| Flood disasters endanger the safety of people and property by not only submerging buildings but also ruining lands.In recent years,flood disasters took place frequently there.They have the features of being sudden,which makes it quite difficult to prevent the happening and being destructive,making it easy to become disasters easily and cause casualties.In addition,it happens frequently and is easily influenced by the microclimate,causing Shennongjia a weak victim under the influence of floods.Shennongjia,located in the central and western mountainous areas,suffered serious damage from mountain torrent in recent years.The mountain torrent threatens the communities usually,which pushes the people in worse condition.In this paper,based on disaster system theory,indicator system of flood disaster loss was established in the varied topography and poor people wide distribution mountainous area.The indicator system divided into four main components,namely disaster-inducing factors,hazard inducing environment,physical exposure,and disaster relief capacity.To optimizing the indicator system,rough set theory was employed to analyze and optimize the input indicators.Applying the RBF neutral network model based on the flood disaster hazard,property damage and casualties were computed in Shennongjia.The results indicated that:(1)Rough sets theory can effectively extract the predictors that closely related to mountain torrent disaster and observably improves prediction accuracy;(2)The RIBF neural network model based on the rough sets theory has higher prediction accuracy compared with the general RBF neural network,of which the relative error generally maintained at about 3%and the relative error curve is gentler.(3)The economy is underdeveloped in Shennongjia.However,considering the lack of disaster prevention and mitigation capacity,the frequent disasters caused the regional socio-economic system to be more vulnerable.Therefore,it is important for the government departments to do a timely prediction of disaster losses,reasonable estimate of the size of the disaster. |
