基于空间增强型XGBoost模型的海河流域卫星降水校正

Key Points

• This research aims to enhance the accuracy of satellite precipitation estimates by addressing systematic errors related to land use and spatial autocorrelation.
• Developed a spatially enhanced XGBoost model incorporating land use types, precipitation information from neighboring stations, and traditional meteorological and terrain factors.
• Applied the model within the Haihe Basin to assess its effectiveness in correcting satellite precipitation data.
• Analyzed the impact of land use and local factors on precipitation correction precision.
• XGBoost model significantly improved the accuracy of IMERG satellite precipitation products in mountainous and plain areas of the Haihe Basin.
• The corrected precipitation distribution aligned well with meteorological station data, capturing local heavy rain event variations.
• Land use significantly contributed to precision improvement (57.8%-66.7%), with neighboring environmental factors showing higher contributions than single-site data, particularly land use contributing 9.7%-14.5% more.