基于SOM的漳浦样区土壤分类研究

The Soil Classification Study Based on SOM in Zhangpu

选取闽东南漳浦沿海低丘台地区为样区,应用MATLAB 7软件编程,利用自组织特征映射网络(SOM网络)的聚类功能,对漳浦样区的9个土壤剖面进行分类,并与中国土壤系统分类结果进行比较,研究结果表明:SOM网络能将变性土与富铝土有效地区分开来,剖面1的归属也较为合理,可以很好地反映、提取土壤剖面的复杂信息,分类效果较好,可以在土壤分类中应用.

Tbe study selects coastal low- hilly area in Zhangpu in southeast of Fujian province as the research zone, appfies the clustering function of Self- Organizing Mapping network（SOM） to classify 9 soil profiles using MATLAB7 software. Through comparing the classification with the one of Chinese soil taxonomy, the results show that the SOM network can distinguish vertisols from ferrisols and classify the profile 1 reasonably, which can reflect the complicated information among each soil profiles and can be applied on soil classification.