基于神经网络和GIS的耕地土壤铊含量分布及其驱动因素

Distribution and driving factors of thallium in cultivated soil based on NNET and GIS

采用多元统计、GIS和神经网络相结合的方法研究了耕地土壤Tl含量的空间分布格局及其主要驱动因素.结果表明,研究区耕地土壤Tl环境质量较好.研究区耕地土壤Tl含量范围为0.01~7.33mg/kg,算术平均值(0.91±0.40)mg/kg,超过湖南省土壤Tl含量背景值(0.61mg/kg),耕地Tl含量累积特征明显,但明显低于2017年全国农用地土壤污染状况详查工作中土壤Tl含量评价参考值(1.0mg/kg).耕地土壤中Tl含量流域分布特征明显,耕地土壤Tl含量呈现“东高西低,南高北低”趋势,高值聚类区主要分布湘江流域南部地区和武水流域.武水流域耕地土壤Tl含量算术均值达到1.10mg/kg,存在潜在污染风险.研究区耕地土壤Tl含量积累和分布格局受水力、工业活动和自然因素的综合影响.高海拔和水网密度区耕地土壤Tl累积明显.黄棕壤的土壤Tl含量显著高于其他类型土壤.

This study systematically collected surface soil samples from farmlands in Hunan Province and employed a combination of multivariate statistics,GIS,and neural networks to investigate the spatial distribution patterns of soil Tl content and its main driving factors.The results indicate that the soil Tl content ranged from 0.01 to 7.33mg/kg,with an arithmetic mean of(0.91±0.40)mg/kg,which exceeded the background value of soil Tl content in Hunan Province(0.61mg/kg).The cumulative characteristics of soil Tl contents were evident despite its significantly lower than the national reference value(1.0mg/kg).The spatial distribution of soil Tl contents exhibited distinct characteristics at the watershed level,showing a trend of“east high,west low,south high,north low”.High-value clusters of soil Tl content primarily presented in the southern part of the Xiangjiang River basin and the Wushui basin.The arithmetic mean soil Tl content in Wushui basin was up to 1.10mg/kg,indicating a potential pollution risk.Hydraulic forces,industrial activities,and other natural factors were important variables driving the accumulation and distribution of Tl content in the cultivated soils in Hunan Province.Tl accumulation in the farmland soils was significant in high-altitude areas and the areas with a dense network of water bodies.The Tl content in yellow-brown soil was significantly higher than in other soil types.