【摘要】 【目的】以天山北麓绿洲农耕区——新疆农八师148团农田为研究对象,以193个耕层土壤(0～30cm)有机质、碱解氮、速效磷和速效钾含量的分析数据为变量进行农田土壤养分精确管理分区研究。【方法】模糊c-均值聚类法被用来进行分区,以棉田产量为外部变量,采用FPI、c-φ多次组合法及基于外部变量的多元回归法来确定适宜的模糊控制参数。【结果】研究区最佳分区数为4,模糊指数为1.6。各管理分区土壤养分的变异系数都较分区前全研究区有所减小,而分区间土壤养分差异显著。研究区的平均混乱度指数为0.19,不同模糊类别交叠程度小,地理空间上土壤的隶属关系相对明确。【结论】通过选取适宜的外部变量,模糊c-均值聚类法可以较好地进行管理分区划分,分区结果可以作为变量施肥的单独作业单元进行耕作管理。更多还原

【Abstract】 【Objective】The objective of this research was to define management zones of oasis cotton field. The variables of organic matter, available N, available P and available K data determined in 193 top soil (0-30 cm) samples were selected as data sources.【Method】Fuzzy c-means clustering algorithm was used to delineate management zones. In order to determine the optimum fuzzy control parameters, the fuzziness performance index (FPI), c-φ combinations and the multiple regression based on external variable were used. Meanwhile, the cotton yield was chosen as the external variable.【Result】The whole field was divided into four management zones. And the fuzziness exponent was 1.6. The zoning statistics showed that variation coefficient of soil nutrients decreased, while the means of the soil nutrients differed markedly between management zones. The average confusion index was 0.19 in all management zones. The overlapping of fuzzy classes at points was low and the spatial distribution of membership grades was unambiguous【.Conclusion】The results indicated that fuzzy c-means clustering algorithm could be used to delineate management zones by selecting the appropriate external variables. The defined management zones can be used for fertilizer recommendation to manage soil nutrients more efficiently.更多还原