Evaluation of satellite rainfall estimates over the Lake Tana basin at the source region of the Blue Nile River

Abstract

Satellite rainfall estimates (SREs) have become alternative sources of rainfall data for several applications. However, the accuracy of the SREs is likely to vary from region to region and must be evaluated on a local basis. This study evaluated the accuracy of three SREs for the Lake Tana basin in northwestern Ethiopia. This basin is characterized by complex topography comprising both lowlands and highlands. The three SREs were the Tropical Applications of Meteorology using SATellite data and ground-based observations (TAMSAT), the Climate Hazard Infrared Precipitation with Stations (CHIRPS), and the Africa Rainfall Climatology (ARC). The SREs were compared with gauge measurements in lowland and highland regions during the period 1995–2010 on a point-to-pixel basis at daily, dekadal (10 days), monthly, and seasonal periodicities. The results show that the three SREs underestimated rainy events, but TAMSAT captured rainfall occurrence relatively well in both regions. ARC better estimated light rain rates (1–5 mm d−1) than did TAMSAT and CHIRPS; however, all the SREs markedly underestimated moderate and heavier rain rates (≥10 mm d−1). TAMSAT and CHIRPS estimated the amount of rainfall reasonably well (high efficiency, low random errors, and bias <10%) at daily, dekadal, and monthly time scales, whereas ARC did not perform satisfactorily (high random errors, low efficiency, and bias >20%) at any time scale. On a seasonal scale, CHIRPS estimated the secondary rainy season (March–May) rainfall better than did ARC and TAMSAT, whereas TAMSAT outperformed both CHIRPS and ARC during the primary rainy season (June–September). Overall, the rainfall detection capabilities and rainfall amount estimates of the SREs were better over the lowlands, and the cumulative rainfall estimates tended to improve with increasing integration time (i.e., from daily to seasonal totals).