Three-dimensional spatial variability of arsenic-containing soil from geogenic source in Hong Kong: Implications on sampling strategies

Soil contamination by trace elements such as arsenic (As) can pose considerable threats to human health, and need to be carefully identified through site investigation before the soil remediation and development works. However, due to the high costs of soil sampling and testing, decisions on risk management or mitigation strategies are often based on limited data at the site, with substantial uncertainty in the spatial distributions of potentially toxic elements. This study incorporates the restricted maximum likelihood method with three-dimensional spatial autocovariance structure, to investigate the spatial variability features of As-containing soils of geogenic origin. A recent case study in Hong Kong is presented, where >550 samples were retrieved and tested for distributions of As concentrations. The proposed approach is applied to characterize their spatial correlation patterns, to predict the As concentrations at unsampled locations, and to quantify the uncertainty of such estimates. The validity of the approach is illustrated by utilizing the multi-stage site investigation data, through which the advantages of the approach over traditional geostatistical methods are revealed and discussed. The new approach also quantifies the effectiveness of soil sampling on reduction of uncertainty levels across the site. This can become a useful indicator for risk management or mitigation strategies, as it is often necessary to balance between the available resources for soil sampling at the site and the needs for proper characterization of contaminant distributions. (C) 2018 Elsevier B.V. All rights reserved.