黄土高原台塬区域麦田土壤养分的管理分区

Management Division of Soil Nutrient of Wheat Field in Tableland Area of Loess Plateau

为探索适宜黄土高原边缘台塬区域麦田土壤养分的管理分区方法,找出对产量影响较大的土壤养分因子,以闻喜县麦田作为研究对象,将地形因子、土壤类型和土壤养分作为输入因子,分别采用K均值聚类、模糊C均值聚类、分级分区3种方法进行管理分区,以变异系数作为评价指标,找出适宜该区域麦田土壤养分的管理分区方法。结果表明,闻喜县的麦田地形因子、土壤养分的空间变异性较大,适宜进行管理分区;较未分区前,K均值聚类、模糊C均值聚类、分级分区均能降低麦田土壤养分的变异系数,其中,分级分区降低幅度最大,达27.07%;地形因子和土壤养分的主成分分析表明,地形因子对主成分的贡献较大,地形因子与土壤养分呈负相关关系。黄土高原边缘台塬区域地形复杂,地形因子掩盖了麦田土壤养分对产量的影响,导致K均值聚类和模糊C均值聚类分区不适宜该区域管理分区。分级分区是适合该区域的麦田土壤养分管理分区方式。

The objective is to explore the suitable management division of wheat yield in the tableland area of Loess Plateau,and make contributions to identify the limitations of soil nutrients on yield.Wenxi county was taken as the research object,using altitude terrain factor,soil types and soil nutrient as input factors,K-means clustering,fuzzy C-means clustering,grading partition were used for management,the variation coefficient and the output of each partition and its regression equation between yield and soil nutrient was used as evaluation indexes,which compared the mean K-means and fuzzy C-means partition and the classification of the graded partitioning results to find the appropriate management of the region partition method.The results showed that the spatial variability of terrain factors and soil nutrients of wheat fields in Wenxi county was large and suitable for management division.Compared with the non-subdivision,K-means clustering,fuzzy C-means clustering and graded subdivision could all reduce the coefficient of variation of soil nutrient,with the largest reduction of 27.07%in the graded subdivision.The principal component analysis of topographical factors and soil nutrients showed that topographical factors contributed a lot to the principal component,and topographical factors were negatively correlated with soil nutrients.The terrain of the plateau area on the edge of the Loess Plateau was complex,and topographic factors covered up the effect of soil nutrients on yield,which led to the inappropriate management division of K-means clustering and fuzzy C-means clustering.Graded partition is a suitable management partition for this region.