Optimal spatial scales for seasonal forecasts over Africa

Abstract

Abstract The availability of seasonal weather forecast information in Africa has potential to provide advanced early warning of rainfall variability, informing preparedness actions to minimise adverse impacts. Obtaining accurate forecast information for the spatial scales at which decisions are made is vital. Here we examine the impact of spatial scales on the utility of seasonal rainfall forecasts in Africa. Using observations alongside seasonal forecasts from the European Centre for Medium-Range Weather Forecasts (ECMWF), we combine measures of local representativity and skill to assess optimal spatial scales for anticipating local rainfall conditions. The results reveal regions where spatial aggregation of gridded forecast data improves the quality of information provided at the local scale, and regions where forecasts have useful skill without aggregation. More generally this study presents a novel approach for evaluating the utility of forecast information which is applicable both globally and at all timescales.