Source apportionment and driving factor identification for typical watersheds soil heavy metals of Tibetan Plateau based on receptor models and geodetector

Abstract

Abstract The identification and quantification of soil heavy metal (HM) pollution sources and the identification of driving factors is a prerequisite of soil pollution control. In this paper, the Sabaochaqu Basin of the Tuotuo River, located in the Tibetan Plateau and the headwater of the Yangtze River, was selected as the study area. The soil pollution was evaluated using geochemical baseline, and the source apportionment of soil HMs was performed using absolute principal component score-multiple linear regression (APCS-MLR), edge analysis (UNMIX) and positive matrix decomposition (PMF). The driver of the source factor was identified with the geodetector method (GDM). The results of pollution evaluation showed that the HM pollution of soil in the study area was relatively light. By comparison, UNMIX model was considered to be the preferred model for soil HMs quantitative distribution in this study area, followed by PMF model. The UNMIX model results show that source 1 (U-S1) was dominated by As, with a contribution rate of 53.31%; source 2 (U-S2) was dominated by Cd and Zn, whose contribution rates are 50.35% and 46.60% respectively; source 3 (U-S3) was dominated by Pb, with a contribution rate of 45.58%; source 4 (U-S4) was dominated by Cr, Cu, Hg and Ni, with contribution rates of 60.58%, 60.07%, 51.58% and 56.45%, respectively. The GDM results showed that the main driving factors of U-S1 were distance from lake (explanatory power q = 0.328) and distance from wind channel (q = 0.168), which were defined as long-distance migration sources. The main driving factors of U-S2 were parent material type (q = 0.269) and distance from Tuotuo river (q = 0.213), which were defined as freeze-thaw sources. The main driving factors of U-S3 were distance from town (q = 0.255) and distance from county road (Yanya Line) (q = 0.221), which were defined as human activity sources. The main drivers of U-S4 were V (q = 0.346) and Sc (q = 0.323), which were defined as natural sources. The GDM results of the 3 models were generally consistent with the analytical results of similar types of sources, especially the results of PMF model and Unmix model can basically verify each other. The research results can provide important theoretical reference for the analysis of HM sources in the soil of high-cold and high-altitude regions.

Keywords

• Source apportionment
• Driving factor identification
• Soil heavy metals
• Tibetan plateau
• Receptor models
• Geodetector