Hydrological Modelling Using Gridded and Ground-Based Precipitation Datasets in Data-Scarce Mountainous Regions
Authors
Khatakho R., Firoz A., Elagib N.A., Fink M.
Reference
Hydrological Processes, vol. 38, n° 12, art. no. e70024, 2024
Description
Satellite- and gridded ground-based precipitation data are crucial for understanding hydrological processes. However, the performance of these products needs rigorous evaluation before their integration into hydrological models. This study evaluates two types of precipitation products based on their hydrological simulation performance. The evaluation focuses on ground-based precipitation datasets (GA and Aphrodite) and satellite-based precipitation products (SPPs). The GA dataset combines rain gauge measurements with the Asian Precipitation—Highly-Resolved Observational Data Integration Towards Evaluation (Aphrodite) dataset to fill gaps in areas with insufficient rain gauge coverage. It is also used for model calibration under Method I. In Method II, models are calibrated with Tropical Rainfall Measuring Mission (TRMM), Climate Hazards Group Infrared Precipitation (CHIRPS), Multi-Source Weighted-Ensemble Precipitation (MSWEP) and Aphrodite product without the station data. The study considers the Koshi River Basin located in the eastern Himalayas encompassing Nepal and China's Tibetan region. The basin supports downstream ecosystems and domestic, hydro-power and irrigation development. Based on ranking of seven performance metrics, CHIRPS emerged as the best performing SPP whereas MSWEP ranked the lowest. When the five precipitation datasets were evaluated, GA performed the best, followed by CHIRPS, TRMM, MSWEP and Aphrodite respectively. In Method I, TRMM achieved the highest Nash−Sutcliffe Efficiency (NSE) value of 0.68, and MSWEP showed poor performance with an NSE value of −0.20. In Method II, CHIRPS showed the strongest performance with an NSE values of 0.82, whereas MSWEP performed slightly lower but still achieved an NSE value of 0.74. Seasonal analysis provided further valuable insights into selecting and blending precipitation datasets by identifying time series that performed best in specific seasons. These findings, alongside model uncertainty analyses, emphasise the influence of precipitation biases and underscore the value of integrating ground-based and satellite data. Ultimately, this study contributes to advancing water resource planning and management strategies in the Koshi River Basin and similar mountainous regions.
