ESTIMATING MILLET PRODUCTION FOR FAMINE EARLY WARNING - AN APPLICATION OF CROP SIMULATION MODELING USING SATELLITE AND GROUND-BASED DATA INBURKINA-FASO

Early warning of impending poor crop harvests in highly variable envir onments can allow policy makers the time they need to take appropriate action to ameliorate the effects of regional food shortages on vulner able rural and urban populations, Crop production estimates for the cu rrent season can be obtained using crop simulation models and remotely sensed estimates of rainfall in real time, embedded in a geographic i nformation system that allows simple analysis of simulation results, A prototype yield estimation system was developed for the thirty provin ces of Burkina Faso, It is based on CERES-Millet, a crop simulation mo del of the growth and development of millet (Pennisetum spp.), The pro totype was used to estimate millet production in contrasting seasons a nd to derive production anomaly estimates for the 1986 season, Provinc ial yields simulated halfway through the growing season were generally within 15% of their final (end-of-season) values. Although more work is required to produce an operational early warning system of reasonab le credibility, the methodology has considerable potential for providi ng timely estimates of regional production of the major food crops in countries of sub-Saharan Africa.