Analysis of Multitemporal NDVI Data for Crop Yield Forecasting in the Sahel

Abstract

Recent investigations demonstrated that inter‐year NOAA‐AVHRR NDVI variations at the middle of the rainy season can provide information on annual crop yields in Sahelian countries. This line of research is presently extended to the consideration of multitemporal NDVI data for several years (1986-1991) pre‐processed by a proven methodology. The investigation was conducted using NDVI and crop yield data from the sahelian sub‐districts of Niger. The results confirm that geographically standardized NDVI data are efficient for crop yield forecasting, but notable differences exist in this prediction capability depending on the beginning of the season. Late beginnings of the growing (rainy) season (after the end of June) allow optimum forecasting only after mid‐August, while early beginnings lead to anticipate the forecasting capability but also to decrease its accuracy. The importance of these findings in the context of an early warning system is finally discussed.