Interannual variability and trends of summertime PM2.5-based air quality in the Intermountain West

Summertime air quality is a growing public health concern in the populated region of Northern Utah. Whereas winter air pollution is highly linked with local atmospheric temperature inversions associated with upper atmospheric high-pressure and radiational cooling in valleys, the relationship between climate factors and the frequency of poor air quality during summer is still unknown. Analyzing the last 20 years of data, we demonstrated that summertime unhealthy days (as defined by PM2.5 air quality index level) in Northern Utah highly correlate with the number of dry-hot days, wildfire size, and an upper atmospheric ridge over the Northwestern United States. The persistent atmospheric ridge enhances lightning-caused fire burned areas in northwestern states and then transports the wildfire smoke toward Northern Utah. Similarly, climate model simulations confirm observational findings, such as an increasing trend of the upper atmospheric ridge and summertime dry days in the northwestern states. Such metrics developed in this study could be used to establish longer-term monitoring and seasonal forecasting for air quality and its compounding factors, which is currently limited to forecasting products for only several days.


Introduction
Northern Utah is characterized by one of the highest population growth rates in the nation (Perlich et al 2017) while also experiencing an influx of tourists in summer and winter. In recent summers, Northern Utah has experienced serious health concerns due to elevated levels of air pollution. As visits to Utah's national parks peak in the summertime (Smith et al 2018), visitors are exposed to poorer air quality, which consists of fine and coarse particulate matter (i.e. PM2.5 and PM10). The inhalable PM is linked to cardiovascular and respiratory health ailments (Utah State Health Assessment 2016). Moreover, on 6 and 16 August 2021, Salt Lake City was ranked as having the worst air quality in the world due to wildfire smoke (Maryon 2021). Climate change is likely to cause an increase in wildfire activity along with drought potential, both of which will exacerbate unhealthy air conditions (Leung and Gustafson 2005, Horton et al 2014, Xie et al 2022. Degraded air quality increases the risk of respiratory ailments, such as asthma and emphysema, resulting in premature mortalities (Pope et al 2006, Beard et al 2012, Baasandorj et al 2017, Wang et al 2019-asthmatics, children, and the elderly are especially vulnerable (Di et al 2017). In Utah, the economic cost of air pollution is in the range of $0.75 to $3.3 billion annually (Errigo et al 2020). Therefore, quantifying the year-to-year variability of poor air quality and identifying primary sources are essential research endeavors to inform and assess public health hazards.
Many studies in northern Utah have focused on the climate factors that affect poor air quality during the winter but have paid less attention to summertime air pollution. During winter, Utah's distinct urban topography, characterized by incised valleys, forms and maintains a shallow pool of stable and relatively cold air near the surface, thereby establishing a temperature inversion in the boundary layer (Wolyn and McKee 1989, Whiteman et al 1999, Billings et al 2006, Gillies et al 2010. Temperature inversions subsequently trap locally produced air pollutants, and air quality degrades over time (Holzworth 1962, Malek et al 2006, Whiteman et al 2014, Green et al 2015. The temperature inversion can often persist for days under the influence of synopticscale high-pressure systems causing multi-day pollution episodes of high PM2.5 concentrations (Gillies et al 2010, Silcox et al 2012, Whiteman et al 2014, Wang et al 2015, Lareau and Horel 2015a, 2015b. Since the main source of wintertime air pollution is local anthropogenic emissions (i.e. transportation, agriculture, and industrial production), policymakers have developed state-level regulations to manage the winter air quality (Utah Department of Environmental Quality 2021).
However, in the summertime, wildfire smoke from local, regional, and distant sources, compounded by prolonged drought conditions, can combine to drive air quality further into the red zone of the air quality index (AQI) (Mallia et al 2015, Liu et al 2021, Xie et al 2022. Although most wildfires are not natural, the magnitude is related to climate and ecological connections. We, therefore, refer to them as 'natural drivers' of degraded air quality; these prevailing factors are problematic given their transitory nature and are difficult to predict in terms of timing, duration, and intensity. In other words, the current monitoring and forecasting activities focus on daily air quality events but lack a seasonal perspective. This bodes a challenge for policymakers who have to develop mitigation strategies. Unresolved questions remain as to frequency of unhealthy events and the year-to-year variability during summertime. It is also unclear how a warming climate may alter the climate and ecological connections regionally and locally (Wuebbles et al 2017). Hence, to assess future risks of enhanced summertime air quality degradation, there is a need to understand the interannual connections that may exist between synoptic-scale climate conditions and regional compounding factors-notably drought and wildfires.
The study, as outlined here, analyzed historical records of a PM2.5-based daily AQI with the climate properties influencing the degradation in air quality. The climate properties included temperature, precipitation, and upper atmospheric circulations along with wildfire activities, as described in section 2. Section 3 establishes what climate factors impact summertime unhealthy days in Northern Utah. Also, we developed some seasonal metrics that could be applicable for monitoring and forecasting air quality. Results are discussed in section 4, and concluding remarks are presented in section 5. Not considered are other air pollutants, such as ground-level ozone (Lin et al 2020), since observational coverage is insufficient to conduct a statistical analysis.

Datasets and method
To minimize observational uncertainty, we analyzed multiple data sources, including station-based observations and gridded hydroclimate products at both daily and monthly resolutions. To calculate summertime unhealthy days, we used the PM2.5 AQI for the Wasatch front (Cache, Weber, Salt Lake, and Utah Counties, see figure 1(a)) owing to its long historical and spatial coverage with daily intervals as well as its associated health risks and weather drivers in this region (Wang et al 2012). We assembled daily data for PM2.5 and the AQI from the daily air quality summary statistics in the United States Environmental Protection Agency (EPA), and precipitation and maximum air temperature from the Global Historical Climatology Network (GHCN) (see supplementary text). For wildfires, we used the total number and the burned acres of lightning-caused wildfires in the Pacific Northwest, Northern California, and the Great Basin from the National Interagency Fire Center (NIFC) annual wildfire statistics repository. Additional datasets included a gridded dataset of daily precipitation and maximum temperature from the Climate Prediction Center (CPC) at a 0.5 • × 0.5 • latitude-longitude resolution (Chen et al 2008), monthly soil moisture (CPC reanalysis) at a 0.5 • × 0.5 • resolution (Fan and van den Dool 2004), monthly biomass burning carbon emissions with small fires (GFED4.1s) at a 0.25 • × 0.25 • resolution (van der Werf et al 2017), monthly aerosol optical depth (AOD) from the Moderate Resolution Imaging Spectroradiometer (MODIS) product at a 1 • × 1 • resolution (Platnick et al 2015), and a monthly atmospheric reanalysis product, i.e. the Japanese 55 year Reanalysis (JRA55), at a 1.25 • × 1.25 • resolution (Kobayashi et al 2015). The time periods for this study were 2001-2020 for wildfires, 2000-2021 for AOD, and 1999-2021 for all other products.
The climate response to external forcings was estimated based on a trend analysis for the 1979-2021 period. Observed products were the daily precipitation and maximum air temperatures in the CPC gridded products (Chen et al 2008), and the monthly geopotential height at 250 hPa in JRA55 (Kobayashi et al 2015). Model simulated products included the downscaled daily products of the Fifth Coupled Model Intercomparison Project (CMIP5) and the monthly products of the Sixth Coupled Model Intercomparison Project (CMIP6, see supplementary text) (Eyring et al 2016, Vano et al 2020. Linear trends and their statistically significant testing were estimated based on non-parametric statistics,

Summertime unhealthy days
The historical record of summertime unhealthy days along the Wasatch front (defined as daily AQI > 50, see supplementary text) shows large interannual variability with an increasing trend in the most recent 20 years (green lines in figure 1). AQI between 50 and 100 is defined as 'moderate'; this indicates acceptable air quality but with potential risks for individuals with compromised health. The number of unhealthy days peaked in 2021, hitting a record of 39 within 5 months. For the May-September season, there were several unhealthy episodes in the most recent two decades (i.e. 39 days in 2021, 36 days in 2018, 29 days in 2012, 25 days in 2008, 24 days in 2020, and 23 days in 2007); these are defined as unhealthy years. By contrast, some years include very few unhealthy days, such as 2002, 2010, 2011, 2013, 2014, and 2019 (referred to as clean-air years). By defining dry-hot days as daily precipitation less than 0.5 mm and maximum air temperature more than 23.9 • C, we found that the time series of summertime unhealthy days correlates strongly with the dry-hot days (correlation coefficient, R = 0.70; p-value < 0.01). The aforementioned relationship results from the distinct valley topography of the Wasatch Mountain Range, which traps air pollutants near the surface during dry-hot days. Such conditions remain until a passing cold front removes the aerosols through active turbulence mixing. The unhealthy days also significantly correlate with either singular hot or dry days (R = 0.55 and 0.61; p-value < 0.01). Since the GHCN weather stations are a historical record of daily precipitation and maximum air temperature, the sum of dry-hot days Correlation maps of the summertime unhealthy days for Wasatch front with the seasonal mean of (a) precipitation anomalies, (b) maximum air temperature anomalies, (c) soil moisture anomalies, and the sum of (d) dry days (daily precipitation <0.1 mm), (e) hot days (daily maximum temperature >23.9 • C), and (f) dry-hot days (daily precipitation <0.1 mm and daily maximum temperature >23.9 • C) during the months of MJJAS. Utah is outlined by a thick line. Correlation coefficients of 0.37, 0.43, and 0.55 correspond to the statistical significance at 90%, 95%, and 99% levels with 19 degrees of freedom on the basis of a two-sided Student's t-test.
might be considered to serve as an alternate metric for the monitoring of poor air quality. This meteorological metric assumes the source of smoke primarily from wildfires, although contributions from other local emission sources are also possible, we discuss further in section 4.

Large-scale climate conditions
Correlation maps of summertime unhealthy days in the Wasatch front highlight close connections with dry, hot, and dry-hot days in the Continental U.S. (figure 2). Specifically, unhealthy days exhibited positive correlations with hot days, revealing two maxima over Utah and the Dakotas (figure 2(e)), while the correlation with dry days was more widely spread over the Northwestern U.S. (figure 2(d)). In contrast, the correlations of unhealthy days with summertime mean precipitation and corresponding soil moisture anomalies were negative around Northern Utah and North Dakota, but weak in the other regions (figures 2(a) and (c)). Similar to the correlation map for the hot days, the summertime-mean air temperature anomalies positively correlate with the unhealthy days with correlation maxima over the Great Basin and the Dakotas ( figure 2(b)). Not surprisingly, figure 2(f) shows a more pronounced and widespread positive correlation between dry-hot days and unhealthy days. These high positive correlations spread around Northern Utah are consistent with the station-based analysis in figure 1 Figure 3 shows composite maps of carbon emissions from biomass burning (GFED4.1s) along with AOD derived from MODIS satellite measurements for unhealthy years, clean-air years, and the difference. The panels indicate enhanced carbon emissions and a higher AOD over Northern California, Idaho, Oregon, Washington, and western Montana during the unhealthy years compared to the clean-air years. These results suggest that active forest fires in the Northwestern U.S. decrease atmospheric visibility (i.e. through wildfire smoke) and will affect air quality along the Wasatch front under favorable atmospheric conditions. Moreover, spatial distributions of enhanced carbon emissions and higher AOD are in good agreement with the highly correlated areas of the dry and unhealthy days (figure 2(d)). Since more dry days suggest scant local storm activity, the unhealthy years along the Wasatch front are characterized by distant active fires. The lack of thunderstorm activity in Utah is consistent with the increased hot and dry conditions and a lack of lightning-igniting fires. Interestingly, the total number of lightning-caused fire occurrences showed a decreasing trend for the most recent two decades in the Northwestern U.S. (Northern California, Northwest, and Great Basin regions defined by NICF) and a weak correlation with summertime unhealthy days along the Wasatch front (R = −0.28, figure S4(a)). By contrast, the unhealthy days in summertime significantly correlated with the burned area of lightning-caused fires (R = 0.68, figure  S4(b)) and even higher with the lightning-caused fire size (burned area divided by total number) in the Northwestern U.S. (R = 0.73, figure 1(b)). The increasing trend of wildfire size is also consistent with the well-known increase of large (timber) wildfire events in the Western U.S. It is reasonable to conclude that the number of lightning-caused fires is less critical than the increasing size and duration of wildfires. These results imply that the summertime air quality along the Wasatch front relies on sources (anthropogenic or natural) of pollutant air far upstream of the state rather than those which are local or regional.
To document relevant large-scale climate conditions, we created composite maps of geopotential height and horizontal wind anomalies at 250 hPa covering the unhealthy years, clean-air years, and their difference (figures 3(g)-(i)). We found the existence of a large high-pressure system in the entire free atmosphere during unhealthy years, with the center of action around Northern California ( figure 3(g)). This anti-cyclone induced northeasterly wind anomalies over Utah. The resulting wind kinematics tend to block moisture transport from the Gulf of Mexico and California towards the Intermountain West and limit the northward extent of the North American monsoon. Moreover, this weakened monsoonal flow into Northern Utah reduces atmospheric convection activity and enhances the frequency of hot and dry days. Additionally, the clockwise circulation driven by the high-pressure ridge enables the transport of wildfire smoke from the Northwestern regions and Northern California toward Northern Utah. Opposite tendencies occur in the clean-air years ( figure 3(h)). Consistent with these atmospheric circulations, geopotential height anomalies at 250 hPa averaged around Northern California (35 • N-45 • N, 120 • W-115 • W) strongly correlated with summertime unhealthy days along the Wasatch front (R = 0.72, figure 1(d)). Consequently, a summertime persistent atmospheric ridge over California is the principal atmospheric feature that is associated with increasing fire size and poor air quality in the Northwestern U.S., and subsequently, Northern Utah. We note that the ridge does not directly reduce the air quality but acts to enhance fire activities in the Western and Northwestern U.S., as well as enabling smoke transport into Northern Utah. Therefore, the ridge is a crucial climate mechanism in Northern Utah's degraded air quality.

Climate change contributions
The upper atmospheric ridge over Northern California and the summation of dry days in the Western U.S. have equally increased over the most recent four decades. Figure 4 shows the linear trends of the geopotential height deviations from the zonal mean at 250 hPa (Z250 * ) along with the summation of dry days (daily precipitation <0.1 mm d −1 ) during the May-September period; these were derived from observational datasets and CMIP simulations. By removing the zonal mean for the seasonally averaged geopotential height anomalies at each latitude and year, we can minimize the thermal expansion component and capture the ridge strength appropriately. There is a significant positive trend of Z250 * on the Northwestern coast of the U.S., indicating a deepened upper atmospheric ridge in recent years ( figure 4(a)). Additionally, the CMIP6 multi-model ensemble shows a significant positive trend of Z250 * (figure 4(c)), albeit with a smaller amplitude and a position shifted more inland. Consistent with a deepened ridge, the summation of dry days exhibits increasing trends in the Northwestern U.S., including Utah, in both the observations and CMIP5 simulations (figures 4(b) and (d). As air temperature warms, as a result of the increased activity of the upper air ridge, the saturation deficit grows even faster in accordance with the Clausius-Clapeyron relationship, leading to low humidity. This environment enhances the probability and intensity of fire events. The increasing occurrence of hot days due to the warming climate suggests that there is a climate change component that acts to enhance wildfire development and resultant summertime episodes of air pollution.

Discussion
Our study has identified the long-range atmospheric transport of PM2.5 from wildfire smoke in the Northwestern U.S. that is directed toward Northern Utah. However, the impact of local emissions on air quality remains uncertain. The localized sources, such as transportation, industry, biomass burning, wind erosion, and residential energy use, also produce PM2.5 and other prevalent pollutants like groundlevel ozone, PM10, carbon monoxide, lead, nitrogen dioxide, and sulfur dioxide (Flocchini et al 1981, Colvile et al 2001, Horel et al 2016, Blaylock et al 2017, Moravek et al 2019, Mitchell and Zajchowski 2022. The localized pollutants are typically distributed and captured near the surface and are associated with local atmospheric dynamics, as classically seen in the winter temperature inversions. However, underlying dynamics in the summertime air quality are distinct from those in the wintertime, where valley subsidence inversions tend to impede long-range transport of pollutant air from biomass burning. Moreover, transported smoky air from wildfires is often effective at reducing sunlight and suppressing photochemical processes, albeit those hot and dry conditions are conducive to locally emitted air pollution through such processes. Therefore, the long-range PM2.5 transport primarily exacerbates Northern Utah's summertime pollution on seasonal timescales, although local sources of PM2.5 driven by local emissions, vertical transport, and wildfire plumes may be relatively important for a single daily event.
It is essential to distinguish between the event trajectory analysis and our climate diagnostic approach. Trajectory analysis tracks the movement of PM2.5 by following air parcels using weather conditions from either actual observations or model simulations (Garcia-Menendez et al 2013, Mallia et al 2015. This analysis is especially effective in tracing the spread of specific wildfire/pollution events from source to exposure, connecting remote wildfire sources with air quality in Northern Utah, and identifying the individual fire event that affects the degraded air quality in the area. However, instead of highlighting a single event, the approach described in our study focuses on the likelihood of a favorable environment in enhancing wildfires and subsequent degraded air quality. When a persistent upper atmospheric ridge develops over Northern California, it increases the likelihood of hot-dry weather. This environment sets the stage for potential large wildfire occurrences in the Pacific Northwest. Although it is challenging to predict the exact timing and location of individual fires, we can expect higher concentrations of PM2.5 transported to Northern Utah from wildfire sources. Such a probabilistic outlook is critical for stakeholders and decision-makers in establishing long-term management plans for air quality remediation in the summer.

Conclusions
Using a derived PM2.5 AQI and applied to Northern Utah, we have demonstrated that summertime unhealthy days are highly correlated with hotdry days along the Wasatch front, wildfire size in the Northwestern U.S., and the upper atmospheric ridge over Northern California. The persistent atmospheric ridge over Northern California enhances the strength and size of fires in this region as well as the Pacific Northwest. A strengthened clockwise circulation transports wildfire smoke from the Northwestern U.S. toward Utah and limits the northward extent of the seasonal Southwestern monsoonal flow. The stable nature of the atmospheric ridge results in a subsidence inversion and inhibits vertical mixing of the air over Utah, thereby trapping the smoke near the surface. Furthermore, the weakened Southwestern monsoonal flow into Utah limits moisture inflow for a convective activity that could help mix out some of the smoke. The combination of these connected climate conditions enhances PM2.5 concentrations.
We have developed a set of metrics for detecting the frequency of summertime air quality, including the dry-hot days using daily precipitation and maximum air temperature from weather stations, along with a high-pressure index using geopotential height anomalies at 250 hPa. These meteorological observations offer alternate metrics in monitoring unhealthy days, i.e. they serve as a basis for seasonal forecasting for summertime air quality. In addition, as was indicated by the CMIP model analysis, the metrics are applicable to future climate change assessments of air quality (Wang et al 2015). The current air quality forecast in Northern Utah is limited to only several days, given that it relies on the weather forecast. Although the CPC in the National Weather Service provides an outlook for seasonally averaged precipitation and temperature, they are an insufficient indicator of summertime air quality in Northern Utah. We found that the sum of dry and hot days in the Northwestern U.S., along with the Z250 * anomalies over Northern California, are more appropriate metrics. Many other meteorological factors also contribute to wildfire activity and summertime air pollution, including wind speed and direction, humidity, geopotential height, and daily averaged temperature (Reddy and Pfister 2016, Holden et al 2018, Srock et al 2018, Zhong et al 2020, Dong et al 2021. The metrics described here could advance seasonal estimates of summertime air quality and become another tool for air quality assessment and monitoring. The trend analysis, covering the most recent four decades, highlighted the buildup of an upper atmospheric ridge over Northern California and increasing dry days in the Northwestern U.S. These conditions can enhance the probability of wildfires, which incidentally often develop into large events and subsequent poor air quality in the Northwestern U.S. Through the analysis undertaken, we established that there are impacts elsewhere, specifically pollution episodes in Northern Utah. A further implication realized is its impact on well-being, as indicated by the CMIP simulations: i.e. future serious health concerns of cardiovascular and respiratory issues, particularly for sensitive groups like the elderly and children. The CMIP simulations captured the upward trends detected in the observation-based products. However, discrepancies were apparent between observations and simulations in the amplitude and position of the high-pressure ridge. Specifically, the model simulated linear trends are weaker than observed in both Z250 * and dry days (figure 4) alongside a large inter-model spread in amplitude and spatial patterns (figures S7-S9). The large uncertainty in climate model simulations originates from several factors, such as low-frequency natural variability, model biases, ensemble sizes, and downscaling methods. The 40 year trend analysis affords a limited perspective on estimating the long-term climate change impacts on summertime air quality. Future work should focus on establishing a more accurate assessment of summertime air quality trends.
The impact of seasonal air quality forecasts on society is consequential, yet creating such forecasts poses a considerable scientific challenge. To overcome this challenge, the climate community is making a substantial effort to advance seasonal-to-decadal climate predictions utilizing global climate models (Boer et al 2016, Becker et al 2022. Although the current climate prediction system does not simulate the direct transport of PM2.5 from wildfires, the research undertaken here advocates that it is possible to assess the seasonal predictability of unhealthy air quality with alternative metrics, such as the frequency of dry-hot days in Northern Utah and the seasonal Z250 * anomalies over Northern California. Recent advances in Earth system models and decadal climate prediction approaches offer the potential to forecast wildfires on multi-year timescales (Chikamoto et al 2017(Chikamoto et al , 2020. Additionally, recent progress in regional and global atmospheric models, coupled with chemistry components, allows for the realistic simulation of fire events and associated aerosol transport (Ye et al 2021, Zhang et al 2022. However, the spatiotemporal coverage of air quality observations is currently insufficient to isolate the impact of long-term climate change from interannual climate variability. Therefore, continuous efforts to improve air quality observations are especially important for the verification and enhancement of Earth system models toward establishing provable seasonal air quality forecasts.
All data that support the findings of this study are included within the article (and any supplementary files).