Coke production is an important source of environmental polycyclic aromatic compounds (PACs), including parent polycyclic aromatic hydrocarbons (PAHs) and their derivatives. The focus near coking plants has primarily been on parent-PAH contamination, with less attention given to highly toxic derivatives. In this study, soil samples were collected from both within and outside of a coking plant. The concentrations of parent-PAHs and their derivatives, including methylated-PAHs, oxygenated-PAHs, and nitrated-PAHs, were examined. Spatial interpolation was employed to determine their spatial distribution patterns. Methods for identifying potential sources and conducting incremental lifetime cancer risk analysis were used. This could achieve a comprehensive understanding of the status of PAC pollution and the associated health risks caused by coke production. The concentrations of total PACs inside the plant ranged from 7.4 to 115.8 mg/kg, higher than those outside (in the range of 0.2 to 65.7 mg/kg). The spatial distribution of parent-PAH concentration and their derivatives consistently decreased with increasing distance from the plant. A significant positive correlation (p < 0.05) among parent-PAHs and their derivatives was observed, indicating relatively consistent sources. Based on diagnostic ratios, the potential emission sources of soil PACs could be attributed to coal combustion and vehicle emissions, while principal component analysis–multiple linear regression further indicated that primary emissions and secondary formation jointly influenced the PAC content, accounting for 60.4% and 39.6%, respectively. The exposure risk of soil PACs was dominated by 16 priority control PAHs; the non-priority PAHs’ contribution to the exposure risk was only 6.4%.

In this study, a Long Short-Term Memory (LSTM) network approach is employed to evaluate the prediction performance of PM_2.5 in the Beijing–Tianjin–Hebei region (BTH). The proposed method is evaluated using the hourly air quality datasets from the China National Environmental Monitoring Center, European Center for Medium-range Weather Forecasts ERA5 (ECMWF-ERA5), and Multi-resolution Emission Inventory for China (MEIC) for the years 2016 and 2017. The predicted PM_2.5 concentrations demonstrate a strong correlation with the observed values (R² = 0.871–0.940) in the air quality dataset. Furthermore, the model exhibited the best performance in situations of heavy pollution (PM_2.5 > 150 μg/m³) and during the winter season, with respective R² values of 0.689 and 0.915. In addition, the influence of ECMWF-ERA5’s hourly meteorological factors was assessed, and the results revealed regional heterogeneity on a large scale. Further evaluation was conducted by analyzing the chemical components of the MEIC inventory on the prediction performance. We concluded that the same temporal profile may not be suitable for addressing emission inventories in a large area with a deep learning method.