Comparative evaluation of different satellite rainfall estimation products and bias correction in the Upper Blue Nile (UBN) basin

• Published in: Atmospheric research Vol. 178-179; pp. 471 - 483
• Main Authors: Abera, Wuletawu, Brocca, Luca, Rigon, Riccardo
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.09.2016
• Subjects: CFSR; CMORPH; Remote sensing of rainfall; SM2R-CCI; TAMSAT; TRMM 3B42V7; Upper Blue Nile; Africa, Blue Nile R; Africa; CMORPH; Upper Blue Nile; Remote sensing of rainfall; TRMM 3B42V7; TAMSAT; SM2R-CCI; CFSR
• ISSN: 0169-8095; 1873-2895
• DOI: 10.1016/j.atmosres.2016.04.017

In a region where ground-based gauge data are scarce, satellite rainfall estimates (SREs) are a viable option for proper space–time rainfall characterization. However, their accuracy and performances vary from region to region, and must be assessed. In this study, five high resolution satellite products (3B42V7, CMORPH, TAMSAT, SM2R-CCI, and CFSR) are compared and analyzed using the available rain gauge data in one of the most topographically and climatologically complex basin of Africa, the Upper Blue Nile basin (UBN). The basin rainfall is investigated systematically, and it is found that, at some locations, the difference in mean annual rainfall estimates between these SREs could be as much as about 2700mm. Considering three goodness-of-fit indexes, correlation, bias and root mean square error (RMSE) between the SREs and ground-based gauge rainfall, CMORPH, TAMSAT and SM2R-CCI outperform the other two. Furthermore, a confusion matrix is used to investigate the detection ability of satellite rainfall products for different rainfall intensities. TAMSAT has the highest (91%) detection skill for dry days, followed by CFSR (77%). On the contrary, SM2R-CCI has the highest accuracy index for medium rainfall ranges (10–20mm). The empirical cumulative distribution (ecdf) mapping technique is used to correct the intensities distribution givenby the SREs. This method provides a means to improve the rainfall estimation of all SREs, and the highest improvement is obtained for CMORPH (bias reduction from −72% to −1%). •We evaluate five satellite rainfall estimation products in Upper Blue Nile basin.•We examine the detection skill of five satellite products using confusion matrix and accuracy index.•The aggregated effect of satellite rainfall estimations on hydrological budget is very high.•Simple bias correction methods, such as ecdf mapping technique, reduce the satellite rainfall biases.•The effect of length of data set used for analysis on the satellite performance (goodness-of-fit indexes) is minimal.