Précision de l'estimation des précipitations au Sahel selon la densité du réseau d'observation pluviométrique

Many earlier studies have shown the very large spatial variability of rainfall in the Sahel at all time steps, from the event to the season. Often, the meteorological network in these countries is sparse, with one to five rain gauges per 10 000 km². It is thus difficult to calculate accurate estimates of the mean rainfall over such a large area. To improve the knowledge of Sahelian systems and the spatial distribution of rainfall, a dense network was set up in an area of 16 000 km² in southwestern Niger between 1991 and 1996. The aim was to calculate accurate rainfall spatial means over an area of 12 000 km² at different time steps (from the season to the ten-day period). With the spatialisation method used (kriging), it was possible to calculate curves of estimation errors of mean rainfall versus the rain-gauge network density. Operational abacuses of the standard estimation error as a function of the spatial mean of rainfall and the network density are proposed.