Hydrological Validation of Satellite-based and Reanalysis Precipitation Datasets in a Himalayan River Basin

The Himalayan River systems sustain the livelihoods of millions of downstream inhabitants by providing water for diverse needs. However, the Himalayas are among the most adversely impacted ecosystems in the world due to global warming and climate change. Many Himalayan basins have reported an increase in extreme precipitation events and floods. As concerns grow about the changing hydrological regime and future water availability, accurately estimating water yield and streamflow under climate change scenarios becomes essential. Hydrological modelling in the Himalayan basins, however, is challenging due to the lack of in-situ measurements. With limited rain gauge networks in these areas, and the majority of observation stations located in valley bottoms, higher-elevation climates remain underrepresented. In recent times, satellite precipitation products (SPPs) and reanalysis products (RAPs) have emerged as alternatives to ground-based observations, offering precipitation estimates with good spatial and temporal resolution. In this study, several SPPs and Re-Analysis Products (RAPs)—including APHRODITE, CHIRPS, ERA5, ERA5-LAND, IMDAA, GPM-IMERG, and PERSIANN—were evaluated for their prediction accuracy and hydrological applications by comparing them with IMD gridded data in the Upper Beas Basin, located in the western Himalayas. The precipitation products (PPs) were assessed based on their ability to capture daily, seasonal, and annual precipitation, as well as extreme precipitation indices (90th, 95th, and 99th percentile rainfall) using statistical metrics such as correlation coefficient (CC), RMSE, R² and relative bias (RB). Their performance in detecting rainfall events was evaluated using Categorical metrics such as POD, FAR, and CSI. Their statistical performance was ranked as: APHRODITE > ERA5-LAND > PERSIANN > ERA5 > GPM > IMDAA > CHIRPS. Overall statistical performances of APHRODITE, ERA5-LAND, ERA5, GPM and PERSIANN were found to be satisfactory. Further, to assess the hydrological utility of these PPs, the SWAT model was employed to generate basin water yield and water balance components using different products. APHRODITE, PERSIANN and GPM satisfactorily reproduced streamflow well (NSE = 0.88, 0.65, & 0.61, RSR = 0.34, 0.59, & 0.62; and PBIAS = -2.18%, -10.86%, & -18.98% respectively). PERSIANN and APHRODITE generated the water yield with an error of 1.68% and 3.91%. Only through hydrological validation, it was revealed that ERA5-LAND, despite ranking second in the statistical evaluation, exhibited poor hydrological performance (PBIAS = -39.14%), while GPM proved to be capable of reproducing streamflow although it performed poorly in statistical evaluation.  Hydrological validation not only revealed such discrepancies but also provided insights into water balance components, water yield, and flow extremes such as high flows and low flows. Therefore, this study recommends hydrological validation in addition to statistical evaluation for selecting reliable precipitation datasets for hydrological modelling in complex mountainous regions.

Keywords: Beas basin, Precipitation products, categorical metrics, Hydrological evaluation, streamflow, water yield