Evaluation and comparison of separated precipitation types from multi-sources data in the Chinese Tianshan mountainous region

Precipitation types primarily include rainfall, snowfall, and sleet, and the transformation of precipitation types has significant impacts on regional climate, ecosystems, and the land-atmosphere system. This study employs the Ding method to separate precipitation types from three datasets (CMFD, ERA5_Land, and CN05.1). Using data from 26 meteorological observation stations in the Chinese Tianshan Mountains Region (CTMR) of China as the validation dataset, the precipitation type separation accuracy of three datasets was evaluated. Additionally, the impacts of relative humidity, precipitation amount, and air temperature on the accuracy of precipitation type separation were analyzed. The results indicate that the CMFD dataset provides the highest separation accuracy, followed by CN05.1, with ERA5_Land showing the poorest performance. Spatial correlation analysis reveals that CMFD outperforms the other two datasets at both annual and monthly scales. Root Mean Square Error (RMSE) and Mean Deviation (MD) values suggest that CMFD is more consistent with the station observational data. The analysis further demonstrates that relative humidity and precipitation amount significantly affect separation accuracy. After bias correction, the correlation coefficients between CMFD, ERA5_Land, and station observational data improved to 0.85-0.94, while the RMSE was controlled within 2 mm. The study also revealed that the overestimation of precipitation was positively correlated with the overestimation of rainfall days, negatively correlated with the overestimation of snowfall days, and that underestimated air temperatures led to an increase in the misclassification of snowfall days. This research provides a basis for selecting climate change datasets and managing water resources in alpine regions.