基于GIS多准则评价与BP神经网络的暴雨洪涝灾害风险辨识--以闽三角地区为例

灾害风险辨识是灾害有效防控的重要环节之一,辨识体系与风险水平之间的非线性复杂关系使研究方法向精细化、智能化转型。闽三角地区是我国重要的沿海经济开放区,独特的“山-海”自然地理格局、起伏破碎的地形、高发的台风暴潮和极端短时降雨特征使其常遭受洪涝灾害侵扰。以闽三角为例,将生态服务价值纳入风险评价体系,构建基于GIS多准则评价与BP神经网络模型的风险辨识方法,旨在完善评价体系的同时,弥补传统评价方法存在的非线性缺陷和主观依赖,并以此为基础,进一步探究该地区风险空间分布规律和空间格局,为灾害风险防控提供思路。结果表明:①基于GIS多准则评价与BP神经网络模型的风险辨识方法能够系统准确的认知暴雨洪涝灾害风险水平与空间分布;②高风险区主要分布在河流沿岸、河口处、湾区,且人口、经济活动较活跃地区,城市化发展快速区与缓慢区相比,更容易遭受洪涝灾害威胁;③洪涝风险以高-高和低-低集聚为主,风险根据空间自相关性特征分为“整体随机”“局部随机-邻域集聚”和“整体集聚”三种类型。最后根据风险特征将闽三角地区高风险分为“厦门集美版块”“泉州湾区版块”“漳州县区版块”,分别提出灾害治理建议。

Risk Identification of Storm and Flood Disaster Based on GIS Multi-criteria Evaluation and BP Neural Network:A Case Study on the Min delta

Flood and waterlogging disaster risk identification is one of the important links for effective sustainable prevention and control of disaster.The research method has been transformed into refined and intelligent because of the nonlinear complex relationship between the identification system and the risk level.The Min Delta area is an important coastal economic open area in China.The unique“Mountain-Sea”natural geographical pattern,undulating and fragmentary terrain,high occurrence of typhoon surge,and extremely short-term rainfall characteristics lead to the region often suffering from the flood disaster.Taking Min Delta as an example,the value of ecological services was included in the risk assessment system,a risk identification method based on GIS multi-criteria evaluation,and BP neural network model was constructed.This paper aims to improve the evaluation system and make up for the non-linear defects and subjective dependence of the traditional evaluation methods.The spatial distribution law and spatial pattern of wind risk were further explored on this basis,and the ideas for disaster risk prevention and control were provided.The results show that:(1)The rainstorm flood disaster risk level and spatial distribution can be recognized systematically and accurately by this method.(2)High-risk areas are mainly distributed along rivers,estuaries,and bay areas,and are more active in population and economic activities.(3)The autocorrelation of flood risk is mainly high-high and low-low agglomeration.According to the spatial autocorrelation characteristics,the risk was divided into three types:‘overall random’,‘local random-neighborhood aggregation’,and‘overall aggregation’.(4)In addition to the strong correlation between population density,economic activity,and disaster risk,ecological service supply also has a high contribution rate to disasters,and the broken and uneven ecological landscape pattern can exacerbate the risk of flood disasters.Finally,according to the spatial characteristics of risk,the high risk in Min delta was divided into“Jimei Section”,“Quanzhou Bay Section”,“Zhangzhou County Section”.