Abstract: Groundwater pollution risk assessment is a critical foundation for the prevention and control of groundwater
pollution. Dense Non-aqueous Phase Liquid ( DNAPL ) pollutants exhibit distinct physicochemical properties and complex migrationtransformation behaviors in groundwater, posing substantial threats to ecological and environmental integrity. Therefore, urgent efforts are required to conduct risk assessments for groundwater contaminated by DNAPLs. In this paper, a novel groundwater DNAPL pollution risk-assessment framework is proposed based on field-scale pollution exceedance duration. By constructing an ideal three-dimensional heterogeneous soil-groundwater model for PCE transport, the spatial-temporal distribution of contamination is simulated. Risk classification criteria are developed using site-boundary concentration exceedance time as the partitioning indicator. Bayesian theory is employed to quantify model uncertainty, and the MCMC algorithm is adopted for parameter calibration. Process-based numerical simulations are applied to analyze the impacts of model uncertainty on riskassessment outputs. The results indicate that the proposed evaluation method can reasonably chracterize risk levels under different remediation scenarios, and the inclusion of model uncertainty can markedly enhance the accuracy and reliability of the risk assessment results.

Key words: groundwater pollution risk assessment, groundwater numerical simulation, model uncertainty, dense non-aqueous
liquid