农田土壤重金属污染来源解析结果的准确性和时效性研究

At present, heavy metal pollution of farmland soil in China has been paid increasingly more attention. Source apportionment is a critical step for the control of heavy metal pollution. The heavy metals in farmland soil have a variety of sources, and the methods for source apportionment include receptor model and stable isotope. The receptor models have been widely used with lack of any check of the accuracy of their results. And the source apportionment was mostly based on the samples collcected in a single time, which could not reveal the change of the potential pollution sources. This project focuses on the farmland in Changxing County, Zhejiang Province. First, the content and distribution of heavy metals in farmland soil are studied in detail through field sampling and laboratory analysis. Then, the potential pollution sources are sampled to acquire their profiles after identification through local survey. Based on these, receptor models were carried out for source apportionment. The lead isotopic composition of some soil samples and the source samples are analyzed, and a Bayesian stable isotope mixing model is employed to apportionment the sources of lead in farmland soil. The result of lead isotope is used as a reference to assess the accuracy of receptor models. Besides, soil heavy metal data in different time are analyzed by receptor models, and the results together with remote sensing images and local survey information are used to quantify the sources of the input of heavy metals into soils between the two sampling times. The study results of this project have theoretical significance and application value for source apportionment of heavy metals pollution in farmland soil.

目前，我国农田重金属污染问题日益受到关注。探明重金属来源是进行农田重金属污染管控的关键步骤。但是，农田重金属来源多样且复杂，目前常用受体模型和稳定同位素等方法对土壤重金属污染进行来源解析。当前，受体模型在土壤重金属污染源解析尚停留在初步应用阶段，对于解析结果的准确性少有报道；此外，土壤重金属污染源解析仅在单一时间节点上进行，无法及时反映区域内污染源时间动态变化过程等。本项目以浙江省长兴县农田土壤为研究对象，采样分析了解农田重金属污染的现状；调研明确潜在污染源，测定源类成分谱；测定源类样品和部分典型土壤样品的铅稳定同位素，利用贝叶斯稳定同位素混合模型估算各源对土壤铅的贡献，以评价受体模型解析结果的准确性；并通过对不同时期的土壤重金属数据进行源解析，结合遥感影像、调查资料等，统计建模，以分析源解析结果的时效性。本项目的研究将为土壤污染源解析提供了重要的理论意义和技术支撑。

土壤是环境的重要组成部分，也是人类赖以生存的自然环境和农业生产的重要资源。然而，由于不合理的社会、经济发展，土壤环境遭受到了污染，对人体健康和安全构成了非常严重的潜在威胁。定性识别源类并量化各源类对于土壤重金属的贡献，是土壤重金属污染防治的前提。然而，国内外对于土壤污染溯源研究多停留在应用阶段，对于源解析结果的准确性可靠性尚无系统研究。本研究以浙江省台州市温岭市某农田土壤为主要研究对象，通过采集不同土层土壤样品，全面摸清研究区域土壤中的重金属污染特征和分布，并通过主成分分析、空间分析等手段，定性识别出区域内的主要污染源类。在此基础上，通过对土壤剖面不同土层的对比分析，证实潜在污染源类对于周围土壤的污染。最后，通过蒙特卡洛模拟方法，通过对土壤重金属进行重采样，模拟研究了不同采样策略对于田块尺度土壤重金属污染评价、来源识别的影响。结果表明，研究区域内农田土壤主要为镉污染，表层0-20cm土壤中镉含量最高为3.81mg/kg，随着土壤剖面深度增加，土壤中各元素含量趋于背景含量，表明深层土壤中无外界输入重金属，由此，可将未受明显人为活动影响的深层（80-100 cm）土壤中的重金属作为研究区域内土壤重金属的自然来源，认为两者有着近似的分布特征。蒙特卡洛模拟表明，采样密度是影响源识别结果的重要因素，当采样密度为每公顷2个采样点时，90%以上的模拟中使用主成分分析等源识别方法获得了正确了的因子数目。