O 3 , CH 4 , CO 2 , CO , NO 2 and NMHC aircraft measurements in the Uinta Basin oil and gas region under low and high ozone conditions in winter 2012 and 2013

1CIRES, University of Colorado, Boulder, Colorado, United States 2NOAA/ESRL, Global Monitoring Division, Boulder, Colorado, United States 3Institute of Arctic and Alpine Research (INSTAAR), University of Colorado, Boulder, Colorado, United States 4University of California, Davis, California, United States 5Scientific Aviation, Boulder, Colorado, United States 6National Institute of Standards and Technology, Gaithersburg, Maryland, United States


Introduction
Vigorous ozone formation in the summer under conditions with high levels of solar insolation and abundant O 3 precursor chemicals has long been recognized (Chameides et al., 1992).High levels of O 3 production during winter has only recently been documented in locations such as the Upper Green River Basin, Wyoming and Uinta Basin, Utah, in association with effluents from oil and gas production operations (Schnell et al., 2009;Oltmans et al., 2014;Rappenglück et al., 2014).In these basins, the strong role of meteorology, topography, and high levels of O 3 precursors from oil and gas extraction activities have been identified as key drivers of the wintertime high O 3 events (ibid.).The presence of snow-cover acts as both an enhancer of available ultraviolet (UV ) radiation by greatly increasing the surface albedo and as a promoter of strong temperature inversions that trap O 3 precursor emissions in a shallow cold pool surface-based layer (Bader and McKee, 1985;Yu and Pielke, 1986;Schnell et al., 2009;Lareau et al., 2012;Oltmans et al., 2014).
In 2012 and 2013, a multi-agency Uinta Basin Winter Ozone Study (UBWOS) took place in the Uinta Basin in NE Utah (Figure 1) (Lyman and Shorthill, 2013;Stoeckenius and McNally, 2014).Measurements included hundreds of ozone profiles through the boundary layer from tethered ozonesondes (Schnell et al., 2016), non-methane hydrocarbon (NMHC) and nitrogen oxide mole fraction vertical profiles (Helmig et al., 2014), surface chemical constituent observations, and meteorological measurements (Lyman and Shorthill, 2013;Edwards et al., 2013;Stoeckenius and McNally, 2014;Edwards et al., 2014).In the Uinta Basin, intense natural gas production is located mainly on the east side of the Basin and oil production primarily on the west side.Natural gas production levels were similar in 2012 and 2013 with ∼18% greater oil production in 2013 (Utah Oil and Gas Program).Emissions from oil and gas production and processing activities as well as combustion emissions associated with vehicle traffic, drilling rigs, compressor stations and gas processing may be significant sources of both volatile organic compounds (VOCs) (Gilman et al., 2013;Warneke et al., 2014) and nitrogen oxides (NO x = NO + NO 2 ), primary O 3 precursors.
This paper presents unique sets of aircraft collected data in the winters of 2012 and 2013 in the Uinta Basin under contrasting meteorological conditions.In the winter of 2012 there were no periods of persistent snow-covered ground in the Uinta Basin and no large enhancements in O 3 while in 2013 the ground was snow-covered during January and February (Oltmans et al., 2014).
Measurements of trace gases both from the in-situ continuous analyzers and the flask packages are reported on the NOAA/WMO scales as dry air mole fractions (moles per mole of dry air), using methods outlined in Karion et al. (2013a) and online (NOAA GMD Aircraft).O 3 and NO 2 mixing ratios (measured with respect to ambient air) are equivalent to mole fractions when referred to dry air and are denoted herein as mole fractions.In 2012, ozone measurements were made with a 2B Technologies UV photometric analyzer with the solid phase scrubber.In 2013, measurements of O 3 were made with a 2B Technology analyzer that used nitric oxide (NO) in place of a solid phase scrubber to avoid potential inference from aromatics and mercury (Ollison et al., 2014).The O 3 instruments were calibrated before and after each campaign against a NOAA GMD standard that is regularly compared with a U.S. NIST standard.The Los Gatos Research NO 2 instrument was factory calibrated prior to the field deployment.
In 2012 ten research flights were conducted over the period 3-18 February with daytime surface temperatures at or above freezing, snow free ground, and a well-mixed boundary layer (Lyman and Shorthill, 2013).In 2013 seven research flights were conducted over an 8-day period from 31 January to 7 February.In 2013 snow-covered ground, sub-freezing surface temperatures (dipping to below -10° C) low wind speeds, and limited cloud cover supported stable shallow cold pools (∼200 m at their deepest) capped by highly stratified temperature inversions.

Results
Since there was no appreciable photochemical O 3 production in winter 2012, flights focused on determining emissions from oil and gas activities, primarily over the main gas field (Karion et al., 2013b).In 2013, flights were conducted during an episode of intense O 3 production in which the spatial and temporal features of O 3 , CH 4 , and O 3 precursor NMHCs mole fractions were characterized.During this 8-day period in 2013, surface ozone hourly mole fractions built up from approximately background levels of 45 ppb to over 140 ppb (Figure 2).The first flight of the study (conducted on 31 January) occurred three days after O 3 rich air from a prior event had been flushed out of the Basin by a frontal system.The final flight on 7 February took place during the early stages of an incursion of a Pacific air mass that again flushed O 3 and oil and natural gas field emissions out of the Basin in the subsequent 2 days.

Vertical profiles of O 3 , CO 2 , CH 4 and CO in relation to temperature profiles
The contrasting vertical temperature structure within the Basin in 2012 and 2013 played a significant role in the way gaseous constituents were distributed both vertically and across the Basin.Typical profiles obtained at a number of locations on 9 February 2012 (Figure 3a) show consistent temperature structure across the Uinta Basin (Figure 3b) with temperature decreasing with altitude.Profiles are plotted with respect to altitude above sea level (asl) to provide consistent altitude registration for the different profiles as terrain varies across the Basin.Also, in 2012 (Figure 3c) O 3 mole fractions fell within a relatively narrow range (∼30-40 ppb) in the sampled altitude range.The highest CH 4 and CO 2 mole fractions (Figure 3d and e) were constrained within the layer that was capped at ∼2000 m asl.
In contrast on 6 February 2013 (Figure 4), the flight day with the highest O 3 mole fractions observed by the aircraft, O 3 in all four profiles was generally >100 ppb below 1650 m asl.Above ∼1800 m asl there was a gradual decrease in all of the constituents with the mole fractions approaching the lower free tropospheric values at ∼2200 m asl where the temperature lapse rate became positive (decreasing temperature with altitude) (Figure 4b).This suggests there was a transition zone resulting from the slow mixing of the higher mole fractions of CH 4 and O 3 out of the top of the boundary layer into the free troposphere.A cold pool layer with a top at 1650-1700 m asl containing the highest constituent mole fractions with a layer of decreasing constituent mole fractions above was characteristic of all of the flights in winter 2013 (Figure S2).

Measurements within the Bonanza power plant plume
On 6 February the flight track passed through the Bonanza power plant (located in the SE quadrant) plume and encountered enhanced CO 2 as observed in profile 4 in Figure 4e at an altitude of 1775-1850 m asl.CH 4 was not enhanced in the plume (Figure 4d) while O 3 was somewhat depleted (profile 4 in Figure 4c).The plume was located above the cold pool layer that topped out at ∼170 m agl.Similar profile structure was seen on other flight days when the aircraft encountered the plume (Figure S2).Notably, the depression of O 3 mole fractions below values of O 3 measured outside the plume was most likely an indication of O 3 titration by NO emitted from the power plant as shown to be the case in other power plant plumes (Luria et al., 1999(Luria et al., , 2003)).The buoyant plume of emissions from the power plant, whose stack height is at 192 m agl above ground level or 1815 m asl, is lofted above the stack height (shown in a photograph on February 2 -Figure S3) but then descends somewhat as the plume cools and stabilizes above the top of the temperature inversion capped boundary layer.The combustion product CO 2 observed in the plume (and CO and NO 2 not shown) on each of these encounters declined abruptly immediately below the plume at the top of the cold pool layer, evidence that the plume was not adding substantially to O 3 precursor mole fractions in the cold pool layer and not likely to O 3 production below the inversion.

Spatial distribution of O 3 , CH 4 , NMHCs and other constituents
In 2012 O 3 mole fractions across the Basin varied over a relatively narrow range (∼30-45 ppb) for the 13 flight days.CH 4 , on the other hand showed a broader variation across the Basin (Figure S4).The location of the highest mole fractions varied from day to day (Figure S4), consistent with the variable wind conditions during the aircraft flights (Karion et al., 2013b).Peak CH 4 fell in the range 2500-3500 ppb.
In 2013 the buildup of O 3 mole fractions within the cold pool layer ≤1650 m asl across the entire expanse of the Basin during the episode of 31 January -6 February is presented in Figure 5 and for CH 4 in Figure 6.Measurements above the top of the boundary layer inversion are not included in these plots.On 1 & 2 February (Figure 5b and c) the highest O 3 amounts were not coincident with the location of the main gas field (located in the south-east quadrant) while on 4, 5 and 6 February (Figure 5d     Air samples collected in flasks over the Uinta Basin in 2013 (blue, red, green or yellow based on quadrant) and 2012 (light grey circles).The quadrants correspond to those shown in Figure 3a.Regression is two-sided accounting for errors in both variables.See Table 1 for slopes.doi: 10.12952/journal.elementa.000132.f007 NMHCs and CH 4 (Figure 7 and Table 1) are tightly correlated, yielding linear regression R 2 results of 0.69 -.0.91.Although the NMHCs and CH 4 are similarly related in the two years, mole fractions are ∼5 times greater in 2013 compared to 2012.Included in Table 1 are the mean NMHC/CH4 ratios that were observed at the Horsepool ground site in 2012 and 2013 (Helmig et al., 2014).There is a remarkably good agreement between these two data sets, with slope values generally agreeing to within ± 20%.The relatively close similarity in these results indicates that the surface observations from the Horsepool fixed ground site can be deemed relatively representative of the Basin.Another inference is that the NMHCs were distributed across the Basin in a similar manner to CH 4 , and that they are co-emitted with CH 4 (Pétron et al., 2012;Peischl et al., 2013) when natural gas is vented or leaks into the atmosphere.The source of these emissions in the SE quadrant is consistent with the presence of a dense array of gas wells, gathering pipelines, compressor stations, and two large processing plants (Chipeta Plant Complex and Stagecoach/QEP) located in this quadrant.It is notable that for most of NMHC species, in particular the lighter NMHC, i.e. propane -pentane, show ∼10-20% higher NMHC/CH 4 slope values for air originating in the SW quadrant compared to the SE.The in general lower NMHC/CH 4 values seen in the SE quadrant point towards a higher CH 4 fraction (dryer gas) in the fugitive emissions from that sector.
The average concentrations of O 3 and CH 4 within the cold pool beneath the temperature inversion for each day of aircraft measurements for the 2013 data presented in Figures 5 and 6 are plotted in Figure 8.On 31 January at the start of the event and just two days after the Basin had been swept out by the passage of a cold front, CH 4 mole fractions over the Basin were close to 3500 ppb This represents a significant enhancement above the mole fractions of ∼1900 ppb measured above the boundary layer (Figure 3), which is similar to the regional background.By 2 February mean CH 4 mole fractions of 6800 ppb were spread throughout the Basin (Figure 6).Between 2 February and 5 February, average CH 4 mole fractions leveled off peaking at just over 7000 ppb, then decreasing to ∼6000 ppb on 7 February.The leveling off and the subsequent decrease of CH 4 with the beginning of the Basin being flushed out is consistent with the time-behavior observed for CH 4 measurements at the surface made at Horse Pool (Edwards et al., 2014-extended data;Helmig et al., 2014).
Over the same time period, mean O 3 mole fractions increased (Figure 8) across the Basin.This was also seen in surface measurements (Helmig et al., 2014) and tether ozonesonde profiles (Schnell et al., 2016).On 31 January average O 3 mole fractions at 1650 m and below were ∼60 ppb increasing steadily to ∼110 ppb by 6 February before decreasing on 7 February with the approaching clean out event.

Discussion
The multi-species aircraft profile measurements presented in this paper develop a picture of the relationship between various oil and gas field emissions within the Basin and their relationship to wintertime photochemical O 3 production.CH 4 and CO 2 were well correlated in the flight data both in 2012 (Figure S5) and 2013 (Figure 9) with the highest levels of both constituents measured in the vicinity of the gas field in the SE quadrant of the Basin.These correlations reflect a collocated source, with CH 4 directly emitted from extraction and gas processing while CO 2 is a product of combustion related activities also taking place in the vicinity of the gas field.High CO 2 values not corresponding to high CH 4 represent CO 2 sources without accompanying CH 4 such as power plant emissions or mobile sources (cars and trucks).This was seen most prominently in the NE quadrant in 2013 (Figure 9) where Vernal is located (the largest population center in the Basin with extensive vehicle traffic) and also in 2012 (Figure S5).Other constituents associated with combustion such as NO 2 and CO were not consistently related to CH 4 or each other in 2013 (Figure S6), likely due to additional chemical processing (NO 2 ) or other sources not related to CH 4 emissions (CO).In 2013 there is a strong tendency for higher O 3 mole fractions with higher CH 4 mole fractions throughout the Basin (Figure 10) reflecting the fact that O 3 precursor NMHCs are co-emitted with CH 4 .However, there is a group of observations where O 3 was relatively constant with increasing CH 4 .These observations are strongly concentrated in the SE quadrant.The highest mole fractions of NO 2 are also found in the SE quadrant (Figure S6).When the observations in Figure 10 are screened for NO 2 mole fractions greater than 10 ppb almost all of the observations where O 3 is relatively constant with increasing CH 4 drop out.This indicates the reduced efficiency of O 3 production in the presence of high NO x mole fractions (Kleinman, 2005;Edwards et al., 2014;Ahmadov et al., 2015) and is consistent with the highest mole fractions of O 3 often being observed not in the immediate vicinity of the main gas field.O 3 mole fractions are moderately consistent with the variability in CO 2 (Figure 9) but do not show a consistent relationship with NO 2 or CO (Figure S7).
The contrasting meteorological conditions during the winters of 2012 and 2013 highlight the way these conditions led to only modest O 3 formation in winter 2012 and to the extreme high O 3 episodes in winter 2013.The contrasting temperature and constituent profiles between 2012 (Figure 3) and 2013 (Figure 4) show that the air confining geographic characteristics of the Basin also require conditions that prevent efficient vertical mixing and ventilation of air out of the Basin as occurred in 2012.It is worth noting that in 2012 when there was little snow on the ground almost all the incoming radiant heating was transferred to an outgoing heat flux with very little loss to latent heating (R. Zamora, private communication; Stoeckenius and McNally, 2014).Although the ground was snow-covered in 2013 (Oltmans et al., 2014) and daytime temperatures were always relatively cold, a cold pool up to ∼200 m thick with nearly constant temperature and constituent mole fractions indicated a mixed layer resulting from daytime heating of the surface and convective mixing.It was within this layer that the highest mole fractions of constituents emitted from the surface as well as elevated O 3 were measured.
In air above this mixed layer, mole fractions of all constituents declined to values representative of background levels.Assuming that CH 4 emissions remain relatively constant in the Basin and due to the long chemical lifetime of CH 4 , the stabilization of CH 4 mole fractions after the first few days of the episode (Figure 8) is consistent with some advection of air out of the Basin.However, O 3 mole fractions continued to increase (Figure 8) across the Basin through the flight period (also seen in other surface and profile measurements; Helmig et al., 2014;Schnell et al., 2016).During the course of the O 3 event, nighttime O 3 minimum values continued to rise (Figure 2 and Schnell et al., 2016), indicating that nighttime chemical loss, surface deposition, and transport out of the boundary layer did not fully remove O 3 from the previous day's production.Although CH 4 (and likely also accompanying NMHC) levels stabilized after a few days, they remained sufficiently high to fuel O 3 production and increasing O 3 mole fractions throughout the event captured by the aircraft flights.
The spatial pattern of emitted constituents in 2013, such as CH 4 (Figure 6) and CO 2 (Figure S8), is indicative of strong emissions in the region of the natural gas field in the SE quadrant of the Basin that spread throughout the Basin.Wind measurements throughout the Basin are supportive of this (Schnell et al., 2016)   and were not concentrated around the primary source of the precursors (Figure 5) where very high NO x levels may influence the O 3 production efficiency.On several days the highest O 3 mole fractions were encountered away from the gas field, consistent with the time scales for O 3 production that are not instantaneous.This O 3 spatial pattern in the Basin was a result of strong photochemical production over the entire Basin, not primarily a result of transport from a more localized source near where the precursors were emitted.
The tight relationship between CH 4 and NMHCs in both ground and aircraft data (Table 1) and the magnitude of the reported methane (Karion et al., 2013b) and NMHC (Helmig et al., 2014) emissions suggests that the eventual dispersal of the basin wide natural gas related emissions contribute to the atmospheric methane and NMHC increases that have recently been reported on a broader geographical scale (Helmig et al., 2016;Turner et al., 2016).

Summary
Extensive aircraft sampling of O 3 , CH 4 , CO, and NO 2 , along with numerous gas species from flask measurements, during the winters of 2012 and 2013, demonstrated the strongly contrasting conditions in both the spatial and vertical dimensions of wintertime O 3 production in these two years.In 2012 O 3 mole fractions were confined to a narrow range of 30-45 ppb throughout the Basin with only small variations with altitude to the top of the measured profiles that extended to above the rim of the Basin.In 2012 CH 4 across the Basin varied from flight-to-flight, although the highest levels were found near the gas field and in a pattern where CH 4 was higher in the direction of the prevailing winds (Karion et al., 2013b).O 3 , on the other hand, was much more uniform over the Basin reflecting the lack of connection with precursor emissions under conditions with minimal O 3 production.
The strongest O 3 production in 2013 occurred in a shallow surface layer with a strong temperature inversion topped near 1650±50 m asl.Above the cold pool was a gradient layer extending to ∼2200 m asl with elevated constituent mole fractions that gradually declined to near lower free tropospheric values.
The structure of constituent profiles measured through the plume of a power plant located in the eastern portion of the Basin did not identify a pathway for mixing chemical precursors found in the power plant plume, which was emitted far above the O 3 maximum, into the boundary layer.Also, since the enhancements in O 3 and emitted constituents were always seen in the layer below the top of the temperature inversion, there is strong evidence that the source of these enhancements was from within the Basin itself.
The contrast seen in the aircraft measurements between 2012 and 2013 demonstrates the dominant role snow cover plays in promoting winter O 3 production in the Uinta Basin.Modeling of the conditions in the 2011-2012 and 2012-2013 winters (Edwards et al., 2013;Neemann et al., 2015;Edwards et al., 2014;Ahmadov et al., 2015) supports the vital role snow cover has on wintertime ozone production as this promotes and sustains strong temperature inversions that leads to highly elevated mole fractions of ozone precursors (Schnell et al., 2009).The snow also enhances UV flux because of its high surface albedo and snow reduces O 3 deposition to the surface.Based on precursor emission levels and meteorological conditions in 2013 and similar conditions observed in 2010 and 2011 in the Uinta Basin (Oltmans et al., 2014), it is likely that high O 3 episodes can be expected in the future under snow-covered conditions if emissions from oil and gas operations remain at 2013 levels in the Uinta Basin in future winters.

Figure 1
Figure 1 Topography of the Uinta Basin.The locations of surface measurement sites and several towns in the basin are also shown.The border between the aqua and purple colors is at 1650 m asl.doi: 10.12952/journal.elementa.000132.f001 Figure 2 Hourly surface ozone mole fractions in January-February 2013.Observations from the Blue Feather (blue) and Ouray (red) sites in the Uinta Basin The period of the aircraft flights from 31 Janurary to 8 February is indicated.doi: 10.12952/journal.elementa.000132.f002

Figure 8 Concurrent
Figure 8 Concurrent CH 4 and O 3 for all observations ≤1650 m asl for seven flight days in winter 2013.Mean (solid line) and median (dashed line) for each aircraft flight day of CH 4 (square symbol, red line) and O 3 (circle symbol, blue line).Observations were obtained from ∼1100-1700 Mountain Standard Time.doi: 10.12952/journal.elementa.000132.f008 where a diurnal wind pattern in which daytime heating at the edge of the Basin moves air from the central portion of the Basin toward the periphery.Based on the CH 4 and aircraft flask-sampled NMHCs, this transport pattern dispersed precursors throughout the Basin.The availability of O 3 precursors throughout the Basin was reflected in the high O 3 mole fractions that encompass the entire Basin Elementa: Science of the Anthropocene • 4: 000132 • doi: 10.12952/journal.elementa.000132 Figure 9 CO 2 mole fraction as a function of CH 4 mole fraction in winter 2013.Aircraft measurements ≤1650 m asl color coded based on the quadrant where the measurement was made.Values are 5 second averages.The quadrants are those shown in Figure 3a.Regression is two-sided accounting for errors in both variables.doi: 10.12952/journal.elementa.000132.f009

Table 1 . Correlation and slope of various non-methane hydrocarbons (NHMCs) and CO with methane determined from flask samples in 2013 a
a Regression is two-sided accounting for errors in both variables.doi: 10.12952/journal.elementa.000132.t001