Contribution of regional-scale fire events to ozone and PM2.5 air quality estimated by photochemical modeling approaches

Two specific fires from 2011 are tracked for local to regional scale contribution to ozone (O-3) and fine particulate matter (PM2.5) using a freely available regulatory modeling system that includes the BlueSky wildland fire emissions tool, Spare Matrix Operator Kernel Emissions (SMOKE) model, Weather and Research Forecasting (WRF) meteorological model, and Community Multiscale Air Quality (CMAQ) photochemical grid model. The modeling system was applied to track the contribution from a wildfire (Wallow) and prescribed fire (Flint Hills) using both source sensitivity and source apportionment approaches. The model estimated fire contribution to primary and secondary pollutants are comparable using source sensitivity (brute-force zero out) and source apportionment (Integrated Source Apportionment Method) approaches. Model estimated O-3 enhancement relative to CO is similar to values reported in literature indicating the modeling system captures the range of O-3 inhibition possible near fires and O-3 production both near the fire and downwind. O-3 and peroxyacetyl nitrate (PAN) are formed in the fire plume and transported downwind along with highly reactive VOC species such as formaldehyde and acetaldehyde that are both emitted by the fire and rapidly produced in the fire plume by VOC oxidation reactions. PAN and aldehydes contribute to continued downwind O-3 production. The transport and thermal decomposition of PAN to nitrogen oxides (NOx) enables O-3 production in areas limited by NOx availability and the photolysis of aldehydes to produce free radicals (HOX) causes increased O-3 production in NOx rich areas. The modeling system tends to overestimate hourly surface O-3 at routine rural monitors in close proximity to the fires when the model predicts elevated fire impacts on O-3 and Hazard Mapping System (HMS) data indicates possible fire impact. A sensitivity simulation in which solar radiation and photolysis rates were more aggressively attenuated by aerosol in the plume reduced model O-3 but does not eliminate this bias. A comparison of model predicted daily average speciated PM2.5 at surface rural routine network sites when the model predicts fire impacts from either of these fires shows a tendency toward overestimation of PM2.5 organic aerosol in close proximity to these fires. The standard version of the CMAQtreats primarily emitted organic aerosol as non-volatile. An alternative approach for treating organic aerosol as semi-volatile resulted in lower PM2.5 organic aerosol from these fires but does not eliminate the bias. Future work should focus on modeling specific fire events that are well characterized in terms of size, emissions, and have extensive measurements taken near the fire and downwind to better constrain model representation of important physical and chemical processes (e.g. aerosol photolysis attenuation and organic aerosol treatment) related to wild and prescribed fires. Published by Elsevier Ltd.