一种降雨智能订正方法、装置、设备及存储介质
By constructing a dual-branch network structure and combining multi-scale spatial feature extraction and cross-modal fusion techniques, the problem of insufficient modeling from a single data source was solved, high-precision rainfall data correction was achieved, and the resolution and accuracy of precipitation data were improved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- 广东省气象数据中心
- Filing Date
- 2026-04-23
- Publication Date
- 2026-07-17
AI Technical Summary
Existing precipitation retrieval methods suffer from insufficient modeling capabilities of single data sources and inadequate utilization of features at different scales, resulting in insufficient accuracy and resolution of precipitation data, making it difficult to meet high-precision requirements.
A dual-branch network structure is used to extract features from X-band phased array radar and meteorological model data. Through multi-scale spatial feature extraction, cross-modal residual embedding, and enhanced upsampling channel attention module, deep and shallow fusion is performed to construct an intelligent correction model for rainfall data, overcoming the problems of insufficient information from a single data source and difficulty in deep coupling of multi-source data.
It significantly improves the spatial resolution and correction accuracy of precipitation data, obtaining more accurate, stable and more generalizable rainfall estimation results, and suppresses the attenuation and clutter interference of X-band radar.
Smart Images
Figure CN122087735B_ABST