基于GeoSOM网络的生态修复分区——以黄河流域山西段为例  

Ecological restoration zoning based on the GeoSOM network:A case study of the Shanxi section of the Yellow River Basin

作　　者：李豪 张蕾 梁晓磊 刘庚 LI Hao;ZHANG Lei;LIANG Xiaolei;LIU Geng(School of Economics and Management,Taiyuan Normal University,Taiyuan 030619,Shanxi,China;Shanxi Key Laboratory of Earth Surface Processes and Resource Ecology Security in Fenhe River Basin,Taiyuan Normal University,Taiyuan 030619,Shanxi,China;School of Geography Science,Taiyuan Normal University,Taiyuan 030619,Shanxi,China)

机构地区：[1]太原师范学院经济与管理学院,山西太原030619 [2]汾河流域地表过程与资源生态安全山西省重点实验室,山西太原030619 [3]太原师范学院地理科学学院,山西太原030619

出　　处：《干旱区研究》2025年第2期321-332,共12页Arid Zone Research

基　　金：山西省哲学社会科学规划课题(2023YY220,2022YJ097);山西省高校科技创新项目(2023L240);山西省基础研究计划(202203021212192)。

摘　　要：生态修复是落实我国生态文明建设的重大工程,划定分区单元是实施国土整治与生态修复差别化建设的前提条件与重要基础,对制定差异化修复措施具有重要理论与指导意义。以黄河流域山西段为例,引入GeoSOM地理自组织特征映射算法对研究区基于栅格格网单元进行生态修复空间聚类,根据Dunn对聚类有效性进行评价并筛选最优方案,最后结合支持向量机进行生态修复分区界线识别,划定生态修复分区。结果表明:GeoSOM网络空间聚类后进行Dunn指数评价把研究区分为4大类,每一类中指标呈现明显的空间分异特征;支持向量机根据聚类结果识别出10个生态修复分区,并对每个分区提出生态修复策略。对传统SOM网络只强调专题属性的特点进行改进,采用的GeoSOM算法进行聚类可同时对专题属性与空间属性进行相似性度量,更符合空间聚类的特点,为生态修复分区方法提供了新的参考。Ecological restoration is a major project to implement China’s ecological civilization construction.Delineating zoning units is a prerequisite and an essential foundation for the differentiated implementation of land remediation and ecological restoration;it holds significant theoretical and guiding value for formulating differentiated restoration measures.Taking the Shanxi section of the Yellow River Basin as an example,this study introduces the GeoSOM(Geographic Self-Organizing Map)algorithm to perform spatial clustering for ecological restoration in the study area based on grid units.The Dunn index evaluated the effectiveness of the clustering and selected the optimal scheme.Finally,the Support Vector Machine(SVM)identified the boundaries of the ecological restoration zones,resulting in the delineation of the ecological restoration areas.The results indicate that,after the GeoSOM network spatial clustering,the study area was divided into four major categories according to the Dunn index evaluation,with each category exhibiting significant spatial differentiation characteristics.Based on the clustering results,the SVM identified ten ecological restoration zones,and ecological restoration strategies were proposed for each zone.This study improves the traditional SOM network,which emphasizes thematic attributes,by using the GeoSOM algorithm that measures the similarity of thematic and spatial attributes,making it more suitable for spatial clustering.The findings provide a new reference for methods of ecological restoration zoning.

关 键 词：GeoSOM网络 生态修复分区 边界识别 黄河流域山西段 

分 类 号：X171.4[环境科学与工程—环境科学]