Accuracy of the CMORPH satellite-rainfall product over Lake Tana Basin in Eastern Africa

Abstract

In this study, we assessed the accuracy of rainfall occurrence, amount and distribution over the Lake Tana basin in Ethiopia, Eastern Africa, as represented in the NOAA satellite-based Climate Prediction Center Morphing technique (CMORPH) rainfall product. This analysis is carried out at high spatial and temporal resolutions (8×8km2 and daily) using observations from rain gauges as a reference for the period covering January 2003 to December 2006. Graphical comparisons and several statistical metrics such as bias, correlation coefficient, and standard deviation of rainfall differences are used to perform the evaluation analysis. Spatial maps of these statistical metrics were developed to assess the spatial dependency in the CMORPH accuracy. The bias is decomposed into different components, hit, missed, and false, in order to gain additional insight into the possible sources of systematic deviations in CMORPH. Overall, CMORPH was able to capture the seasonal and spatial patterns of rainfall over the basin, but with varying degrees of accuracy that depend on topography, latitude and lake-versus-land conditions within the basin. The results show that CMORPH captured rain occurrence relatively well in both wet and dry seasons over the southern part of the basin but it significantly overestimated those over the lake and its southern shore. The bias of CMORPH in the study area is characterized by seasonal and spatial variations (−25 to 30% in wet season and −40 to 60% in dry season). False as well as missed rains contribute significantly to the total rainfall amounts over the basin. Significant levels of the differences are observed at the daily resolution of CMORPH. The relation between CMORPH and gauge rainfall amounts is stronger (correlation mostly >0.4) in the wet season than in the dry (mostly <0.4).