Based on the IMERG multi-source fusion precipitation product, this precipitation fusion product integrates high-quality microwave precipitation products from FY-3C, 3D, and 3E satellites; Using machine learning methods to invert precipitation products from FY-3C, 3D, 3E, and 3F satellites, and concatenating FY-3 precipitation products with other microwave observations in IMERG. All passive microwave products after splicing are extrapolated using cloud motion vectors and fused with infrared precipitation products from stationary satellites such as GOES, MetSat, and Himawari through Kalman filtering. The high-quality microwave precipitation products of FY-3 series satellites can effectively supplement the errors caused by microwave observation gaps in IMERG, providing a more reliable dataset for global precipitation research.
Based on the IMERG multi-source fusion precipitation product, this precipitation fusion product integrates high-quality microwave precipitation products from FY-3C, 3D, and 3E satellites.
Using machine learning methods to invert precipitation products from FY-3C, 3D, 3E, and 3F satellites, and concatenating FY-3 precipitation products with other microwave observations in IMERG. All passive microwave products after splicing are extrapolated using cloud motion vectors and fused with infrared precipitation products from stationary satellites such as GOES, MetSat, and Himawari through Kalman filtering.
The root mean square errors of Fengyun-3 C, D, and E satellites are about 9.9%, 18%, and 19.5% lower than IMERG products, respectively. High quality PMW precipitation products based on FY-3 satellites can fill the gap in PMW observations in the IMERG constellation. After integrating FY-3 and IMERG data, the proportion of high-quality PMW data in the total data significantly increased, and the daily PMW data proportion remained stable at around 50%. Based on the root mean square error (RMSE) of annual hourly observation data from approximately 4400 meteorological stations in China, Europe, and the United States, the accuracy of global precipitation data has improved by 16%. The overestimation rate of precipitation for IMERG products has decreased by 32%, and the overestimation rate of intensity has decreased by 63%. In addition, the underestimation of precipitation frequency in IMERG data during the period of 09:00-12:00 has improved by 67%. The accuracy and false alarm rate of precipitation detection have also significantly improved.
This work is licensed under
CC BY 4.0 (Creative Commons Attribution 4.0 International License).
10.11888/Atmos.tpdc.301183
