Source-Resolved PM2.5 Metals Emissions, Exposure, and Health Risks in California Using an Observation-Constrained Hybrid Model

Abstract Metals constitute a small fraction of PM 2.5 mass but can contribute disproportionately to its toxicity. Assessing their health risks requires spatially resolved estimates of individual metal concentrations and emission source contributions. However, sparse monitoring and uncertain source-specific metal profiles remain major barriers to such assessments. We developed an observation-constrained hybrid model integrating source-oriented chemical transport modeling and Bayesian inversion to constrain source-specific profiles and estimate emissions and population exposure for 23 PM 2.5 metals across 11 source sectors in California. The inversion improved modeling agreement with observations, reducing mean fractional errors from 0.835 to 0.511. For the noncarcinogenic effects, miscellaneous sources contributed 9.1% of emissions, 24.1% population-weighted exposure, and 57.7% of the hazard index, with Cd driving the hazard contribution. Industrial and miscellaneous sources each accounted for approximately 35% of excess lifetime cancer risk across six carcinogenic metals. Census-tract clusters characterized by substantial joint contributions from these sectors had the highest mean risks and were prominent in metropolitan Los Angeles. The framework provides observation-constrained source profiles and spatial exposure estimates for 23 metals, supporting epidemiological studies and community-level health assessments.