Hydrogel particle classification and identification method and device for dual-polarization weather radar

Abstract

The invention discloses a hydrogel particle classification and identification method and device for a dual-polarization weather radar, and the method comprises the steps: taking a plurality of polarization parameters, such as a horizontal reflectivity factor, a differential reflectivity, a differential phase shift rate and a cross correlation coefficient, in a dual-polarization weather radar product as input; the input number is processed, analyzed and discriminated by using deep learning, and identification of precipitation particles such as crystal (CR), hairy rain (LR), rain (RN), heavy rain (HR), wet snow (WS), dry snow (DS), ice crystal (IC), vaterite (GR), hail (IH), rain and hail (RH) and the like is realized.

Description

technical field [0001] The invention relates to the technical field of weather radar information processing, in particular to a method and device for classifying and identifying hydrometeor particles for dual-polarization weather radar. Background technique [0002] In the central and western regions of China, the mountainous terrain often leads to frequent occurrence of destructive local meteorological disasters such as hail, which greatly affects the cultivation of grain and cash crops, and causes major damage to the local economy. The main cause of meteorological disasters represented by hail is the emergence of strong convective weather systems. In order to reduce the economic losses caused by meteorological disasters, technical means of artificial weather modification operations must be adopted to restrain the formation and development of these convective weather systems. The key to inhibiting the formation and development of convective weather systems is the early iden...

AI Technical Summary

Problems solved by technology

[0006] In the traditional shadow operation command business, the hydrometeor particle identification method based on dual polarization observation parameters is mainly based on fuzzy logic. This method needs to manually set and select the membership function boundary based on historically accumulated observation data. On the one hand, there are more human factors in the selection of parameters. On the other hand, once the recognition model is established, it is difficult to automatically adjust the corresponding recognition parameters according to the accumulation of new business observation data. The algorithm is poor in adaptability and cannot be improved.

The improved recognition method based on fuzzy neural network, although it can determine the boundary of membership function through sample training, partially solves the above problems, but the business usage is not ideal

Images