基于多源数据的盐碱地精确农作管理分区研究

Classification of management zones for precision farming in saline soil based on multi-data sources to characterize spatial variability of soil properties

为了便于对盐碱地实施变量管理和精确农作,以海涂围垦区盐碱土为研究对象,以NDVI数据、盐分数据以及作物产量数据作为分区变量,对一面积为15hm2的盐碱地农田进行了基于多个数据源的精确农作管理分区研究。利用模糊c均值聚类方法进行分类分区,引入了模糊聚类指数(FPI)和归一化分类熵(NCE)作为最佳分区数目的判断标准,通过单项方差分析对分区结果进行比较和评价。研究发现,对本研究区,最佳的分区数目为三个。不同管理分区之间土壤化学性质(EC1:5,有机质,速效磷,速效钾,全氮,碱解氮以及阳离子交换量)的均值都存在着统计意义上的显著差异性,其中子区3具有最高的肥力水平和作物生产能力而子区1最低。利用所选取的三个变量,模糊c均值聚类算法可以较好地进行精确农作管理分区划分。分区结果不但可以指导采样,而且可以作为变量管理的决策单元用于田间变量管理作业中,为精确农业变量投入的实施提供有效手段和决策依据。

Recent research in precision agriculture has focused on use of management zone as a method to more efficiently apply crop inputs and precise soil management. In this paper, the variables of NDVI data, soil salinity data and cotton yield were selected as clustering variables and fuzzy c-means clustering algorithm was used to define management zone in an about 15 hm^2 field in a coastal saline land, fuzzy performance index and normalized classification entropy were used to determine the optimal number of clusters. The results revealed that the optimal number of management zones for the study area was 3. To assess whether the defined management zones can be used to characterize spatial variability in soil chemical properties, 224 georeferenced soil sampling points were examined by using One-way variance analysis. It was found that there exist significantly statistical differences between the chemical properties of soil samples in each defined management, and management zone 3 presented the highest nutrient level and potential crop productivity, whereas management zone 1 worst. The results reveal that fuzzy c-means clustering algorithm can be used to delineate management zones by using the given three variables. The defined management zones can not only be useful for the sampling design, but provide an effective decisionmaking support for variable input in precision agriculture.