 6. Farm planning Facilitate discussion of farm • Present framing questions. management plans for upcoming • Discuss management options in breakout groups. season • Present and discuss group plans to plenary. • Address farmer questions and needs for additional support. For this activity, we are practicing a workshop that combines all 6 steps, using 1981-2018 data from Melkasa. However, we are using a forecast from a past year (2018) to learn how to lead the process. The purpose is to learn how to lead the process and not to plan for the current year. The full forecast training, communication and planning workshop is needed the first time farmers are exposed to the downscaled forecast in the Flexible Forecast format. As discussed in the presentation and Reference Guide, it can either be done at one time, or split between training (steps 1-4) during a slow part of the agricultural calendar, and a shorter planning workshop (steps 5-6) shortly before the start of the growing season. The entire training and planning process would typically take about 8 to 12 hours, spread over two days. However, for this activity we have only 4-6 hours to practice the process, including discussion time. Materials needed: • Flip chart • Markers • Meter stick • Tape • Printouts (one large copy for the group, and one A4-size copy per participant): 1 Hansen JW, Dinku T, Robertson AW, Cousin R, Trzaska S and Mason SJ (2022) Flexible forecast presentation overcomes long- standing obstacles to using probabilistic seasonal forecasts. Front. Clim. 4:908661. doi: 10.3389/fclim.2022.908661 Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 142 • seasonal rainfall historical time series graph • seasonal rainfall probability of exceedance graph • seasonal rainfall historical time series graph with El Nino highlighted 143 Ethiopia • seasonal rainfall probability of exceedance graph with El Nino highlighted • seasonal rainfall forecast graph • On a flipchart or poster-sized paper, blank table with three columns (Year, Crops and Rainfall) and five rows. Leave extra space in the Rainfall column because you will later add measured July-September rainfall amounts. Year Crops Rainfall 2022 2021 2020 2019 2018 Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 144 • On a flipchart or poster-sized paper, blank graph with Year as the x axis, and July-September rainfall as the y axis: • On a flipchart or poster-sized paper, blank graph with July-September rainfall as the x axis. The y axis will have two labels as shown below: “Years with at least this much rain” “Probability of at least this much rain” Step 1: Introduce the worskshop purpose and key concepts This step involves explaining the purpose of the workshop, and making sure participants understand and agree about the meaning of key concepts such as climate and how it differs from weather, variability, frequency, uncertainty, probability and forecast. Workshop purpose Explain that the purpose of this workshop is to talk with farmers about climate variability, about forecasts for the next rainy season, and about how this type of information might be useful for farm management decisions. Key concepts Several concepts are crucial to understanding and using probabilistic climate information appropriately, it is a good idea to discuss the concepts with farmers at the beginning, and perhaps leave a short definition or description available as a reminder (for participants who can read) during training and planning meetings. 145 Ethiopia When translating into the closest word or phrase in the local language, are farmers likely to have a different meaning in mind for that word? Does the selected word or phrase fully convey the concept, or do you need to define how we (climate service providers, communicators or facilitators) are using the words? To avoid misunderstanding, it is important to explain and agree with participating farmers what we mean when we use key terms and concepts. • First, explain the time scale of seasonal forecasts. Do your farmers think about long-term climate change when they encounter the concept of “climate?” Do your farmers have weather forecasts in mind when seasonal forecasting is introduced. If your farmers do not already have a good understanding of climate variability and seasonal climate forecasts, then discuss these concepts. • Second, define and explain these concepts: • Variability refers to the fact that climate conditions have been different in different years in the past. For example, growing season rainfall in 2022 was different from rainfall in 2021, which was different from rainfall in 2020. • Frequency expresses variability with numbers. For example, in two out of the past ten years I was not able to produce enough crops to feed my family until the next harvest. • Uncertainty deals with what will happen in the future. Because the climate has been variable in the past, I am uncertain about what the weather will be like in next growing season. • Probability expresses uncertainty with numbers. For example, there are two chances in five that I will not produce enough crops to feed my family until the next harvest. • Forecast. A forecast is new information that changes the probabilities about the future. A forecast reduces uncertainty about the future, but doesn’t eliminate it completely. Explain that the workshop will show farmers how to use probability and graphs to describe past climate variability, and to understand what a seasonal climate forecast means for their local climate. • Optional: Forecast and decision analogies. You could discuss examples of ways that participants use uncertain information outside of farming, and how new information that shifts the probabilities is a form of forecast. Use examples that are relevant to the farmers that you work with. One illustration that might work in some places is guessing (or betting) which team will win a sporting event, such as football. Past records of wins and losses against a particular team (or similar teams) give an idea of the probability that your favorite team will win the next game. Suppose you learn that the star player on your team (or the opposing team) is injured and can’t play. This new information provides a forecast; it changes the probability that the team will win the next game. • Optional: Indigenous climate indicators. If a community has lived in the same location for a long time, it has likely developed some knowledge of how to anticipate weather conditions by observing the sky or biological phenomena. Showing respect for farmers’ indigenous climate knowledge can be a useful way to build trust and foster communication. This discussion should emphasize several points: • Both farmers and meteorologists make forecasts by observing the environment around them. You are not trying to replace traditional climate knowledge. Instead, you are introducing new information that farmers can consider alongside their traditional knowledge. • Forecasts from farmers and meteorologists provide useful information about the future, even though it has some uncertainty. Forecasts about future climate conditions do not have to be certain to be useful. • Meteorologists use information about the environment in distant locations, for example what is happening with temperatures in distant oceans. This helps meteorologists make predictions farther in the future. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 146 Step 2: From memory to variabiility Producers generally have a good understanding of their local climate based on their collective experience. This step starts with discussing climate conditions in recent years, and lead farmers to trust and correctly interpret time series graphs of their local climate. Memory of recent years • Display a table with three columns (Year, Crops and Rainfall) and five rows, listing five latest years for which rainfall data are available, starting with the most recent. • For the upcoming agricultural season, ask participants to classify each of the past 5 years as good, medium or poor based on agricultural performance. Repeat in terms of rainfall: wet, medium or dry. Record their consensus responses on the table. Note and discuss any obvious differences between crops and rain. • On the table with farmers’ memory of rainfall conditions, write the amount of rainfall measured for each of the past five years. Year Crops Rainfall 2022 2021 2020 2019 2018 Participatory time series graph activity This participatory activity may not be needed if participants already have experience reading simple graphs. • Involve participants in plotting the rainfall amounts on a time series graph. Show a blank graph on a flip chart page, with Year as the x axis and rainfall as the Y axis. Ask volunteers to take turns marking the amount of rainfall measured in each year. They will measure from the bottom, and place a horizontal mark above the corresponding year, and connect the marks with vertical lines to make a rectangle. Write the amount of rainfall above each mark. • Discuss how the rainfall amounts in the graph relate to the way they classified those years, in terms of rainfall and in terms of crop yields. Interpret a time series graph • Display a large printout or projection of the seasonal total rainfall time series graph, and distribute small (A4) printouts to participants. By this point, participants should be comfortable with this format of graph. Remind them of what the vertical axis (total rainfall for a particular season or set of months) and the horizontal axis represent (year). 147 Ethiopia • Discussing the following questions will help participants interpret the time series graph: • How many years are in the graph? • If participants did the participatory graphing activity: Do the last five years in the graph look like the graph that they just made? They should recognize that the computer-generated graph has the same information as the graph that they made, but with more years of data. • What year had the most rainfall (for the June-September season)? How much rain was measured that year? • What year had the least rainfall? How much rain was measured that year? • What years do you remember because they were either unusually wet or unusually dry? Discuss your experience during those years. • Discuss whether the graphs show more variability or less variability than participants expected. You may discuss whether the time series graphs show a long-term increasing or decreasing trend in rainfall, particularly if farmers have already expressed a belief that rainfall has been changing in a particular direction. • Optional: If most participants are familiar with division and percentage, lead them through an activity to use the time series graph to find the probability that seasonal rainfall will exceed a given threshold. Either suggest a threshold, or ask participants to identify some threshold rainfall amount, above which they would consider a season to be wetter than normal. Ask them to count how many years had greater than this rainfall amount. If any of the participants know how to divide, ask them to calculate the probability that rainfall will be above the threshold. If not, do the calculation for them. • For example, suppose you were using the graph below for this activity, and you selected a threshold of 650 mm. Out of the 38 years of data in the graph, there were 8 years, or 21%, or about two out of ten, that had rainfall greater than 650 mm. Suggestion: Holding a straight-edge horizontally across the graph, with the top edge of the straight-edge lined up with the threshold amount on the vertical axis, makes it easier to count the number of points above or below a given threshold. You can demonstrate this process on the large printout. Suggestion: If the participants are not familiar with percentage or fractions, you can round the probability to the nearest 10%, and explain the result in terms of the number of years out of ten in which rainfall exceeds the given threshold. • Optional: If time permits and participants are interested, lead them through the same process to find the probability that rainfall will be less than 450 mm or another threshold below which they would consider a season to be dryer than normal. Step 3: From variability to probability This step introduces participants to a probability of exceedance graph, and trains them to interpret it correctly. Participants will learn that probability is a way to quantify the uncertainty about the future, and that probability is calculated from relative frequency (meaning the number of times that something occurred in the past, divided by the number of observations that are available). They will learn that the probability of exceedance graph contains the same information as a time series graph, but arranged in a format that makes it easier to see the probability that seasonal climate conditions will fall above or below any threshold that might be relevant to their decisions. Participatory probability graph activity • Start with a blank graph with quantity (e.g., seasonal rainfall) on the x-axis and frequency (“Years with at least this much rain,” 0 to 5) on the y-axis. When you first display it, hide the second y axis (“Probability of at least this much rain”). If there is interest and sufficient time, ask volunteers to sort the rainfall amounts over the past 5 years (September-December season) from lowest to highest, and mark the point on the new graph. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 148 • For the first point: What was the driest year on the graph? How many mm of rainfall were measured? Out of the five years, how many had at least that much rain? (The correct answer is all 5 years.) Line up the ruler horizontally at 5 years, then measure to the right that many cm, and make a dot. Write above the dot what year it was. The second point is similar: What was the second driest year? How much rain? Out of 5 years, how many had at-least that much rain? (The correct answer is 4 years.) Make a dot at the correct point. Repeat for the third driest, the fourth driest, and finally the wettest. The last person to make a dot can also connect all the dots. To turn this graph into a probability distribution, we need to do two things. • First, we turn number of years (frequency) into percent of years (relative frequency). Probability can also be expressed as percent. Something with a 100% probability will definitely happen. Something with a 50% probability will happen 50% of the time – if you repeat it enough times. Note: we chose 5 because it is easy to calculate the relative frequency (in percent): 5 out of 5 years (5/5) = 100%, 4/5 = 80%, 3/5 = 60%, 2/5 = 40%, 1/5 = 20%. Write these percentages next to the numbers on the vertical axis. • Next, we go from frequency to probability. Show the hidden label or write a new label for the vertical axis, next to the old one: “Probability of at least this much rain.” (You could also explain this after you put up the probability graph printout based on all 30 years of data.) • Explain that frequency (in the past) is related to probability (in the future). 149 Ethiopia Note: Research shows that presenting information as relative frequencies rather than equivalent probabilities has a positive effect on many quantitative reasoning or estimation tasks. This is why we first express the graph as number of years, then percent of years, and finally percent probability. Interpret a probability graph • Explain that looking at more years in the past makes the probability graph more accurate. Then show a complete probability of exceedance graph, based on all available years of data. This graph shows the probability associated with any given amount of rain, or the amount of rain associated with a given probability. • Lead participants through an activity to interpret the probability of exceedance graph, similar to the exercise that you did in Module 2. Whatever threshold you used with the time series graph in Step 2, lead participants through the process of finding the probability of experiencing more than this amount using the probability of exceedance graph. Discuss which is easier to use. • Remind participants that the time series and probability graphs have the same information, just arranged differently. Allow time for discussion until most of the participants appear to understand the probability of exceedance graph. Although this is probably the most complicated step, experience shows that farmers can understand probability graphs with training and repetition. Step 4: Understand a seasonal forecast as a shift in the probability distribution Once farmers understand the probability of exceedance graph, and how it is related to historical data for their location, you can discuss how a seasonal forecast changes the forecast. Two activities can help: • Show how El Nino (or La Nina) conditions shift the probability distribution. • Use a familiar location with wetter or drier climate to illustrate what a forecast climate might mean. El Niño shifts the probability distribution This activity can be effective if most farmers are already familiar with El Nino and La Nina. It requires preparing a time series graph that uses a different symbol or color to highlight the El Nino (or La Nina) years; and a probability of exceedance graph with just the El Nino (or La Nina) years, plotted alongside the full climatological distribution. Describe the El Niño phenomenon. When meteorologists talk about “El Nino,” this refers to unusually warm temperatures in the eastern Pacific, near the equator. For a long time, Fishermen in Peru and Ecuador noticed unusually warm waters every few years. Because it was usually strongest at the end of the year, near Christmas, they called it “El Nino,” which is Spanish for Little Boy, and refers to baby Jesus because of Christmas. When meteorologists talk about El Nino, they are talking about something that happens in the Pacific Ocean, on the other side of the world from Ethiopia. Use a globe to show locations of Ethiopia and the eastern Pacific near the equator. El Niño (and La Niña) an important example of how the oceans influence the climate in many parts of the world, including Ethiopia. The reason why it is possible to make climate predictions several months in advance is because the oceans affect the atmosphere above them, but the oceans change slowly. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 150 Suggestion: Using a globe can help farmers visualize the Pacific Ocean in relationship to Ethiopia. If farmers have heard of El Nino but have an inaccurate understanding, it is important to avoid correcting them in a way that embarrasses them. • Present the time series graph with El Niño years highlighted. Discuss whether El Niño years have tended to be wetter or drier than years that were not El Nino. Ask what they think the probability of exceedance graph would look like if it only included El Nino years. • Present the probability of exceedance graph for El Nino years, plotted with the climatological distribution. Discuss how to interpret this type of forecast in terms of shifted probabilities. Discuss how knowing that the next season will be an El Niño would influence their expectations. Use a familiar location to discuss what a wetter or drier climate might look like This activity requires selecting a location that the farmers would be familiar with, and that has a wetter or drier climate. Plot seasonal data in time series and probability of exceedance formats for both locations (your location, and the wetter or dryer location), using different symbol shapes and colors. • Explain: Imagine that we used rainfall data from a wetter location to develop a probability distribution. How would it compare with your location? (If they interpret the probability of exceedance graph correctly, they should expect the curve will be shifted to the right of their location. If you use a dryer location, it will be shifted to the left.) Be careful that this doesn’t lead to confusion between a forecast that shifts the probability distribution, and a forecast referring to a different geographic location. 151 Ethiopia Step 5: Present the current seasonal forecast Note: In this activity, we are using the 2018 forecast as an example. This means that all of the discussion in steps 5 and 6 should focus on interpreting what a forecast means and identifying what farmers might do differently, and not planning for the next agricultural season. • Present a seasonal forecast for your location from the EMI Seasonal Forecast Maproom. Someone from EMI could present and discuss the forecast, if they are available and if they have been trained to interpret and communicate the Flexible Forecast format. • When you first introduce farmers to the new format, lead farmers through the process of finding the probability of experiencing more than this amount using the probability of exceedance graph, for whatever threshold you used with the time series graph in Step 2 and probability of exceedance graph in Step 3. Discuss what the probability is based on the historical distribution and based on the forecast. Discuss why they are different. • Make sure that everyone understands: • EMI produces the seasonal forecast; • what variable(s) and time period the forecast covers; and • how to interpret the forecast as a shift in probabilities. Note: When you go through this process with farmers, additional seasonal forecast variables might be available besides total rainfall. Any forecast variables that may be relevant to farmers’ decisions should be presented. Suggestion: If threshold values of seasonal climate variables have been identified for particular crops, cultivars or other management options, the probabilities of experiencing more or less than the threshold value can be derived for the historical distribution and for the forecast. Potential examples could include: • length of the rainfed growing season, relative to the time from sowing to the end of grain fill for particular cultivars; • Water Requirement Satisfaction Index (WRSI) required for sufficient harvest yield to cover the costs of production; or • chill units needed to trigger flowering, also known as vernalization, in some cool-season cereals or fruits. This would help reinforce the relevance of the forecast to farm management decisions. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 152 Step 6: Plan how to adjust farm management based on forecast In this step, you will facilitate participants to decide what seasonal management decisions, if any, they will change in response to the forecast. Group discussion allows participants to learn from other farmers, and get feedback on their ideas. While agricultural extension personnel and researchers can provide information about management options and answer questions, the process aims to support farmers’ decision making. Note: When this step is done as part of the training, using a forecast from a past year, be sure to understand that participants are discussing management options as a way to learn how to interpret the new forecast format, and to identify the types of decisions that they think they should consider changing if and when the actual forecast for the upcoming season is presented. • Guide participants to discuss the question, “What, if anything, will I do differently this season because of the forecast?” Explain that whether they want to adjust their plans for the upcoming season, and the kinds of adjustments that they make, may depend on: (1) how much the forecast shifts probabilities that are important for decisions, (2) whether management options are available that would be better suited than existing plans to the forecast conditions. Note: Discussion among farmers can help them identify ideas and clarify their own plans. This can be done either in the large group, or in breakout groups that then summarize their plans to the larger group. Participants may or may not want to adjust some of these plans in response to the seasonal forecast that they have been given. They may agree as a group to change some of their plans, while some decisions might be different for different farmers. A few discussion questions can help guide participants’ thinking and discussion: • What does the forecast say about rainfall for the upcoming season? It would probably be: (a) increased probability of wet conditions, (b) increased probability of dry conditions, or (c) very little change from the historical probability distribution. • For the forecast variable(s), are there any thresholds that would affect the performance of your farm, crops or livestock; or influence what management options you would choose? If so, how does the forecast affect the probability of experiencing rainfall above/below the threshold? • How do you expect the forecast to affect your farm (or crops or livestock) management decisions? • Think about a recent year in which the growing season was wetter than usual, or a year in which the season was dryer than usual. Is there anything you would have done differently on your farm if you had known what rainfall would have been like? • What if anything, will you do differently this season because of the forecast? • Do you need any further information in order to decide how to best manage your farm based on the seasonal forecast? • Do you need any further assistance in order to implement your plans for the upcoming season? • If participants plan in small groups, ask a representative from each to present their list of management decisions they would change based on the forecast. • Record the types of decisions that participants say they plan to change in response to the seasonal forecast. 153 Ethiopia Appendix A Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 154 APPENDIX A Self-Check Answers Self-Check Section 1.1 1. Recall the differences between weather, climate and climate change Weather are atmospheric conditions over a couple of hours or days in given location. Usually those vary between locations and times. Climate are the statistical properties of weather, reflecting how it is usually like in a given region. They include averages but also variability (its amplitude and extremes, frequency of harmful events etc.). Climate change are changes in those statistical propertiesWhat are the two dimensions of climate. How do weather and climate differ in those two dimensions 2. What are the two dimensions of climate? How do weather and climate differ in those two dimension? Climate varies in space and time. Weather will show quick changes over time (e.g. sunny to rain within 30 min) and over space (it can rain on side of the valley or town and not the other). Both side of the valley or town may exhibit the same climate – larger areas: districts, regions, may exhibit very similar climate. Statistical properties of weather, aka climate, are computed and will remain stable over several years or decades. Recall three main variables used to describe climate in a given location Most common variables are: Temperature, Precipitation (rainfall) and wind or cloud cover/solar radiation. Self-Check Section 1.2 1. What are the two main classes of factors that define the characteristics of climate in Ethiopia? Cite three elements in each class. Local climate results from the interaction of global factors and local factors. Examples of global factors include: i. latitude- distance to the equator which receives on average more solar energy while the poles receive the least. This contrast produces a lot of winds and ocean currents. ii. seasonality – results from the tilt of Earth’s axis and rotation around the sun: part of the year the North or he South pole do not receive radiation from the sun (northern and southern winter respectively). This affects the contrast listed in i. ii. contrast between oceans and land: oceans heat and cool more slowly than continents. This also creates winds between oceans and continents’’ Local factors include:’ i. Mountains, which have the capacity to channel the winds around them or upwards. When forcing the winds upwards they also force the air to drop the water vapor it contains which produces rain or snow. ii. Large inland lakes or rivers, provide additional sources of humidity that can be transported by for the winds blowing over them and cool the temperature iii. Vegetation that can intercept rain than evaporate it later, also providing humidity for the atmosphere and cooling the temperature 2. Which local factor is the most important in Ethiopia? Cite 2-3 ways it impacts climate in Ethiopia. Mountains are the most important local factor. Highest rainfall and lowest temperatures are observed in the mountainous regions of Ethiopia. 3. What are the three main regions of Ethiopia with respect to the seasonal cycle of rainfall? Which class of factors is the most important for the seasonal cycle of rainfall? North-West and West of the country has one rainy season, Kiremt, centered on July-September South and South-East of the country have two rainy seasons, roughly in March to May and October to January (corresponding to Belg and Bega), while Kiremt is dry. 155 Ethiopia East and North-East of the country also have two rainy seasons, during Belg and Kiremt, separated by a short pause in rain and followed by a long dry season during Bega. These seasonal cycles of rain are mostly due to the global factor defining the overall seasonality of the global system. Rain follows the maximum of solar radiation with a about 3 weeks’ lag. Kiremt corresponds to the northernmost position of the maximum of solar radiation, hence the rains during the Kiremt season in the center and northern parts of Ethiopia. Solar maximum crosses the southern regions twice, hence the rains in Belg and Bega. [Additional rains in the Eastern region in Belg are related to winds from the South- East that do not reach the western part of Ethiopia due to the mountain range. These winds do not bring a lot of moisture, hence the rains in Belg are weak]. 4. El Nino/La Nina is a phenomenon in the tropical Pacific. Explain the mechanism by which it affects climate in Ethiopia (hint: which factor is important in this mechanism?) ENSO disrupts the large-scale circulations (winds) between oceans and lands around the equatorial Pacific. By disrupting the circulation in the Pacific, it disrupts the circulations over the Indian Ocean and the Atlantic Ocean. Those circulations are important for the rain in Ethiopia as the oceans are important sources of water vapor, which is transported inland by the winds and falls as rain or snow. Disruption in the global circulation transmits the disruption in one area around the globe (this phenomenon is called ‘teleconnections’). Self-Check Section 1.3 1. What is the difference between climate data and climate information? Give one example of each. Climate data are large arrays of values, recorded by the Meteoroogical Services, with no or minimal analysis. Climate information is obtained from these data after analyzing them for a given purpose. Example of climate data: daily rainfall observed in Addis Ababa over the past 30 years. Example of climate information: average seasonal cycle of rainfall in Addis Ababa (i.e. how much it rains on average in each month); the amount of rain during very strong and very rare rainy events that occur on average once in 10 years is another example of climate information. 2. List the advantages and disadvantages of the different types of data. What precautions need to be taken when using information based on each type of data? Station data reflect the real conditions observed in a given location. However, stations can be very distant one from another (several hundreds of kilometers) and there may be no data for areas between stations. Gridded data are data that provide information over the entire globe and have been obtained by filling the gaps where station information is not available. There are three main methods to fill these gaps: • Simple interpolation between stations. This type of approach may miss some of the local factors that will affect climate locally (e.g. a mountain range) and give erroneous values for temperature or rain outside of the stations. Few stations exist in the middle of the oceans, leaving vast areas without any observations. • Satellite data use the data recorded by satellites orbiting high above the Earth. While they record observations continuous in space over very large areas at a time, these data are not without flaws: some data are not observed directly but only inferred by complex calculations (think rainfall, satellites are above clouds which are not transparent, thus satellites cannot directly ‘see’ and measure rainfall); some data are not always available for recording (think temperature at the surface of the Earth beneath the clouds, here again the satellites do not ‘see’ the surface); finally, not every area of the Earth is observed by a satellite at every instant: geo-stationary satellites are ‘stationed’ above a certain area and due to the sphercity of the Earth do not see the entire globe (think: the satellite cannot ‘see’ the areas on the other side of the globe); orbiting satellites make passes over each area of the Earth every 15 days. In summary: satellite data combines the observations from different types of satellites and use complex computations to provide data over every square kilometer of the Earth and every instant so these are often not direct observations. • Model data use complex equations to fill the gaps between observations and in the upper atmosphere to take into account local features such as mountain ranges, lakes, land-sea contrasts etc. While the data are produced everywhere on Earth, including over ocean and high up, the models are only an approximation of reality and may produce data that differ slightly in values, seasonal cycle etc. from what is observed in a station on the ground. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 156 There is no ideal dataset and the choice is often dictated by the use: are we interested in climate of a particular location where there is a dense network of stations, the choice could be station data, remembering that our study and conclusions may not be valid outside of that area. Are we interested in large areas, not well covered by stations: satellite data might be our choice, noting that values may slightly differ from observations on the ground. Model data are often used when oceanic areas and upper atmosphere need to be analyzed. It is important to remember that different types of data may lead to discrepancies in the results. 3. Read carefully the caption in Figure 1.3-6. What is plotted on the figure? Is it spatial or time dimension? Over what period of time? What data have been used? What analysis has the data undergone? What are the main types of variations presented? What is the main caveat of the result presented? i. The figure shows anomalies of temperature in C ii. The graph is in time dimension: it is not a map and we have years on the x axis. iii. The time period is between 1880 and 2010. There is one value per year (thin line) iv. Station data have been used: daily records of maximum and minimum temperature v. The data have undergone several analyses: daily maximum and minimum temperature have been transformed into mean daily temperature; then annual average has been computed for each station; then spatial average has been computed taking into account all the station available in a given year; then long term time average of this global average temperature has been computed over the entire period and subtracted to obtain anomalies. vi. The main feature of the graph is the increase in the global mean temperature, since 1880. This increase is not occurring at a steady rate, there are periods when it occurs faster (e.g. between 1920 and 1940 then after 1980) and slower (before 1920, or during 1940-60). Within this overall trend and the different periods one can notice variations from year to year, highlighting the interannual variability. vii. The main caveat of this graph is that it is based on station data and that large portions of the globe, especially oceans, are not captured. The number of observing stations has also changed over time, increasing over time over land, although a decrease in the number of stations has been noted in Africa in the second half of the 20th century. Self-Check Section 1.4 1. What are the two main types of forecasts? Cite advantages and drawbacks of each. Statistical forecast: easy to implement and to interpret; relies on relationships (e.g. between seas surface temperatures and rainfall) observed in the past which can be a problem in a changing climate, where those relationships can evolve. Dynamical forecasts: they are based on physical laws and don’t rely on past relationships but their resolution remains coarse (approximately hundred by hundred kilometers) thus capture well global circulation (and impacts of distant phenomena such as ENSO) but they may not capture very well the influence of local features (e.g. mountain ranges, lakes etc.) that contribute to local climate. 2. What are the main sources of uncertainty? Can uncertainty be avoided? How do forecasts deal with the uncertainty? Uncertainty in climate forecasts arises from: i. Our imperfect knowledge of the weather and climate conditions around the globe at the start of the forecast (called initial conditions) ii. Our imperfect knowledge of all the factors that influence climate during the entire period of the forecast iii. Our imperfect representation of all the phenomena that occur within the climate system and their interaction (think: how much water is going to evaporate above a swat of forest or prairie, how much a cloud is going to affect this evaporation by shading the area from the sun, we do not have measurements of these phenomena over all forests and prairies) 3. Would you base your on-farm decisions on a deterministic (i.e. giving precise values), high resolution seasonal forecast (e.g. at the level of the farm)? Explain why or why not. Precise values at very high resolution given by a forecast should be used with caution. There is a lot of uncertainty in climate forecasts (see the question above) and forecasts need to include some information about this uncertainty. Probabilistic forecasts, or forecasts giving an uncertainty interval (e.g. 900+/- 300mm of rainfall in the season) are one way to indicate how confident we are that something is going to happen. Given the large uncertainties, we should not attach too much confidence in forecasts that predict precisely 923mm of 157 Ethiopia rainfall in the season. In addition, forecasts are usually valid at large scale i.e. over hundreds of kilometers and: it is unlikely that two adjacent small farms will experience very different season, unless they differ significantly in local features (are on different slopes of a mountain, on different sides of a lake or within a forest and rangelands); our ability to predict different seasonal climate in such two farms is, in general, severely hampered by the lack of observations that would capture those differences and allow to include them in the forecast. Thus, most often forecasts advertised ‘at farm level’ will give similar forecast for many adjacent farms. The only advantage would be to receive the information directly on the phone but it should be provided with a measure of uncertainty/confidence and an assessment of how skillful the forecast is. National Meteorological Services develop the capacity to provide their forecasts directly to the farmer (or Agricultural Extension). Such forecasts usually reflect regional conditions and include levels of uncertainty and skill assessment. Self-Check Section 2.1 1. Which maproom is best for identifying the driest and wettest months of a given location? Monthly climate analysis 2. Increase in temperature is expected to decrease coffee production. How would you check this risk over a given coffee growing aera? By doing seasonal analysis for temperature (max or min). 3. You have been asked to check if a given wheat variety, which requires about 450 mm of seasonal total rainfall, can be grown in a given woreda. Which maproom, which “Seasonal daily statistics”, and which “Yearly seasonal Statistics” would you use to check this? Use the Extreme Rainfall Analysis maproom for total rainfall with probability of exceedance statistics. Precise values at very high resolution given by a forecast should be used with caution. There is a lot of uncertainty in climate forecasts (see the question above) and forecasts need to include some information about this uncertainty. Probabilistic forecasts, or forecasts giving an uncertainty interval (e.g. 900+/- 300mm of rainfall in the season) are one way to indicate how confident we are that something is going to happen. Given the large uncertainties, we should not attach too much confidence in forecasts that predict precisely 923mm of rainfall in the season. In addition, forecasts are usually valid at large scale i.e. over hundreds of kilometers and: it is unlikely that two adjacent small farms will experience very different season, unless they differ significantly in local features (are on different slopes of a mountain, on different sides of a lake or within a forest and rangelands); our ability to predict different seasonal climate in such two farms is, in general, severely hampered by the lack of observations that would capture those differences and allow to include them in the forecast. Thus, most often forecasts advertised ‘at farm level’ will give similar forecast for many adjacent farms. The only advantage would be to receive the information directly on the phone but it should be provided with a measure of uncertainty/confidence and an assessment of how skillful the forecast is. National Meteorological Services develop the capacity to provide their forecasts directly to the farmer (or Agricultural Extension). Such forecasts usually reflect regional conditions and include levels of uncertainty and skill assessment. Self-Check Section 2.2 1. Which maproom is best for identifying the driest and wettest months of a given location? Monthly climate analysis 2. Increase in temperature is expected to decrease coffee production. How would you check this risk over a given coffee growing aera? By doing seasonal analysis for temperature (max or min). 3. You have been asked to check if a given wheat variety, which requires about 450 mm of seasonal total rainfall, can be grown in a given woreda. Which maproom, which “Seasonal daily statistics”, and which “Yearly seasonal Statistics” would you use to check this? Use the Extreme Rainfall Analysis maproom for total rainfall with probability of exceedance statistics. 4. The above wheat variety cannot tolerate more than two consecutive dry spells in the growing season. Which maproom, which “Seasonal daily statistics”, and which “Yearly seasonal Statistics” would you use to check this? Use the Extreme Rainfall Analysis maproom for Number of dry spell and choosing the probability of exceedance statistics. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 158 5. The EMI has announced that the coming season would be under neutral ENSO condition. How would you check the potential impact of this condition over your woreda? By using the “Probability of Seasonal Rainfall Conditioned on ENSO” maproom, choosing “Neutral” for ENAOS state, and then checking for probabilities of wet and dry conditions. Self-Check Section 2.3 1. How do you find the probability of having onset before or after a give date (say July 1st)? Select onset and then click on the map at area of interest. Then examine the probability of exceedance curve. 2. What are the white areas on an onset or cessation map? These areas show that either the conditions for onset/cessation were not stratified or it is not the rainy season for those locations 3. What is the difference between seasonal total rainfall in the “Seasonal Analysis” maproom and the seasonal total in the Agriculture maproom? The total in Agriculture maproom is calculated between the onset and cessation dates. Self-Check Section 2.4 1. Which product would you use to check the severity of a dry dekad or month? Standard Precipitation Index (SPI) 2. Which one more useful between “Cumulative anomalies” and “Cumulative anomalies in percentage” and why? Cumulative anomalies in percentage because percentage difference offers a better comparison than an absolute difference. For instance, An anomaly of 100mm would have different impact for different rainfall regions. 3. From the cumulative graph, where would the blue line be relative the dark line for a very wet year? It would be at the top (above the black line) close to the 95th percentile. Self-Check Section 2.5 1. What is the difference between the tercile presentation of seasonal forecast and the presentation in the Flexible Forecast Maproom? The tercile presentation is fixed for three vague categories (below normal, normal, and above normal) while one can select any rainfall threshold in the Flexible Forecast maproom. 2. Using the Probability exceedance graph, how would you tell if the forecast is above or below normal ? Above normal if the forecast curve is to the right of the climatology curve, and below normal if the forecast curve is to the left side of the climatology. 3. How would you find the most likely rainfall amount in the forecast? It is the value with the highest probability in PDF curve. Self-Check Section 3.1 1. Would you use a weather or a seasonal climate forecast to decide whether to spray for a crop pest? Weather forecast 159 Ethiopia 2. Which would you use to decide what cultivar to plant? Seasonal forecast Self-Check Section 3.4 1. For a risk averse farmer who is selecting a farm management strategy under climate risk, would CE be: c) less than expected return Self-Check Section 3.5 1. Which is more cost-effective? b) one year out of six. Increasing the frequency of payouts increases the cost of the insurance. Agricultural insurance is most cost-effective when it pays out only in those years when conditions are too severe for farmers to manage through other means. Self-Check Section 4.1 1. Will farmers need more training and guidance to use weather forecasts or seasonal climate forecasts? They need more help to use seasonal forecasts. This is because seasonal forecasts are probabilistic, they are experienced less frequently, and they are not associated with memories of particular weather events. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 160 Appendix B Activity Worksheets The following pages contain Activity Worksheets for each of the four modules outlined in this Reference Guide. Activity 1.5. Understanding and Interpreting Common Maps and Graphs Map examples and exercises *All maps shown in this section are accessible from Ethiopia Meteorological Institute (EMI) maproom and are based on Ethiopia’s ENACTS climate data or digital elevation data. • Example 1: Ethiopia’s annual average rainfall The map below shows the average annual rainfall for the country of Ethiopia. The regions of Ethiopia are demarcated by solid black lines. Figure 1.5-1: Average annual rainfall In Ethiopia over the period1981-2016. Source - Ethiopia Meteorological Institute (EMI). Link: http:l/213.55.84.78:8082/maproom/Climatology/Climate_Analysis/seasonal.html?resolution=.1& YearStart= 1981&YearEnd=2016&seasonStart=Jan&seasonEnd=Dec&var=Rain&yearlyStat=Me an accessed 08/27/2021 Exercises Working individually or in small groups answer the following questions: Which direction is north? (up) What is plotted in the horizontal direction on this map? What are the units? (longitude, degrees) What is plotted in the vertical direction on this map? What are the units? (latitude, degrees) Can you estimate the scale of the map, even though it is not given? (The east to west extent of Ethiopia is 33 to 48 degrees or 15 degrees and in this image, this corresponds to about 15 cm. So, to first approximation, 1 cm ~1 degree or about 110 km) Using the legend/colorscale, which color corresponds to the wettest regions? (red) The driest? (light blue) What color correspondes to the regions with between 900 and 1200 mm of rainfall annually? (light green) Where, within the country is the wettest region? (the southwest) Are the colors on this colorscale equally spaced? (not exactly- the separation between isohyets from 800 mm and above seems to be about 400 mm, but for drier regions, the rainfall interval between isohyets is smaller) Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 162 Example 2: August rainfall over Ethiopia based on data from 1981-2016. Source - Ethiopia Meteorological Institute (EMI). Link: http://213. 55.84.78:8082/maproom/Climatology/Climate_Analysis/monthly.html?T=Aug&YearSt art=1981&YearEnd=2016&seasonStart=Jun&seasonEnd=Sep&var=.rfe_merged Figure 1.5-1: Average rainfall in August in Ethiopia over the period 1981-2016. Source - Ethiopia Meteorological Institute (EMI). Link: http://213. 55.84.78:8082/maproom/Climatology/Climate_Analysis/monthly.html?T=Aug&YearSt art=1981&YearEnd=2016&seasonStart=Jun&seasonEnd=Sep&var=.rfe_merged accessed 08/27/202 Exercises Working individually or in small groups answer the following questions: 1. How is this colorscale for August rainfall different from the colorscale in Example 1 for annual rainfall? (The colorscale here is just in green instead of a rainbow pattern and the wetter regions are shown with a darker green color). 2. How do the regions of wettest and driest rainfall in this map compare to Example 1? (The driest region is still in the southeast, and the wetter regions are still in the west, but there is significant rainfall in NW Ethiopia during August. For the annual total, the region of peak rainfall was in the southwest). 3. Are the colors on this colorscale equally spaced? (yes — each isohyet or shift in green color is spaced at an interval of 45 mm) 4. Now do the largest August rainfall values compare to the annual totals in the wettest regions? The wettest values in August are on the order of 350mm, whereas the wettest annual total rainfall is on the order of 20004- mm. The August rainfall may constitute 20% or more of the annual total in some regions in the north, but is a smaller share of the annual total in the southwest and the more arid regions. 163 Ethiopia Example 3: February - May and June - September rainfall maps Figure 0-2: Spatial distribution of average seasonal rainfall amounts in Ethiopia over the period 1981-2016.. Left panel: February to May; right panel: June to September. Source –Ethiopian Meteorological Institute (EMI). http://213.55.84.78:8082/maproom/ Climatology/Climate_Analysis/seasonal.html?resolution=.1&YearStart=1981&Yea rEnd=2016&seasonStart=Feb&seasonEnd=May&var=Rain&yearlyStat=Mean http://213.55.84.78:8082/maproom/ Climatology/Climate_Analysis/seasonal.html?resolution=.; accessed 08/27/2021 Exercises Working individually or in small groups answer the following questions: 1. How do these maps of rainfall compare to the map of annual rainfall in Ethiopia (Example 1)? (During FMAM. there is proportionally more rainfall in the south than in the north. relative to the annual total. By contrast, during JJAS. the south and southeast are very dry. but the north and west are very wet). 2. What additional information do these maps provide regarding the annual total? (These maps offer some indication of what share of the annual rainfall occurs during the Kiremt (JJAS) and BeIg (FMAM) seasons. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 164 Example 4: Ethiopia annual mean temperature Figure 0-3: Annual mean temperature in Ethiopia, based o the date over the period 1981-2016. Source - Ethiopia Meteorological Institute (EMI). Link: http:l/213.55.84.78:8082/maproom/Climatology/Climate_Analysis/seasonal.html?resolution=.1& YearStart= 1981&YearEnd=2016&seasonStart=Jan&seasonEnd=Dec&var=Tmax&yearlyStat=Mean Exercises Working individually or in small groups answer the following questions: 1. Which colors correspond to the hottest temperatures? (pink and dark red) 2. Where is this hottest region? (the northeast- in the Afar region) 3. Which color corresponds to the coolest temperatures? (dark blue) 4. Where is this coolest region? (the south center in central Oromia) 165 Ethiopia Example 5: Ethiopia elevation. Figure 0-4: Elevation in Ethiopia. Source — Ethiopia Meteorological Institute (EMI). Link: http://213.55.84.78:8082/maproom/Climatology/Climate Analysis/seasonathtml?resolution=.18, YearStart=19818,YearEnd=20168seasonStarl=Jan&seasonEnd=Dec&vai=Rain&yearlyStat=Me 7nettavers=bathtstavers=Reaion&lavera=Zonetalis=2; accessed 08/27/2021 Exercises 1 Working individually or in small groups examine the map of Ethiopia’s elevation above sea level in Figure 1.5.5. 1. What are the units of elevation shown in this map? (meters above sea level “m7 2. What color corresponds to the lowest elevation on land? (light blue - corresponds to land below sea level) 3. Where is this lowest region in Ethiopia located? (In the northeast Afar region- Denakil Depression) 4. What color corresponds to the highest elevation? (reddish pink) 5. Where is this region located? (in the south-central region of Oromia and a few locations in Amhara) 6. What is the elevation of the highest mountains in Ethiopia, according to this map? (between 3000 and 4000 meters — note the actual highest single peak in Ethiopia is over 4500 m above sea level, but very little of Ethiopia’s area is over 4000 m. so these high peaks don’t appear clearly as separate regions on a map of this resolution). 7. Where are the warmest regions relative to elevation? Where are the coolest regions relative to elevation? (Higher elevations tend to have cooler temperatures than lower elevations —i.e. temperature decreases with elevation) 8. Do higher elevations tend to have more rainfall or less rainfall than lower elevations? (Higher elevations tend to have more rainfall than low elevations — i.e. rainfall increases with elevation) Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 166 Interpreting Graphs (Primarily Histograms and Time Series) In addition to maps which convey information in space, graphs are frequently used in climate science to convey information which varies over time. Although, some graphs are also used to convey spatial information. In this primer, we will only deal with two dimensional graphs, although three dimensional graphs also exist. The key elements of a graph are as follows: Independent variable: A variable that is not influenced by another variable (eg. time, space). Dependent variable: A variable that is influenced by another variable (eg. temperature, rainfall – these vary in time and space). Horizontal (x) axis: The line drawn as a reference in a horizontal direction above or below which to plot data. The data plotted along the x or horizontal axis is generally the independent variable. Vertical (y) axis: The line, or a point, a bar, drawn as a reference in a vertical direction. The data plotted along the y or vertical axis is generally the dependent variable. 2nd vertical axis: Some graphs have a second vertical axis and plot a second dependent variable. This type of graph can be insightful when one wants to understand the relationship between two variables (example: temperature and rainfall or rainfall and streamflow). Scale or axis label: The distance between increments on a graph in the units of the x or y axis. For example, a graph that shows temperature as a function of month may have twelve monthly time steps as the values for the x axis, with a range of temperature from hot to cold temperatures on the y-axis. For regions in the tropics, where temperatures do not vary that much over the course of the year, the effective scale of the vertical temperature axis may be relatively small and may only range from 20 to 35 C. By contrast, for mid-latitude and high latitude locations where the contrast in temperature between winter and summer is much more dramatic, the range on the temperature scale might be much greater and might span from -10C to 30C. Each axis on a graph should have its own axis labels and scale markings delineating unit values. Key/Legend: As with maps, a key or legend is usually either an insert in a graph or a feature above or below a graph that explains what is being measured and how. There are many types of graphs — many of which can be made easily in Microsoft Excel, Apple Numbers, or any number of mathematical programming languages, such as R, Python, ba etc.. The example graphs shown below were created on the basis of climatological data from 1981-2010 from the World Meteorological Organization meteorological stations in the respective cities (specifically, their airports). Here are some common types of graphs used in climate work: • Column graphs: Dependent variable data for each value of the x axis is shown as a vertical column of varying heights. 167 Ethiopia Example 1: Average Monthly Rainfall in Addis Ababa, Ethiopia. Axis labels (demarcations) of rainfall amounts and months Figure 1.5-1: Average monthly rainfall in Addis Ababa. by month. Source: World Meteorological Organization. Link: https://worldweather.wmo.int/en/city.html?cityld=162 Exercises 1. What is plotted on each axis? What are the units for the independent and dependent variables? (x-axis/independent variable is time and the unit is months. Y-axis/dependent variable is rainfall and the unit is mm). 2. Which are the two wettest months in Addis Ababa? (July and August - months 7 and 8) 3. Which are the two driest months in Addis Ababa? (November and December— months 11 and 12) 4. Which months average more than 100 mm/month? (June. July. August and September—months 6-9) • Line graphs Each data point is mapped as a pair of coordinates in x.y space and a line is drawn connecting each point to the next point. Some line graphs include multiple series. Some line graphs are connected by straight lines, which others use smoothed interconnections. Example 2: average monthly rainfall in selected East African cities. Source: World Meteorological Organization. Links: https://worldweather.wmo.int/en/city.html?cityld=162 https://worldweather.wmo.int/en/city.html?cityld=167 https://worldweather.wmo.int/en/city.html?cityld=251 Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 168 Figure 1.5-2: Seasonal cycle of rainfall in selected East-Afrian cities. Source: World Meteorological Organization. Links: https:llworldweather.wmo.intlenlcity.html?cityld=162; https:llworldweather.wmo.int/en/city. html?cityld=167; https:llworldweather.wmo.int/en/city.html?cityld=251 Exercises Working individually or in small groups answer the following questions: Which city’s rainfall climatology is shown in black? Blue? Green? (Addis Ababa, Mekele and Nairobi, respectively) 1. In which months do Mekele and Addis Ababa experience maximum rainfall? (July/August) 2. In which months does Nairobi experience maximum rainfall?( April/May and November) 3. What are the driest months in Addis Ababa and Mekele? (November/December) 4. What are the driest months in Nairobi? (June-September) 5. Which months of the year experience more than 200 mm of rainfall on average for each city? (Addis Ababa: July and August; Mekele: only August; Nairobi: April) 6. Which months of the year experience less than 50 mm of rainfall on average for each city? (Addis Ababa: October-February; Mekele: September-June; Nairobi: June-September) 7. How does the rainfall in Mekele differ from that of Addis Ababa? (Mekele has a similar seasonality, but is drier in every month and its period of significant monthly rainfall is shorter). 8. While Addis Ababa and much of northwestern Ethiopia have one rainy season, some other regions of Ethiopia, particularly in the south and east of Ethiopia have two rainy seasons as shown by the rainfall climatologies for the Somali and Afar regions below. 169 Ethiopia Example 3: Somali region rainfall climatology. Figure 1.5-3:Rainfall climatology for the Somali region. Source: - Ethiopia Meteorological Institute (EMI). Link: http:l/213.55.84.78:8082/maproom/Climatology/Climate_Analysis/monthly. html?yearlyStat=Mean&T=Jan®ion=irid s%3ASOURCES%3AFeatures%3APolitical%3AEthiopia_ adm1%3AEthiopia_level1%3Agid%409%3Ads&YearStart= 1981&YearEnd=2016&seasonStart=Jul&seasonEnd=Dec&var=.rfe_merged Example 4: Afar region rainfall climatology. Source: - Ethiopia National Meteorological Agency. Link: http://213.55.84.78:8082/maproom/Climatology/Climate_Analysis/monthly.html?yearlyStat=Mea n&T=Jan®ion=irids%3ASOURCES%3AFeaturee/03APoliticark3AEthiopia_adml%3AEthiopi a_leve11°/03Ag id%402/o3Ads&YearStart=1981&YearEnd =2016&seasonSta rt=J ul&season End = Dec&var=.rfe_merged Figure1.5-4: Seasonal cycle of rainfall for the Afar region. Source: – Ethiopia Meteorological Institute (EMI). Link: http://213.55.84.78:8082/maproom/Climatology/Climate_Analysis/monthly. html?yearlyStat=Mean&T=Jan®ion=irid s%3ASOURCES%3AFeatures%3APolitical%3AEthiopia_ adm1%3AEthiopia_level1%3Agid%402%3Ads&YearStart= 1981&YearEnd=2016&seasonStart=Jul&seasonEnd=Dec&var=.rfe_merged Note that in the above graphs, the key shows that the 5’” percentile is shown in green, the 50”’ is shown in blue and the 969 percentile is shown in red, whereas the gray columns show the median (50”’ percentile) value. The concept of percentile implies probability of non-exceedance: only 5% of the data are below the 5” %ile, while 95% of the data are below the 95” %Ile and the median or 50’ %Ile is in the middle of the observed data. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 170 • Histograms: Histograms are often used to demonstrate the frequency of values of the dependent variable in different ranges. The x axis will show the ranges of the independent variable and the y axis will show the frequency, count or proportion. Histograms may be line or column graphs. Example 5: Frequency distribution of annual rainfall in Addis Ababa The graph below uses the same data above but shows the number of months with average monthly rainfall between 0-50mm, 50-100mm, 100-200mm and 200+mm. Figure 1.5-5: Frequency of average monthly rainall in Addis falling in the categories of 0-50mm/month, 50- 100mm/month, 100-200 mm/month, more than 200mm/month. Source: World Meteorological Organization. Exercises Working individually or in small groups answer the following questions: 1. Which category has the largest number of months? (0-50. Rainfall is under 50 mm for five months of the year. From the earlier graphs, we know that these months are October- February). 2. Which categories have the smallest number of months? (The 100-200 mm and 200+ mm categories have two months each. The 200+mm category consists of July and August and the 100-200 mm category consists of June and September). • Climographs: Climographs are a specialized type of graph that show the average monthly temperature and precipitation for a given location throughout the year. These figures will generally be presented as graphs with two vertical axes (temperature scale on one and precipitation scale on the other). Often the temperature will be plotted as a line graph and the precipitation as columns. 171 Ethiopia Example 6: Climograph for Addis Ababa. Figure1.5-7: Average seasonal cycle of rainfall and temperature in Addis Ababa. Source: World Meteorological Organization. Links: https://worldweather.wmo.int/en/city.html?cityId=162 Exercises Working individually or in small groups answer the following questions: 1. What do you notice about the relationship between temperature and rainfall? (Temperature peaks during the relatively dry months of April and May and is suppressed during the rainy season in July and August). 2. Which three months have average temperatures over 17 C? (March, April. May) 3. Which months have average temperatures below 16 C? (July. August. November. December) • Time series, anomalies and trends: A time series is a plot of a variable as a function of time. Time is plotted on the x axis and the variable(s) of interest are plotted on the y-axis. Unlike the climatological information shown in examples 1-6. time series data generally present unique values of a variable (rather than an average) for sequential months or years (eg. the annual rainfall for each year from 1981- 2021, the July price of maize in Ethiopia for 1991-2020). In climate science and in other fields, the word anomaly is used to define a measurable deviation from the average value. In climate science, anomalies are generally calculated in reference to a long-term average (or climatological) value. The anomaly value is the observed value for the given year and location minus the long-term average at that location. Positive rainfall anomalies indicate wetter than average conditions, whereas negative rainfall anomalies indicate drier than average conditions. Positive temperature anomalies indicate warmer than average temperatures, whereas negative temperature anomalies indicate cooler than average temperatures. When the pattern of anomalies shifts progressively from negative to positive or positive to negative, a time series is said to have a “trend”. Example 7: Time series of the anomalies of average annual maximum daily temperature for Addis Ababa Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 172 Example 7: Time series of the anomalies of average annual maximum daily temperature for Addis Ababa Figure1.5-7: Time series o the anomalies of average mean temperature in Addis Ababa, (1981-2014): . Source –Ethiopia Meteorological Institute (EMI). Link: http://213.55.84.78:8082/maproom/Climatology/Climate_ Analysis/seasonal.html?region=irids%3ASOURCES%3AFea tures%3APolitical%3AEthiopia_adm1%3AEthiopia_ level1%3Agid%401%3Ads&resolution=irids%3ASOURCES%3AF eatures%3APolitical%3AEthiopia_ adm1%3AEthiopia_level1%3Ads&YearStart=1981&YearEnd=2016&seasonStart=J an&seasonEnd=Dec&var=Tmax&yearlyStat=Mean Exercises Working individually or in small groups answer the following questions: 1. What is described on each axis? (x-axis shows the year. y-axis shows the anomaly of maximum temperature) 2. What do negative anomalies and positive anomalies mean in this context? (Negative anomalies imply the average daily maximum temperature for that year was cooler than the long term average, whereas positive anomalies imply that the average daily maximum temperature was hotter than the long-term average.) 3. What is the earliest year in this time series with a positive Tmax anomaly? (1984) 4. What is the latest year in this time series with a negative Tmax anomaly? (2007) 5. Which we the coldest and hottest years in this record and how different were their temperatures? (1982 and 2009. respectively. 1982 has a negative anomaly of about -1.6 degrees C and 2009 has a positive anomaly of 1 C, so the difference is 2. 6 C) 6. What do you notice about the overall pattern of positive and negative anomalies? (The earlier part of the record (80s and 90s) is dominated by negative anomalies, whereas the later part (21s’ century) is dominated by positive anomalies. This indicates a warming trend.) • Probability graphs Some graphs show the probability of a particular outcome with respect to a climate variable. In this case. a range of values of the climate variable is plotted on the x-axis and the probability is plotted on the y-axis. Probability values range from 0 to 1 (0% to 100%). One statistical tool that is often shown is the “probability density function” or PDF. For many types of data, the more common values are close to the long-term average and large either positive or negative anomalies are rare. 173 Ethiopia Example 8: June-September rainfall time series for Addis Ababa (1981-2016) and corresponding probability density function. Figure 1.5-1: Time series of June-toSeptembert seasonal rainfall in Addis Ababa. Source – Ethiopia Meteorological Institute (EMI). Link: http://213.55.84.78:8082/maproom/Climatology/Climate_Analysis/seasonal. html?region=irids%3ASOURCES%3AFea tures%3APolitical%3AEthiopia_adm1%3AEthiopia_ level1%3Agid%401%3Ads&resolution=irids%3ASOURCES%3AF eatures%3APolitical%3AEthiopia_ adm1%3AEthiopia_level1%3Ads&YearStart=1981&YearEnd=2016&seasonStart=J un&seasonEnd=Sep&var=Rain&yearlyStat=Mean In this graph of the 36 years of data, there is one JJAS rainfall below 600 mm, 3 instances of rainfall between 600 and 700mm, 10 instances of rainfall between 700 and 800 mm, 15 instances of rainfall between 800 and 900 mm, 6 instances of rainfall between 900 and 1000mm and one instance of rainfall above 1000mm. These data correspond to the probability density function shown below. The counts in each category are divided by the length of the record (36) to determine the probabilities. Figure 1.5-2: Probability density Function of JJAS seasonal rainfall in Addis Ababa based on the times series in figure 1.5.9. . Source – Ethiopia Meteorological Institute (EMI) • Cumulative density functions and probability of non-exceedance Cumulative density functions (CDFs) or probability of non-exceedance graphs show the historical probability of a variable being below a specified value in the given time range. CDFs and probability of non-exceedance graphs have a value of 0 for small values of the x axis (which may be temperature or rainfall or some other variable) and a value of 1 for large values of the x axis. Probability of exceedance graphs show the opposite trend because they display the historical probability that a variable’s value exceeds the stated threshold. So, for thresholds smaller than the smallest in the data, the probability is one and for thresholds larger than the largest in the data, the value is 0. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 174 Example 9: Probability of Exceedance Graph for June-September rainfall for Addis Ababa Figure 1.5-3: Probability of exceedance of June-to-September rainfall n Addis Ababa. Based on the same data as above. Source: EMI Exercises Working individually or in small groups answer the following questions: 1. What is the most likely category of June-September rainfall for Addis Ababa? (between 800 and 900 mm) 2. What is the probability that the June-September rainfall will exceed 600 mm? 900 mm? (around 98% and 20% respectively) 3. What is the probability that the June-September rainfall will not exceed those same thresholds? (around 2% and around 80% - the sum of the probability of exceedance and the probability of non-exceedance is always 100%. Example 10: Probability of exceeding 600mm of rainfall in June . Source — Ethiopia National Meteorological Agency. Link: http://213.55.84.78:8082/maproom/Climatology/Climate_ Analysis/extRainfall.html?seasonStart= Jun&probExcThresh1=600 175 Ethiopia Example 10: Probability of exceeding 600mm of rainfall in June . Source — Ethiopia National Meteorological Agency. Link: http://213.55.84.78:8082/maproom/Climatology/ Climate_Analysis/extRainfall.html?seasonStart= Jun&probExcThresh1=600 igure 1.5-11: Map of the probability of exceedance of 600mm of cumulative rainfall in July-September season. Same data as above. Source EMI. Example 10: Probability of exceeding 600mm of rainfall in June . Source — Ethiopia National Meteorological Agency. Link: http://213.55.84.78:8082/maproom/Climatology/ Climate_Analysis/extRainfall.html?seasonStart= Jun&probExcThresh1=600 Exercises Working individually or in small groups answer the following questions: 1. In which regions of the country is the probability of exceeding 600 mm during June- September highest? (the northern and western regions) 2. In which regions of the country is the probability of exceeding 600 mm rainfall during June- September lowest? (the southern and eastern regions) 3. Explain this geographic distribution in light of earlier maps and graphs. (Throughout this primer, we have learned that northern and western Ethiopia tend to have a single rainy season from June-September, and we have further learned that these areas tend to be at higher elevation and prone to higher total rainfall than the lowlands of the southern and eastern regions. It’s therefore quite expected that the probability of June-September seasonal rainfall exceeding a chosen threshold in the northern and western regions would be much higher than the probability of exceeding the same threshold in the lowlands of the southern and eastern regions.) Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 176 Activity 3.1. Agricultural calendar An agricultural calendar is a planning tool that summarizes the timing of agricultural activities for a particular location. In the case of annual crop farming, an agricultural calendar identifies normal time windows for sowing, crop growth and harvesting. Because crop growth stages and key agricultural activities are sensitive to weather conditions, it is useful to overlay the calendar with graphs of the seasonal climatology of rainfall and temperature. You will create an agricultural calendar for the location where you work. This exercise will summarize information about the seasonal cycles of climate and of agricultural activities. It will be part of your plan to integrate climate services into your work with farmers. You will do this as an individual exercise. The Excel file, A3-1 Crop calendar Melkasa.xls has an example of a crop calendar for several crops at Melkasa. In the cells under the months, each column represents a dekad, or 1/3 of a month. Note that the dates included in this example have not be verified and are not necessarily accurate. The file has two tabs: one with climate seasonal cycles graphed in Excel, and one with climate seasonal cycle graphs from the EMI Dekadal Analysis Maproom pasted and aligned with the table of dates. Procedure: 1. For your location, select one or more important farming or cropping systems. For each crop, list the normal range of dates for sowing, crop growth and harvest. If there is a critical period when crops are most sensitive to weather-related stress, such as flowering, list the normal range of dates. List the normal range of dates for the most important farm activities, for example: land preparation, sowing, fertilizer application, weeding, harvesting, any post-harvest processing and marketing. If you include livestock or other farm enterprises, list the normal range of dates for the most important management activities. 2. Open A3-1 Crop calendar Melkasa.xls, and save it with a different name. Select the Excel graph tab if you have a table of dekadal average rainfall and maximum and minimum temperature values, or Graph images if you only have graph images saved from the EMI Dekadal Climate Analysis Maproom. 3. Enter the climate seasonal cycle data if available, or insert the Maproom images. If you are using Maproom images, you will need to crop them to remove the x axis, and resize them to fit the available space and line up with the months and dekads at the top of the table. 4. Replace the existing crop information with the crops and activities that are important for your location. Use Fill colors to shade the range of dates for crop growth, important stages and important farm activities. 5. Save your calendar as an image. The easiest way to do this is to select the portion of the spreadsheet that contains your calendar, and print it to a PDF file. Insert it below. You will also use it in your presentation of your plan to integrate climate services into your extension activities, on the final day of the training. Activity 3.2. Use a decision tree to represent a farmers cultivar selection and fertilizer rate decision A decision tree provides a useful way to describe management decisions under uncertainty, when the decision has a small set of options, and when the risks that influence the decision outcomes can be described by a small set of states of nature. This small group activity is designed to: • Introduce you to decision trees as a tool to describe simple decision problems that are affected by climate risk; • Enable you to calculate the expected return on a decision option when information about probabilities and outcomes of different seasonal climate conditions is available; and • Demonstrate how a probabilistic seasonal forecast can lead a farmer to choose a different management option. You will make a decision tree that represents a farmer’s choice among two maize cultivars and three fertilizer levels. You will gain experience calculating expected returns based on the probability of experiencing different seasonal climate conditions. 177 Ethiopia Procedure: 1. Draw a decision tree that represents the cultivar decision, the fertilizer rate decision, and the uncertain seasonal rainfall conditions. You may either use shapes and lines in PowerPoint, or draw neatly by hand. Use the returns listed in Table A3-2 below. Table A3-2. Returns (or gross margin) from each decision option and state of nature (birr/ha). Cultivar Fertilizer Below- Near- Above- Expected return based on (kg N/ha) normal normal normal climatology forecast tercile tercile tercile BH 540 0 5,500 7,000 8,500 30 7,300 8,900 9,800 60 6,600 9,300 10,800 BH 660 0 6,900 8,200 9,400 30 7,200 8,900 10,200 60 6,800 9,200 10,600 2. Calculate the expected return for each decision option, and enter the values in the climatology column of Table A3- 3. Draw a decision tree that replaces the chance nodes and states of nature with end nodes that show expected returns. What is the most profitable management option? 4. Assume that a seasonal forecast shifts the rainfall probabilities to 50% wet, 30% middle and 20% dry for the upcoming season. Repeat steps 2 and 3 using the forecast probabilities, and enter the values in the forecast column. 5. Briefly discuss how the forecast influences the decision. Activity 3.3. Simulate crop management options with SIMAGRI To make good agronomic management decisions, farmers and their advisors must be able to anticipate how crops respond to management decisions, genetics, soils and uncertain future weather. Crop simulation models can help agricultural advisors understand how management decisions, genetics, soils and uncertain future weather interact to determine production. When they are used properly, with understanding of their capabilities and limitations, that can be used to translate historical, monitored and forecast climate information into expected crop performance. You will gain experience using a crop management decision support tool, SIMAGRI, to simulate maize yields for several management options. You will analyze how fertilizer rate and seasonal rainfall interact to influence crop yields. This small group activity is designed to: • Introduce you to the interface and capabilities of the SIMAGRI decision support tool; • Demonstrate insights that crop models can provide about the interactions between climate and crop management; and • Demonstrate how crop simulation models provide insights into how crop management and climate interact to influence productivity. Procedure: 1. Run 10 management scenarios in SIMAGRI using the input data below: • Station: Awassa • Crop: Maize • Years: 1981-2018 • Soil: ETET000_10 (ASAW, L, shallow) • Initial soil water: 50% of AWC • Initial NQ3: Low (23 kg N/ha) • Planting date: 20 April Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 178 • Planting density: 6 plants/m2 • Critical growing period to relate rainfall with yield: May-July Scenario Cultivar Fertilizer (kg N/ha) AOOO CIMT01 8H540 0 A030 30 A060 60 A090 90 A120 120 8000 CIMT17 8H660-FAW- 40% 0 8030 30 8060 60 8090 90 8120 120 2. Look at the different types of graphs in SIMAGRI. Which crop management option strategy seems to perform best? Several Excel spreadsheet templates are provided to simplify analysis and graphing results. 3. Download the simulated yields as a comma-separated value (CSV) file into a folder on your computer. Make sure you remember where you download it, as you will use it for several activities. 4. Open the Excel spreadsheet, A3-3 SIMAGRI activity.xis. Import the simulated yield CSV file into the tab, simulated yields, cell B-5. The Excel file already has graphs that show time series of yields, and yields for each cultivar as a function of N fertilizer rate. Save the 2 graphs as a picture, and paste them in this activity document. 5. Plot yield as a function of fertilizer rate, averaged for years with May-July rainfall in the below-normal and in the above-normal terciles. • In A3-3 SIMAGRI activity.xis, go to the wet vs dry years tab. It already has May-July rainfall, based on the same weather station used to simulate yields at Awassa. • Copy simulated yields, along with their scenario names, from cells D5 to M43 in the Simulated yields tab. Paste them in cell C9 in the wet vs dry years tab. • Highlight A9 to L47, and sort it by May-July rainfall, from lowest to highest. Once you have sorted the simulated yields according to seasonal rainfall amount, the formulas at the top calculate the mean for all years, and for the driest third, the middle third and the wettest third (i.e., the below-normal, near normal and above-normal terciles). Graphs at the right should show simulated yield response to fertilizer N rate, separate for the wet and the dry terciles. Paste the graph below, for either of the two cultivars. Activity 3.4. Use an enterprise budget to analyze crop management options An enterprise budget is a useful tool for estimating the profitability of a farm enterprise, and for comparing the profitability of alternative ways of managing farm enterprises. You will get experience with enterprise budget calculations. You will also see how the optimal level of production inputs such as fertilizer is lower when cost and net return are considered than when only yields are considered. Although using enterprise budget calculations to estimate the profitability of agricultural management options might seem tedious at first, it is a valuable skill for a professional who seeks to help farmers make better management decisions. This is small group activity is designed to: • Provide experience with the basic calculations needed to conduct an enterprise budget analysis; • Demonstrate how enterprise budgets translate information about productivity into information about profitability for different management options; • Demonstrate how production costs matter to farm management decisions; and 179 Ethiopia • Illustrate why management that maximizes profit is often less intensive than management that maximizes productivity. Table A3-4 lists the input data, management assumptions and crop yield that you will use for these calculations. After you get experience with the calculations, you will use an Excel template that already has the formulas needed to perform enterprise budget calculations on the crop yields that you simulated with SIMAGRI in Activity 3.3. You will use the enterprise budget calculations to compare the performance of different crop management options. Table A3-4. Information on inputs, management and expected yield for a maize production enterprise. Inputs Item Units Value Grain price birr/kg 12 Fertilizer price birr/kg 16 Fertilizer type urea Fertilizer N concentration g N/g fertilizer 0.26 Seed price BH 540 birr/kg 43 BH 660 birr/kg 30 Average seed weight g/seed 0.4 Labor price birr/day 150 Harvest and post-harvest labor per metric ton day/Mg 4 Fixed labor required day 6 Plowing price birr/ha 4000 Management Fertilizer N rate kg/ha 30 Cultivar BH 540 Stand density plants/m2 6 Seeds per plant (if extra seeds are planted then thinned) 3 Expected yield Grain yield kg/ha 2000 Yield calibration (to adjust simulate yields to match farmers’ fields) 0.7 Procedure: 1. Calculate cost of production. Note that units are in red in the equations below. • Calculate planting cost (birr/ha): = stand density plants/m2 x seeds per plant x seed weight g/seed x seed price birr/kg x 10,000 m2/ha x (1 kg/ 1000 g) • Calculate fertilizer cost (birr/ha): = (N fertilizer rate kg N/ha / N concentration kg N/kg fertilizer) x fertilizer price birr/kg Calculate harvesting and post-harvest processing cost (birr/ha). Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 180 Add all production costs. 2. Calculate gross receipts. 3. Calculate gross margin. 4. Calculate breakeven yield. 5. Calculate breakeven crop price. 6. Calculate an enterprise budget for the maize management scenarios you simulated with SIMAGRI. • Import the crop yields that you simulated with SIMAGRI into the Excel spreadsheet, A3-4 enterprise budget. xis. It is easiest to import them into a blank tab, and then copy and paste the results into the shaded area (cell B7) in the Simulated yields tab. • Enter the input costs from Table A3-4, in the Input data tab, and rows 1-4 in the Simulated yields tab. • Look at the other tabs. Adjusted yields multiplies simulated yields by the yield calibration factor, and includes graphs of results. There are separate tabs for Cost of production and Gross receipts. The Gross margins tab has the calculations and graphs that are most important. Use the information in that tab for the rest of this activity. • Save the bar graph of gross margins for all scenarios as a picture, and paste below. 7. What is the most profitable management option based on these input data and simulation results? 8. For one of the cultivars, make a graph that shows gross margin and grain yield as a function of fertilizer rate. Note that a graph is already made for the first six columns of crop simulation results, which should be BH 540 if you followed the sequence in the sequence in Activity 3-3. 9. What would be the most profitable management option if the grain price were 20% higher? ... if it were 20% lower? 10. Pick one other price that you think might change the most profitable management option, and test it by increasing it 20%, and then decreasing it 20%. How do these changes affect the most profitable management option? How do they affect a farmer’s average profit? 11. OPTIONAL: Plot gross margin and grain yield as a function of fertilizer rate, averaged for years with May-July rainfall in the below-normal and in the above normal terciles. The procedure is similar to what you did in Activity 3-3, step 5. • Open A3-4 SIMAGRI enterprise budget-wet & dry.xis. This file has the same calculations as A3-4 enterprise budget.xis. It already has Awassa May-July rainfall in the Simulated yields tab. 181 Ethiopia • Copy simulated yields from the Simulated yields tab of A3-4 enterprise budget.xis to the Simulated yields tab of A3-4 SIMAGRI enterprise budget-wet & dry.xis. Be sure you do not over-write the rainfall column. Copy or enter the crop, cultivar, seed price, fertilizer rate and stand density into cells C1 to L5 in the Simulated yields tab. Copy or enter the other enterprise budget input data into the Input data tab. • In the Simulated yields tab, highlight A7 to L47, and sort it by May-July rainfall, from lowest to highest. Once you have sorted the simulated yields according to seasonal rainfall amount, the formulas at the top calculate the mean for all years, and for the driest third, the middle third and the wettest third (i.e., the below-normal, near-normal and above-normal terciles). Graphs at the right should show gross margin response to fertilizer N rate, separate for the wet and the dry terciles. Paste the graph below, for either of the two cultivars. 12. Based on this activity, briefly discuss what agricultural extension personnel should consider when they recommend the use of fertilizer and other costly production inputs to farmers. Activity 3.5. Use risk efficiency analysis to identify prefered fertilizer strategies Risk efficiency analysis methods have been developed to evaluate management options based on a few general assumptions, for a group of decision makers who may have different risk preferences. The simplest type of risk efficiency analysis assumes that any decision maker prefers higher expected return and lower risk, and uses graphs to show the tradeoff between expected return and risk. When expected return is expressed as the mean, and risk can be expressed as standard deviation, it is referred to as E-S (expected return - standard deviation) analysis. Using the results of SIMAGRI simulations (Activity 3.3) and enterprise budgeting (Activity 3.4), you will use E-S analysis to identify what management strategies farmers might prefer, and which, if any, would not be preferred by any farmer. This small group activity is designed to: • Provide experience with the E-S approach to identifying risk efficient management options when mean and standard deviation of returns are available for different farm management options. • Demonstrate the tradeoff farmers face between increasing expected profit and reducing risk. • Demonstrate how risk aversion leads to management that is generally less intensive than management that maximizes expected profit. Procedure: 1. Make a graph of mean (y axis) and standard deviation (x axis) of each of the 10 management scenarios that you simulated in Activity 3.3 and analyzed in Activity 3.4. • The Excel file, A3-4 enterprise budget.xis, that you used in Activity 3.4 includes an E-S graph in the E-S graph tab, based on your SIMAGRI crop yield simulations and enterprise budget analysis. HINT: If you use this graph, it will be easier to analyze the results if you go to Format Axis, and set the minimum values of the standard deviation (x axis) and mean (y axis) to be just below the minimum values in your date. • Save your graph and paste it below. 2. Mark the management options that are part of the E-S efficient set in the graph, and list them below. 3. Discuss which management options would likely be preferred by a commercial farmer who uses agricultural insurance, and which options would likely be preferred by a smallholder farmer who faces significant resource constraints. Explain why. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 182 Activity 3.6. Analyze a rainfall insurance contract Agricultural insurance protects policyholders from specific types of risks, by providing a monetary payout when a covered loss occurs. This benefits farmers either by protecting productive assets such as livestock, or by enabling them to adopt improved production practices and access credit and other resources. Knowing that insurance will provide payouts in the event of extreme climate conditions increases farmers’ confidence to invest in improved agricultural practices, and increases the confidence of credit providers to lend to smallholder farmers. In this way, farmers can benefit from insurance even in years when they do not receive a payout. Index based agricultural insurance now makes it possible to insure relatively poor smallholder farmers. In this small group activity, you will examine a simple rainfall index insurance contract. It is designed to: • Illustrate insurance contract design options that could be adjusted to meet farmers’ needs. • Illustrate the concept of basis risk, and demonstrate how index insurance sometimes does not compensate farmers who experience a loss. Assume that a farmer in Awassa purchases an insurance contract with the following parameters: • Index: EMI merged total rainfall for May-July at Awassa (38.4E-38.5E, 7N-7.1N) • Target crop: Maize • Annual premium: 1,600 birr • Maximum payout: 10,000 birr • Trigger: 270 mm • Exit: 220 mm Procedure: 1. To analyze this contract, you will need to have the following data in an Excel spreadsheet: • Available rainfall data from EMI’s historical daily precipitation analysis maproom (1983-2014). • Maize yields simulated with SIMAGRI for the same location and set of years. For this activity, assume that simulated yields represent the farmer’s actual yields. It is up to you to decide which crop management scenario to use for the yield simulations. Please list the input data that you used to simulate yields. You are encouraged to figure out how to answer the questions on your own. However, if you get stuck, you may use the Excel file, A3-6 index insurance activity.xis, which has graphs and formulas to calculate payouts. You would need to paste the data in the shaded area in cells AB to C39. 2. With a table or graph, show what years had payouts, and the amount of each payout. 3. In what percentage of years would the farmer receive a payout? 4. How does the expected value of the payout compare with the premium? 5. Make graphs of the time series of the rainfall index value and crop yields, and paste them here. 6. Discuss how well years with payouts match years with low crop yields, assuming that simulated yields represent the farmer’s actual yields. 7. Discuss whether you would recommend this insurance contract to your farmers, and why. 8. How could you improve this insurance contract? 183 Ethiopia Activity 4.1. Seasonal forecast training and planning workshop In this activity, you will learn how to lead a group of farmers in a seasonal forecast training, presentation and planning workshop. This process uses historical climate graphs, and the probability-of-exceedance format for seasonal forecasts, which is available through the EMI Flexible Forecast Maproom. For this activity, we are practicing a workshop that combined all six steps, using data from near Melkasa. However, we are using a forecast from a past year (2018) and not the current forecast. The purpose is to learn how to lead the process and not to discuss plans for the current year. You will divide into small groups. We hope to get a few Development Agents who will play the role of farmer participants. If this is not possible, group members will need to take turns playing the role of farmers. Make sure that all group members are actively involved in leading parts of the workshop. Pay attention to the schedule below. Step Purpose Time 1 Workshop purpose and key Help farmers know what to expect. Define 10:30-11:00 concepts key terms that might have different meanings in different languages. 2 Understand past variability Ground concepts, climate data and 11:00-11:45 graphs in participants’ experience. Discussion (Plenary) 11:45-12:00 3 Introduce probability of Relating probability distribution to 12:00-13:00 exceedance graph time series prepares participants to understand forecast as a shifted distribution. 13:00-14:00 4 Seasonal forecast as a shift in Build understanding that a forecast is 14:15-15:00 probability distribution a shift in historical probabilities, not foreseeing the future. Builds trust in forecasts by reducing mystery. Afternoon coffee break 15:00-15:15 Discussion (Plenary) 15:15-15:30 5 Present the seasonal forecast Interpret a new forecast format. Connect 15:30-16:45 forecast to understanding of past climate variability. Plan seasonal farm Consider what management decisions, management based on the if any, to change in response to the forecast seasonal forecast. Identify potential options for future forecasts. The full forecast training, communication and planning workshop is needed the first time farmers are exposed to the downscaled forecast in the Flexible Forecast format. As discussed in the presentation and Reference Guide, it can either be done at one time, or split between training (steps 1-4) during a slow part of the agricultural calendar, and a shorter planning workshop (steps 5-6) shortly before the start of the growing season. The entire training and planning process would typically take about 8 to 12 hours, spread over two days. However, for this activity we have only 6 hours to practice the process, including debriefing and discussion time. Materials needed: • Flip chart • Markers (at least 2 colors) • Meter stick • Printouts (one large copy for the group, and one A4-size copy per participant): Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 184 o seasonal rainfall historical time series graph o seasonal rainfall probability of exceedance graph o seasonal rainfall historical time series graph with El Nino highlighted o seasonal rainfall probability of exceedance graph with El Nino highlighted o seasonal rainfall forecast graph • On a flipchart or poster-sized paper, blank table with three columns (Year, Crops and Rainfall) and five rows: Year Crops Rainfall • On a flipchart or poster-sized paper, blank graph with Year as the x axis and rainfall as the Y axis: • On a flipchart or poster-sized paper, blank graph with seasonal rainfall as the x axis. The y axis will have two labels as shown below. 185 Ethiopia Step 1: Introduce the workshop purpose and key concepts Workshop purpose Explain that the purpose of this workshop is to talk with farmers about climate variability about forecasts for the next rainy season, and how this type of information might be useful for farm management decisions. Key concepts Several concepts are crucial to understanding and using probabilistic climate information appropriately, it is a good idea to discuss the concepts with farmers at the beginning, and perhaps leave a short definition or description available as a reminder (for participants who can read) during training and planning meetings. When translating into the closest word or phrase in the local language, are farmers likely to have a different meaning in mind for that word? Does each word fully convey the concept, or do you need to define how we (climate service providers, communicators or facilitators) are using the words? In climate services, we often use words that have a particular technical meaning that is different from how the general public might use that word. To avoid misunderstanding, it is important to explain and agree with participating farms what we mean when we use key terms and concepts. It is important to discuss key concepts, especially if the local language does not have words for some of the key concepts. First, it is important to explain the time scale of seasonal forecasts. In many places, farmers and the general public think about long-term climate change when they encounter the concept of “climate.” In other places, farmers may have weather forecasts in mind when seasonal forecasting is introduced. If your farmers do not already have a good understanding of climate variability and seasonal climate forecasts, then discuss these concepts. • Weather is what happens on a particular day at a particular place. • Climate refers to longer times and larger regions. (Climate is the “statistics of weather,” but I’m not sure whether your farmers will understand what “statistics” means.) Illustrate by contrasting climate (i.e., long- term average temperature and rainfall) between two familiar locations. • Climate variability has to do with how seasons (periods of several months) differ from year to year (e.g., short rainy season or growing season conditions). • Climate change deals with long-term (multiple decades or longer) changes in things like average temperature and average rainfall, which are driven in part by changes in the atmosphere due to human activity. Second, define and explain these concepts: • Variability deals with what happened in the past. For example, rainfall in 2020 was different from rainfall in 2019, which was different from rainfall in 2018. • Frequency expresses variability with numbers. For example, in three out of the past ten years I was not able to produce enough maize to feed my family until the next harvest. • Uncertainty deals with what will happen in the future. Because the climate has been variable in the past, I am uncertain about what the weather will be like in next growing season. • Probability expresses uncertainty with numbers. For example, there are two chances in five that I will not produce enough maize to feed my family until the next harvest. • Forecast. A forecast is new information that changes the probabilities about the future. A forecast reduces uncertainty, but doesn’t eliminate it completely. Explain that the workshop will show farmers how to use probability and graphs to describe past climate variability and understand a seasonal climate forecast. Forecast and decision analogies (optional). You could discuss examples of ways that participants use uncertain information outside of farming, and how new information that shifts the probabilities is a form of forecast. Use examples that are relevant to the farmers that you work with. One illustration that might work in some places is guessing (or betting) which team will win a sporting event, such as football. Past records of wins and losses Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 186 against a particular team (or similar teams) give an idea of the probability that your favorite team will win the next game. Suppose you learn that the star player on your team (or the opposing team) is injured and can’t play. This new information provides a forecast; it changes the probability that the team will win the next game. There are some additional points that the facilitator can emphasize: • Both farmers and meteorologists make forecasts by observing the environment around them. Agricultural extension staff from MoA (and meteorologists from EMI, if they participate in the workshops) respect farmers’ traditional knowledge. We are not trying to replace traditional climate knowledge. Instead, we are introducing new information that farmers can consider alongside their traditional knowledge. • Forecasts from farmers and meteorologists provide useful information about the future, even though they have some uncertainty. Forecasts about future climate conditions do not have to be certain to be useful. • Meteorologists use information about the environment in distant locations. for example what is happening with temperatures in distant oceans. This helps meteorologists make predictions farther in the future. Step 2: Understand past variability Relate time series graphs to farmers’ collective memory 2. Display table with hree columns (Year, Crops and Rainfall) and five rows, listing five latest years for which rainfall data are available, starting with the most recent. Year Crops Rainfall 3, For the upcoming agricultural season, ask participants to daffy each of the past 5 years as good, medium or poor based on crop yields. Repeat in terms of rainfall: wet, medium or city. Record their consensus responses on the table. Note and discuss any obvious differences between crops and rain. 4. On the table with farmers’ memory of rainfall conditions, write the amount of rainfall measured for each of the past five years. 5. Involve participants in plotting the rainfall amounts on a time series graph. Show a blank graph on a flip chart page, with Year as the x axis and rainfall as the Y axis. Write the amount of rainfall above each mark. Connect the marks with straight lines. 6. Discuss how the rainfall amounts in the graph relate to the way they classified those years, in terms of rainfall and in terms of crop yields. 187 Ethiopia Interpret a time series graph 1. Display a large printout or projection of the seasonal total rainfall time series graph, and distribute small (A4) printouts to participants. By this point, participants should be comfortable with this format of graph. Remind them of what the vertical axis (total rainfall for a particular season or set of months) and the horizontal axis represent (year). Note that the year number in Ethiopia’s calendar is different from the year in the Gregorian calendar that most other countries use. 2. Discussing the following questions will help participants interpret the time series graph: a. How many years are in the graph? b. Do the last five years in the graph look like the graph that they just made? They should recognize that the computer-generated graph is the same as the graph that they made, but with more years of data. c. What year had the most rainfall (for the June-September season)? How much rain was measured that year? d. What year had the least rainfall? How much rain was measured that year? 3. Discuss whether the graphs show more variability or less variability than participants expected. You may discuss whether the time series graphs show a long-term increasing or decreasing trend in rainfall, particularly if farmers have already expressed a belief that rainfall has been changing in a particular direction. 4. Lead farmers through an activity to interpret the probability of exceedance graph, similar to the exercise that you did in Module 2. Suggest a threshold rainfall of 650 mm, or ask farmers to identify some threshold rainfall amount, above which farmers would consider a season to be wetter than normal. Ask them to count how many years had greater than this rainfall amount. If the threshold is 650 mm, there were 9 years out of 32, or 28%, or about three out of ten, that had rainfall about that amount. Suggestion: You can demonstrate this process on the large printout. Hold a meter stick or straight-edge horizontally across the graph, with the top edge of the straight-edge lined up with the threshold amount on the vertical axis. If any of the farmers know how to divide, ask them to calculate the probability that rainfall will be above the threshold, just as you did in Module 2. If not, do the calculation for them. Suggestion: If some of the farmers are not familiar with percentage or fractions, you can round the probability to the nearest 10%, and explain the result in terms of the number of years out of ten in which rainfall exceeds the given threshold. Optional: If time permits and farmers are interested, lead them through the same process to find the probability that rainfall will be less than 550mm or another threshold below which farmers would consider a season to be dryer than normal. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 188 Step 3: Introduce !the probability of exceedance graph Develop a probability graph Start with a blank graph with quantity (e.g., seasonal rainfall) on the x-axis and frequency (“Years with at least this much rain,” 0 to 5) on the y-axis. When you first display it, hide the second y axis (“Probability of at least this much rain”). If there is interest and sufficient time, ask volunteers to sort the rainfall amounts over the past 5 years (September-December season) from lowest to highest, and mark the point on the new graph. For the first point: What was the driest year on the graph? How many mm of rainfall were measured? Out of the five years, how many had at least that much rain? (The correct answer is all 5 years.) Line up the ruler horizontally at 5 years, then measure to the right that many cm, and make a dot. Write above the dot what year it was. The second point is similar: What was the second driest year? How much rain? Out of 5 years, how many had at-least that much rain? (The correct answer is 4 years.) Make a dot at the correct point. Repeat for the third driest, the fourth driest, and finally the wettest. The last person to make a dot can also connect all the dots. Remind participants that relative frequency (in the past) is related to probability (in the future). To turn this graph into a probability distribution, we need to do two things. First, we turn number of years (frequency) into percent of years (relative frequency). Probability can also be expressed as percent. Something with a 100% probability will definitely happen. Something with a 50% probability will happen 50% of the time - if you repeat it enough times. Note: we chose 5 because it is easy to calculate the relative frequency (in percent): 5 out of 5 years (5/5) = 100%, 4/5 = 80%, 3/5 = 60%, 2/5 = 40%, 1/5 = 20%. Write these percentages next to the numbers on the vertical axis. Next, we go from frequency to probability. Show the hidden label or write a new label for the vertical axis, next to the old one: “Probability of at least this much rain.” (It might be better to explain this after you put up the probability graph printout based on all 50 years of data.) 189 Ethiopia Note: Research shows that presenting information as relative frequencies rather than equivalent probabilities has a positive effect on many quantitative reasoning or estimation tasks. The frequency of experiencing any climatic category or exceeding any climatic quantity is easily derived from a time series sorted by climatic outcome. We do it interactively for only the past 5 years, so participants understand how it is derived from the time series. This is why we first express the graph as number of years, then percent of years, and finally percent probability. Interpret a probability graph Explain that looking at more years in the past makes the probability graph more accurate. Then show a complete probability of exceedance graph, based on all available years of data. This graph shows the probability associated with any given amount of rain, or the amount of rain associated with a given probability. Lead farmers through an activity to interpret the probability of exceedance graph, similar to the exercise that you did in Module 2. Whatever threshold you used with the time series graph in Step 2, lead farmers through the process of finding the probability of experiencing more than this amount using the probability of exceedance graph. Discuss which is easier to use. This is probably the most complicated step for farmers. However, experience shows that they can understand probability graphs with training and repetition. Remind them that the time series and probability graphs have the same information, just arranged differently. Allow time for discussion until most of the farmers appear to understand the probability of exceedance graph. Step 4: Understand a seasonal forecast as a shift in the historical probability distribution Once farmers understand the probability of exceedance graph, and how it is related to historical data for their location, you can discuss how a seasonal forecast changes the forecast. Two activities can help: • Use a familiar location with wetter or drier climate to illustrate what a forecast climate might mean. • Show how El Nino (or La Nina) conditions shift the probability distribution. El Nino shifts the probability distribution This activity can be effective if most farmers are already familiar with El Nino and La Nina. It requires preparing a time series graph that uses a different symbol or color to highlight the El Nino (or La Nina) years; and a probability of exceedance graph with just the El Nino (or La Nina) years, plotted alongside the full climatological distribution. Describe the El Nifio phenomenon. When meteorologists talk about “El Nino,”this refers to unusually warm temperatures in the eastern Pacific. near the equator. For a long time, Fishermen in Peru and Ecuador noticed unusually warm waters every few years. Because it was usually strongest at the end of the year. near Christmas. Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 190 they called it “El Nino.” which is Spanish for Little Boy, and refers to baby Jesus because of Christmas. When meteorologists talk about El Nino, they are talking about something that happens in the Pacific Ocean. on the other side of the world from Tanzania. Use a globe to show locations of Tanzania and the eastern Pacific near the equator. El Nino (and La Niiia) an important example of how the oceans influence the climate in many pads of the world, including Ethiopia. The reason why it is possible to make climate predictions several months in advance is because the oceans affect the atmosphere above them. but the oceans change slowly. Suggestion: Using a globe can help farmers visualize the Pacific Ocean in relationship to Ethiopia. If farmers have heard of El Nino but have an inaccurate understanding, it is important to avoid correcting them in a way that embarrasses them. Present the time series graph with El Nifio years highlighted. Discuss whether El Nifio years have tended to be wetter or drier than years that were not El Nino. Ask what they think the probability of exceedance graph would look like if it only included El Nino years. Present the probability of exceedance graph for El Nino years, plotted with the climatological distribution. Discuss how to interpret this type of forecast in terms of shifted probabilities. Discuss how knowing that the next season will be an El Nino would influence their expectations. 191 Ethiopia Use a familiar location to discuss what a wetter or drier climate might look like This activity requires selecting a location that the farmers would be familiar with, and that has a wetter or drier climate. Plot seasonal data in time series and probability of exceedance formats for both locations (your location, and the wetter or dryer location), using different symbol shapes and colors. Explain: Imagine that we used rainfall data from a wetter location to develop a probability distribution. Now would it compare with your location? (If they interpret the probability of exceedance graph correctly, they should expect the curve will be shifted to the right of their location. If you use a dryer location, it will be shifted to the left.) Be careful that this doesn’t lead to confusion between a forecast that shifts the probability distribution, and a forecast referring to a different geographic location. Step 5: Present the current seasonal forecast In this activity, we are using the 2018 forecast as an example. This means that all of the discussion in steps 5 and 6 should focus on thinking about what a forecast means and what farmers might do differently, and not planning for the next agricultural season. Present a seasonal forecast for your location from the EMI Seasonal Forecast Maproom. Someone from EMI could present and discuss the forecast, if they are available and if they have been trained to interpret and communicate the Flexible Forecast format. When you first introduce farmers to the new format, lead farmers through the process of finding the probability of experiencing more than this amount using the probability of exceedance graph, for whatever threshold you used with the time series graph in Step 2 and probability of exceedance graph in Step 3. Discuss what the probability is based on the historical distribution and based on the forecast. Discuss why they are different. Make sure that everyone understands: • EMI produces the seasonal forecast; • what variable(s) and time period the forecast covers; and • how to interpret the forecast as a shift in probabilities. When you go through this process with farmers, additional seasonal forecast variables might be available besides total rainfall. Any forecast variables that may be relevant to farmers’ decisions should be presented. If threshold values of seasonal climate variables have been identified for particular crops, cultivars or other management options, the probabilities of experiencing more or less than the threshold value can be derived for the historical distribution and for the forecast. Potential examples could include: • length of the rainfed growing season, relative to the time from sowing to the end of grain fill for particular cultivars; Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 192 • Water Requirement Satisfaction Index (WRSI) required for sufficient harvest yield to cover the costs of production; or • chill units needed to trigger flowering, also known as vernalization, in some cool-season cereals or fruits. This would help reinforce the relevance of the forecast to farm management decisions. Step 6: Plan how to adjust farm management based on the forecast Guide farmers to discuss the question, “What, if anything, will I do differently this season because of the forecast?” Explain that whether they want to adjust their plans for the upcoming season, and the kinds of adjustments that they make, may depend on: (1) how much the forecast shifts probabilities that are important for decisions, (2) whether management options are available that would be better suited than existing plans to the forecast conditions. Discussion among farmers can help them identify ideas and clarify their own plans. This can be done either in the large group, or in breakout groups that then summarize their plans to the larger group. Farmers may or may not want to adjust some of these plans now that they have the seasonal forecast. They may agree as a group to change some of their plans, while some decisions might be different for different farmers. A few discussion questions can help guide farmers’ thinking and discussion: • What does the forecast say about rainfall for the upcoming season? It would probably be: (a) increased probability of wet conditions, (b) increased probability of dry conditions, or (c) very little change from the historical probability distribution. • For the forecast variable(s), are there any thresholds that would affect the performance of your farm/crops/ livestock, or influence what management options you would choose? If so, how does the forecast affect the probability of experiencing rainfall above/below the threshold? • How do you expect the forecast to affect your farm (or crops or livestock) management decisions? • Think about a recent year in which the growing season was wetter than usual, or a year in which the season was dryer than usual. Is there anything you would have done differently on your farm if you had known what rainfall would have been like? • What if anything, will you do differently this season because of the forecast? • Do you need any further information in order to decide how to best manage your farm based on the seasonal forecast? • Do you need any further assistance in order to implement your plans for the upcoming season? If extension officers or other experts are present, give them an opportunity to respond to farmers’ questions and requests, and offer their own suggestions to the farmers. 193 Ethiopia Climate Risk Management in Agricultural Extension | REFERENCE GUIDE 194 Climate Risk Management for Agricultural Extension for Ethiopia 2nd Edition REFERENCE GUIDE