Abstract:
In a current situation using hydrological models to represent the characteristics of hydrological regime is the priority interims for water resource development and planning purposes. However, the input for such model is hardly available and became a challenge for investigating dynamics in the watersheds due to absence of obtaining accurate weather input data. Recently there is an approach to utilize satellite data for hydrological models and also, main objective of this study was to compare and evaluate three high global resolution satellite precipitation products (CFSR, PERSIANN and CHIRPS). The study was conducted in one of complex basins in Ethiopia of Upper Blue Nile Basin on upper main Beles subbasin. The study compares bias corrected satellite products with observed precipitation independently and by using HBV hydrological model and using statistical parameters to evaluate and compare. These parameters are coefficient of determination (R2), Nash-Sutcliff error (NSE) and percent of bias (PBIAS). Therefore, these three high resolution satellite products performed reasonably in comparing precipitation products as well as, in simulating hydrological model. According to the monthly comparison, seasonal and annual comparison of precipitation products of CFSR, CHIRPS and PERSIANN showed that R2 (0.93, 0.95 and 0.83), NSE (0.75,0.79 and 0.62) and PBIAS (-0.18, -0.032 and -0.031) of monthly respectively, seasonal comparison of precipitation products performs R2 (0.85, 0.84 and 0.73), NSE (0.86, 0.94 and 0.75) and PBIAS (0.67, -1.01 and -0.23) for seasonal respectively and annual comparison R2 (0.85, 0.77 and 0.74), NSE (0.75, 0.77 and 0.58) and PBIAS (1.08, -0.032 and 2.22) for annual. On the other hand, these products were used independently in simulating HBVlight hydrological model and the results of model revealed with R2 (0.72, 0.63 and 0.59), NSE (0.71, 0.63 and 0.68) and PBIAS (10%, 2.57% and -0.8) for CFSR, PERSIANN and CHIRPS products for calibration period and R2 (0.72, 0.76 and 0.66), NSE (0.72, 0.70 and 0.51) and PBIAS (0.15, 18.2 and 0.34%) for validation period. This study also reveals that all the three products can effectively detect rainfall events. This study provides use full information about three mainstream satellite products in the main beles. But in overall comparison and evaluation by using statistical parameters CFSR reanalysis precipitation product performed well. Therefore, CFSR reanalysis product can be selected to substitute observed precipitation after possible identification and bias corrections and HBV-light hydrological model is applicable at the study area.