Understanding the performance of global precipitation products for hydrological modeling in the data-scarce morphologically complex central Himalayan region

Abstract

Abstract Developing a hydrological model is challenging in ungauged river basins. Hydrological modeling requires historical precipitation estimates. Global precipitation products (GPPs) have equipped hydrologists with a significant resource for hydrological applications such as flood modeling and forecasting. This study compares the quality and hydrological utility of four different GPPs in the Gandak river basin (Nepal and India), a representative of the morphologically complex Himalayan region. The study reveals that among the four GPPs, the SM2RAIN had the least average Root Mean Squared Error (RMSE) of 5.83 mm/day, and TRMM 3B42RT had the highest RMSE of 11 mm/day. When it comes to R-squared, SM2RAIN had the highest value (0.37), and TRMM 3B42 showed the lowest at 0.07. Similarly, Bias for SM2RAIN showed the most negligible average bias, which was − 5.34%, and TRMM 3B42 indicated the highest bias of 28.71%. POD for SM2RAIN was the highest at 0.83 and the lowest for TRMM 3B42RT (0.67). Similarly, when a hydrological model using MIKE 11 NAM model was developed and calibrated with observed rainfall data, TRMM 3B42 (daily), TRMM 3B42RT, APHRODITE, and SM2RAIN, the NSE were calculated to be 0.74, 0.81, -0.25 and 0.55 respectively for the calibration period and 0.67, 0.16, 0.4 and 0.50 respectively for the validation period. The TRMM 3B42RT data sets were not found to be suitable for hydrological modeling in the Gandak river basin as mostly the NSE value was found to be negative. The study reveals that the best-suited product among the four GPPs for hydrological simulations in the central Himalayan region is SM2RAIN, followed by the TRMM 3B42.