Satellite and gauge-based precipitation dataset assessment in Thua Thien Hue (Vietnam)

Abstract

Thua Thien Hue, a central province of Vietnam, has a monsoon tropical climate and complex interaction of weather patterns and topography and, particularly, very sparse of in-situ precipitation observations to model the hydrological characteristics for flood monitoring. So, this study evaluates the performance of four satellite-based precipitation datasets (CHIRPS, GSMaP, GPM and MSWEP) against gauge-based precipitation observations in Thua Thien Hue from 2020 to 2023. Tthe accuracy of each dataset is evaluated based on Taylor diagrams, Mean Absolute Error (MAE), Root Mean Square Error (RMSE), correlation coefficient (R), Critical Success Index (CSI), Probability Of Detection (POD) and False Alarm Ratio (FAR). Results show that MSWEP exhibits the highest correlation (R=0.58), lowest RMSE (23 mm/day), and best agreement with observed rainfall, making it the most reliable dataset. GSMaP follows, with strong correlation (R=0.63) but higher RMSE, indicating good temporal alignment but greater variability in extreme events. In contrast, CHIRPS and GPM have weaker correlations (R<0.40) and higher RMSE (>50 mm/day), leading to frequent underestimation of precipitation. The findings highlight systematic biases in satellite precipitation estimates and emphasize the need for regional calibration and bias correction. The study suggests that MSWEP is a useful source of data and should be prioritized for hydrological modeling in the region.