Development of a predictive model for lead, cadmium and fluorine soil-water partition coefficients using sparse...

Area of Science:

• Environmental Science
• Soil Science
• Geochemistry

Background:

• Understanding soil adsorption characteristics is crucial for predicting contaminant fate and transport.
• Identifying key soil properties that influence adsorption can simplify environmental risk assessments.
• Cadmium, lead, and fluorine are important soil contaminants with varying adsorption behaviors.

Purpose of the Study:

• To identify easily obtainable soil properties for predicting adsorption coefficients (K and n values).
• To develop and validate sparse multiple linear regression (SMLR) models for estimating K and n values for cadmium, lead, and fluorine.
• To clarify the relationships between soil properties and the adsorption of these elements.

Main Methods:

• Application of sparse multiple linear regression (SMLR) analysis.
• Construction of predictive models using nine soil parameters (pH, CEC, SSA, C, SOM, WHC, sand, silt, clay).
• Validation of SMLR models using root mean square error (RMSE) on 17 soil samples and 512 parameter combinations.

Main Results:

• SMLR models accurately predicted K and n values for cadmium, lead, and fluorine adsorption.
• Fluorine adsorption correlated with soil organic matter, while cadmium and lead adsorption were influenced by soil pH.
• Over 65% of predicted K values and 76% of predicted n values were within acceptable ranges of measured values.

Conclusions:

• Soil physical and chemical properties can effectively predict the adsorption behavior of cadmium, lead, and fluorine.
• The developed SMLR models offer a practical tool for estimating adsorption properties in environmental studies.
• Easily measured soil parameters provide a reliable basis for assessing heavy metal and fluoride adsorption in soils.