UWA School of Agriculture and Environment, The University of Western Australia, M079, 35 Stirling Highway, Perth, WA 6009, Australia.
Sci Total Environ. 2018 Jun 15;627:1106-1120. doi: 10.1016/j.scitotenv.2018.01.332. Epub 2018 Feb 5.
Urban environments are dynamic and highly heterogeneous, and multiple additions of potential contaminants are likely on timescales which are short relative to natural processes. The likely sources and location of soil or sediment contamination in urban environment should therefore be detectable using multielement geochemical composition combined with rigorously applied multivariate statistical techniques. Soil, wetland sediment, and street dust was sampled along intersecting transects in Robertson Park in metropolitan Perth, Western Australia. Samples were analysed for near-total concentrations of multiple elements (including Cd, Ce, Co, Cr, Cu, Fe, Gd, La, Mn, Nd, Ni, Pb, Y, and Zn), as well as pH, and electrical conductivity. Samples at some locations within Robertson Park had high concentrations of potentially toxic elements (Pb above Health Investigation Limits; As, Ba, Cu, Mn, Ni, Pb, V, and Zn above Ecological Investigation Limits). However, these concentrations carry low risk due to the main land use as recreational open space, the low proportion of samples exceeding guideline values, and a tendency for the highest concentrations to be located within the less accessible wetland basin. The different spatial distributions of different groups of contaminants was consistent with different inputs of contaminants related to changes in land use and technology over the history of the site. Multivariate statistical analyses reinforced the spatial information, with principal component analysis identifying geochemical associations of elements which were also spatially related. A multivariate linear discriminant model was able to discriminate samples into a-priori types, and could predict sample type with 84% accuracy based on multielement composition. The findings suggest substantial advantages of characterising a site using multielement and multivariate analyses, an approach which could benefit investigations of other sites of concern.
城市环境是动态且高度不均匀的，与自然过程相比，在相对较短的时间内，可能会有多种潜在污染物的添加。因此，应该可以使用多元素地球化学组成结合严格应用的多元统计技术来检测城市环境中土壤或沉积物污染的可能来源和位置。在澳大利亚西部大都市珀斯的罗伯逊公园，沿着相交的横断面采集了土壤、湿地沉积物和街道灰尘样本。对这些样本进行了多种元素（包括 Cd、Ce、Co、Cr、Cu、Fe、Gd、La、Mn、Nd、Ni、Pb、Y 和 Zn）的近全浓度分析，以及 pH 值和电导率分析。罗伯逊公园内某些位置的样本中含有高浓度的潜在有毒元素（健康调查限值以上的 Pb；生态调查限值以上的 As、Ba、Cu、Mn、Ni、Pb、V 和 Zn）。然而，由于主要的土地用途是娱乐性开放空间，超过指导值的样本比例较低，而且最高浓度往往位于较难进入的湿地盆地内，因此这些浓度的风险较低。不同组别的污染物的不同空间分布与与该场地历史上的土地利用和技术变化有关的不同污染物的输入一致。多元统计分析加强了空间信息，主成分分析确定了与空间相关的元素的地球化学关联。多元线性判别模型能够根据多元素组成将样本分为预先确定的类型，并且可以以 84%的准确率预测样本类型。研究结果表明，使用多元素和多元分析来描述一个场地具有很大的优势，这种方法可以有益于其他关注地点的调查。
