不同空间聚类算法在识别优质集聚基本农田图斑中的应用及对比研究  

Application and Comparison of Spatial Clustering Algorithms in Prioritizing Protected Plots of Basic Farmland

作　　者：杨建新 刘静 龚健 刘玉铃 朱江洪 YANG Jianxin;LIU jing;GONG jian;LIU Yuling;ZHU Jianghong(School of Public Administration,China University of Geosciences(Wuhan),Wuhan 430074,China;Key Laboratory of the Ministry of Land and Resources Law Evaluation,Wuhan 430074,China;Institute of Territorial Space Planning in Qinghai Province,Xining 810006,China)

机构地区：[1]中国地质大学(武汉)公共管理学院,湖北武汉430074 [2]自然资源部法治研究重点实验室,湖北武汉430074 [3]青海省国土空间规划研究院,青海西宁810006

出　　处：《中国土地科学》2022年第1期107-117,共11页China Land Science

基　　金：国家自然科学基金青年项目(42101275);国家自然科学基金面上项目(42071254,41871172)。

摘　　要：研究目的:探索基于不同空间聚类指数排序识别空间集聚的高质量永久基本农田保护图斑,以期为当前国土空间规划中永久基本农田调整补划和布局优化提供方法借鉴。研究方法:分别利用规则空间聚类算法Local Moran’s I和Getis-Ord Gi*以及非规则空间搜索聚类算法AMOEBA构建能在地块尺度上同时反映耕地质量及其空间集聚信息的表征指数,设计相应的排序优选方案,进而快速识别指定数量的永久基本农田保护图斑,并进行对比及邻域敏感性分析。研究结果:(1)基于AMEOBA算法输出聚类指数并设计相应的排序方案能在研究区识别空间上更为集聚同时具有较高质量的永久基本农田保护图斑;(2)基于邻接关系的邻域定义方式能在研究区取得比基于空间距离的邻域定义方式更好的识别结果。研究结论:应用空间聚类算法计算同时指示耕地质量高低及其空间聚散性的地块级表征指数,并设计相应优选排序方案识别高质量集聚永久基本农田保护图斑是有效可行的。The purpose of this study is to prioritize spatially aggregated high-quality basic farmland spots for permanent protection by calculating and ranking spatial clustering index generated by spatial clustering algorithms.The research methods are as follows.Three spatial clustering methods,i.e.,Local Moran’s I,Getis-Ord Gi*and AMOEBA,are respectively applied to calculate the clustering indices and design their corresponding ranking scheme,based on which permanent basic farmland plots of a certain amount are prioritized.Permanent basic farmland spots of a certain amount identified by the three designed ranking schemes and by the quality-based ranking scheme were comprehensively analyzed and compared in two aspects,i.e.,spatial agglomeration and average quality,to validate the feasibility of the ranking schemes.Also,the neighborhood sensitivity of the results was investigated.The results show that 1)ranking farmland spots in order of the spatial clustering index generated by the AMOEBA algorithm can help identify more spatially concentrated permanent basic farmland plots with higher quality in the study area;2)the contiguity-based neighborhood definition is more suitable than the distance-based neighborhood definition in prioritizing spatially aggregated high-quality plots of permanent basic farmland.In conclusion,the results verify the feasibility and suitability of spatial clustering algorithms in guiding practitioners to prioritize spatially aggregated high-quality permanent basic farmland in a simple and straightforward way.

关 键 词：空间聚类 排序 耕地质量 永久基本农田 空间邻域 敏感性分析 

分 类 号：F301.24[经济管理—产业经济]