Dependence of Runoff Simulation on Selection of Precipitation Products and Calibration Strategies: A Case Study in the Ganjiang River Basin

The accuracy of precipitation inputs critically impacts hydrological simulations. Unlike strongly physics-constrained weather models, hydrological models require parameter calibration to optimize unmeasurable, heterogeneous parameters against observed streamflow. This study evaluates four precipitation products [CN05.1 Gridded Precipitation Dataset (CN05.1), China Meteorological Forcing Dataset (CMFD) v2.0, Gridded Precipitation Dataset for the Chinese Mainland (CHM_PRE), Multi-Source Weighted-Ensemble Precipitation (MSWEP)] for runoff simulation in China’s Ganjiang River basin using the Soil and Water Assessment Tool (SWAT) hydrology model. Two calibration scenarios are implemented: Scenario I uses parameters calibrated with rain gauge data to assess quality of each precipitation product; Scenario II independently calibrates parameters for each product to mimic ungauged conditions. The results show that: (1) CHM_PRE and CN05.1, based on dense Chinese rain gauges, perform best, even surpassing rain gauge-driven simulations in many metrics. CMFD ranks next, while MSWEP systematically performs the worst. (2) Scenario II significantly improves simulations, especially for CMFD and MSWEP, demonstrating how calibration can compensate for product biases. (3) All products simulate monthly runoff better than daily runoff but tend to underestimate peak flows. The study confirms that high-quality regional rain gauge-based products (e.g., CHM_PRE, CN05.1) are reliable for runoff modeling and, when combined with product-specific calibration, provide a robust approach for applications in ungauged basins.