|
| Analysis of influencing factors and bioavailability prediction of soil heavy metals based on RF and MLR |
| Received:May 12, 2023 |
| View Full Text View/Add Comment Download reader |
| KeyWord:soil heavy metal;influencing factor;bioavailability prediction;random forest（RF）;multiple linear regression（MLR） |
| Author Name | Affiliation | E-mail | | PAN Yongxing | College of Environmental Science and Engineering, Guilin University of Technology, Guilin 541004, China | | | CHEN Meng | College of Environmental Science and Engineering, Guilin University of Technology, Guilin 541004, China
Guangxi Collaborative Innovation Center for Water Pollution Control and Water Security in Karst Area, Guilin 541004, China | cattlepen@163.com | | WANG Xiaotong | College of Earth Science, Guilin University of Technology, Guilin 541004, China | | | LIU Nan | College of Environmental Science and Engineering, Guilin University of Technology, Guilin 541004, China | |
|
| Hits: 1545 |
| Download times: 1879 |
| Abstract: |
| Taking a typical lead - zinc mining area in northern Guangxi as the research object, the single factor pollution index, risk assessment code（RAC）, multiple linear regression（MLR）, and random forest（RF）methods were used comprehensively to analyze the influencing factors of accumulation and bioavailability prediction of soil heavy metals（Pb, Zn, Cu, and Cr）quantitatively. The results showed that the Cr content was relatively evenly distributed spatially and did not exceed the background value（the coefficient of variation was 0.51）. The average values of the Cu, Pb, and Zn contents exceeded the background values（52.58, 280.31 mg·kg-1, and 654.71 mg· kg-1, respectively）, and the total amount and bioavailability were greater in front of the Sidi River mountain and at the entrance of the subterranean river, which presents a certain risk to the soil ecological environment. Among the factors influencing total heavy metal distribution and bioavailability, CEC, clay, SOM, and iron-aluminum oxides had a greater effect on Cr; SOM, clay, pH, and iron-aluminum oxides had a greater effect on Cu; pH, EC, and clay had a greater effect on Pb; and CEC, pH, soil texture, and iron-aluminum oxides had a greater effect on Zn. The bioavailability prediction results showed that both RF and MLR could better predict the total amount and secondary phases of soil heavy metals, with an R2 interval of 0.44-0.93 for RF and 0.30-0.72 for MLR. The RF prediction results were most accurate. |
|
|
|
