| description abstract | AbstractSeasonal climate forecasts for an early warning of climate anomalies are produced by regional climate outlook forums (RCOF) worldwide. This study presents a verification of one of the earliest RCOF products, the precipitation outlook for the West African monsoon peak period (July?September). The basis of this outlook is countrywide precipitation forecasts from various statistical (downscaling) models, which are subjectively reinterpreted by experts on the basis of information from observed SST pattern analysis and global forecasts. The forecast quality was analyzed from 1998 to 2013 using a novel database of rain gauge measurements established for several West African countries, among other references. The analysis indicated skill for above normal and below normal on different spatial scales but also showed typical limitations of seasonal forecasting such as lack of sharpness and poor skill for near normal. A specific feature of the RCOF product is a strong overforecasting of near normal, very likely a result of the risk aversion of experts. To better illustrate the usefulness of the outlooks, they were evaluated with respect to a binary warning system by determining the maximum economic value Vmax. This verification indicated moderate valuable precipitation warnings for dry (Vmax = 0.39) and wet (Vmax = 0.34) years for four climatological zones (Sahel, Sudan?Sahel, Sudan, and Guinean) and five river basins (Volta, Senegal, and three Niger subbasins) but with strong regional differences (0.14 < Vmax < 0.54). The bootstrap analysis illustrated large uncertainties, indicating the relevance of uncertainty margins when seasonal forecast products with small sample sizes like RCOF outlooks are evaluated. | |
