Hydrological evaluation of merged satellite precipitation datasets for streamflow simulation using SWAT: A case study of Potohar Plateau, Pakistan

Abstract

Hydrological modeling is a challenging task in poorly gauged catchments, especially in developing countries like Pakistan. Open access precipitation and temperature datasets with different spatial and temporal resolutions provide alternative sources in data-scarce regions. However, individual satellite precipitation datasets (SPDs) have significant uncertainties. Motivated by data scarcity issues, significant spatial and temporal gaps in in-situ observations, and poor performance of individual SPDs in hydrological models, this study evaluates the performance of two merged precipitation datasets (MPDs) to simulate daily streamflow using Soil and Water Assessment Tool (SWAT) in Potohar Plateau, Pakistan. These two MPDs are based on Dynamic Clustered Bayesian Averaging (DCBA) and Dynamic Bayesian Model Averaging (DBMA), respectively, and have 0.25° spatial resolution and daily temporal resolution. Precipitation data from rain gauges (RGs) are also used, and results were compared with MPDs simulated streamflow. Multi-site calibration and validation are performed at seven stations, and the performance of RGs, DCBA, and DBMA in streamflow simulation was evaluated using the coefficient of determination (R2), Nash-Sutcliffe Efficiency (NS), and Percent BIAS (PBIAS). The results demonstrated that precipitation input from RGs presented better performance (very good to good) in streamflow simulation, even with its sparse distribution. The performance of DCBA showed better agreement with the results from RGs; however, DBMA presented satisfactory results on occasional bases. It is concluded from the current study that MPDs combines the advantages of individual SPDs and have higher potential for hydrological applications, significantly reduce the uncertainties of individual SPDs, and can be used as alternatives to RGs in poorly gauged or ungauged catchments.