Performance of the Global Forecast System's Medium-Range Precipitation Forecasts in the Niger River Basin

Abstract

Weather forecast information has the potential to improve water resources management, energy, and agriculture. This study evaluates the accuracy of medium-range (1–15 day) precipitation forecasts from the Global Forecast System (GFS) over watersheds of eight major dams in the Niger river basin. The Niger basin lies in three latitudinal/climatic sub-regions: Sahel (latitude > 12° N) with annual rainfall of rainfall 400–600 mm, Savannah (latitude 8°–12° N) with annual rainfall of 900–1200 mm, and Guinea Coast (latitude 4°–8° N) with annual rainfall of 1500–2000 mm. The GFS forecast tends to overestimate rainfall in the Guinea Coast and western parts of the Savannah, but estimates well in the Sahel. The overall performance of daily GFS forecast was found to be satisfactory for two watersheds, namely, Kainji (the largest watershed in the basin, predominantly located in the Sahel), and Markala (the second largest watershed, located partly in the Sahel and partly in the Savannah). However, the performance of daily GFS forecast was found to be unsatisfactory in the remaining six watersheds, with GFS forecasts characterized by large random errors, high false alarm, high overestimation bias of low rain rates, and large underestimation bias of heavy rain rates. The GFS forecast accuracy decreases with increasing lead time. The accuracy of GFS forecasts could be improved by applying post-processing techniques involving near-real time satellite rainfall products.