中文摘要：

为了实现基本农田划定过程中的质优、集中并重的目标,以鹤峰县为例,结合产能核算、空间聚类、景观破碎度指数等方法探讨构建了一种基本农田划定的新方法。首先基于耕地地形条件、水利设施、产能状况、建设占用可能性和规划约束构建了5个因素12个因子的指标体系和域值型、空间扩散型的指标属性标准化方法以表征研究区耕地质量的空间差异;然后结合空间距离邻近度和景观破碎度指数修正K均值空间聚类法的空间距离和分类数两个参数,通过分析各分类数和耕地破碎度指数的变化趋势、耕地破碎度指数变化率与图斑最小面积的关系确定最佳分类数和最小剔除图斑面积;经K均值空间聚类再结合上级下达鹤峰县基本农田指标确定入选基本农田的耕地面积为16 172.35hm2,质量指数均大于73分,景观破碎度指数为1.26,实现基本农田划定的质优、集中。其结果为基本农田保护规划理论体系的完善和实际工作的深入开展提供参考和借鉴。

英文摘要：

Farmland protection is the foundation of national food security and social stability.In China,rapid industrialization and urbanization has catalyzed farmland resource deterioration.To assure a stable supply of farmland,we need a comprehensive plan.In a long-term food production plan,both quantity and quality of land are important.Prime farmland,which is defined as high-quality farmland in terms of fertility,location,and equipped irrigation system,is the most critical category of farmland as it can guarantee stable grain productivity.Due to its important role in food production in China,prime farmland should be reserved for agricultural use and thus requires continual monitoring and defragmentation.Conventional demarcation method of prime farmland planning,which has set a target in both quantity and quality of the farmland,often overlooks the productivity and location aspects,and thus has resulted in suboptimal efficiency.This paper aims to explore new demarcation method of prime farmland based on productivity and spatial clustering.To describe farmland quality,indexes for each land unit are constructed by three elements that consist of twelve factors.These factors,according to their attributes,can be categorized into productivity and spatial diffusion type.As each indicator has a value on a different scale,to eliminate the scale effect,the value of each indicator is normalized,based on which a comprehensive value is calculated for each land unit by weight-adding models.The comprehensive values of each unit are then used to describe the spatial differences of farmland quality in a region.We use K-means spatial clustering method and landscape fragmentation measure in the defragmentation process.The K-means spatial clustering method reduces fragmentation by accounting for both spatial proximity and attribute similarity of the land units.The optimal number of land categories can be determined by the point where the marginal reduction in the fragmentation index is maximized as the number of categories varies.Once the