Modeling arsenic in European topsoils with a coupled semiparametric (GAMLSS-RF) model for censored data

Area of Science:

• Environmental Science
• Geochemistry
• Risk Assessment

Background:

• Arsenic (As) is a toxic metalloid with industrial uses, posing health risks via ingestion and inhalation.
• Elevated soil arsenic concentrations result from agricultural practices and environmental factors.
• Existing soil arsenic databases, like LUCAS, have high detection limits hindering accurate analysis.

Purpose of the Study:

• To develop a novel method for modeling arsenic contamination in European soils.
• To overcome limitations of existing soil arsenic databases and analytical methodologies.
• To create high-resolution arsenic concentration maps for risk assessment and policy support.

Main Methods:

• Introduction of the GAMLSS-RF model, combining Random Forests with Generalized Additive Models for Location, Scale, and Shape.
• Development of a semiparametric model capable of handling non-linear interactions and censored data.
• Calibration with multiple databases and validation of a spatial model for European-scale mapping.

Main Results:

• Generated European-scale arsenic concentration maps at 250m spatial resolution.
• Identified significant variability in arsenic levels across Europe.
• Observed lower concentrations in Northern countries and higher concentrations in Portugal, Spain, Austria, France, and Belgium.

Conclusions:

• The GAMLSS-RF model effectively addresses limitations in censored data and existing databases for soil arsenic analysis.
• The developed maps provide a valuable probabilistic tool for assessing arsenic contamination risks.
• This approach supports informed policy-making for environmental and public health protection regarding arsenic exposure.