基于多源探测数据的液态降水现象及量级综合判识研究

Research on liquid precipitation phenomenon and scale comprehensive identification based on multi-source detection data

文章针对液态降水现象进行分析,利用自动气象站、双偏振天气雷达等设备观测资料,采用XGBoost、GBDT等机器学习算法,开展探测数据与降水量级的相关性分析,建立基于多源数据的降水类型及量级判识模型,实现对无雨、小雨、中雨、大雨、暴雨五类降水类型的识别,最终生成判识格点产品,使降水类型判识空间分辨力得到一定程度的提升。

This paper analyzes the phenomenon of liquid precipitation,using observation data from automatic weather stations,dual polarization weather radars,and other equipment,using XGBoost GBDT and other machine learning algorithms to carry out the correlation analysis of detection data and precipitation magnitude,establishes a precipitation type and magnitude recognition model based on multi-source data,realizes the recognition of five types of precipitation,namely,no rain,light rain,moderate rain,heavy rain,rainstorm and above,and finally generates recognition grid products,so as to improve the spatial resolution of precipitation type recognition to a certain extent.