资源环境科技发展态势分析平台

Understanding the performance of satellite-retrieved extreme rainfall rates is a prerequisite for developing appropriate flood and drought monitoring systems in a region where rain gauge data are sparse and unevenly distributed. In this study, seven satellite rainfall estimates (SREs) that primarily use the infrared (PERSIANN, PERSIANN-CDR, and TAMSATv3) and the microwave (CMORPHv1, TRMM 3B42RT, TRMM 3B42v7 and IMERG-v06B) data were evaluated in the highland and lowland regions of Upper Awash Basin, Ethiopia, from 2003 to 2015. Three categorical error metrics, seven extreme precipitation indices, and the Empirical Cumulative Distribution Function (ECDF) plot were used to compare satellite retrieved extreme rainfall rates against rain gauge observations using a pixel-to-point approach. Results showed that TAMSATv3 outperformed all the remaining SREs in detecting and estimating low rainfall in both the highlands (elevation>2000 m) and lowlands (elevation<2000 m) of the basin. For high rainfall, the IMERG-v06B followed by TRMM 3B42v7 and TRMM 3B42RT products relatively produced the best result in both regions. Overall, the performance of SREs in detecting and estimating extreme rainfall rates was better in the highlands than in the lowlands, and this may be associated with the density of rain gauge networks. Additionally, all seven SREs can correctly detect low rainfall but their performances tend to worsen as the extreme precipitation quantile increases. In both regions, on average, the microwave-based SREs showed the highest result in capturing high rainfall. In contrast, the infrared-based SREs had a better performance for low rainfall. This study suggests that TAMSATv3 product can be used for drought monitoring studies in the basin; however, a lot of effort (e.g. bias-adjustment, and improvements of SREs algorithm) could be needed to integrate SREs into hydrological models for flood forecasting and early warning purposes.