Seasonal Hydrometeorological Forecasts for Water Managment in West- and Northeast Africa: Development, Operationalisation and Performance of a Regional Prediction System

It is the knowledge of the coming months that can be crucial for the management and control of water reservoirs for hydropower generation or for irrigation. This is particularly important in semi-arid regions of Africa that are characterized by distinctive dry seasons, i.e. where rainfall is limited to few months only. In addition, observation data in Africa are usually extremely sparse and computational power for forecasting systems is difficult to access. We present the spatial disaggregation and bias-correction of the globally available ECMWF’s newest seasonal forecast system SEAS5 and its tailored operational processing to support local water resources management and decision-makers. The forecast horizon is up to 7 month lead time, and our final forecasts have 0.1° spatial resolution. For the retrospective years 1981 till 2016 our ensemble consists of 25 members, while for the ongoing forecasts since 2017 there are 51 members available, allowing probabililistic predictions The performance of the regional prediction system is presented for 1) the Tekeze-Atbara and Blue Nile basins in Eithiopia/Sudan, and 2) the Volta and Niger basins in West Africa. The evaluation against the reference ERA5 data shows significant reduction in biases from the monthly averages as well as consistent and lead-independent forecasts characteristics like wet/dry frequencies. The performance metrices considered comprise accuracy (mean absolute error skill score), overall performance (continuous ranked probability skill score), sharpness (interquantile range skill score) and reliability. The operationalized system provides seasonal predictions each month to support water management on regional and local levels.