基于主分量模糊c-均值算法的区域土地利用分区方法探讨——以广东省大埔县为例

ANALYSIS ON REGIONAL LAND USE ZONING BASED ON FUZZY C-MEANS ALGORITHM OF PRINCIPAL COMPONENTS ——A CASE STUDY OF DAPU COUNTY IN GUANGDONG PROVINCE

土地利用分区本质上是一个分类问题,应用模糊分类的思想探讨土地利用最优分区数目策略,可为区域土地利用最优区划提供决策辅助。然而,目前关于土地利用分区的研究多是应用传统硬分类的聚类方法,分类数目带有很强的主观性,忽视了最优分区数目的客观存在。本研究应用因子分析、c-均值和模糊c-均值算法,并在遵循有效最大隶属原则的基础上,提出了一种土地利用最优分区数目的定量计算方法,并以广东省大埔县土地利用分区为例进行了实证分析,结果表明该方法能够较好的解决小样本数据集的土地利用分区数目最优的难题,同时对海量样本数据集多分类的效果可以提供后评估支持。

Land use zoning is essentially a classification problem,so how to apply the idea of fuzzy classification to discuss the optimal regionalization number strategy of land use shall be valuable for providing decision assistance for regional land use optimal zoning.But current research on land use zoning is usually conducted by adoption of traditional hard classification clustering method,of which the classification number is of a strong subjectivity,ignoring the existence of the objectivity of optimal zoning number.This study proposed a quantitative calculation method for land use optimal regionalization number by applying factor analysis,c-means and fuzzy c-means algorithm complying with the efficient maximum membership principle.After a case study of Dapu County in Guangdong Province,the results showed that the method was a better solution in solving the puzzle of optimal classification number determination in land utilization zoning with small sample dataset,meanwhile a post-assessment support could be provided for the multi-classification effect of massive sample dataset by this method.