基于灰关联谱聚类模型的福建省暴雨灾害评价研究

[目的]研究基于灰关联谱聚类模型的福建省暴雨灾害评价指标体系。[方法]根据福建省灾害性气象年鉴,综合考量致灾因子、孕灾环境、承灾体以及区域防灾水平,建立福建省区域暴雨灾害评价指标体系,并通过基于灰关联分析的谱聚类模型对福建省各区域进行暴雨灾害风险区划,最后通过实例研究对聚类模型效果以及应用进行了分析。[结果]为挖掘各区域特征内在联系,提高评价单元防灾联动性能,通过基于灰关联分析的谱聚类模型对福建省各区域进行暴雨灾害风险区划,并引用组合权重评判各指标优劣,对聚类模型进行调整。通过实例研究,得出了67县域暴雨灾情等级,分析各类别的内部特征,提取了每个类别的主要关联因子指标,并与已有暴雨灾情统计结果进行了对比,验证了该模型的有效性和可行性。[结论]该方法是可行的,其评价结果有更好的区分度与决策精度。

Study on Evaluation of the Rainstorm Disaster in Fujian Province Based on Spectral Clustering Model with Grey Relational Analysis

[Objective] The research aimed to study assessment index system of the rainstorm disaster in Fujian Province based on spectral clustering model with grey relational analysis.[Method] According to meteorological disaster yearbook in Fujian Province,by comprehensively considering disaster-inducing factor,disaster-inducing environment,disaster-sustaining body and regional disaster-prevention level,evaluation index system of the regional rainstorm disaster in Fujian was established.By spectral clustering model based on grey relational analysis,risk zoning of the rainstorm disaster was conducted in each area of Fujian.Finally,effect and application of the clustering model were analyzed by instance research.[Result] In order to dig immanent connection among regional characteristics and improve disaster-preventing linkage performance of the evaluation unit,a spectral clustering model based on grey relational analysis was used to conduct risk zoning of the rainstorm disaster in Fujian Province.Moreover,combination weight was introduced to judge each evaluation index,so as to adjust clustering model.By case study,rainstorm disaster levels in 67 counties were obtained.Interior characteristics of each type were analyzed,and main relational factor of each type was extracted.It was compared with statistical result of the rainstorm disaster,verifying validity and feasibility of the model.[Conclusion] The method was feasible,and its evaluated result had better differentiation and decision accuracy.