一种土地生态敏感性评估的加权聚类方法

A weighted clustering model for land eco-environmental sensitivity evaluation

针对土地生态敏感性评价工作中,传统的分区评价方法多依赖于专家经验、主观性强且区域适用性差而导致评价结果差异性的问题,该文提出一种结合变异系数(CV)和自组织映射神经网络(SOM)相耦合的客观加权聚类评价方法。以长江经济带中游城市黄冈市为研究区,以土地生态敏感性为研究目标,从遥感、地形和土壤等多源数据中提取因子建立"目标-准则-指标"三层评价指标体系,构建CV-SOM耦合模型进行实验开展评价研究。研究结果表明,相较于传统的综合指数评价方法,CV-SOM模型的评价结果呈现出了更好的空间分区特性和区域连续性。同时,对生态敏感性的空间分布和特定影响因素进行了特征分析,以期为政府国土空间规划和土地环境保护提供理论支持和决策参考。

Aiming at the problem that traditional zoning evaluation methods in land eco-environmental sensitivity evaluation mostly rely on expert experience,strong subjectivity and poor regional applicability,which leads to the difference of evaluation results,an objective weighted clustering evaluation method combining coefficient of variation(CV)and self-organizing mapping neural network(SOM)was proposed in this paper.Taken Huanggang City,a city in the middle reaches of the Yangtze River Economic Belt,as the research area,with the land eco-environmental sensitivity as the research target,extracted factors from multi-source data such as remote sensing,topography and soil to establish a three-tier evaluation index system of"Target-Criteria-Index",and used the CV-SOM coupling model to conduct experimental research.The research results showed that the evaluation results of the CV-SOM model show better spatial zoning characteristics and regional continuity compared to the traditional comprehensive index evaluation method.At the same time,the characteristics of the spatial distribution of ecological sensitivity and specific influencing factors were analyzed.The experimental results were expected to provide theoretical support and decision-making reference for the government’s land and space planning and land environmental protection.