基于神经网络的农田土壤重金属空间分布分析

以江苏省南通市为研究区，通过野外采样，利用采样点实测数据，在借助神经网络模型进行空间插值的基础上。结合3S技术对农田土壤重金属的空间动态分布进行分析，进而确定农田土壤重金属污染状况。运用Arcgis进行的分析结果显示．在该地区Pb和As造成的污染最严重，其他重金属污染相对较轻，其中南通市区、海门市和启东市重金属富集最严重；南通大部、通州、如东大部分地区含量较少，含量最少的地区是如皋市和海安县。造成这种空间分布变异性的主要原因是经济发达的地区污染源较多而且集中，而经济欠发达区污染源则相对较少而且分散。重金属污染物通过水系导致农田污灌也是重金属产生空间变异性的重要原因。

Analysis of Spatial Distribution of Soil Heavy Metals in Farmland Based on Artificial Neural Networks

Using Nantong city in Jiangsu province as research region, collecting data by field sampling and experimental analysis, resorting to Artificial Neural Networks modeling and 3S technology to traverse the problems, the spatial dynamic distribution of farmland soil heavy metals and their pollution level were described. The results prove that ANN modeling can not only learn intelligently the mapping relationship between spatial position and heavy metal content, but also predict robustly heavy metals content in every spatial interpolating dot. Based on ANN spatial interpolation, by using Arcgis analysis, two kinds of soil heavy metals, Ph and As, are the most serious pollutants in local farm- land soil, others relatively mild; and farmland soil in large region in Nantong district, Tongzhou city and Rudong city belongs to light pollution level, but some fractions, such as Nantong in town area, Haimeng city and Qidong city, soil heavy metals enrichment is very severe, its pollu- tion problem must be highly regarded. The content of soil heavy metals in Rugao city and Haian city is least. The spatial heterogeneity of soil heavy metals mainly contributes to many more concentrative contaminant sources in developed regions than in developing regions, besides, heavy metal pollutants caused by farmland irrigation is also important. The analysis on spatial distribution and pollution assessment of soil heavy metals fit well with the status in the research region, which can serve effectively for local crops layout, developing high quality primary products, etc.