基于BP模型的南通市农田土壤重金属空间分布研究

Study on spatial distribution of farmland soil heavy metals in Nantong City based on BP-ANN modeling

利用采样点实测数据,借助BP神经网络模型并结合GIS技术对江苏省南通市农田土壤重金属的空间动态分布进行了详细地描述。结果表明,BP神经网络模型能够智能地学习各个采样点的空间位置与该点各重金属含量之间的映射关系,并能够稳健地对各个空间插值点处的土壤重金属含量进行预测。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。在运用神经网络模型进行空间插值了解重金属空间动态分布的基础上,可以根据污染的分布状况确定农产品的生产布局和规划。

The present paper is aimed at presenting its authors＇ description of the spatial dynamic distribution of heavy metal content in the farmland soils by selecting Nantong, Jiangsu, as the targeted area. In the given research, all the statistical data are processed on a BP Artificial Neural Network Model and problem-solving tasks through GIS technology. The results of our description prove that ANN modeling can not only enable us to grasp intelligently the mapping relationship between spatial position and heavy metal content, but can also help to predict robustly the heavy metals content in every spatial interpolating dot. by using Arcgis analysis. As the result of our study , it is found that there are two kinds of soil heavy metals, Pb and As, are the most serious pollutants in the local farmland soils with others relatively mild; in the whole main region of Nantong district. Tongzhou city and Rudong city, as a matter of fact, prove to belong to lower pollution areas, though some sections, such as the township areas in Nantong, Haimeng and Qidong, heavy metals pollution prove to be very severe. However, the least soil heavy metals content region is located in Rugao town and Haian town. Thus, the description of the spatial dynamic distribution of heavy metal content in the farmland soils proves to be conformed with the actual tested facts perfectly in the regions to be covered. Thus, all the results of our study helps to provide a new thought and model for such heavy metals distribution objectives and serve for crop-planting and growing layout perfectly.