Use of satellite-derived data to improve biophysical model output: An example from Southern Kenya

The use of satellite data products produced by the National Oceanic and Atmospheric Administration (NOAA) and the National Aeronautics and Space Administration (NASA) was explored to determine if these products could be used to provide plant growth model output for large landscapes. The use of satellite-derived data is generally advantageous because it is spatially dense (i.e., many measurements for a large landscape). Gridded daily temperature (0.1 x 0.1 degree) and rainfall (8x8 km), derived from the METEOSAT satellite, were used as inputs into the PHYGROW plant growth model for 30 pastoral households in southern Kenya. After model runs were completed, cokriging was used to determine if model output, coupled with NASAï'½s Normalized Difference Vegetation Index (NDVI) product (a greenness index), could be used to create forage production maps for a large landscape. Cokriging is a geostatistical technique that allows one to take advantage of spatial autocorrelation (i.e., things closer together in space are usually similar than those farther apart), and the similarity between a small number of data points (model output in our case) and a one that is spatially dense (NDVI). Using cokriging, the majority of ten-day averages for year 2000 had moderate to high similarity between model output and NDVI. A comparison of model output and estimates from cokriging indicated that cokriging generally did a good job of estimating forage available for the 30 pastoral households. Mapped surfaces of the cokriging output successfully identified areas of drought in year 2000. Institutions at all levels could use these mapped surfaces as part of their GIS, which can then be linked to economic models, natural resource management assessments, or used for drought early warning systems.