Bias correction of daily chirps-V2 rainfall estimates in Ghana

Johnson, R. (2022). Bias correction of daily chirps-V2 rainfall estimates in Ghana. Kumasi, Ghana: Kwame Nkrumah University of Science and Technology, (105 p.).

A wide range of economic sectors in the Ghana, including agriculture, health care, and energy, heavily rely on climate data; as a result, having access to reliable climate data is crucial for research and economic growth yet rainfall gauge data in Ghana scarcely available, therefore, researchers tend to depend on satellite estimates for hydrological studies and impact assessments. However, biases in satellite rainfall estimates and the ability for these rainfall products to effectively capture rainfall indices poses major issues for researcher and various key stakeholders. In this study, CHIRPS-v2 rainfall estimates were bias corrected using four (4) different bias correction algorithms (Linear Scaling (LS), Local Intensity Scaling (LOCI), Quantile Mapping (QM) and Bias Correction and Spatial Disaggregation (BCSD) methods) using 28 selected stations across Ghana and spatio-temporally over the entire country. At the station level the Linear Scaling method produced the best results, although after correction no significant changes were observed especially on a daily scale, using the day to compute seasonal indices yielded improved results. Spatio-temporally, The BCSD approach outperformed the other bias corrective correction strategies, most likely because it can capture the development of the average rainfall while matching statistical moments. The rainfall seasonal indices were then calculated from bias corrected CHIRPS-v2 data and the spread and the distributing of the various indices were well represented. Moreover, the extreme rainfall analysis produced results consistent with gauge values measured at the same time duration. Bias correction was able to minimize the errors and uncertainties that existed within the daily CHIRPS-v2 dataset, making it more suitable to derive agro-advisories.