Abstract: The snowstorm occurs every year in the middle and eastern part of Inner Mongolia,China and it often induces serious harm.Therefore,it is necessary to improve the forecasting,early warning and service level of such weather disasters and reduce disaster losses as much as possible.Because the hazard degree caused by snowstorm is different in various seasons,it is necessary to do the analysis by seasons,so the snowstorm is classified into three categories based on the seasons namely,the late autumn and early spring snowstorm,the midwinter snowstorm,and the cold rain or wet snow type of snowstorm occurred in the late spring and early summer,which have been analyzed in this paper.Based on the factor analysis method and grey correlation degree analysis it is found that the snowstorms,regardless of the types,happened in different seasons were basically all the hazardous weathers due to the gale,the snowfall and the cold wave or even a combination of the three.The leading factors were different in different seasons which led to different hazard degrees.The understanding of the characteristics of the snowstorms in various seasons will help the weather forecasters in snowstorms prediction and early warning.The BP neural network method and SVM method were further used to evaluate the snowstorm hazard level with the data about the snowstorm events that have occurred in the last several decades.Through comparing the effect of evaluated results,it showed that the computational efficiency of SVM method was better and it makes the earlywarning possible in the snowstorm forecasting based on the numerical forecast product.

Key words: snowstorm, risk assessment, BP, SVM