Research On Recognition Method Of Severe Convective Weather Based On VQ/HMM

Severe convective weather has great harm。Therefore it makes sense to carry outthe research of recognition of severe convective weather. On the basis of previouswork, this paper proposes a new recognition method of severe convective weather.The main contents are as follows:(1) This paper proposes a segmentation method for cells combining the rectanglemarking method with the extraction of contours of cell’s nuclear. On this basis,extract features including height, overhang, vertical integrated liquid(VIL) and hailindex, and then a statistical analysis of these features’ trend is given. Theexperimental results show that there are clear differences on the height features andthe hail index between the hail and rainstorm, and the values of the hail are far higherthan the values of the rainstorm. The features’ changes follow the law of “rise-stability-down ".(2) This paper proposes a recognition method of severe convective weather basedon the features’ statistical properties and majority voting method. Experiments showthat the method is effective for classification of hail and rainstorm.(3) This paper proposes a recognition method of severe convective weather basedon vector quantization, classifying the hail and rainstorm by establishing codebookseparately. The experimental results show that the vector quantization can make fulluse of the characteristic parameters’ distribution in space, and the recognition resultsare good.(4) This paper proposes a recognition method of severe convective weather basedon VQ/HMM. The experimental results show that the method use of the time seriesproperties of the severe convective weather effectively, and could be very good todescribe the features’ time-varying characteristics.