EVALUATION OF CLIMATE FORECAST AND SYSTEM REANALYSIS SATELLITE WEATHER DATA FOR THE PREDICTION IN THE UNGAUGED BASIN: CASE STUDY OF OMO-GIBE RIVER BASIN, ETHIOPIA

Hydrological data inadequacy at ungauged watersheds has been a problem for modeling the water resources of Omo-Gibe Basin, Ethiopia. Climate Forecast and System Reanalysis (CFSR) Satellite climate products can be utilized as an alternative source of climate data in such regions where the usual ground observed climate data are insufficient. Therefore, the aim of this study was to evaluate the hydrological performance of CFSR weather data as an input to the HBV-Light hydrological model over the Omo-Gibe River Basin, Ethiopia. The evaluation was executed in three different steps. Initially, the CFSR data was compared to ground climate data using efficiency descriptors (ENS, 𝑅2, PBIAS) and categorical statistics (Probability of Detection and False Alarm Ratio) for the same time window of 1987 to 2010. Secondly, the runoff predictive performance of the CFSR driven HBV -Light model was calibrated and validated at gauged catchments. Finally, streamflow was generated at ungauged catchments through the method of regionalization. The soil descriptors (Soil permeability, Soil available water content, Soil bulk density), vegetation descriptors (NDVI) and topographical descriptors were some of the data used in this study. Prior to the regionalization, Principal component analysis (PCA) was carried out on physical catchment characteristics (PCCs) to improve the regression model between the model parameters and PCCs. The result of direct comparison revealed that CFSR correlated to ground observed rainfall data at catchments with an efficiency of (R ², NSE ≥ 0.7) except for Walga, Wabe, Gorombo, and Sokie-Weybo where CFSR have shown reasonable efficiency with an acceptable bias of 10%, 8%, 𝑎𝑛𝑑 − 11%. However, the CFSR efficiency was increased after bias correction applying the linear scaling bias correction method. CFSR data-driven rainfall-runoff model has shown a good performance (R², NSE ≥ 0.7) in simulating the observed flow at well-recorded catchments. The performance of the regional model verified in the validation catchments was found satisfactory and recommended to be modified introducing any dataset not included in this study like anthropogenic activities, geological and mining data.