Research On The Model Of Rainstorm Disaster Evaluation And Prediction

Rainstorm disaster as one of our common natural disasters, which induces secondary disasters, such as flood and waterlogging,influences the normal industrial production and causes great harm to people’lives and property at the same time bringing serious economic loss. Fujian province is located in the southeast coast of China, which belongs to Tropical and subtropical marine monsoon climate, is affected by the typhoon. What’s more, the rivers confluence also aggravates rainstorm flood to each district.Therefore, the establishment of relational rainstorm disaster comprehensive evaluation model in Fujian province has remarkable significance to disaster regionalization and decision basis.The traditional comprehensive evaluation gets risk zoning result by natural fracture of evaluation value. It performs the comprehensive effect of each area but could not reflect the effect of each index in the zoning result or get the common characteristic aspect among areas. At the aspect of the above analysis and consideration, the paper estabilshed spectral clustering model based on gray correlation to get the rainstorm disaster risk zoning result through unsupervised learning of this clustering model. Because of the difference on relative degree of specific index, the combination weights which is based on AHP method and Entropy method as the adjustment factor are applied to the clustering model. In various weighting methods, AHP method, based on hierarchy and hierarchical nature of a system, is a subjective weighting method, through the expression of this model is lack of objectivity. Entropy method with poor interpretation extracts data information by the discrete degree of the data, which is objective weighting method but could not reflect the views of policy makers.In this paper the social and economic assessment data of Fujian provice from2000to2010, the precipitation data of the meteorological foundation sites in recent50years and the MICAPS system data of each meteorological station in resent11years were collected and collated. First of all, through studying the formation mechanism of rainstorm disaster and relational theory, after pre processing of the indicator data the paper focused and researched on the rainstorm conditions and socioeconomic data in spatial processing way. Secondly, we got the combination weights under the optimization model of weight vectors. Fujian province rainstorm disaster regionalization result was got in the method of GIS technology and arcmap show. Once more, relational factors of each level of risk zoning were analysed and the linkage characteristics of disaster prevention were excavated. At the aspect of the four indicators of standard layer including the risk of disater factors, susceptibility of disaster environment, the vulnerability of hazard bearing bodies and the capacity of prevention and reduction, the study offered space process and spatial anlyst. Finally, through the analysis of risk time sequence of each risk level, meteorological indexes time series and their change characteristic, we provided trend and forecasting analysis of risk and meteorological factors.