# Assessing PM2.5 pollution in the Northeastern United States from the 2023 Canadian wildfire smoke: an episodic study integrating air quality and health impact modeling with emissions and meteorological uncertainty analysis

## TL;DR

This study used modeling to assess PM2.5 pollution from Canadian wildfires in the US Northeast, showing how it affected air quality and health, even in areas without monitors.

## Contribution

A new integrated modeling system (WRF-CMAQ-BenMAP) was developed to estimate wildfire-related PM2.5 and health impacts in regions lacking air quality monitoring.

## Key findings

- CMAQ simulations matched PM2.5 observations well (R² ∼0.6, slope ∼0.9).
- Wildfire emissions dataset choice can alter PM2.5 simulations by up to 40 µg m⁻³.
- Asthma-related emergency department visits increased up to ∼40% during the event.

## Abstract

Between June 6 and 8, 2023, wildfires in Quebec, Canada generated massive smoke plumes that traveled long distances and deteriorated air quality across the Northeastern United States (US). Surface daily PM2.5 observations exceeded 100 µg m−3, affecting major cities such as New York City and Philadelphia, while many areas lacked PM2.5 monitors, making it difficult to assess local air quality conditions. To address this gap, we developed a WRF-CMAQ-BenMAP modeling system to provide rapid, spatially continuous estimates of wildfire-attributable PM2.5 concentrations and associated health impacts, particularly benefiting regions lacking air quality monitoring. CMAQ simulations driven by two wildfire emissions datasets and two meteorological drivers showed good agreement with PM2.5 observations, with linear regression results of R2∼0.6 and slope ∼0.9. We further quantified uncertainties introduced by varying emissions and meteorological drivers and found the choice of wildfire emissions dataset alone can alter PM2.5 simulations by up to 40 µg m−3 (∼40%). Short-term health impacts were evaluated using the BenMAP model. Validation against asthma-associated emergency department (ED) visits in New York State confirmed the framework’s ability to replicate real-world outcomes, with ED visits increased up to ∼40%. The modeling results identified counties most severely affected by wildfire plumes, the majority of which lack regulatory air quality monitors. Our approach highlights the value of integrated modeling for identifying vulnerable populations and delivering timely health burden estimates, regardless of local monitoring availability.