Accurate precipitation data are crucial for hydrological modeling, particularly in regions with significant climatic variability such as Monsoon Asia. This study assesses the performance of five precipitation products, integrated multi-satellite retrievals for GPM (IMERG), climate hazards group infrared precipitation with station data (CHIRPS), Global Data Assimilation System (GDAS), modern-era retrospective analysis for research and applications, version 2 (MERRA2), and MERRA2 with in-situ corrected precipitation (MERRA2-C) as input forcing in the Noah-MP land surface model (LSM). We also investigate how errors in these precipitation datasets propagate to simulated soil moisture (SM) and evapotranspiration (ET) by analyzing the metric \(\alpha\) , which is defined as the ratio of the normalized and centered root mean square error (NCRMSE) values of simulated variables to precipitation. The results show that MERRA2-C consistently outperforms the others with the lowest RMSE (3.14 mm/day), the highest Kling–Gupta Efficiency (KGE = 0.56), and the lowest false alarm ratio (FAR = 0.08). Error propagation analysis reveals that high precipitation errors do not always result in proportionally high SM and ET errors, as the propagation rate varies by climate zone, land cover, and topography. Errors propagate more to SM (mean \(\alpha\) = 0.60) than to ET (0.52), particularly in regions with greater porosity and steep terrains, while ET is more affected by precipitation errors in urban areas. Additionally, small precipitation errors and lower detection capability of light precipitation events in dry regions can lead to significant deviations in SM, while ET is more susceptible in wetter regions. These findings highlight the importance of not only improving precipitation data quality but also refining the LSM’s response to these inputs to enhance the accuracy and reliability of hydrological simulations.
