The Hail Cloud Model Identification Based On Radar Reflectivity Image And Multivariate Statistical Method Research

Severe convective weather refers to thunderstorms,lightning,heavy rain,hail and other disastrous weather.Hail as a strong convective weather,has great destructive power and strong explosive force.This is the main disastrous weather in Akesu area,which has brought serious harm to the development of agricultural economy in spring and summer in Akesu area.Under the guidance of socialism with Chinese characteristics in the new era of xi jinping,the economic aspect of aksu region has developed rapidly.Thus,to create a stable and harmonious society,the development of aksu region is particularly important.Therefore,the forecast and early warning of hail and other strong convective weather are the basis for improving the agricultural economic development of aksu.In this paper,the image of Hail Cloud detected by Doppler weather radar is used for a series of enhancement processing,then data mining is carried out to excavate the available information variables,and then the hail cloud is identified by multiple statistical methods.First,the hail cloud image and non-hail cloud image are enhanced with histogram,but the effect is not obvious,Furthermore,the image enhancement algorithm with wavelet transform is used to enhance the cloud image,which is more clear than the edge contour of the histogram enhancement algorithm.Finally,using the operator to sharpen the hailstone and hailless image,the effect is the most obvious in the three image enhancement methods.The edge processing of the enhanced cloud image is compared with the traditional edge detection operator,and the advantage of edge detection is compared.Then according to the xinjiang aksu region hail radar reflectivity image reflectance of different data as the basis,to extract the hail discriminant index,according to the ratio of different scales in the image reflectance image can be divided into 17 orders of magnitude,calculate the mean,variance,skewness,kurtosis four characteristic value;Based on the gray scale symbiosis matrix,the second order moment,contrast,correlation and entropy are extracted.Finally,the model is established to identify the hail and no hail images,which are applied to the maximum probability test,FCM clustering algorithm and SVM algorithm.Among them,the enhancement of histogram equalization,the enhancement algorithm on wavelet transform,the Laplacian cloud image sharpening,the hail cloud radar image data extraction method,the maximum probability criterion,fuzzy C-average method,the SVM algorithm is studied.