Abstract：A Probabilistic risk assessment (PRA) was used to characterize variability and uncertainty of 17 parameters related to human exposure and 5 related to soil physicochemical properties in deriving the soil cleanup levels of 8 pollutants in a coking plant. Results showed that the ratios of PRA clean-up level to DRA clean-up level were 1.11~2.49, except for Nap in subsoil (0.9), therefore, the cleanup levels derived by using deterministic risk assessment (DRA) method were in general more conservative than these by using PRA in this case. If contaminants in soil were reduced to the PRA clean-up levels, all the exposure risks would be acceptable but at different levels, and Ben in topsoil was more likely to cause harm to human health than others. The sensitivity analysis revealed that adult exposure duration (EDa) and child exposure duration (EDc) both contributed most to the uncertainty (sensitive ratio: 35%~59.8% for EDa and 6.2%~20.2% for EDc) in the derivation of all the clean-up levels. Soil physic-chemical parameters, such as soil organic carbon content (foc), volumetric air content in soil (θair,vad) and volumetric water content in soil (θwater,vad), were more sensitive to volatile Ben and Nap and less to semi-volatile BaA~DBA. Contrastingly child soil ingestion rate (Irs-c) were more sensitive to BaA~DBA and less to Ben and Nap.
