Ambient Measurements of Hazardous Air Pollutants in the United States Routinely Exceed Predictions from Screening-Level Exposure Models

Hazardous air pollutant (HAP) emission regulations in the US often rely on modeled estimates of ambient exposures. Model accuracy compared to real-world measurements of HAPs is crucial for understanding and mitigating exposure and associated health harms. While previous work shows ambient measurements are higher than regulatory model estimates, the implications for health risk assessments are rarely discussed. We provide a comprehensive comparison of the modeled and measured concentrations at 489 US monitoring sites for 79 HAPs. We quantify how model-measurement discrepancies affect the estimation of the exposure and risk of adverse health effects. Measurements were higher than modeled concentrations in 74% of comparisons over all monitors, chemicals, and years assessed, with measurements a median 2 (IQR 1–9) times higher than model estimates. Measurements exceeded noncancer adverse health effect thresholds, while the model did not (model false negatives) for nine pollutants. Adjusting for model bias in two industrial centers, we found the number of people with multipollutant exposure above the US EPA’s acceptable excess lifetime cancer risk increased by a factor of 30 times in Houston, Texas, and 13 times in Baton Rouge, Louisiana. Our results imply that assessments relying exclusively on models like those we evaluated likely underestimate the spatial extent and magnitude of health hazards and risk.