The Fires, Asian, and Stratospheric Transport –Las Vegas Ozone Study ( FAST -LVOS)

. The Fires, Asian, and Stratospheric Transport –Las Vegas Ozone Study ( FAST -LVOS) was conducted in May and June of 2017 to study the transport of ozone (O 3 ) to Clark County, Nevada, a marginal non-attainment area in the southwestern United States (SWUS). This 6-week (20 May–30 June 2017) ﬁeld campaign used lidar, ozonesonde, aircraft, and in situ measurements in conjunction with a variety of models to characterize the distribution of O 3 and related species above southern Nevada and neighboring California and to probe the inﬂuence of stratospheric intrusions and wildﬁres as well as local, regional, and Asian pollution on surface O 3 concentrations in the Las Vegas Valley ( ≈ 900 m above sea level, a.s.l.). In this paper, we describe the FAST -LVOS campaign and present case studies illustrating the inﬂuence of different transport processes on background O 3 in Clark County and southern Nevada. The companion paper by Zhang et al. (2020) describes the use of the AM4 and GEOS-Chem global models to simulate the measurements and estimate the impacts of transported O 3 on surface air quality across the greater southwestern US and Intermountain West. The FAST -LVOS measurements found elevated O 3 layers above Las Vegas on more than 75 % (35 of 45) of the sample days and show that entrainment of these layers contributed to mean 8 h average regional background O 3 concentrations of 50–55 parts per billion by volume (ppbv), or about 85–95 µg m − 3 . These high background concentrations constitute 70 %–80 % of the current US National Ambient Air Quality Standard (NAAQS) of 70 ppbv ( ≈ 120 µg m − 3 at 900 m a.s.l.) for the daily maximum 8 h average (MDA8) and will make attainment of the more stringent standards of 60 or 65 ppbv currently being considered extremely difﬁcult in the interior SWUS.


Introduction
Ground-level ozone (O 3 ) is one of six "criteria" air pollutants identified as serious threats to human health and welfare and made subject to National Ambient Air Quality Standards (NAAQS) by the US Clean Air Act (CAA) (Karstadt et al., 1993). Ozone is not directly emitted into the atmosphere by anthropogenic activities but is a secondary product of photochemical reactions between nitrogen oxides (NO x = nitric oxide (NO) + nitrogen dioxide (NO 2 )) and carbon monoxide (CO), methane (CH 4 ), or volatile organic compounds (VOCs). Thus, efforts to control ambient O 3 concentrations have sought to reduce anthropogenic emissions of these precursors, and more stringent NO x emission controls have contributed to a 36 % decrease in the annual mean daily maximum 1 h NO 2 , the benchmark used to ascertain NAAQS attainment, across the US between 2000 and 2019 (https://www.epa.gov/air-trends/ nitrogen-dioxide-trends, last access: 16 August 2021). In response, the US mean fourth-highest annual maximum daily 8 h average (4MDA8) O 3 , the metric by which compliance with the O 3 NAAQS is determined, declined from 82 to 65 parts per billion by volume (ppbv), or 21 % over the same period (https://www.epa.gov/air-trends/ozone-trends, last access: 16 August 2021).
The air quality improvements in the US over the last 2 decades represent a major success, but the gains have not been uniform, with large decreases in the southeastern US (SEUS: Alabama, Florida, Georgia, North Carolina, South Carolina, and Virginia), where NO 2 and O 3 declined by 39 % and 25 %, respectively, but much smaller changes in the southwestern US (SWUS: Utah, Arizona, Colorado, and New Mexico), where a 38 % reduction in NO 2 led to an O 3 decline of just 11 %. The 4MDA8 O 3 decreased from 83 to 63 ppbv in the SEUS but only decreased from 77 to 69 ppbv in the SWUS. The weaker response to NO 2 reductions in the SWUS is attributed in part to increased oil and gas development (Pozzer et al., 2020) and in part to the much higher background O 3 in this region (EPA, 2013;Lefohn et al., 2014;Cooper et al., 2015). This background is derived from a variety of non-controllable ozone sources (NCOSs) including O 3 produced by photochemical reactions of anthropogenic emissions outside the US borders or by soils, vegetation, lightning, or wildfires and by naturally occurring O 3 transported downward from the stratosphere (Jaffe et al., 2018). The high mean elevations of Colorado (2078 m a.s.l.), Wyoming (2047 m a.s.l.), Utah (1864 m a.s.l.), Nevada (1681 m a.s.l.), and Idaho (1528 m a.s.l.), which make up the heart of the Intermountain West (IMW), i.e., the area of the US bounded by the Cascade (≤ 4392 m a.s.l.) and Sierra Nevada (≤ 4421 m a.s.l.) mountains to the west and the Front Range of the Rocky Mountains (≤ 4401 m a.s.l.) to the east (Fig. 1a), make this entire region particularly vulnerable to both stratospheric intrusions (Lin et al., 2012a) and pollution transported across the Pacific Ocean from East Asia (Lin et al., 2012b).
Background O 3 cannot be directly measured, but an upper limit can be estimated from measurements at remote "baseline" locations sufficiently removed from anthropogenic sources. Jaffe et al. (2018) estimated that the monthly mean MDA8 O 3 in the higher elevations of the IMW is around 50 ppbv in late spring, with the 4MDA8 O 3 concentrations exceeding 60 ppbv in some locations. These high baseline concentrations constitute more than 70 % of the 70 ppbv NAAQS set in 2015 (EPA, 2014) and limit the ability of western air quality managers to maintain surface O 3 concentrations below this standard through local control strategies (Cooper et al., 2015;Uhl and Moore, 2018;Faloona et al., 2020).
Concern about the potential impacts of stratospheric intrusions, Asian pollution, and other NCOSs on NAAQS attainment in the greater Las Vegas area motivated the first Las Vegas Ozone Study (LVOS) conducted by the NOAA Chemical Sciences Laboratory (CSL) in late May and June of 2013 (Langford et al., 2015b) in partnership with the Clark County (Nevada) Department of Air Quality (CC-DAQ). The LVOS 2013 field campaign was organized around the truck-mounted TOPAZ (Tunable Optical Profiler for Aerosol and oZone) differential absorption lidar (Alvarez et al., 2011), which is part of the NASA-sponsored Tropospheric Ozone Lidar Network (TOLNet). The lidar was deployed for 6 weeks to a decommissioned US Air Force radar base on the summit of Angel Peak (AP: 36.32 • N, −115.57 • E; 2680 m above mean sea level, a.s.l.), about 45 km west of Las Vegas, in the Spring Mountains (Fig. 1c). The lidar measurements were augmented by in situ measurements of O 3 and CO and basic meteorological parameters.
The LVOS 2013 campaign documented several episodes in which the appearance of stratospheric intrusions, Asian pollution, or wildfire plumes at AP was followed by MDA8 O 3 concentrations greater than 70 ppbv at multiple regula- tory monitors in Clark County and in some cases by exceedances of the 2008 O 3 NAAQS of 75 ppbv then in effect. Since stratospheric intrusions do not typically reach the relatively low elevations of Las Vegas (620 m a.s.l.), it was hypothesized that some of the high-O 3 episodes resulted from entrainment of mid-tropospheric O 3 layers (Langford et al., 2017) by the deep-convective boundary layers that form over the Mojave Desert (Seidel et al., 2012). This hypothesis could not be confirmed, however, because of the placement of the lidar above the valley floor, and a more extensive follow-up study, the Fires, Asian, and Stratospheric Transport-Las Vegas Ozone Study (FAST-LVOS), was conducted over the same time period in 2017 to (i) test the entrainment hypothesis, (ii) determine the representativeness of the LVOS 2013 results, and (iii) better characterize the different sources of surface O 3 in the Las Vegas Valley (LVV).
In this paper, we present an overview of the FAST-LVOS campaign with brief examples highlighting the influence of stratospheric intrusions, Asian pollution, biomass burning, and both local and regional pollution on surface O 3 in Clark County and the greater IMW. The companion paper by Zhang et al. (2020) uses the GFDL-AM4 and GEOS-Chem global models to simulate these measurements and quantify the impacts of these processes on high-O 3 events in southern Nevada and the greater SWUS and IMW.

Background
Previous airborne and ground-based lidar measurements have shown that elevated O 3 layers are common features of the lower free troposphere between ≈ 3 and 6 km a.s.l. above California (Langford et al., 2012;Ryerson et al., 2013;Faloona et al., 2020) and southern Nevada (Langford et al., 2015b(Langford et al., , 2017 in late spring and early summer, and it is well established that the US west coast is one of the global hotspots for deep stratosphere-to-troposphere transport (STT) of O 3 in springtime (Wernli and Bourqui, 2002;Sprenger and Wernli, 2003;James et al., 2003;Skerlak et al., 2014Skerlak et al., , 2015Breeden et al., 2021). This occurs primarily through the formation of tropopause folds, tongues of lower stratospheric air that descend isentropically beneath the jet stream circulating around upper-level lows. These intrusions of dry, O 3 -rich air often form near the end of the North Pacific storm track and descend behind the mixture of clean or polluted tropospheric air in the dry airstream (Wernli, 1997;Cooper et al., 2004;Trickl et al., 2014). They are thus an important mechanism for transport of East Asian pollution from the upper troposphere down to the boundary layer above the western US (Brown-Steiner and Hess, 2011;Lin et al., 2012a, b). Tropopause folds typically follow one of two pathways as they reach the middle troposphere. Many intrusions, particularly those formed by deep closed lows, continue to curve cyclonically as they descend and wrap up above the surface low in the lower troposphere (Danielsen, 1964). These deep LC2 (lifecycle 2) intrusions (Thorncroft et al., 1993;Polvani and Esler, 2007) often reach the top of the boundary layer and sometimes even the surface at higher elevations (Schuepbach et al., 1999;Stohl et al., 2000;Bonasoni et al., 2000;Trickl et al., 2020). These intrusions were first described by Reed and Danielsen (1958) and have long been observed as steeply sloping tongues in ozone lidar curtains (Browell et al., 1987;Ancellet et al., 1994;Vaughan et al., 1994;Langford et al., 1996;Eisele et al., 1999).
Much less attention has been paid to the so-called LC1 (lifecycle 1) intrusions that are sheared anticyclonically from elongated troughs to form quasi-horizontal filaments in the middle and upper troposphere (Appenzeller and Davies, 1992;Appenzeller et al., 1996;Vaughan et al., 2001;Albers et al., 2021). These streamers are well-known features of satellite water vapor imagery (Manney and Stanford, 1987) and can stretch for thousands of kilometers. They usually roll up horizontally into a series of diminishing vortices to be irreversibly mixed into the free troposphere (Wirth et al., 1997;Vaughan and Worthington, 2000;Vaughan et al., 2001;Colette and Ancellet, 2006), but they can also be dissipated by moist convection (Langford and Reid, 1998). They are a significant source of background O 3 in the free troposphere (Albers et al., 2021) but usually remain well above the boundary layer and thus do not influence surface O 3 directly.
Both types of intrusions are most common in winter, when cyclonic activity is at a maximum, but stratospheric intrusions that form in late May and early June are more likely to affect compliance with the ozone NAAQS since (i) more O 3 is available for transport from the lower stratospheric reservoir in springtime (Albers et al., 2018), (ii) surface concentrations are higher because of increased photochemical production, and (iii) deeper afternoon mixed layers (Seidel et al., 2012) can potentially entrain some intrusions that would otherwise pass overhead. Stratospheric influence plays an important role in driving the observed year-to-year variability in springtime ozone air quality over the western US (Lin et al., 2015). The potential surface impacts of stratospheric intrusions and co-transported pollution decrease in summer after the jet stream migrates poleward into Canada, and deep convection lifts pollution higher into the upper troposphere over East Asia (Hudman et al., 2004;Brown-Steiner and Hess, 2011).

FAST-LVOS measurements and models
The FAST-LVOS field experiment was also organized around the TOPAZ lidar, which was upgraded with a new data acquisition system in 2015. This upgrade more than doubled the useful operating range of the lidar from ≈ 3 to 8 km, which allowed it to reach higher altitudes than was possible during LVOS 2013, even while operating from a much lower-elevation site at the North Las Vegas Airport (NLVA; 36.2 • N, −115.2 • E; 680 m a.s.l.), located about 8 km NW of downtown Las Vegas. NOAA CSL also brought a vertically staring Doppler lidar to the NLVA to characterize mixing between the boundary layer and free troposphere and a mobile laboratory with a more extensive suite of in situ measurements to the summit of AP. These primary measurements were supplemented by measurements from a single-engine Mooney TLS Bravo aircraft operated by Scientific Aviation, Inc. (SA) and by ozonesondes launched by the NOAA Global Monitoring Laboratory (GML) during four 2 to 4 d intensive operating periods (IOPs) initiated when the synoptic conditions showed that stratospheric intrusions or Asian pollution transport events were likely.

NOAA CSL TOPAZ lidar
The TOPAZ truck arrived in Las Vegas on the morning of 17 May and was deployed to a secure CCDAQ enclosure on the north side of the NLVA. In addition to the lidar, the truck was equipped with an automated weather station (Airmar 150WX) and a commercial UV absorption O 3 monitor (2B Technologies model 205) that sampled air 5 m above the ground. The autonomous NOAA CSL vertically staring 1.5 µm micro-Doppler lidar was placed near the TOPAZ truck to measure aerosol backscatter and vertical wind vari-ance for the estimation of boundary layer depths (Bonin et al., 2018). The two lidars were located near the CCDAQ wind profiler (Fig. 2, top) and visibility camera and about 500 m NNW of the National Weather Service (NWS) KVGT meteorological tower.
TOPAZ uses a tuneable Ce:LiCaF solid-state laser and a unique transceiver configuration to profile O 3 and particulate backscatter (β) from just above the surface to ≈ 8 km above ground level (a.g.l.). The lidar points vertically but uses a large steerable mirror on top of the truck to sequentially deflect the co-axial laser beams and return signals along a series of elevation angles. The scanner line of sight was oriented to the east (parallel to the 7-25 runway) and successively tilted along paths 2, 6, 20, and 90 • above the horizon. This cycle was repeated every 8 min, and the vertical projections of the slant profiles merged with the zenith profile to create vertical ozone and backscatter profiles starting 27.5 ± 5 m above the ground, with the lowest measurements displaced about 800 m downrange. Ozone number densities were retrieved using two wavelengths (≈ 287 and 294 nm) with 30 m range gates and a smoothing filter that increased from 270 m wide at the minimum range (815 ± 15 m) to 1400 m wide at the maximum range. The ozone and backscatter profiles were computed simultaneously using an iterative procedure (Alvarez et al., 2011) incorporating the O 3 absorption crosssections of Malicet et al. (1995) and temperature and pressure profiles interpolated from the 3 h National Centers for Environmental Prediction (NCEP) North American Regional Reanalysis (NARR) to account for the temperature dependence of the O 3 cross-sections and to convert the calculated O 3 number densities to mixing ratios. The effective O 3 vertical resolution increased from ≈ 10 m near the surface to 150 m at 500 m a.g.l. and 900 m at 6 km a.g.l. The maximum range was limited by solar background radiation during the day and decreased from about 8 to 6 km near midday. Backscatter from aerosols, smoke, and dust was also retrieved with 7.5 m resolution at 294 nm.
Total uncertainties in the 8 min O 3 retrievals are estimated to increase from ±3 ppbv below 4 km to ±10 ppbv at 8 km. The upgraded TOPAZ system was extensively compared with measurements from the same SA Mooney aircraft used here as well as the NASA Alpha Jet (Hamill et al., 2016) in the 2016 CABOTS field campaign (Langford et al., 2019) and with electrochemical concentration cell (ECC) ozonesondes and ground-based TOLNet lidars in the 2016 Southern California Ozone Observation Project (SCOOP) . Both intercomparisons showed excellent agreement within the stated uncertainties in the measurement techniques, particularly when the spatial and temporal differences introduced by the different sampling methods are considered.

NOAA CSL Mobile Laboratory
The van-based CSL mobile laboratory  was equipped with instruments to measure O 3 , CO, water vapor (H 2 O), carbon dioxide (CO 2 ), CH 4 , NO, NO 2 , total reactive nitrogen oxides (NO y ), nitrous oxide (N 2 O), and meteorological parameters. The NO, NO 2 , NO y , and O 3 measurements were made with a custom-built cavity ring-down spectrometer (CRDS) (Washenfelder et al., 2011;Wild et al., 2014;Womack et al., 2017); a commercial (Picarro) wavelength-scanned CRDS to measure CO 2 and CH 4 (Peischl et al., 2013); and a modified commercial (Los Gatos) instrument using off-axis-integrated cavity output spectroscopy (OA-ICOS) to measure N 2 O, CO, and H 2 O (Coggon et al., 2016). Additional details about these instruments can be found in the references. Ozone was also measured by a commercial (2B Tech, Model 205) ultraviolet photometer, and these data were used to fill in the gaps during brief periods when the CRDS instrument was offline. There were no aerosol measurements.
The mobile laboratory carried the same automated weather station (Airmar 150WX) as the TOPAZ truck and was also equipped with differential GPS and sonic anemometers to measure absolute wind speed and direction while the van was moving. The laboratory was parked on the southeastern edge of the AP summit overlooking Kyle Canyon, the primary corridor for upslope transport from the LVV, for most of the campaign. This is the same location (Fig. 2, bottom) occupied by the TOPAZ truck during LVOS 2013 and provides an unobstructed fetch to the west, south, and east. Northerly winds (≈ 335 and 65 • ) were perturbed by one of the nearby buildings, but winds from this direction were uncommon during the study. A few drives were conducted between AP and the LVV on 23-25 May during IOP1 and again on 15 and 17 June. These measurements will be described elsewhere. The van was also relocated to the NLVA for an intercomparison with the instruments on the SA Mooney on 15 June. All of the measurements described here are from AP.
The first LVOS campaign in 2013 used the relationships between the O 3 , H 2 O, and CO measured on AP to infer the history of the air masses sampled on the summit (Langford et al., 2015b). This was made possible because of the relative remoteness and high elevation of the site, which minimized the influences of nearby NO x or CO emissions and surface deposition, and by the lack of rainfall and extreme aridity of the Mojave Desert, which makes water vapor a semi-conserved tracer. Since CO has a tropospheric lifetime of ≈ 60 d and originates primarily from soil emissions and combustion processes (Holloway et al., 2000), concentrations are lower in the stratosphere than in the free troposphere and terrestrial boundary layer. Conversely, O 3 concentrations are much higher in the stratosphere than in the troposphere. Thus, O 3 and CO tend to be negatively correlated in mixtures of stratospheric and free-tropospheric air, which is also very dry, but positively correlated in mixtures of free-tropospheric air and urban pollution or biomass burning plumes. The marine boundary layer is far removed from most natural and anthropogenic sources of CO and O 3 and has low concentrations of both (Clark et al., 2015). Thus, the relationships between these three parameters can be used to separate the influences of stratospheric intrusions and Asian pollution from regional pollution and wildfires and distinguish air that descended from the lower stratosphere or upper troposphere from air advected inland from the Pacific Ocean.
This empirical approach was much improved by the addition of N 2 O, NO, NO 2 , NO y , CH 4 , and CO 2 measurements in the 2017 FAST-LVOS campaign. Nitrous oxide has a much longer tropospheric lifetime than CO (> 100 years) and originates primarily from natural and fertilized soils (Tian et al., 2019) and the oceans (Tian et al., 2020). It is thus well mixed throughout the free troposphere, with much lower concentrations in the stratosphere, and can be used as a tracer for recent agricultural and oceanic influences and stratospheric intrusions (Hintsa et al., 1998;Assonov et al., 2013). The CH 4 measurements provide another useful tracer for Asian pollution (Xiao et al., 2004) as well as oil and gas (Peischl et al., 2018), agricultural (Peischl et al., 2012), and biomass burning (Delmas, 1994) influences. The NO, NO 2 , and NO y measurements can be used to identify air masses influenced by biomass burning or by local, regional, and Asian pollution. In particular, the short lifetime (≈ 2-4 h) of NO x (Laughner and Cohen, 2019) makes these measurements a useful tracer for pollution from the LVV. The CO 2 measurements can also help identify pollution and biomass burning influences, although interpretation of these measurements is complicated by the strong diurnal variations created by photosynthetic uptake.

Meteorological measurements
The temperature, pressure, relative humidity, wind speed, and wind direction from the automated weather stations in the TOPAZ truck at the NLVA and the mobile laboratory on AP are summarized in Fig. S1 in the Supplement. The NLVA measurements were supplemented by the NWS measurements from the KVGT tower, and vertical wind information was also obtained from the radar wind profiler, ozonesondes, and aircraft. The automated micro-Doppler lidar measurements near the TOPAZ truck were used to calculate hourly averaged boundary layer heights (Bonin et al., 2018), which ranged from ≈ 2 to 4 km in the afternoon. Mixed layer heights were also derived from the potential temperature and relative humidity profiles acquired by the GML ozonesondes (8 km distant) and from the afternoon (00:00 UT or 17:00 Pacific Daylight Time, PDT) radiosondes launched from the Harry Reid (formerly Mc-Carran) International Airport (NWS station identifier KVEF) about 15 km to the south. The figure also shows solar radiation measurements from a site in Henderson near KVEF (cf. Fig. 1f) that were obtained from the University of Utah MesoWest network (https://mesowest.utah.edu, last access: 16 August 2021) and from the Spring Mountain Youth Camp (SMYC), which occupies the former cantonment area of the decommissioned Air Force base and lies ≈ 800 m west and 120 m below the AP summit. The SMYC measurements were obtained from the Western Regional Climate Center (WRCC) (http://www.wrcc.dri.edu/weather/smyc.html, last access: 16 August 2021).

Supplemental measurements
The supplemental ozonesonde and aircraft sampling during the intensive operating periods (IOPs) provided important context for the lidar and surface measurements. The four IOPs were conducted on 23-25 May, 31 May-2 June, 10-14 June, and 27-30 June (no ozonesondes were launched on 14 and 27 June). The NOAA Global Monitoring Laboratory (GML) launched a total of 30 ozonesondes (1 to 4 ozonesondes per day) from a park adjacent to the CCDAQ Joe Neal monitoring site located about 8 km northwest of the NLVA during the four IOPs. The ozonesondes (Sterling et al., 2018) measured O 3 concentrations to altitudes well above the 8 km range of the lidar and recorded temperature, relative humid-ity, and wind profiles, which helped characterize the synoptic context.
The SA Mooney conducted daily flights between the NLVA and Big Bear, CA (cf. Fig. 1b), during the four FAST-LVOS IOPs, logging a total of 90 flight hours over 15 d. The aircraft carried a pilot and flight scientist, along with a 2B Technologies Model 205 O 3 monitor, an Aerodyne Research cavity attenuated phase shift (CAPS) NO 2 monitor, and a Picarro 2301f wavelength-scanned cavity ring-down spectrometer (WS-CRDS) to measure CO 2 , CH 4 , ethane (C 2 H 6 ), and H 2 O (Trousdell et al., 2016). The 2B O 3 data were sampled at 2 s intervals, which corresponds to a mean distance of 150 m at the typical level flight speed of 75 m s −1 . The standard TLS flight plan ( Fig. 1b and c) began with a vertical profile to about 5 km a.s.l. above North Las Vegas after take-off. The aircraft then flew to AP and spiralled down around the summit to ≈ 3 km a.s.l. (cf. Fig. 2, bottom). From there, the aircraft headed to Jean, NV (924 m a.s.l.), where the southernmost CCDAQ O 3 monitor is located, and conducted another profile. The pilot then followed the I-5 corridor before diverting south to Big Bear, CA, where the aircraft landed and refueled. The return leg began with a profile above Barstow, CA, before following the I-5 corridor back to Clark County with additional profiles above Jean, AP, and North Las Vegas if fuel permitted. The default flight plan was modified as necessary to account for air traffic control requirements.

Ancillary measurements
The CCDAQ maintains a network of continuous air monitoring stations (CAMSs) for O 3 and other parameters (e.g., NO 2 , fine (PM 2.5 ) and coarse (PM 10 ) particulates, meteorology) in the LVV and surrounding areas (cf. Fig. 1). The CCDAQ also operates an upper-air station consisting of a radar wind profiler and profiling radiometer at the NLVA (cf. Fig. 2, top) and automated visibility cameras at the NLVA and M-Resort (cf. Fig. 1). The CAMS network included 11 active O 3 monitors during the FAST-LVOS campaign, with the Joe Neal (C75), Walter Johnson (C71), and JD Smith (C2002, since deactivated) monitors located within 8 km of the NLVA (cf. Fig. 1c). The hourly averaged measurements from the TOPAZ monitor at the NLVA agreed with the Joe Neal and Walter Johnson measurements to within 3 % on average, with linear regression coefficients of determination of R 2 = 0.87 (NLVA-C75), R 2 = 0.76 (NLVA-C71), and R 2 = 0.75 (C75-C71) when the Doppler lidar showed the boundary layer was well mixed. The Joe Neal and Walter Johnson O 3 measurements are discussed in Sects. 8 and 9 along with those from the more distant Apex (C22, 32 km), Jean (C1019, 49 km), Indian Springs (C7772, 58 km), and Mesquite (C23, 121 km) monitors located outside the LVV. The CCDAQ also operated a temporary O 3 monitor at the SMYC; these measurements averaged ≈ 8 % lower (R 2 = 0.91) than both the CRDS and 2B mea-

Meteorological contexts
The FAST-LVOS campaign can be divided into two meteorologically distinct periods, a late spring period from mid-May to mid-June and an early summer period from mid-June through the end of June. The Synoptic Discussion for May 2017 from the National Centers for Environmental Information (NCEI) (https://www.ncdc.noaa.gov/ sotc/synoptic/201705, last access: 16 August 2021) describes the jet stream as being very active in the late spring period, with a series of closed lows and upper-level troughs crossing the contiguous US every few days. These cyclonic systems spawned a series of stratospheric intrusions that appear as potential vorticity (PV) enhancements (Duncan et al., 2021) in the 300 hPa (≈ 9.5 km a.s.l.) NASA MERRA-2 Reanalysis plots in Fig. 3a-d. GOES water vapor images coinciding with the PV analyses are displayed in Fig. S2. The first extratropical cyclone, designated L 1 in Fig. 3a, passed through Nevada before the official start of the campaign on 20 May, but the next three low-pressure systems, labeled L 2 , L 3 , and L 4 in Fig. 3b-d, were targeted by IOPs on 23-25 May, 31 May-2 June, and 11-14 June. The solar radiation data in Fig. S1 show the appearance of cirrus ahead of the cold fronts as well as perturbations in the normal diurnal variations in pressure, temperature, dew point, and winds created by thermally driven regional (e.g., plains-mountain) and mesoscale (e.g., valley and slope) circulations (Stewart et al., 2002). There was no measurable precipitation, but the temperatures on AP dropped below freezing on the nights of 17-18 May and 11-12 June, and the cold fronts decreased the depth of the afternoon boundary layers above the NLVA to ≈ 2 km, or roughly the elevation of AP above the valley.
The jet stream retreated north into Canada in the wake of L 4 , and the NCEI Synoptic Discussion for June 2017 confirms that the expanding subtropical ridge seen in Fig. 3e dominated the weather across the SWUS for the rest of the campaign (https://www.ncdc.noaa.gov/sotc/synoptic/ 201706, last access: 16 August 2021). The ridge brought clear skies, dry conditions, and record high temperatures to Clark County, and June 2017 was the third-hottest in Las Vegas since official record keeping began in 1948, and the extreme temperatures and dry conditions exacerbated multiple wildfires across the SWUS. The high temperatures at the NLVA exceeded 41 • C (106 • F) each day during the last 2 weeks of the study, and the daily records for Las Vegas were tied or exceeded on 5 straight days from 20 to 24 June. The official high of 47.2 • C (117 • F) at Harry Reid International Airport on 20 June, the last day of astronomical spring, tied the all-time Las Vegas record. Coincidentally, this record was also tied on the final day of the first LVOS campaign (30 June 2013). The daily high temperatures abated slightly to 42-43 • C during the last few days of the campaign and IOP4, when a weak cold front associated with the shallow trough (L 8 ) in Fig. 3f passed through the LVV. The temperatures on AP were typically 10-15 • C lower due to its higher elevation.

TOPAZ profiles
TOPAZ operated for an average of ≈ 12 h a day over 45 consecutive days (17 May-30 June) during FAST-LVOS, accumulating a total of 4026 profiles or 537 h of observations. Approximately 60 % of the profiles were acquired between the hours of 09:00 and 17:00 PDT, but there were several extended runs including a 60 h continuous session from 11-13 June. The O 3 and β profiles are summarized as timeheight curtain plots in Figs. 4 and 5, respectively. The scalloped appearance of the individual curtains is caused by the diurnal variation in background solar radiation, which determines the measurement signal-to-noise and thus the maximum achievable altitude. The dark-gray curves show the boundary layer heights from the Doppler lidar, and the red boxes outline the four IOPs. The colored stripe along the bottom of Fig. 4 shows the NLVA measurements from the in situ monitor in the truck. Preliminary TOPAZ measurements from the afternoon of 17 May caught the remnants of a deep cyclonic intrusion from the closed low (L 1 ) in Fig. 3a, and free-tropospheric layers and filaments were present in nearly all of the profiles measured between mid-May and mid-June. The low backscatter in Fig. 5 shows that these layers were not created by biomass burning, and most of the layers were higher (> 4 km a.g.l) and more persistent than would be expected for pollution lofted from the Los Angeles Basin by the "mountain chimney effect" , but this process may have contributed to some of the lowerlying "residual layers" in the ozone curtains (e.g., 16 June). This suggests that most of the higher-O 3 layers were stratospheric intrusions or Asian pollution plumes, a conclusion supported by the AM4 and GEOS-Chem simulations (Zhang et al., 2020). Free-tropospheric O 3 layers appeared less frequently after the jet stream retreated into Canada, and photochemical production increased rapidly as the subtropical ridge moved into the SWUS. The highest O 3 concentrations were usually measured in the boundary layer after mid-June, although Fig. 4 shows that this pattern was briefly interrupted on 18-19 June, when the anticyclonic circulation transported clean marine air deep into the IMW. High O 3 was present both in and above the boundary layer during the last 3 d of the campaign, when IOP4 was conducted.
The backscatter curtains in Fig. 5 show that TOPAZ measured relatively low backscatter during most of the campaign. The aerosol loading was usually highest in the boundary layer, but comparisons to the near-IR Doppler lidar measurements and ozonesondes show that the gradients in the UV profiles were usually too weak to reliably define the boundary layer height. The backscatter above the top of the boundary layer increased abruptly on 19 June, when smoke from fires in Arizona and Mexico drifted into the LVV, and high backscatter was measured both in and above the boundary layer on 20 June, when smoke from the much closer Holcomb Fire reached Las Vegas. Note that the boundary layer height determination relied heavily on the vertical velocity variance profile when dispersed smoke was present. These measurements will be described in more detail elsewhere along with those from 23-26 June, when smoke from other regional wildfires reached Las Vegas and possibly contributed to the high surface O 3 measured on those days.

Mobile laboratory measurements
The nearly continuous 1 min averaged in situ O 3 measurements from the AP mobile laboratory are summarized in the top panel of Fig. 6; the lower panels plot the corresponding CO, CO 2 , CH 4 , N 2 O, NO, NO 2 , and NO y measurements. As noted above, the lack of rainfall and extreme aridity of the Mojave Desert allow us to use water vapor as a semiconserved tracer, and each of the time series is colorized by the co-measured H 2 O to show whether the sampled air came from the lower stratosphere or upper troposphere (dry: violet-blue) or from the terrestrial (moderate: green-yellow) or marine (moist: yellow-red) lower troposphere. The letters A-G identify specific transport episodes that are examined in more detail below. The mobile laboratory measured 1 min O 3 mixing ratios in excess of 80 ppbv on 8 of the 43 measurement days in the campaign and mixing ratios in excess of 70 ppbv on 17 d. The MDA8 O 3 averaged 60 ± 8 ppbv and exceeded the NAAQS on 4 d. Some of the highest O 3 concentrations were measured in very dry air with low NO 2 and NO y during the night and early morning (e.g., 8-9 June, episode C, and 11-12 June, episode D), but high O 3 was also measured in more humid air with elevated NO 2 and NO y on some afternoons (e.g., 2 June, episode B). The AP wind measurements show that all of the short-lived nocturnal peaks coincided with strong southwesterly winds, while the broader afternoon peaks were associated with weaker southeasterly upslope flow. The lowest O 3 concentrations of the campaign were measured on 18-19 June in very moist air transported inland by the anticyclonic flow around the subtropical ridge seen centered over southern Nevada in Fig. 3e.

Comparisons and validation
The vertical scanning capability of the TOPAZ lidar allows direct comparisons with nearby surface in situ monitors. Figure 7a compares time series of the O 3 mixing ratios retrieved at 27.5 ± 5 m a.g.l. (red) with the in situ measurements sampled 5 m a.g.l. at the truck (gray). The two series are in excellent agreement (±1 %, R 2 = 0.91) when the co-located Doppler lidar showed the boundary layer to be > 2500 m deep (black). The TOPAZ lidar was not normally run overnight, but Fig. 7a shows that the only significant differences between the time series occurred during a 24 h run through the night of 31 May-1 June (IOP2) when O 3 near the truck was titrated by NO emitted from nearby combustion sources. These emissions did not affect the TOPAZ concentrations retrieved about 20 m high and 800 m down range. The nocturnal losses were much smaller on 25 May and 7-12 June, when winds from the passing cold fronts dispersed the NO x emissions and disrupted the shallow nocturnal inversion layers. Figure 7b is similar to Fig. 7a but compares the NLVA in situ measurements with the TOPAZ mixing ratios at 2000 m a.g.l., the elevation of AP. Although TOPAZ frequently measured higher O 3 aloft on 7-12 June when the boundary layer was relatively shallow (Fig. S1), these measurements are also in good agreement (±1 %, R 2 = 0.76) when the boundary layer was deeper than 2500 m. Figure 7c compares the in situ measurements from the NLVA (gray) with those from AP (red). The most obvious difference between the time series is the lack of surface deposition and NO x titration on AP, which was exposed to free-tropospheric air during the night. The black points show that mixing ratios were similar (±4 %, R 2 = 0.46) when the boundary layer above the NLVA was > 2500 m. One notable exception is the measurements from 23 June, when the measurements in the LVV may have been influenced by smoke from the Brian Head Fire in southwestern Utah (cf. Fig. 5f). Although difficult to see in Fig. 7c, the AP O 3 concentrations typically lagged those at the NLVA on days with welldeveloped upslope flow, including the 4 AP exceedance days (14,16,29,and 30 June). A key assumption of the FAST-LVOS experimental design was that the air sampled on the summit of AP was representative of the air entrained by the mixed layer above the LVV. The similarity between the 2000 m a.g.l. TOPAZ retrievals and the mobile laboratory O 3 measurements in Fig. 7d shows that this was generally the case when the boundary layer was > 2500 m deep. The time series are in good agreement (±4 %, R 2 = 0.60) under these conditions if the measurements from 23 June are excluded.

Intensive operating periods
The aircraft and ozonesonde profiles acquired during the four IOPs provided spatial context for the lidar measurements and helped distinguish stratospheric intrusions from Asian pollution. The planning and successful execution of these intensives relied on the ability of the RAQMS (96 h) and RAP-Chem (48 h) models to predict stratospheric intrusion and pollution transport events more than 48 h in advance so that the aircraft and ozonesonde teams could return to Las Vegas from their home stations. Examples of the model forecasts are shown in Fig. 8, which displays the O 3 and CO forecasts for 12:00 UT 12 June (cf. Fig. 3d). The left panels of Fig. 8 show the 96 h total O 3 (top) and CO (bottom) at 320 K (≈ 500 hPa or 5.8 km a.s.l. above Las Vegas) from RAQMS initialized at 12:00 UT on 7 June, and the right panels show the 36 h RAP-Chem O 3 and CO 500 hPa forecasts initialized at 00:00 UT on 11 June. Note that the higher-resolution RAP-Chem model gets its boundary conditions from the RAQMS model. This example shows that the RAQMS model captured both the timing and location of the stratospheric intrusion-Asian pollution event 4 d out or 3 d before the start of IOP3 on 10 June. The retrospective MERRA-2 PV analyses in Fig. 3 and FLEXPART stratospheric ozone (STTO3) and Asian CO (ASIACO) tracer distributions (https://csl.noaa.gov/projects/ fastlvos/FAST-LVOSfinalreport604318-16.pdf, last access: 7 December 2021) in Figs. 9 and 10, respectively, show that stratospheric air and/or Asian pollution was present in the middle and upper troposphere above Nevada during each of the IOPs. The correspondence between the 96 h RAQMS forecasts in Fig. 8 and the retrospective FLEXPART distributions in the third columns of Figs. 9 and 10 is particularly impressive.
The curtain plots in Fig. 4 show that the TOPAZ lidar measured high O 3 in the middle and upper troposphere during each of the IOPs; the expanded 3 d curtain plots in Figs. 11-14 show these measurements in more detail. The superimposed horizontal ribbons represent the NLVA and AP surface measurements and the nearly vertical ribbons the ascending profiles from the ozonesondes. Those portions of the descending profiles within 20 km of the NLVA are also plotted. The NLVA and AP surface wind measurements and continuous Doppler lidar boundary layer heights are also overlaid on the curtains, with the boundary layer heights derived from the ozonesonde potential temperature profiles (open black circles) plotted for comparison. The heavy arrows at the top of each plot correspond to the MERRA-2 analyses in Fig. 3. The labels A, B, D, and G match the corresponding peaks in Fig. 6.
The SA aircraft profiles overlap with the midday ozonesonde profiles and are omitted from the curtain plots for clarity, but Fig. 15 shows longitudinal transects of the sections of the outbound (left) and inbound (right) flight legs between the NLVA and AP on 25 May (IOP1), 2 June (IOP2), 12 June (IOP3), and 28 June (IOP4). The flight tracks are colorized by the O 3 measurements, and the plots also show the mean lidar and ozonesonde profiles and surface O 3 measurements from the roughly 1 h interval bracketing the NLVA and AP profiles. Ozonesondes were only launched between 09:00 and 15:00 PDT and thus did not overlap with the return flight legs. These plots underscore the good agreement between the different FAST-LVOS O 3 measurements and support the assumption that the air above AP was usually representative of the air above the LVV. Similar plots colorized by H 2 O and CH 4 are displayed in Figs. S2 and S3. Figure 16 shows  the measured O 3 mixing ratios along the outbound flight legs from the NLVA to Big Bear and the inbound flight legs back to Jean. The AP profiles and TOPAZ measurements are omitted from these plots as are the afternoon profiles above the LVV. In the following sections, we briefly describe each of the IOPs and compare the upper-air measurements with the FLEXPART STTO3 and ASIACO tracer distributions. More complete descriptions of these measurements are planned for future publications.

IOP1: 23-25 May
The first IOP was timed to coincide with the arrival of the trough (L 2 ) poised above the northern IMW in Fig. 3b. The PV analysis for the morning (12:00 UT) of 25 May shows a deep cyclonic intrusion wrapping around L 2 , and the FLEX-PART STTO3 plots in the first column of Fig. 9 show a deep cyclonic intrusion descending almost vertically to 700 hPa above northern Nevada and California. The PV analysis in Fig. 3b also shows a thin anticyclonic streamer stretching across southern Nevada from the tip of L 1 , now an elongated trough over the eastern US, and FLEXPART shows this shallower feature on the 300 and 400 hPa surfaces. The ASIACO tracer distributions in the first column of Fig. 10 show Asian pollution mingled with the cyclonic intrusion and in a deep narrow band stretching across northern Nevada between the cyclonic and anticyclonic intrusions. Analysis of satellite CO measurements and the GFDL-AM4 and GEOS-Chem model simulations by Zhang et al. (2020) also found a large Asian pollution component in these layers.
The curtain plots for 23-25 May in Fig. 11 show layers with more than 100 ppbv of O 3 above the LVV on all 3 d, but the layers were above the ≈ 6 km a.s.l. altitude ceiling of the SA Mooney on the first 2 d. The aircraft was able to reach the more diffuse band with 75 to 80 ppbv of O 3 seen above 4.5 km a.g.l. on 25 May, however, and the plots in Figs. 15 and 16 show that this layer extended to the south at least as far as Barstow (cf. Fig. 1). Figure S3 shows that the air in this layer was extremely dry, but the absence of a corresponding CH 4 enhancement in Fig. S4 suggests that this layer was primarily of stratospheric origin. The lidar, aircraft, and ozonesonde measurements also show that the high O 3 aloft was separated from the boundary layer by a layer of continental air with much lower O 3 concentrations, and there was no obvious local mixing of the O 3 aloft into the boundary layer. The GFDL-AM4 simulations also found little evidence for stratospheric or Asian pollution influences in Clark County surface air but estimated Asian contributions of 8-15 ppbv of O 3 to the surface along the areas of northern California, Idaho, and Wyoming lying beneath the bands seen in the 700 hPa STTO3 and ASIACO FLEXPART analyses.
The GFDL-AM4 and GEOS-Chem models did show large contributions from local and regional pollution in the lowest few kilometers above Clark County during IOP1, and the TOPAZ curtain plots also show moderately high (70-80 ppbv) O 3 in the boundary layer on 23 May and what appears to be a residual layer with 70-80 ppbv of O 3 and high β above the boundary layer on the morning of 24 May. The lidar also found 60-70 ppbv of O 3 with high β in the boundary layer on the morning of 25 May that decreased abruptly when the winds rotated to the southwest just after noon. These measurements correspond to episode A in Fig. 6, and the outbound aircraft measurements plotted in the top panels of Figs. S2 and S3 show that the air above the LVV and AP A. O. Langford et al.: The Fires, Asian, and Stratospheric Transport-Las Vegas Ozone Study (FAST -LVOS) Figure 10. Same as Fig. 9, but for the ASIACO tracer.
was relatively moist and enriched in CH 4 . The mobile laboratory also found moist air with moderate O 3 on AP but also measured elevated CO, CH 4 , CO 2 , and N 2 O and the highest NO y concentrations measured during the campaign. The scatterplots in Fig. 17 show that O 3 was positively correlated with all of these tracers, and a 96 h HYSPLIT back trajectory launched 2 km above the NLVA (Fig. 18, solid red line) meandered around the southern San Joaquin Valley (SJV) within the shallow boundary layer in the valley, which is typically less than 1 km deep in summer (Faloona et al., 2020), for more than 48 h before exiting through the Tejon Pass and crossing the Mojave Desert to Las Vegas. This can explain the high N 2 O and CH 4 concentrations, which likely came from agricultural and oil and gas sources in the SJV. The air parcel was also exposed to urban sources in the SJV, and the high CO 2 and NO 2 concentrations show that the parcel entrained additional urban pollution as it passed through the LVV en route to AP.
This interesting case illustrates yet another way that passing troughs can influence surface air quality in the LVV. Figure 18 shows that the trajectory initially followed the cyclonic circulation from L 2 (cf. Fig. 3b) southward along the California coast (hence the high moisture content) before crossing the Coast Range into the SJV on 23 May. The parcel then followed the circulation northward into the LVV as the trough moved east. The influence of the synoptic flow on  Fig. 3b. The 300 hPa surface lies at approximately 9 km a.g.l. The label "A" is the same as in Fig. 6. the low-level winds is also seen in the transport to AP, which occurred in the early morning and several hours before the thermally forced upslope flow became established.

IOP2: 31 May-2 June
The second IOP focused on one in a series of shallow upperlevel lows that spun off of the Aleutian Low in late May and early June. The MERRA-2 analysis for 06:00 UT 2 June in Fig. 3c shows the low (L 3 ) approaching the Pacific Northwest. A narrow filament of high-PV air emanating from the low stretches anticyclonically above the Nevada-Utah bor- Figure 12. Same as Fig. 11, but for IOP2. The heavy black arrow corresponds to the 06:00 UT 2 June 300 hPa plot in Fig. 3c. der, but the corresponding FLEXPART STTO3 distributions in the second column of Fig. 9 suggest that the intrusion was quite shallow. The ASIACO distributions in Fig. 10 show bands of Asian pollution on both sides of the stratospheric filament, with the pollution descending to at least 400 hPa above California. The TOPAZ and ozonesonde measurements in Fig. 12a show two downward-sloping bands of high O 3 between the top of the convective mixed layer and ≈ 9 km a.g.l. on the afternoon of 1 June, with the lowermost band appearing to merge with the mixed layer on the afternoon of 2 June. The backscatter curtains in Fig. 12b also show slightly higher β in the uppermost layer at 5 km a.g.l. than in the lower one at 4 km a.g.l.
The aircraft and ozonesonde profiles in Fig. 15 also show the two O 3 layers above the LVV in the morning, but the aircraft measurements show little indication of the lower layer in the profile above AP, and Fig. 16 shows it to have been rather tenuous between Las Vegas and Barstow. Figures S2  and S3 show that both layers were very dry (3 %-8 % RH), but only the uppermost layer was elevated in CH 4 . This, together with the backscatter measurements in Fig. 12b, sug-Figure 13. Same as Fig. 11, but for IOP3. The heavy black arrow corresponds to the 12:00 UT 12 June 300 hPa plot in Fig. 3d. gests that the upper layer was mostly Asian pollution, while the lower layer was mostly stratospheric. The TOPAZ curtain plot and afternoon aircraft profiles show that the convective boundary layer completely entrained the lower (stratospheric) layer and reached the bottom of the upper (Asian pollution) layer, where some entrainment must also have taken place.
The mobile laboratory measurements from AP in Figs. 6 and 17 show a pronounced O 3 peak on the afternoon of 2 June (episode B), but the other tracers show that this peak was caused by continental air and not the stratospheric intrusion or Asian pollution plume, consistent with the modelbased attribution of the 2 June O 3 peak to regional pollution (Zhang et al., 2020). The sampled air had typical tropospheric N 2 O concentrations and elevated CO, CO 2 , NO 2 , and NO y concentrations consistent with anthropogenic pollution. The high NO 2 / NO y ratio points to local sources, and the wind barbs in Fig. 12 show that the air was transported up Kyle Canyon from the LVV by the afternoon upslope flow. The HYSPLIT trajectory in Fig. 18 (green line) approaches southern Nevada from the southeast without passing over Figure 14. Same as Fig. 11, but for IOP4. The heavy black arrow corresponds to the 00:00 UT 29 June 300 hPa plot in Fig. 3f.
California or any of the fires burning in Mexico and Arizona at the time. Similar local upslope events occurred frequently in late June (14, 16, 22, and 30 June) after the subtropical ridge became established.

IOP3: 10-14 June
The third IOP was triggered by another deep closed low (L 4 ) that moved into the western US in the second week of June. This cyclonic system was unusually deep for mid-June, and the PV plot for 12:00 UT on 12 June in Fig. 3d appears very similar to that for 00:00 UT 18 May (Fig. 3a). The cold front brought strong (> 10 m s −1 ) south-southwesterly winds to Clark County and freezing temperatures (−5 • C) to AP on the night of 11-12 June (Fig. S1). The FLEXPART STTO3 analyses in Fig. 9 show a deep cyclonic stratospheric intrusion descending to at least 700 hPa over southern Nevada, and the corresponding ASIACO analyses in Fig. 10 show cyclonic bands of pollution descending ahead of the stratospheric air. The RAQMS and RAP-Chem forecast plots in Fig. 8 appear similar.  The third IOP was extended to 5 d (10-14 June), with aircraft sorties flown on all 5 d and ozonesondes launched on 4 (10-13 June). The curtain plot in Fig. 13a shows a continuous TOPAZ run of 60 h lasting from 08:00 PDT on 11 June to 21:00 PDT on 13 June. The lidar and superimposed ozonesonde profiles show a complex network of O 3 filaments descending into the lower troposphere, with more than 150 ppbv of O 3 approaching to within 3.3 km of the surface at ≈ 03:00 PDT on 12 June and ≈ 250 ppbv measured at 5.5. km a.g.l. around 09:00 PDT. The aircraft, ozonesonde, and lidar profiles in Fig. 15 show more than 120 ppbv of O 3 between 6 and 8 km a.g.l. in the early afternoon of 12 June, and Fig. 16 shows the intrusion sloping downward to the south, with more than 140 ppbv measured along the 5.5 km a.s.l. flight leg between Barstow and Jean. The profiles near the NLVA and AP also show elevated CH 4 in the air above and below the high-O 3 layer (Fig. S4), which along with the elevated backscatter beneath the tongue in Fig. 13b shows the Asian pollution descending ahead of the stratospheric air. Figure 13a shows that the O 3 maximum of 84 ppbv recorded around 23:00 PDT on 11 June by the mobile laboratory on AP (episode D) occurred close to the deepest penetration of the intrusion in the lidar curtain. The sampled air was extremely dry (≈ 0.1 % H 2 O), and the stratospheric origin was confirmed by the low N 2 O concentrations in Fig. 6 and the negative correlations of O 3 with CO, CH 4 , and N 2 O in Fig. 17. The NO y converter was offline during this episode, Figure 18. NOAA HYSPLIT 96 h back trajectories launched 2 km above North Las Vegas during episodes A-G. The small and large filled symbols are spaced 12 and 24 h apart, respectively. The trajectories were calculated using the 40 km NCEP Eta Data Assimilation System (EDAS) meteorology. but the NO x concentrations were near the detection limit. Similar chemical characteristics were measured in the air sampled during the first of the two O 3 peaks that appeared on the morning of 5 June and are mentioned in Sect. 5.2. Figure 18 (purple line) shows the HYSPLIT back trajectory descending from the northwest over the previous 72 h.
Interestingly, this large intrusion did not lead to particularly high surface O 3 in the LVV on 11-13 June since the high winds from the cold front dispersed most of the locally produced O 3 and other pollutants ahead of the descending stratospheric air (Langford et al., 2012). The curtain plots do show a short-lived increase in the lowest 2 km on the afternoon of 11 June, when the surface O 3 concentrations briefly climbed from 50 to 70 ppbv at the NLVA. However, the GFDL-AM4 model suggests that the intrusion layer was mixed into regional pollution on 13-14 June and pushed the MDA8 O 3 to exceed the 70 ppbv NAAQS level at sites across the SWUS, including several sites in Clark County (Zhang et al., 2020). Once stratospheric air is mixed into local pollution, it loses its key stratospheric characteristics (e.g., low CO and low H 2 O), challenging diagnosis of its surface impacts directly based on observations. This case study demonstrates the importance of integrating observational and modeling analysis for the unambiguous attribution of high-O 3 events in the SWUS (Zhang et al., 2020).

IOP4: 27-30 June
The final IOP targeted a weak trough (L 8 ) that crossed the Canadian border near the end of June. The MERRA-2 analysis for 06:00 UT 29 June (Fig. 3f) shows the PV maximum on the western flank of the trough. The analysis also shows the latest (L 8 ) in a series of cut-off lows (COLs) (Nieto et al., 2005) that formed off the coast of California in late June. These COLs typically meandered around the eastern Pacific for several days before rejoining the main flow, and Fig. 3f shows a narrow filament connecting L 9 to the remnants of L 7 (cf. Fig. 3e), which now lies above the Midwest. The FLEX-PART STTO3 tracer in Fig. 9 shows both a cyclonic intrusion from L 8 descending to 500-600 hPa above Oregon and the narrow filament connecting the two COLs above southern Nevada on the 400 hPa surface. Figure 10 shows that there was significant Asian pollution between the filament and the cyclonic intrusion.
The TOPAZ curtain plot in Fig. 14 shows high (> 80 ppbv) O 3 above the boundary layer on 29 and 30 June, with up to 90 ppbv measured in the boundary layer on 30 June. The SA profiles from these days (not shown) found very dry air with depressed CH 4 concentrations in the O 3 -rich layers above 4 km a.g.l., suggesting that they originated in the stratosphere. The most interesting measurements are those from 28 June, however, which show a thin layer with more than 100 ppbv of O 3 but low β disappearing into the convective mixed layer in the early afternoon, the most striking example of boundary layer entrainment observed during FAST-LVOS. The companion paper by Zhang et al. (2020) refers to this as an "unattributed event" since the layer was not resolved by either the AM4 (C192: ≈ 50 × 50 km 2 horizontal, ≈ 200 m vertical) or GEOS-Chem global models and cannot be resolved by FLEXPART for that matter. The spatial inhomogeneity of the layer is evidenced by the aircraft measurements in Figs. 15 and 16. Figure S3 shows that the air was very dry both within and above the mystery layer, and the combination of extreme dryness and low β rules out biomass burning or regional pollution. The layer was also rich in CH 4 , suggesting that it was not a stratospheric intrusion. Figure 14 shows that the winds on the summit of AP were mostly southerly during the morning and early afternoon of 28 June, and the mobile laboratory sampled air with much lower O 3 concentrations than that profiled by the lidar. The winds briefly rotated to the northwest in the early evening, and the mobile laboratory recorded a drop in relative humidity and increase in O 3 around 18:30 PDT (episode G). The relationships in Fig. 17 suggest that this spike was caused by Asian pollution, a conclusion supported by the HYSPLIT back trajectory (dashed blue line) in Fig. 18, which descends from the upper troposphere over the Pacific Ocean before passing over the SJV well above the boundary layer. Although the transient peak in Fig. 6 was detected more than 6 h after the O 3 layer was entrained, it does support the conclusion that Asian pollution was present in the vicinity of AP on 28 June.

Other transport examples
Not all of the interesting transport episodes occurred during one of the four IOPs, and the labels C, E, and F in Figs. 6, 17, and 18 highlight three other examples. Figure 6a shows that the highest 1 min O 3 concentrations (≈ 95 ppbv) measured on AP during FAST-LVOS 2017 were sampled during episode C around 02:00 PDT on the morning of 9 June. The high O 3 lasted for about an hour but occurred several hours after TOPAZ had ceased operations for the night. However, the 8 June curtain plot in Fig. 4d shows a 1 km deep layer of high O 3 just above the top of the boundary layer throughout the observing session that appears to have persisted through the night and into the next morning, when the lidar operations resumed. The unusual dryness of the air sampled on AP and relationships between O 3 and the other measured species (Fig. 17) suggest that this layer was also Asian pollution that recently descended from the upper troposphere. The episode C measurements completely overlap with the measurements from 28 June (episode G) in many of the scatterplots, suggesting a common origin. The HYSPLIT trajectories in Fig. 18 are also similar, and the peak concentrations measured by TOPAZ, ≈ 105 ppbv on 8 June and ≈ 115 ppbv on 28 June, are also comparable.
The 1 min mobile laboratory measurements from the morning of 18 June (episode E) include both the highest H 2 O (12 g kg −1 ) and lowest O 3 concentrations (29 ppbv) measured on AP during FAST-LVOS 2017. Figure 4f shows that TOPAZ also measured unusually low O 3 throughout the tropospheric column. As noted above, this episode is attributed to transport of clean marine air deep into the IMW by the anticyclonic flow around the subtropical ridge (Fig. 3e), and the HYSPLIT back trajectory (Fig. 18, dashed red line) suggests that this moist background air was lifted into the lower free troposphere above the Pacific Ocean more than 4 d earlier. Figures 6 and 17 show that the mobile laboratory also measured very low O 3 , CO, CO 2 , CH 4 , and NO 2 concentrations on AP (the NO y converter was offline). The filled black circles in Fig. 17 show that the O 3 , N 2 O, CO 2 , and CH 4 concentrations measured at AP on the morning of 18 June were comparable to the mean (±1σ ) June 2017 values measured at the high-elevation NOAA GML Mauna Loa Observatory (https://gml.noaa.gov/dv/data/, last access: 16 August 2021). These measurements are thought to represent the background concentrations in the free-tropospheric air reaching the US west coast.
The final example (episode F) coincides with the highest backscatter ( Fig. 5f) measured above the NLVA during FAST-LVOS 2017. These measurements were made about 24 h after the start of the Holcomb Fire near Big Bear on the afternoon of 19 June (cf. Fig. 1). This fire grew rapidly but was battled aggressively, and the expansion was checked at ≈ 600 ha (≈ 1500 acres) by the evening of 21 June. The anticyclonic path followed by the HYSPLIT trajectory (dashed green line) in Fig. 18 passed directly over the Holcomb Fire and possibly over older fires in Arizona and northern Mexico. The curtain plots in Figs. 4f and 5f show that there was ≈ 85 ppbv of O 3 in the smoke that appeared at ≈ 3 km a.g.l. on the morning of 20 June but only ≈ 50 ppbv of O 3 in the denser smoke that appeared between 4 and 6 km a.g.l. in the afternoon and evening. Some of the smoke from the second plume was entrained by the convective boundary layer, and the CO time series from the mobile laboratory on AP shows a corresponding rise to more than 140 ppbv. However, Fig. 17e shows that there was little NO 2 in the second plume, which is consistent with the absence of any O 3 enhancement in either the sampled air or the lidar measurements.

Baseline ozone during FAST-LVOS
The top five panels of Fig. 19 show the hourly averaged and MDA8 O 3 measurements from the remote sampling sites operated by the WMRC and the NVDEP during FAST-LVOS. All of these sites are located well away from populated areas, with the WMRC located ≈ 310 km WNW of the LVV and the NVDEP monitors located between 180 and 250 km to the north (cf. Fig. 1). The bottom panel shows the measurements from the monitor maintained by the CCDAQ in the town of Mesquite ≈ 120 km NE of Las Vegas, near the Nevada-Utah border. MDA8 O 3 concentrations exceeding 65 and 70 ppbv are highlighted in orange and red, respectively. The mean MDA8 O 3 mixing ratio (±1σ , the standard deviation of the mean) is shown in each plot. These measurements show a pronounced elevation trend consistent with descent of O 3 from the free troposphere and destruction at the surface. The former is reflected in the higher mean values and episodic increases at the higher elevations and the latter in the low nighttime values at the lower elevations.
The WMRC operates the highest-elevation long-term O 3 monitors in the continental US, with both the WMS (4.34 km a.s.l.) and BCO (3.88 km a.s.l.) located roughly 1 km higher than the CASTNET site in Gothic, CO The mean MDA8 O 3 at WSSU, the highest-elevation NVDEP site at 2.3 km a.s.l., was about 17 ppbv lower than that at WMS, with a 4MDA8 O 3 of 61 ppbv. The large episodic increases measured by the WMRC monitors during the IOPs did not reach the lower-lying NVDEP sites, and there were no NAAQS exceedances by any of these nonregulatory monitors. Nevertheless, the mean MDA8 O 3 at these three remote sites ranged from ≈ 50 to 55 ppbv, or 70 %-80 % of the NAAQS, between mid-May and the end of June, with 4MDA8 O 3 concentrations ranging from 59 to 62 ppbv (84 %-88 % of the NAAQS). An earlier analysis of A. O. Langford et al.: The Fires, Asian, and Stratospheric Transport-Las Vegas Ozone Study (FAST -LVOS) springtime (March-May) measurements from 2012 and 2013 at CGSP (Fine et al., 2015a) found very similar mean MDA8 O 3 concentrations of 51 ± 6 and 50 ± 8 ppbv, respectively. These measurements are comparable to the background (i.e., formed from natural sources plus anthropogenic sources in countries outside the US) monthly mean springtime MDA8 O 3 concentration of 50 ppbv estimated by Lin et al. (2012a, b) for higher elevations in the IMW using the GFDL AM3 model for the year 2010 but are significantly higher than the mean MDA8 O 3 of 40 ± 7 ppbv derived by Zhang et al. (2011) using GEOS-Chem model runs for the years 2006-2008 and the upper limit of 40-45 ppbv for the seasonal mean MDA8 estimated by Dolwick et al. (2015) using Community Multiscale Air Quality (CMAQ) and Comprehensive Air Quality Model with Extensions (CAMx) runs for the year 2007. These differences should be viewed with caution since they do not account for changes in Asian precursor emissions or differences in trans-Pacific pollution transport and STT caused by the position of the jet stream (Lin et al., 2015;Albers et al., 2018;Breeden et al., 2021).

Implications for air quality attainment
The EPA has designated Las Vegas, Nevada, a marginal nonattainment area for O 3 (https://www3.epa.gov/airquality/ greenbook/jnc.html, last access: 16 August 2021). The warm and sunny conditions that lead to rapid photochemical production of O 3 also drive the upslope flow into the Spring Mountains that causes O 3 and other pollutants to accumulate in the northwestern LVV. The highest O 3 concentrations are usually measured at the Joe Neal (C75) and Walter Johnson (C71) monitors (cf. Fig. 1), which had 2017 ozone design values, i.e., the 3-year running average of the fourth-highest MDA8 O 3 , of 74 and 72 ppbv, respectively. Figure 20 plots the hourly averaged and MDA8 O 3 measurements from these two sites during FAST-LVOS, along with the corresponding measurements from the Apex (C22), Jean (1019), and Indian Springs (C7772) monitors (cf. Fig. 1). The thin blue line in each plot represents the average of the RRVA, CGSP, and WSSU MDA8 O 3 measurements from Fig. 19, which we use as an estimate for the southern Nevada mean baseline MDA8 O 3 . The Joe Neal and Walter Johnson monitors also recorded the highest O 3 concentrations in Clark County during FAST-LVOS and exceeded the 2015 NAAQS on 7 and 6 d, respectively. Most of these exceedances coincided with the higher temperatures that followed the poleward expansion of the subtropical ridge in mid-June (cf. Fig. S1). Several of the other monitors located within a 15 km radius of the NLVA also exceeded the NAAQS on at least 1 of these days. There was only one exceedance each at the slightly more distant Apex (32 km NE) and Green Valley (22 km SE) monitors and no exceedances at the Boulder City (40 km SE), Jean (49 km SSE), or Indian Springs (58 km NW) monitors (cf. Fig. 1c). Note that the Jean monitor was offline from 9-12 June. All of the CCDAQ monitors measured unusually low O 3 on 18-19 June.
The MDA8 O 3 measured by the Indian Springs monitor was very similar to the NVDEP baseline values, with the largest differences arising on 16-17 June, when the anticyclonic marine incursion had reached the more northerly NVDEP sites but had not yet reached Clark County. The mean NVDEP and Indian Springs time series agree to within 3 % on average (R 2 = 0.69) if the measurements from these 2 d are excluded. This suggests that the Indian Springs measurements may be a good proxy for the LVV baseline O 3 during the FAST-LVOS campaign, and the solid black line in Fig. 20f represents the difference between the Joe Neal MDA8 O 3 concentrations and the Indian Springs monitor. The dotted line is similar but shows the difference for the Walter Johnson monitor. The measurements from days where the MDA8 O 3 concentrations exceeded 65 or 70 ppbv are highlighted as before. The MDA8 O 3 concentrations at Joe Neal and Walter Johnson were significantly higher than the baseline concentrations on 23 May, with the Walter Johnson and Joe Neal monitors reaching 71 and 69 ppbv, respectively. Although the lidar, aircraft, and ozonesonde observations showed high O 3 aloft throughout the first IOP, there was no obvious mixing of the high O 3 into the boundary layer. There were multiple exceedances in the San Joaquin Valley and South Coast air basins on 22-24 May, however, and the elevated O 3 in the LVV on 23 May is most likely due to regional transport and local production (cf. Sect. 6.1). Ozone was also elevated in the LVV on 2-3 June, and the measurements in Fig. 12 Fig. 19. Nearly all of the NAAQS exceedances at the high elevation sites occurred before the middle of June, when stratospheric intrusions and Asian pollution plumes appeared frequently in the middle and lower troposphere, but nearly all of the exceedances in the LVV occurred after the middle of June, when higher temperatures and stagnant conditions increased photochemical production of O 3 in the boundary layer. The high temperatures also exacerbated regional wildfires, including the Holcomb and Brian Head fires, whose plumes reached the LVV in the third week of June (cf. Figs. 4f and 5f). Note that these fires did not increase surface O 3 at the Jean, Indian Springs, or Mesquite monitors, suggesting that the smoke plumes were NO x -limited, and significant O 3 pro-  , and Asian pollution event of 21 June (Langford et al., 2017). These major events affected monitors outside of Las Vegas as well as those in the valley, and the Jean monitor exceeded 70 ppbv on 8 d in 2013 but not once in 2017 (the Indian Springs monitor was not installed until 2015).

Summary and conclusions
The 6-week-long FAST-LVOS field campaign collected a wealth of lidar, surface, aircraft, and ozonesonde measurements that greatly improve our understanding of O 3 transport in Clark County, Nevada, and the greater southwestern US and Intermountain West in late spring and early summer. Daily lidar observations found high-O 3 layers in the lower and middle free troposphere above Las Vegas on more than 75 % (35 of 45) of the measurement days. The highest tropospheric concentrations were measured on 11-12 June, when both the lidar and an ozonesonde launched nearby found up to 250 ppbv within 5.5 km of the surface. Several of these elevated O 3 layers were also sampled by a high-elevation research monitor located more than 300 km to the westnorthwest, in the White Mountains of California (Burley and Bytnerowicz, 2011), and the MDA8 O 3 concentrations at this site averaged more than 68 ppbv over the course of the study. Simulations from the FLEXPART, GEOS-Chem, and AM4 models (Zhang et al., 2020) show that the elevated layers were created by stratospheric intrusions or long-range transport from Asia and, in most cases, a mixture of the two. Aircraft measurements made during the four IOPs support this conclusion, and the aircraft mapped the spatial distribution of these plumes above the Mojave Desert in southern Nevada and California. FAST-LVOS also documented several finescale anticyclonic streamers in addition to the deep cyclonic intrusions that are the focus of most studies.
The FAST-LVOS measurements also captured several examples of O 3 being entrained from aloft by the convective boundary layer, and correlations between the different tracers measured on AP (2.7 km a.s.l.) showed the distinctive signatures of both stratospheric air and Asian pollution on multiple occasions. The AP measurements also found wildfire influences and evidence for regional transport of anthropogenic pollution from the San Joaquin Valley in addition to local upslope transport of pollution from the LVV. Although none of the stratospheric intrusions or Asian pollution events directly caused any of the NAAQS exceedances in the LVV during FAST-LVOS, they clearly added to the surface concentrations on at least 1 (14 June) of the exceedance days and contributed to the mean MDA8 O 3 of 50-55 ppbv measured at the remote sites in rural Nevada during the study. These mean baseline concentrations represent 70 %-80 % of the 2015 NAAQS of 70 ppbv, making it more difficult for western air quality managers to maintain surface O 3 concentrations below the NAAQS in the IMW during late spring and early summer (Cooper et al., 2015;Uhl and Moore, 2018). Data availability. FLEXPART simulations presented in this paper are available upon request to the corresponding author (andrew.o.langford@noaa.gov). Field measurements during FAST-LVOS are available at https://csl.noaa.gov/projects/fastlvos/data. html (last access: 31 December; NOAA, 2021).
Author contributions. AOL and CJS conceived this study and planned the field campaign; AOL, CJS, RJA, SPS, AMW, IP, PDC, CWS, JP, TBR, SSB, ZCJD, GK, SAC, and DJC prepared the instruments and carried out the field measurements; SB and TAB analyzed the Doppler lidar data; KCA compiled and archived the mobile laboratory measurements; LZ performed the GFDL-AM4 simulations and all analysis under the supervision of ML; RBP and MP conducted the RAQMS and RAP-Chem model forecasts, respectively; JB and SE executed the FLEXPART model calculations; AOL did the analyses and wrote the article with inputs from all coauthors.
Competing interests. At least one of the (co-)authors is a member of the editorial board of Atmospheric Chemistry and Physics.