Abstract
A semi-deterministic model that allows assessment of groundnut yields from NOAA AVHRR data has been developed for the Peanut Basin in Senegal. Linear regression, using yield as the dependent variable and integrated NDVI during the reproductive period of the groundnuts, minus integrated NDVI during a period before the growing season as the independent variable, allowed 64% of the groundnut yield variance to be explained. This is comparable to similar studies concerned with other crops. Variable integration periods were used to account for variation in the temporal location of the reproductive periods. Compared to models using fixed integration periods this significantly increased both the level of explained yield variance and the interannual stability of the model. Pathfinder data were used to fill gaps in the time series of LAC data, using a linear transformation of Pathfinder NDVI values.
Acknowledgments
The author wishes to express his sincere thanks to Michael Schultz Rasmussen, Institute of Geography, University of Copenhagen, for his inspiration, support and guidance. Furthermore, the colleagues at the Centre de Suivi Ecologique (CSE) in Dakar deserves a great many thanks for their help with the fieldwork in 1999 and for making my stay in Senegal an unforgettable experience. Also thanks to Madiagne Diagne, Institut Senegalais de Recherche Agricole (ISRA) for inspiration and for providing the data from Projet Espace. This study was financed by DANIDA through the State of the Senegalese Environment (SSE) project.