Meteorological Influences on Tropospheric Ozone over Suburban Washington , DC

During the NASA DISCOVER-AQ (Deriving Information on Surface conditions from Column and Vertically Resolved Observations Relevant to Air Quality) campaign conducted in July 2011 across metropolitan Washington, DC, 25 daytime ozonesondes were launched from the Howard University Beltsville Campus (HUBC), representing the highest temporal density of measurements ever taken over the region. Time-height depiction of ozone concentration reveals considerable variability near the surface and in the mid/upper troposphere. Meteorological influences are assessed using the synoptic and mesoscale information, which revealed that periodic stratospheric intrusion of ozone-rich air was the main mechanism for elevated levels in the upper troposphere. It is also shown that high concentrations of ozone prevalent in the free troposphere were due to a combination of advection and aged air masses, which accounted for over 40% of the total budget. Poor air quality episodes with high surface concentration were due local emissions of precursors and associated synoptic-scale conditions. Breaks in the elevated ozone concentrations in the boundary layer and free troposphere occurred from frontal passages and accompanying storms. In the most unusual case (26 July, 2011), the combined effect of storms followed by two frontal passages and high pressure positioned west of the region led to decrease column ozone as much as 70%. The unusually deep extent of the reduction appears to be related to the temporary cutoff of air advected from fire locations and polluted regions, which dominated the upper troposphere throughout the period.


INTRODUCTION
Elevated levels of tropospheric ozone (O 3 ) can be very hazardous to human health, therefore its surface concentrations are continuously monitored in order to issue air quality alerts to the general public (Bell et al., 2004).Tropospheric O 3 is primarily formed via photochemical reactions entailing nitrogen oxides (NO), carbon monoxide (CO), and volatile organic compounds (VOCs).Conversely, it can also be produced via interaction of regional pollution with convection and photochemical reactions resulting from lightning-generated nitric oxide (NO) (Thompson et al., 2007a, b;Thompson et al., 2013).Having an understanding of the prevalent weather conditions before, during and after poor air quality events is important as weather patterns influence production and transport of O 3 and its precursors.
For instance, the direction and strength of the wind combined with the amount of incoming solar radiation and cloud cover can affect O 3 production rates and transport through advection.Under favorable conditions with strong sunlight and high temperatures, emissions of VOCs and the combination of NO and nitrogen dioxide (NO 2 ) produce O 3 , with the highest concentrations tending to occur at distances a few tens of miles downwind of urban centers.
Nevertheless, this understanding cannot be limited to just monitoring surface conditions, as downward transport of air from the entire atmospheric column can contribute to relative O 3 maxima near the surface and above (Neu et al., 1994;Zhang and Rao, 1999;Aneja et al., 2000;Yorks et al., 2009;Morris et al., 2010;Tong et al., 2011).For example, stratospheric intrusions of O 3 , ignited through tropopause folding events, are one of the responsible mechanism for high O 3 variability throughout the entire column (Yorks et al., 2009;Hu et al., 2011;Thompson et al., 2013).Therefore, vertical characterization of O 3 profiles from surface to the lower stratosphere, combined with knowledge of the prevalent atmospheric conditions is of utmost importance.
We describe the synoptic and mesoscale meteorological drivers influencing O 3 profile variability in the Washington, DC area during the NASA Deriving Information on Surface conditions from Column and Vertically Resolved Observations Relevant to Air Quality (DISCOVER-AQ) campaign conducted in July 2011 across metropolitan Washington, DC.The observations described were obtained at Howard University Beltsville Campus (HUBC), which is located to the northeast of Washington, DC (39.05N, -76.88W), and is situated in a suburban area with about 5% of the land occupied by buildings.Local emissions of O 3 precursors from Washington, DC, Baltimore, MD as well as those transported from the Ohio Valley and Canada impact the study site (Hu et al., 2011), making it an ideal location for taking observations, and studying a variety of urban air quality processes.Thompson et al. (2013) describe a similar analysis from HUBC and a secondary site near Edgewood, MD (39.4N; 76.3W) during the same period, with the objective to achieve satellite comparisons and contrast the tropospheric O 3 budgets.This paper complements Thompson et al. (2013), but focuses on the specific synoptic and mesoscale features that influenced O 3 concentration.Unlike previous studies described at HUBC, where O 3 soundings were launched only during specific poor air quality events (Yorks et al., 2009;Morris et al., 2010), the measurements described here are of unprecedented temporal resolution and obtained over a range of various weather conditions, not only when high O 3 events were forecasted.We specifically examine the transport patterns that help maintain mid and upper tropospheric O 3 maxima, along with parcel trajectories and laminar analyses that characterize the temporal and spatial variability of elevated tropospheric O 3 .The application of a temporally dense dataset, however, allows for the examination of both polluted and unpolluted cases, as well as observations of the meteorological mechanisms that vary with both phases of O 3 incidence.

METHODS AND DATA
Atmospheric profiles of O 3 are known to exhibit a layered or "laminar" structure, and it is believed waves such as gravity waves (GW) and Rossby waves (RW) are responsible for the dynamical origins of these layers (Teitelbaum et al., 1996;Pierce and Grant, 1998).For example, laminae associated with RW usually represent downwind advection of a) O 3 pollution and precursors at the various atmospheric levels and b) O 3 -rich air from stratospheric/tropospheric exchange (STE).STE is defined as the irreversible transport across the isentropic tropopause, and can represent a significant injection of O 3 and other reactive species into the mid and lower troposphere (Holton et al., 1995).This transport can be triggered by convective systems (timescales as short as a few hours) or baroclinic motion (timescales of days).STE can also occur from tropopause folds in the vicinity of jet stream maxima, cut-off lows, mesoscale convective complexes, thunderstorms, and gravity wave breaking (Stohl et al., 2003).On the other hand, GW normally represents vertical mixing, which means it is associated with convective storms, lightning and boundary layer (BL) mixing.Consequently, it is ideal to distinguish between RW and GW elements to resolve their relative contributions.
The Laminae Identification (LID) method (Thompson et al., 2007a, b;Yorks et al., 2009;Thompson et al., 2010Thompson et al., , 2011Thompson et al., , 2013) ) allows an understanding of the dynamical origins of these laminae by exploiting the relationship between O 3 concentration and potential temperature profiles, and it is thus applied here.O 3 profiles were obtained by balloon-borne ozonesondes equipped with electrochemical concentration cell (ECC) sensors that have an accuracy of about 5% in the troposphere (Thompson et al., 2007a;Yorks et al., 2009;Thompson et al., 2013).The O 3 sonde preparations used the procedures described by Komhyr et al. (1995) while corresponding pressure, temperature, and relative humidity (PTU) profiles were obtained by coincident Vaisala RS-92 radiosondes.O 3 mixing ratio and potential temperature profiles were interpolated to account for the differences in height resolution from the balloon borne measurements.Small-scale fluctuations were removed by applying a boxcar smoothing method, followed by construction of normalized profiles of both O 3 and potential temperature that were scaled in order to account for the differences in their corresponding vertical gradients (Pierce and Grant, 1998).Then, a moving correlation (r-correlation) was then performed and applied every 5 km at 1 km intervals and results were filtered based on correlation values.Teitelbaum et al. (1996) and Pierce and Grant (1998), established that poor correlations (between -0.3 to 0.3) are associated with RWs, while strongly correlated cases (≥ +0.7) are associated with GWs (Thompson et al., 2007(Thompson et al., , 2013)).We follow this convention.
These analyses, along with the variation of the tropopause height, were used to help determine whether or not stratospheric intrusion occurred at a given time.O 3 in vertical layers that do not correspond to BL mixing, GW or RW are labeled as background (BG) or residual, meaning that they are associated with air that has aged enough such that its dynamical origins cannot be simply distinguished unless coupled by backtrajectory analysis, or potential vorticity and other tracers.
To further analyze STE cases, folding events were identified using tropopause heights (as shown in Fig. 1), fields of potential temperature (Fig. 2), and relative humidity (RH, Fig. 3) that were derived from sounding data.Since high potential vorticity (PV) values are associated with stratospheric air, sounding data were contrasted against PV derived from European Centre for Medium-range Weather Forecast (ECMWF) ReAnalysis-(ERA)-Interim (Fig. 4).We used this same dataset to evaluate the daily frequency of stratospheric intrusions over the HUBC site.We used ECMWF ERA-Interim PV at 250 mb since we estimated it to be the lower bound for the division of the stratosphere and troposphere.This value was determined by observing which pressure levels have the best match to the tropopause heights observed using the cold point approach on the PTU profiles.Various values of potential vorticity units or PVU (1 PVU = 10 -6 m 2 s -1 K kg -1 ) are used to indicate the separation between stratosphere and troposphere, but for the purpose of this study, we use 1.5 PVU (Holton, 1985;Stohl et al., 2003;Chen, 2005;Kim et al., 2008) as long as these fields match the condition of RH ≤ 20%, which is established as a condition for stratospheric intrusion.Events satisfying the criteria are shown on Fig. 5.The laminae distribution is embedded on Fig. 6 (GW laminae) and Fig. 7 (RW).

RESULTS AND DISCUSSIONS
Fig. 1 shows the O 3 mixing ratio (in ppbv) contour from soundings collected between 1 and 29 July 2011 at HUBC at an interval of 10 ppbv and ranging from 5 to 150 ppbv.Twenty-five daytime launches are shown, most of them coinciding with times when a NASA P-3 aircraft was spiraling over the HUBC site.Because of the codependency with the airplane overpasses, the data were gathered in a non-regular frequency (generally two sondes per flight day, but not necessarily during consecutive days).However, the figure includes all 25 launches consecutively to emphasize the changes in the tropospheric O 3 structure throughout the campaign.Comparisons of HUBC sondes and the P-3 below 2 km are in good agreement (Thompson et al., 2013).
Most O 3 soundings reached altitudes of 32 km or above.However, data are shown from the surface up to 20 km in order to highlight O 3 interactions between the troposphere and stratosphere.Here, we employ a traditional definition of the tropopause, identified as the lowest level at which the lapse rate decreases to 2 °C km -1 or less (World Meteorological Organization-WMO), which is represented by the magenta line in Fig. 1.Using this method, the average tropopause height over all launches was 14.75 km, corresponding to a mean potential temperature surface of 360 K.Because the thermal distinction between the troposphere and stratosphere sometimes corresponds to an atmospheric layer rather than just a single point, we follow an analogue procedure used by Thompson et al. (2003Thompson et al. ( , 2007aThompson et al. ( , b, c, 2011Thompson et al. ( , 2013)), in which a tropopause transitional layer (TL) is broadly defined.For the purposes of this study, the TL corresponds to the region between 12-16 km.
What we denote as the mid troposphere lies between 5-12 km and the lower troposphere is defined as anything below 5 km.The BL height was averaged at 1 km for all launches and the calculated percentages correspond to the total laminae found within the surface to the BL height.We also defined the free troposphere as a layer comprised between 1 and 12 km.At the surface level, four events occurred in which O 3 concentrations reached or exceeded 76 ppbv (Fig. 1), a value that serves as an approximate threshold for poor air quality conditions.O 3 concentrations between 5-10 km for the HUBC site normally average between 70 to 100 ppbv.These data, however, show several instances in which O 3 concentrations significantly exceeded this range, reaching values between 90 and 160 ppbv.Each case of O 3 enhancement was followed by reductions of tropospheric O 3 at varying tropospheric depths, which were primarily triggered by frontal passages through the region (05 July 10-14 July, 16 July, 20-22 July).Although our primary focus falls on meteorological causes associated with O 3 enhancement in the mid-upper troposphere, we also explore the meteorological nature of anomalously low events, and particularly a substantial reduction of column O 3 on 26 July.The discussion of these study cases is found below.

Mean Statistics
Fig BG O 3 is the greatest contributor to the column-integrated O 3 budget (Fig. 8) followed by RW, accounting for 41.7% and 31.7% of the tropospheric column O 3 during the entire observation period.On the other hand, GW-induced O 3 is the least influential mechanism, present only in 14 out of 25 of the profiles (Fig. 8) and 11.7% of the total budget.BL processes account for as much as 14.9% of the total budget.These numbers are in agreement with observations described by Yorks et al. (2009) who conducted studies at our same site during the summers of 2004-2007.They are also in agreement with Thompson et al. (2013), in which they found the corresponding budgets for the HUBC site as 18.9% for BL, 13.1% for GW, 27.3% for RW and 40.7% for background.Even though the methodology to calculate the budgets is similar for both studies, some differences arise due to how the BL height and tropopause height are defined.STE was the most influential mechanism influencing total column concentrations by virtue of both RW and GW, though we note the possibility of the influence transport from polluted areas may play, which we will discuss in more detail in the subsequent sections.

Meteorology of Period 1: 01-05 July
On 01-02 July the Mid-Atlantic region was dominated by the influence of anticyclonic flow at the surface and an upper-level ridge.This anticyclone followed an occluded surface low that was moving through eastern Canada, carrying with it a warm front that stretched southward across Maryland and bringing stagnant winds with relatively dry conditions.These would imply that corresponding high O 3 concentrations were mainly the result of local pollution.Fig. 1 depicts the evolution of the first elevated surface O 3 event of the month, which reached its peak on 02 July.The atmospheric sounding data reveal peak O 3 concentrations greater than 90 ppbv within the first 2 km.
Elevated O 3 concentrations were also observed above 6 km on 01 July (Fig. 1).Potential temperature contours (Fig. 2) show that tropopause did not strictly follow the 360 K isentrope on these first two days.Coincidentally, Fig. 4 shows how PV values increases between the 300-200 hPa levels and were accompanied by sinking motion (black contour lines).This scenario suggests that a STE occurred that was triggered by a tropospheric fold, specifically driven by RW breaking (Figs. 5, 7 and 8) and resulting in the transport of O 3 -rich air into the upper and mid troposphere.The corresponding profiles obtained from soundings on 01 July, show that the column was mostly dominated by RW and BG induced O 3 , which correlate with very low levels of RH.Aircraft tracers -mainly CO and the collective of oxidized forms of nitrogen or NO y (NO, NO 2 , nitric acid (HNO 3 ), and organic nitrates), support this interpretation (Thompson et al., 2013).STE events associated with tropopause folds typically occur west of an upper-level trough, and are enhanced by the presence of a jet (Shi et al., 2012).The 250 hPa analysis for this day (not shown) demonstrates that Maryland sat along the western edge of an upper-level trough and is also directly under a jet streak.
While surface O 3 levels were greater on 02 July, midtropospheric O 3 (6.5-10km) plummeted from maximum values of 100 ppbv to 60 ppbv, except for a sharp increase in O 3 at the 6 km level that is persistent both days and associated with low RH and RW-induced laminae.Evidence of subsidence within the column is shown on Fig. 4 (black contours) during both days, especially at upper levels, consistent with a folding event.The 1500 UTC (11:00 am, local Eastern Time) sounding on 02 July shows O 3 in the 9-13 km region as mostly background, with highly variable RH profiles sometimes exceeding 50% within 8 to nearly 15 km.
During the afternoon of 03 July, a warm front passed through the region, triggering a local microburst that was followed by a cold frontal passage on 04 July.The front produced showers and severe thunderstorms over Washington, DC, Maryland and portions of Virginia, with wind gusts on the order of 60 mph as well as large hail.The National Weather Service (NWS) high resolution Sband Doppler Next Generation Radar (NEXRAD, Fig. 9), depicted a period of heavy precipitation over HUBC that lasted over 40 minutes on 03 July (2207 through 2248 UTC, 6:07-6:48 pm, local Eastern Time), followed by periods of intermittent light showers over the following two days.These conditions resulted in a surface "clean-up" that can be observed from the 05 July O 3 sounding (Fig. 1).On the other hand, Fig. 4 shows a corresponding increase in PV above 650-hPa on 05 July that was accompanied by subsiding air, while Fig. 5 shows an increase in STE potential on 04-05 July that may suggest a STE event.The derived tropopause height (Fig. 2) also does not follow the 360 K surface on 05 July, indicating a possible tropospheric fold.
Though tropospheric O 3 enhancement for this period was mostly the result of STE, horizontal transport of O 3 precursors may have also contributed to mid-tropospheric O 3 enhancement (5-10 km) on 05 July.Enhanced atmospheric attenuation due to fires and smoke were reported over a substantial portion of the south-central Canadian plains, as well as western states such as New Mexico and Colorado.Forty-eight to seventy-two hour back trajectories (supplemental materials) for 04 July show air being advected from New Mexico and Arizona.Trajectories beginning 05 July (supplemental materials) show air masses originating closer to fires in Saskatchewan and Alberta.

Meteorology of Period 2: 06-16 July
Good surface air quality prevailed around the Maryland region, due to the presence of a stationary cold front 07-08 July.Late in the day on 08 July, a storm developed producing the greatest precipitation amounts recorded at the site during the campaign.Despite this, mid and upper-level winds were steering O 3 precursors (smoke) from Canadian fires occurring in Ontario towards the Great Lakes region.Of particular interest, were large fires south of Reindeer Lake (57.5°N, 102.0°W) on 07-08 July, which moved southeast toward Maryland arriving 10-11 July.We analyzed the 532 nm backscatter signal obtained by the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) satellite, corresponding to these cases and could observe two strong signals: one located close to Saskatchewan, Canada (between 65.91N, 100.59W-71.53N,103.73W) that spreads between 4-9 km, and a second one close to the Great Lakes Region (54N, 92W) between 8-10 km (not shown).At the same time, fires over the USA continued, most notably in Georgia and Northwest New Mexico, around Las Conchas and Pacheco.
On 10 July, a high-pressure system and a large-scale upper-level ridge covered most Continental United States (CONUS), bringing clear conditions and calm winds out of the southwest.The Bermuda High was pushed to the east as a front moved off the coast of the United States.These conditions, which are known to be conducive to high surface O 3 in Maryland, enabled the second high surface O 3 episode of the month.Atmospheric soundings collected on 10-11 July at HUBC during the mid-morning and afternoon show O 3 peak levels ranging between 88 and 105 ppbv between the surface and the BL (Fig. 1).In the mid-troposphere, they exceeded 150 ppbv near 10 km on 11 July (Fig. 1).A very dry layer developed between 3-6 km, coupled with a slight temperature inversion near 3 km that persisted through 11 July.Analysis of the individual laminae profiles corresponding to the on the 1400 UTC (10:00 am, local) 10 July sounding (complemented by Figs.7-8), show that the column was mostly dominated by background O 3 .However, a GW-induced layer located between 6-7 km, with values exceeding 100 ppbv, was observed that gave way by the 1800 UTC (2:00 pm, local) radio/ozonesonde launch to more RW-dominant conditions.Both soundings show a mixture of RW and background O 3 throughout the column with some GW filaments, but the highest O 3 concentrations (O 3 exceeded 100 ppbv at mid and upper levels) were coupled with RW activity.
STE was likely the principal mechanism for O 3 enhancement in the mid and upper levels on 10-11 July.Fig. 4 shows an increase in PV content between 09-10 July, which is accompanied by subsiding (09 July) and then rising (10 July) motion respectively.These high PV fields in the mid-troposphere (Fig. 4) may have been triggered by the previous thunderstorm activity.Fig. 5 shows, however, that some of the highest STE frequencies for the entire dataset correspond to 10-11 July.Fig. 4 also shows a reduction in PV during 11 July, when mid-upper air values of O 3 range from 100 ppbv between 6-10 km during the 21 UTC (5:00 pm, local) launch.The impact of the fires and transport on 7-8 July cannot be fully ruled out as precursor elements.However, a P-3 aircraft flown concurrently during the NASA DISCOVER-AQ did not capture relevant smoke concentrations at that altitude.Figs.10(a After the precipitation event on 11 July, smoke transport from fires in the Mississippi Valley region was observed at HUBC.Although ozonesonde data was not collected on 12 July, a "code orange" urban alert (unhealthy for sensitive groups) was issued in Maryland and Washington, DC that afternoon, with air quality levels improving during the evening.On 13 July, a cold front moved through Maryland bringing showers and thunderstorms again to the region.Precipitation from this system was more widespread than in the previous events (Fig. 9), and also was longer lived with heavier rains.
The cold fronts on 11 and 13 July served to clean out much of the eastern seaboard, allowing a return to moderate levels of surface air quality in regions previously affected by fires.However on 14 July, fire activity picked up again.Large wildfires in New Mexico and the surrounding area, as well as the numerous wildfires in the Ontario and Manitoba produced dense smoke and high aerosol optical depth (AOD) values observed by the MODIS Terra (not shown).Poor air quality was forecasted for Maryland on 14 July, yet surface levels remained between the healthy and moderate range.At HUBC, O 3 sonde operations resumed on this day and soundings at 12 and 18 UTC (8:00 and 2:00 pm local Eastern Time) showed peak surface O 3 concentrations of 49 ppbv and 58 ppbv, respectively.Similarly, O 3 levels between 5-10 km remained relatively low (max around 70 ppbv) compared with measured values observed on 11-12 July.There is evidence of some GW-induced laminae (Fig. 6), but Fig. 4 shows a small area of PV intruding close to 250 hPa, which may indicate small Rossby Wave Breaking (RWB) and a possible short-lived stratospheric intrusion that was not as deep or extensive compared with all other cases.STE frequency increased on the 14 July, lasting through 17 July (Fig. 5), with alternating regions of air ascent and subsidence (Fig. 4) throughout the column.
Surface O 3 levels remained low through 20 July.However beginning with 16 July, sounding observations showed a substantial O 3 increase within the 6-12 km layer.PV reanalysis (Fig. 4) shows some evidence of stratospheric injected air (below the 250 hPA level), despite of the lack of substantial variation in the 360 K-isentrope heights.The upper levels were dominated primarily by ascending air (white contours on Fig. 4), which may have hindered Fig. 10.NOAA HYSPLIT Back trajectories, corresponding to select launches between 10-16 July.These show runs at 5, 7 and 9 km.vertical transport to the surface.Yet, areas in which the O 3 mixing ratio exceeds 110 ppbv (around 10.5 km) are seen, in contrast with the less than 70 ppbv measurements made at the same altitude on 14 July.

Meteorology of Period 3: 17-29 July
Beginning 18 July, poor air quality conditions at the surface returned to the Mid-Atlantic region.Smoke transport from active fires observed in northern Ontario and Quebec was also taking place.An anticyclone centered over the region brought southerly winds off the Atlantic while increasing column moisture levels.A very strong upperlevel ridge was shifting eastward strengthening southwesterly flow, adding another ingredient to the poor air quality recipe.
From 20-24 July, the Mid-Atlantic experienced the national record-breaking heat wave in which 1,966 daily high maximum temperature records were broken or tied (Payne et al., 2014).Smoke from the Canadian fires had also drifted towards the Northeast US, while fires in central and eastern Canada, mainly Manitoba and Ontario, continued advecting very dense plumes towards the Northeast and Mid-Atlantic states.
Surface O 3 concentrations increased dramatically around the region, with some stations reporting an unhealthy air quality rating (code red).At the HUBC site, O 3 levels did not increase enough to reach a code red.Nevertheless, concentrations above 80 ppbv were observed during four consecutive days within the BL and soundings measured up to 110 ppbv during the afternoon on 20 July from 0-2 km.Yet, there was very strong mid to upper tropospheric enhancement in O 3 concentrations.Fig. 4 does show evidence of moderate PV air along the column during 20 July, though not as markedly as seen in the cases previously addressed.PV levels remained low along the column and did not pick up until after 23 July.Fig. 5 also shows lower than average STE frequencies for profiles obtained between 21-26 June.
At 1400 UTC (2:00 pm, local Eastern Time) on 20 July, most of the O 3 within 5-12 km is of background origin (Fig. 8).Four additional soundings were obtained between the 21-22 June, depicting an alternating trend between background O 3 and RW-induced O 3 -controlled profiles.Figs. 11(a),11(b) and 11(c) show HYSPLIT back trajectories during this time period.The analysis shows air masses that had been advected through Saskatchewan (Fig. 11(a)) and Ohio Valley (Fig. 11(b)), prior to arrival at HUBC on 20-21 June, respectively.Transport likely played a more significant role than STE and BL processes during this event, though not necessarily from O 3 precursors due to fire activity (which would have been the case for air masses advected from Canada).Instead, areas characterized by continuous production of anthropogenic pollutants (Ohio Valley region) were a likelier stimulant.
Air quality improved substantially after 23 July.A stationary front was cast over Maryland for most of that day producing showers and thunderstorms.Heavy rain produced flash flooding and was accompanied by damaging winds near the DC area.Low levels of O 3 concentrations (not exceeding 70 ppbv) prevailed at heights below 10 km in the wake of the frontal passage and precipitation event.
As can be noted in Fig. 1 and Fig. 8, tropospheric column O 3 was reduced extensively compared with all of the other cases.This same behavior can be observed in the O 3 column data obtained at the Edgewood ground station (Thompson et al., 2013), which is north of HUBC.Calculations of integrated column O 3 (in Dobson units or DU) from the surface up to 10 km show 23.0 DU on 1941 UTC (3:41 pm) 21 July, followed by a dramatic decrease of nearly 70% on 26 July (12.2DU), building up further by 27 July.
Previous research (Hu et al., 2010) has examined the effects of deep, tropical convection on concentrations of O 3 throughout the troposphere.They observed that deep convection reduced O 3 concentrations in the upper troposphere (12-15 km) by 50% and enhanced it by 39% in the lower 2 km.At HUBC, representative of a midlatitude atmosphere, a number of meteorological processes contributed along with the associated precipitation and dynamics from the storm, including a Canadian airflow "blockage" that was formed due to the interaction of the frontal system and a high pressure that developed over the north central USA.Fig. 12 shows NOAA's Weather Prediction Center (WPC) surface pressure and precipitation maps for July 24-July 27.Following the stationary front on 25 July, a cold front proceeding from the Northwest also approached the region.The atmosphere was already very moist due to the precipitation associated with the stationary front, with relative humidity values above 70% up to 700 mb (not shown), yet, on the next day (26 July), another cold front arrived from the northwest accompanied by an outflow boundary along the Maryland/Pennsylvania border as Maryland sat at the edge of an upper-level trough that was moving over New England.The Bermuda High had shifted slightly eastward and was obstructed near the east coast by a front stretching off of a low-pressure system moving out of Canada, bringing north/northwest winds into the Mid-Atlantic.During the frontal passages on 25 July, an area of high pressure lingered in the north central USA, covering portions of both USA and Canada.This area moved eastward until 27 July, thus heavily influencing air flows.The position of the high ---northwest of the region ---brought clockwise circulation and helped to steer air masses proceeding from the northern regions.Thus, air masses during this period were advected more from the north-central United States, where air quality levels were ranging between good and moderate.We can observe this change in air masses comparing Figs. 11(a), 11(b) and 11(c) (all polluted cases) with the particular case of Fig. 11(d) (air mass proceeding from cleaner areas, or where fire activity was at minimum during this period).This air mass shift is evident in both back trajectory and in-situ data at HUBC (Payne et al., 2014).Sounding data from the NWS-Sterling was used to look at the horizontal wind profiles corresponding to the precipitation events during the DISCOVER-AQ campaign, when no launches were performed at the HUBC site.It seems that from all storms, the system on July 25 exhibited the weakest horizontal winds up to the 12 km level.Consequently, O 3 concentration around the mid-troposphere remained low compared with cases in which polluted air mass advection was present.
By 27 July, the cold front had passed through Maryland and was followed by an area of high surface pressure accompanied by an upper-level ridge.During this same day, light to moderate smoke was observed from southern Maryland through northern South Carolina as a consequence of the North Point wildfire in Wyoming and fires in Canada.Fig. 5 shows that STE frequency increases for 27-28 July.By 28 July, surface air quality had deteriorated in the east coast, where particulate matter (PM) 2.5 and O 3 reached moderate and unhealthy levels.O 3 increased during the afternoon in the Mid-Atlantic, while fires and smoke from Quebec continued to be reported.The effect of these last fires, along with a building high-pressure system and increasing temperatures is observed in Fig. 1, as O 3 increased again both in the lower and mid-upper tropospheric levels after the cleanup event.These were the last soundings conducted during the campaign.Another heat wave hit the east by the end of this period, with temperatures exceeding 100°F (37.8°C) in most places in the Mid-Atlantic.O 3 levels in the region once again ranged from moderate to unhealthy.
At this point, it is noteworthy to mention that Figs. 6 and 8 and show fragments of GW between 8 and 10 km and one above 12 km during the last portion of the campaign.GW lamina is normally associated with regional convection and lightning.Although the impacts of lightning are not fully explored in this publication, it is noteworthy to mention that Thompson et al. (2007a) pointed out that a complex mixture of stratospheric, lightning, and long-range transport influences the O 3 variability in the DC-area.This in itself is a future area of study.

SUMMARY
Here, we described the synoptic and mesoscale meteorological drivers influencing O 3 profile variability in the Washington, DC vicinity during summer 2011.We applied the LID method (Thompson et al., 2007a, b;Yorks et al., 2009, Thompson et al., 2010, 2011, 2013), which allows to resolve the relative contribution from planetary waves in the atmospheric column ozone.Specifically, we highlight how RW and GW are responsible to originate and maintain mid and upper tropospheric O 3 maxima through downwind advection of O 3 pollution and precursors at the various atmospheric levels and O 3 -rich air from STE.
The mean statistics derived from the laminar analysis show that background O 3 tends to dominate the laminar budget, followed by RW, accounting for 41.7% and 31.7% of the tropospheric column O 3 , respectively.These changes resulted primarily from stratospheric intrusion.However, there was a strong influence from advection of pollutants and STE produced at other locations.The BL dynamics and photochemistry accounted for 14.9%, while GW contributed with 11.7% of the total budget.These results are similar to the ones obtained by Thompson et al. (2013).
There were four periods of mid-upper tropospheric enhancement over HUBC during July 2011, which are primarily dominated by STE.Most of the variability was triggered by convective systems, baroclinic motion and tropopause folds in the vicinity of jet stream maxima.The first was an intrusion case on 01-02 July, caused by a tropopause fold over Maryland in the presence of a jet streak and an upper level trough.We showed that for the subsequent cases (05 July 10-14 July, 16 July, 20-22 July) there is the possibility that horizontal transport may have contributed to O 3 enhancement.Even though coincident aircraft spirals during DISCOVER-AQ show low concentrations of CO and smoke at those levels, we do not discard the possibility of stratospheric air or pollution being transported from upwind, especially in cases like 20-22 July, when the ECMWF reanalysis does not show strong evidence of high PV air at the upper levels.The dynamical processes associated with frontal passages and changes in air mass characteristics also contributed to reductions in O 3 levels not only at the surface levels but also throughout the column.Example of this was the dramatic decrease of nearly 70% in free tropospheric ozone on 26 July.
We have demonstrated how weather can serve as the primary driver for transport of O 3 and its precursors.Nevertheless, monitoring and understanding of weather conditions should not be limited to just surface as downward transport of air from the entire atmospheric column can contribute before, during and after poor air quality events.Therefore, vertical characterization of O 3 (and aerosol) profiles from surface to the lower stratosphere, combined with knowledge of the prevalent atmospheric conditions is necessary and should not be decoupled from chemistry.

Fig. 1 .
Fig. 1. O 3 mixing ratio contour (in ppbv) from 25 daytime O 3 sonde measurements collected between 1 and 29 July 2011 at HUBC Maryland during the DISCOVER-AQ campaign.The x-axis indicates the date and launch time in UTC (e.g., MMDDHHMM), which also matches the aircraft over-flight of HUBC.The data (e.g., time axis) are rendered sequential in order of time of launch but not linear in time to emphasize the changes in the tropospheric O 3 structure throughout the campaign.Magenta colored line represents the tropopause height, which was obtained applying the World Meteorological Organization (WMO) definition.Pressure levels are overlaid.

Fig. 2 .
Fig. 2. Potential temperature as derived from O 3 sonde data (in Kelvin) during July 2011 at HUBC Maryland.As in Fig. 1, the magenta colored dots represent the thermal tropopause and pressure levels (in hPa) are superimposed.

Fig. 4 .
Fig. 4. Reanalysis using PV fields from ECMWF ERA-Interim with superimposed vertical velocity, also derived from ERA-Interim.Black contours denote areas of sinking motion (negative vertical velocity), while white contours denote rising motion.PV is in PV units (PVU).

Fig. 5 .Fig. 6 .
Fig. 5. Daily frequencies of stratospheric intrusions over the DC area (and HUBC) using ECMWF ERA-Interim reanalysis PV and RH data over the month of July 2011.A stratospheric air intrusion is defined by PV greater than 1.5 PVU and RH less than 20%.An event satisfying the criteria is counted and divided by the total number of days in the campaign.

Fig. 7 .
Fig. 7. Same as Fig. 6, but for the RW laminae distribution.Note the shaded regions correspond to the absolute value for correlations between 0 to +0.3 (instead of -0.3 to 0.3).
. 8 complements Fig. 1, as it shows the integrated tropospheric O 3 (ITO 3 ) between 1-20 km, but in Dobson Units (DU) for each of the 25 obtained profiles.The highest ITO 3 corresponds to the 1501 UTC (11:01 am, local Eastern Time) 01 July sounding, with values exceeding 60 DU, and the lowest on 1127 UTC (7:27 am, local Eastern Time) on 26 July when ITO 3 values barely reached 20 DU.Fig. 8 also displays information about the contribution of each laminae based on the definitions of GW, RW and BG O 3 presented on the previous section.

Fig. 8 .
Fig. 8. Integrated tropospheric O 3 (in Dobson units) between 1-20 km as obtained from daytime O 3 sonde measurements at HUBC Maryland during the DISCOVER-AQ campaign.Each bar also shows the contribution of each of the mechanisms responsible for tropospheric O 3 enhancement.It is noticeable how the BG and RW laminae are by far dominate most of the budget, while the GW contribution is minimal and observed in 14 of the 25 the profiles.

Fig. 9 .
Fig. 9. (Above): shows NOAA Next Generation Radar (NEXRAD) images for the 2225 UTC (6:25 pm, local) 03 July , 1923 UTC (3:23 pm, local) 13 July, and 1925 UTC (3:25 pm, local) 25 July We can observe the difference on the spatial extent and relative strength of all three storms as they were impacting the HUBC, MD area.
)-10(b) show 72 hour National Oceanic and Atmospheric Administration (NOAA) Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model air mass back trajectories between 5-10 km, coinciding with selected launches on 11-10 July.These originate from both Canada and the midwestern U.S. respectively, which could imply injection of O 3 or O 3 precursors during this time frame.

Fig. 11 .
Fig. 11.NOAA HYSPLIT Back trajectories, for the record-breaking heat wave (20-22 July corresponding to panels a, b and c) and column clean-up event (26 July).

Fig. 12 .
Fig. 12. NOAA Weather Prediction Center (WPC) surface pressure and precipitation maps corresponding to the O 3 depletion scenario between 24-27 July.