综合污染物迁移机制和空间统计模型的土壤有机污染物空间分布预测

The spatial distribution of organic pollutants in soil is the basis of understanding the transport mechanism, evaluating the risk of contamination and developing relative environmental decisions. The spatial patterns of organic pollutants in soil were influenced by variety of environmental processes (such as diffusion, adsorption, volatilization, degradation, leaching). While mapping the spatial distribution of organic pollutants using traditional interpolation methods (inverse distance weighted, kriging, spline, et al.), only the concentration and spatial position information of the sampling point were considered, therefore, it is difficult to characterize the complexity and variability of the pollutant distribution. In this study, a site contaminated by organic pollutants was selected to investigate the spatial distribution characteristics of organic pollutants in soil, to explore the relationship of environmental factor (source location, soil physical and chemical properties, climatic conditions, et al.) and the spatial pattern of soil organic pollutants, to determine the contribution of each environmental factor to the spatial heterogeneity of organic pollutants and to identify the key environmental factors. Based on the idea of multilevel modeling, pollutant transport model, multiple linear regression models, geostatistical method, and other related methods were selected to simulate the different levels of the spatial variation of soil pollutants. Based on the results of different prediction models, a integrative prediction of spatial distribution of soil organic pollutants was built. The integrative prediction models combine the advantages of pollutant migration mechanism model and spatial statistical models and overcome the deficiencies of the traditional organic pollutants mapping using single prediction model. If the modeling process is reasonable, our approach will improve the accuracy of spatial prediction of soil organic pollutants.

土壤有机污染物空间分布是认识污染物迁移规律、评价污染风险和制定相关环境决策的基础。有机污染物空间分布是多种环境过程综合作用的结果。现有的空间分布预测方法（空间插值）只是考虑了采样点浓度和空间位置信息，很难表征污染物分布的复杂性和变异性。本研究拟以某有机物污染场地为例，研究土壤有机污染物的空间分布特征，探索有机污染物空间分布特征与环境因素的关系，确定不同环境因素对污染物空间分布异质性的贡献，识别关键环境因素。基于多层次建模思路，分别运用污染物迁移模型，多元回归模型，克里格等方法，模拟不同层次的污染物空间分布变异性。集成多种模型的模拟结果，构建土壤有机污染物空间分布预测方法。期望将污染物迁移机制模型和空间统计模型的优势相结合，克服传统单一预测模型的不足，提高土壤有机污染物空间分布制图的精度。

土壤污染物空间分布是认识污染物迁移规律、评价污染风险和制定相关环境决策的基础。场地污染物受污染源分布、地层条件、土壤性质等因素影响，在空间上的变异性较大。以某化工厂为例，利用Voronoi、空间自相关等空间统计方法，研究分析了场地土壤中氯代烃污染物在水平和垂直向的分布特征，识别污染物分布的热点区域，探讨了影响氯代烃土壤污染的空间分布的主要因素。结果表明，应用空间统计方法有助于准确认识污染物在不同地层的格局以及地层间分布的相互关系。.为了克服传统插值方法的平滑作用，我们利用地统计条件模拟方法进行土壤污染空间分布概率制图。并以某土壤重金属污染区为例，验证评估了土壤污染分布概率预测方法的效果。总的来看，概率方法能够避免对超标点位的平滑作用，提高污染区的预测精度。应用BP神经网络进行场地土壤污染分布预测。案例研究结果表明BP神经网络模型的土壤重金属污染分布预测结果同时具有传统插值方法和条件模拟方法的特性，既能反映重金属的空间分布规律，又表现了重金属含量在局部区域的变异性。.为提高土壤污染分布区域的预测精度，提出一种土壤污染调查加密布点优化方法。该方法的基本思路为在分析初步调查数据的基础上，利用地统计条件模拟预测土壤污染概率；结合污染概率和土壤污染物含量局部变异确定加密布点的优先区域；根据污染物浓度的空间变化趋势布设加密调查点位。以某场地土壤中镉污染为例，验证了布点优化方法的效果。结果表明，土壤污染调查加密布点方法显著提高了污染区面积和污染区空间位置的估计精度，同时降低土壤污染调查的样本量。.项目研究成果可用于指导污染场地调查的点位布设，污染物空间分布规律分析与污染分布制图，提高场地土壤污染调查的精度和效率。