基于双偏振雷达数据的降水类型识别与预测算法研究  

Research on precipitation type identification and prediction algorithm based on dual polarization radar data

作　　者：余代辉 伦绪勇 兰世怀 李圳坤 蒙延雷 Yu Daihui;Lun Xuyong;Lan Shihuai;Li Zhenkun;Meng Yanlei(Qiandongnan Miao and Dong Autonomous Prefecture Meteorological Bureau,Kaili 556099;Cengong Meteorological Bureau,Cengong 557801)

机构地区：[1]黔东南苗族侗族自治州气象局,凯里556099 [2]岑巩县气象局,岑巩557801

出　　处：《气象水文海洋仪器》2025年第1期11-14,共4页Meteorological,Hydrological and Marine Instruments

基　　金：贵州省气象局科研业务项目“气象雷达数据资料管理系统设计与实现”(556000)资助。

摘　　要：文章基于双偏振雷达数据,提出了一种新的降水类型识别与预测算法。该算法利用双偏振雷达反演的水平反射率因子、差分反射率、交叉相关系数、差分传播相移等极化参量,结合模糊逻辑和支持向量机方法,实现了对降水类型的精细识别。构建了基于长短期记忆网络的降水类型预测模型,可提前30 min预测降水类型的演变。算法在实际降水个例中进行了验证,识别准确率达90%以上,预测准确率达85%以上,展现出较好的应用前景。研究结果对于提升定量降水估测和临近预报能力具有积极作用。Based on dual polarization radar data,a new precipitation type identification and prediction algorithm is proposed.This algorithm utilizes polarization parameters such as horizontal reflectance factor,differential reflectance,cross correlation coefficient,and differential propagation phase shift obtained from dual polarization radar inversion,combined with fuzzy logic and support vector machine methods,to achieve precise identification of precipitation types.A precipitation type prediction model based on long short-term memory network has been constructed,which can predict the evolution of precipitation types 30 minutes in advance.The algorithm has been validated in actual precipitation cases,with a recognition accuracy of over 90%and a prediction accuracy of over 85%,demonstrating good application prospects.The research results have a positive impact on improving the quantitative precipitation estimation and forecasting capabilities.

关 键 词：双偏振雷达 降水类型识别 降水类型预测 模糊逻辑 

分 类 号：P415.2[天文地球—大气科学及气象学]