Random forest insights in prioritizing factors and risk areas of soil polycyclic aromatic hydrocarbons in an urban agglomeration area.

Changes in the soil environment due to urbanization pose a challenge to predict the health risks from soil PAHs on a regional scale. The sources and distributions of PAHs in the soil of the Changsha-Zhuzhou-Xiangtan Urban Agglomeration in China were analyzed using positive matrix factorization (PMF), GIS mapping, and the Random Forest (RF) model to estimate the spatiotemporal variations in health risks. The measured concentration of PAHs in the soil ranged from 20.1 to 15,233.1 μg/kg with a mean of 493.2 μg/kg. PMF and diagnostic ratios identified the primary sources of soil PAHs as transportation and coal combustion (59.1 %), soil-air exchange processes (20.5 %), and petroleum products (20.4 %). Transportation and industrial coal combustion contributed to 81.9 % and 81.7 % of the total health risks to children and adults, respectively. Locations of major point sources were identified by PMF and GIS mapping and were distributed at the congestive overpasses in urban areas and industrial enterprises carrying out large-scale coal combustion. The RF model suggested an increase, by 33.2 %, of areas with ILCRs exceeding the risk threshold if the population increases by 50.0 % in 2035. Areas with a PD > 500 p/km2, RD > 2 km/km2, and DNS < 15 km pose a high probability of health risks on local residents. The integrated approach enhanced the reliability of apportioning sources and risk assessment, serving as a reference for early warnings and a basis for soil surveys in urban areas.