Impact of wildfire smoke on ozone concentrations using a Generalized Additive Model in Salt Lake city, Utah, USA, 2006-2022.

We investigated the impact of wildfires on maximum daily 8-hour average ozone concentrations (MDA8 O 3 ) at four sites in Salt Lake City (SLC), Utah for May to September for 2006-2022. Smoke days, which were identified by a combination of overhead satellite smoke detection and surface PM 2.5 data and accounted for approximately 9% of the total number of days, exhibited O 3 levels 6.8 to 8.9 ppb higher than no-smoke days and were predominantly characterized by high daily maximum temperatures and low relative humidity. A Generalized Additive Model (GAM) was developed to quantify the impact of wildfire contributions to O 3 . The GAM, which provides smooth functions that make the interpretation of relationships more intuitive, employed 17 predictors and demonstrated reliable performance in various evaluation metrics. The mean of the residuals for all sites was approximately zero for the training and cross-validation data and 5.1 ppb for smoke days. We developed 3 approaches to estimate the contribution of smoke to O 3 from the model residuals. These generate a minimum and maximum contribution for each smoke day. The average of the minimum and maximum wildfire contributions to O 3 for the SLC sites was 5.1 and 8.5 ppb, respectively. Between 2006-2022, an increasing trend in the wildfire contributions to O 3 was observed in SLC. Moreover, trends of the fourth-highest MDA8 O 3 before and after removing the wildfire contributions to O 3 at the SLC Hawthorne site in 2006-2022 were quite different. Whereas the unadjusted data do not meet the current O 3 standard, after removing the contributions from wildfires the SLC region is close to achieving levels that are consistent with meeting the O 3 standard. We also found that the wildfire contribution during smoke days was particularly high under conditions of high temperature, high PM 2.5 concentration, and low cloud fraction.