Accuracy of satellite rainfall estimates in the B lue N ile B asin: L owland plain versus highland mountain

Authors:
Open Access
Publisher website
Google Scholar
Abstract
The demand for accurate satellite rainfall products is increasing particularly in Africa where ground‐based data are mostly unavailable, timely inaccessible, and unreliable. In this study, the accuracy of three widely used, near‐global, high‐resolution satellite rainfall products (CMORPH, TMPA‐RT v7, TMPA‐RP v7), with a spatial resolution of 0.25° and a temporal resolution of 3 h, is assessed over the Blue Nile River Basin, a basin characterized by complex terrain and tropical monsoon. The assessment is made using relatively dense experimental networks of rain gauges deployed at two, 0.25° × 0.25°, sites that represent contrasting topographic features: lowland plain (mean elevation of 719 m.a.s.l.) and highland mountain (mean elevation of 2268 m.a.s.l.). The investigation period covers the summer seasons of 2012 and 2013. Compared to the highland mountain site, the lowland plain site exhibits marked extremes of rain intensity, higher mean rain intensity when it rains, lower frequency of rain occurrence, and smaller seasonal rainfall accumulation. All the satellite products considered tend to overestimate the mean rainfall rate at the lowland plain site, but underestimate it at the highland mountain site. The satellite products miss more rainfall at the highland mountain site than at the lowland plain site, and underestimate the heavy rain rates at both sites. Both sites have uncertainty (root mean square error) values greater than 100% for 3 h accumulations of <5 mm, or daily accumulations of <10 mm, and the uncertainty values decrease with increasing rainfall accumulation. Among the satellite products, CMORPH suffers from a large positive bias at the lowland plain site, and TMPA‐RP and TMPA‐RT miss a large number of rainfall events that contribute nearly half of the total rainfall at the highland mountain.

This publication has 30 references indexed in Scilit: