基于模糊k-均值算法模型的土壤数值化分类——以河南省境内分布的雏形土为例  

Numerical Soil Classification Based on Fuzzy k-means Algorithm Model——A Case Study of Cambisols in Henan Province

作　　者：孙亚洲 陈杰[1] 吴克宁[2] 李玲[3] 韩杏杏 王海洋 

机构地区：[1]郑州大学水利与环境学院,郑州450001 [2]中国地质大学(北京)土地科学技术学院,北京100083 [3]河南农业大学资源与环境学院,郑州450001

出　　处：《土壤》2017年第6期1262-1267,共6页Soils

基　　金：国家科技基础性工作专项(2008FY110600);国家自然科学基金项目(40971128)资助

摘　　要：本研究基于《中国土系志·河南卷》中的97个雏形土土壤样本点,根据土壤系统分类标准遴选并确定用于划分雏形土纲中不同层级分类单元的16个土壤诊断层和诊断特性,并依据这些不同指标的属性特征值和不同层级指标的权重建立属性数据集。基于此数据集运用模糊k-均值算法模型,确定6个聚类中心土壤,计算样本土壤与中心土壤之间的分类距离,并对研究区97个土壤样本实施数值化分类。研究表明,基于97个样本点与各中心土壤之间的分类距离进行土壤数值化分类,并通过检验数值化土壤分类结果与谱系式层级土壤分类单元的一致性,可以评价传统的谱系式层级土壤分类系统的合理性。According to Chinese Soil Taxonomy（CST）, we selected 16 soil diagnostic horizons and diagnostic characteristics, which can be used to classify cambisols at different levels based on 97 cambisol soil profiles in Soil Series of China （Henan）, and we built a data matrix with different characteristics and hierarchy weights of these indexes. Then we extracted 6 centroid soil families by fuzzy k-means algorithm model, and calculated the taxonomic distance between sampled and centroid soil families, meanwhile numerically classified the 97 cambisol soil profiles. We find that the rationality of the traditional hierarchic CST can be evaluated according to the consistence of traditional hierarchic classifying units and the numerical soil classification derived from the taxonomic distance between sampled and centroid soil families.

关 键 词：模糊k-均值算法模型 土壤数值化分类 雏形土 

分 类 号：S159.9[农业科学—土壤学]