Multi-source precipitation data fusion and prediction correction method and system
By constructing a classification model for rainy and dry conditions, combining land surface measurements, satellite remote sensing, and numerical model data, and integrating underlying surface factors, a spatiotemporal fusion and correction model for precipitation is built. This solves the problem of low precipitation prediction accuracy in existing technologies and achieves high-precision precipitation forecasting.
CN122241605APending Publication Date: 2026-06-19CHINA HYDROELECTRIC ENGINEERING CONSULTING GROUP CHENGDU RESEARCH HYDROELECTRIC INVESTIGATION DESIGN AND INSTITUTE +1
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA HYDROELECTRIC ENGINEERING CONSULTING GROUP CHENGDU RESEARCH HYDROELECTRIC INVESTIGATION DESIGN AND INSTITUTE
- Filing Date
- 2026-04-24
- Publication Date
- 2026-06-19
Smart Images
Figure CN122241605A_ABST
Abstract
This invention discloses a method and system for multi-source precipitation data fusion and forecast correction, comprising: acquiring historical precipitation data, including measured precipitation data from land surfaces, satellite-retrieved precipitation data, numerical model-predicted precipitation data, underlying surface factors, and meteorological factors; using measured precipitation from land surfaces as the true label to input a rainy and dry classification model to obtain rainy sample sequences from the satellite-retrieved precipitation data; training a precipitation spatiotemporal fusion model using satellite-retrieved precipitation and underlying surface factors as input and the rainy sample sequences as the true label; obtaining fused precipitation data based on the trained precipitation spatiotemporal fusion model; training a precipitation spatiotemporal correction model using the fused precipitation data as the true label and numerical model-predicted precipitation and meteorological factors as input; and obtaining corrected fused precipitation data based on the trained precipitation spatiotemporal correction model. This invention improves the accuracy of precipitation data fusion and prediction in watersheds at high altitudes.
Need to check novelty before this filing date? Find Prior Art