Determining thresholds of arsenic and mercury in stream sediment for mapping natural toxic element anomaly using data-driven models: a comparative study on probability plots and fractal methods

Natural toxic elements such as arsenic and mercury are widespread distributed in nature and the over accumulation of them usually has adverse effects on both environment and human beings. Therefore, to assess potential toxicity of these elements in natural environment such as water, soil, and stream sediment, it is significant to determine thresholds between anomalous and background geochemical data for identifying and mapping target areas with toxic element geochemical anomaly. In terms of data-driven, this study aims to demonstration of using the four statistical models that are box plot (BP), cumulative distribution function (CDF) diagram, concentration-number (C-N) fractal, and concentration-area (C-A) fractal methods to determine the threshold values of As and Hg in stream sediment for mapping their potential toxic anomaly in Pangxidong district, southwestern Guangdong, China, and compare the results critically. Firstly, a wide range of threshold values have been estimated by the four investigated data-driven methods. Then, the three threshold values of As and Hg that separate background data from low, moderate, and high geochemical anomalies determined by BP, CDF diagram, C-N fractal, and C-A fractal models were compared with the three level baselines defined in the Environmental Quality Standard for Soil in China (EQSSC). Furtherly, the geochemical anomaly maps of As and Hg were outlined by using the three threshold values to investigate the performance of the four data-driven methods in practical application of mapping natural toxic element anomaly. Finally, the contaminations of As and Hg in the study area were evaluated based on the suitable geochemical anomaly maps derived by the applicable data-driven method (i.e. CDF diagram). The main conclusions are that (1) of the four investigated data-driven methods, CDF diagram is the most suited for assist in the estimation of proper threshold values of As and Hg dividing geochemical background data from anomalies in the Pangxidong district; and (2) the contaminations of As mainly come from natural sources (i.e., bedrocks), while the contaminations of Hg are most likely to originate from anthropogenic factors (i.e., mining activities) in the Pangxidong district. The target areas, particular to the southern Wuling Reservoir, which have been identified with high As and Hg anomaly, probably should be taken effective measures to prevent the occurrences of pollution and toxicity from these elements.