JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Explainable machine learning for arsenic remobilization potential in the vadose zone: Leveraging readily available soil properties

  • 0Water Cycle Research Center, Korea Institute of Science and Technology (KIST), Seoul 02792, Republic of Korea; Division of Energy and Environment Technology, KIST School, Korea University of Science and Technology, Seoul 02792, Republic of Korea.
Journal of Hazardous Materials +

|

April 27, 2025

|

April 27, 2025

View abstract on PubMed

Summary

This summary is machine-generated.

Repeated wet-dry cycles can mobilize significant amounts of arsenic (As) from the vadose zone, posing a groundwater contamination risk. A machine learning model predicts this arsenic remobilization using soil properties.

Area Of Science

  • Environmental Science
  • Soil Science
  • Geochemistry

Background

  • The vadose zone is crucial for preventing groundwater contamination by elements like arsenic (As).
  • Previous research indicates that repeated wet-dry cycles can remobilize substantial amounts of retained arsenic from soil.
  • Underestimating this remobilization risks underestimating potential groundwater contamination.

Purpose Of The Study

  • To quantify arsenic remobilization in the vadose zone under wet-dry cycles.
  • To develop a predictive model for arsenic remobilization based on soil properties.
  • To understand the influence of soil characteristics on arsenic mobility.

Main Methods

  • Utilized 22 unsaturated soil columns to simulate vadose zones with diverse soil properties.
  • Subjected As-retaining soil columns to repeated wet-dry cycles to measure remobilization.
  • Developed a machine learning model (random forest) to predict As remobilization using soil properties.

Main Results

  • Remobilization of 13.9–150.6 mg/kg of As (37.0–74.6% of retained As) was observed.
  • A predictive model was established using soil properties: organic matter (OM), iron (Fe) content, uniformity coefficient, D30, and bulk density.
  • Shapley additive explanation revealed D30 and OM content as key predictors of As remobilization.

Conclusions

  • Wet-dry cycles significantly remobilize arsenic from the vadose zone, highlighting a groundwater contamination risk.
  • The developed machine learning model effectively predicts arsenic remobilization potential using accessible soil properties.
  • Soil properties like D30 and OM content are critical factors influencing arsenic mobility in the vadose zone.

Keywords: