Mapping and communicating soil contamination risk: case study of soil lead in Greater Glasgow

Areas with elevated levels of soil contamination need to be mapped before risk can be mitigated. Effective communication of the expected risk is necessary. This study predicted soil lead (Pb) concentrations using quantile regression forests, an expansion of random forests, for a 200m 2 grid covering Greater Glasgow. The median and 90% confidence intervals for predicted soil Pb values at grid points were classified by the soil screening level they exceeded and then mapped. The probability of a location exceeding a high soil Pb threshold, and any hotspots, were also mapped to assist different types of end-users in decision making.