Remote sensing and GIS modeling for selection of a benchmark research area in the inland valley agroecosytems of West and Central Africa

Thenkabail, P., Nolte, C. & Lyon, J. (2000). Remote sensing and GIS modeling for selection of a benchmark research area in the inland valley agroecosytems of West and Central Africa. Photogrammetric Engineering and Remote Sensing, 66(6), 755-768.

This paper presents and illustrates a methodology for rational selection of benchmark research areas (or benchmark watersheds) for technology development research activities in the inland valley (IV) agroecosystems of West and Central Africa. This was done through a two-tier characterization approach. The Level 1 characterization involved macro-scale sub-continental- level secondary agroclimatic and soil datasets to produce 18 agroecological and soil zones (AESZ), each of over 10 million hectares, spread across West and Central Africa. The Level II characterization involved the use of Landsat TM or SPOT high-resolu tion visible (HRV) "windows” within each Level I AESZ, as well as other spatial datasets to determine locations of the representative benchmark research areas. The focus here is a methodology for Level II characterization for benchmark research-area selection using SPOT HRV data, secondary GIS datasets, and detailed ground-truth data with GPs locations. The spatial datalayers were analyzed in a GIS modeling framework. The study was conducted in an area of 0.39 million hectares around Gagnoa, southwestern C6te d'lvoire which is located in as number ( humid forests with acrisols). A toposequence oriented land-use/land-cover mapping was suggested and implemented. The spatial distribution of the 16 land-use classes was mapped across toposequence: uplands (40.1 percent of total geographic area), valley fringes (40.3 percent), valley bottoms (18 percent), and others (1.6 percent). The broad land-use/land-cover classes as a percentage of total geographic area (3931 12 hectares) comprised (1) 58.2 percent of areas in pristine humid forests, (2) 23 percent of areas in humid forest-cropland mosaic, and (3) 15.4 percent of areas in significant farmlands in humid forests. Expert knowledge was incorporated through an appropriate weighting criterion for classes in various land-use/land-cover data layers and other spatial data layers. GIs modeling was then performed on various spatial data layers leading to the selection of representative benchmark research areas. It is expected that the research conducted or technologies developed in these benchmark research areas can then be extrapolated or transferred to other areas within the same agroecological and soil zones like AESZ number 16.