Evaluating the Performance of Multiple Satellite-Based Rainfall Products in a Data-Scarce Region

Rainfall is an important form of input data for hydrological modelling applications, climate change impact analysis, and integrated water resources management. However, spatially varied and temporally consistent ground-based and long-term observed rainfall data are usually lacking for such analyses, especially in developing countries where spatial variability in rainfall is predominantly persistent. In the absence of such a data source, gridded satellite-based rainfall products (Climate Hazards Center Infrared Precipitation with Stations (CHIRPS v3), Integrated Multi-satellitE Retrievals for GPM (Global Precipitation Measurement) Final Run (IMERG V07), Climate Prediction Center MORPHing technique (CMORPH) and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks Climate Data Record (PERSIANN_CDR)) can be used as surrogates, but it is crucial to evaluate their reliability and accuracy first. Such studies are particularly important in data-sparse regions, such as the Genale River watershed, located in the semi-arid region of Ethiopia. This study evaluates the performance of satellite-based rainfall products using ground-based rainfall data at different locations and integrating citizen-based science. Model performance was evaluated using categorical metrics (POD, FAR, and CSI) and statistical metrics (PBIAS, RMSE, R², and NSE) to assess event detection capability and prediction accuracy. IMERG V07-Final outperforms the other three sources of rainfall products, including the best detection ability of rainfall occurrence, with an NSE value of 0.87, an R² value of 0.99, an RMSE value of 49.18, and a rainfall detection POD value of IMERG V07-Final, which is the best compared to all other rainfall stations in rainfall detection, and the FAR value of CMORPH and IMERG-Final, which produce fewer false rainfall detections than CHIRPS v3 and PERSIANN-CDR. Overall, the study identifies the best satellite-based rainfall products that can be used for the Genale River watershed in the absence of spatially distributed rain gauge data. It is concluded that IMERG V07-Final, CMORPH, PERSIANN-CDR, and CHIRPS v3, respectively, could be used as alternatives for rain gauge data, especially for monthly hydrological modelling and drought-monitoring applications of the Genale River watershed and similar areas.