The role of open lead interactions in atmospheric ozone variability between Arctic coastal and inland sites

The return of sunlight to the Arctic in the polar spring is associated with episodic atmospheric boundary layer ozone depletion to near zero levels (e.g. Oltmans, 1981; Barrie et al., 1988; Oltmans et al., 2012). These ozone depletion events (ODE) are linked to active bromine chemistry occurring in the Arctic boundary layer (Barrie et al., 1988; Simpson et al., 2007b) due to the release of molecular bromine, from halide enriched surfaces (Fan and Jacob, 1992; Abbatt et al., 2012), particularly surface snowpacks (Pratt et al., 2013) and airborne aerosol particles (Frieß et al., 2011). The activation of halogens and occurrence of ODEs alters the oxidation of pollutants in the Arctic boundary layer (Simpson et al., 2007b). In particular, halogen activation is associated with enhanced deposition of mercury occurring concurrently with ODEs (Schroeder et al., 1998; Steffen et al., 2008). While snow-covered sea ice regions are commonly associated with both the activation of halogens (e.g. Wagner and Platt, 1998; Simpson et al., 2007a; Peterson et al., 2016) and the initiation of ODEs (e.g. Gilman et al., 2010; Jacobi et al., 2010; Oltmans et al., 2012), knowledge of the factors controlling the spatial extent of ODEs, both in sea ice and polar coastal regions, remains limited.

Several studies have used networks of ozone measurements to examine the spatial extent of ODEs (e.g. Jacobi et al., 2010; Jones et al., 2013; Van Dam et al., 2013; Halfacre et al., 2014). Jones et al. (2013) used a network of ozone monitors in coastal Antarctica to show that ODEs can horizontally extend up to 1200 km, as well as 200 km inland. Halfacre et al. (2014) used three years of autonomous buoy measurements across the Arctic to estimate spatial scales for ODEs of up to 1000 km, with smaller air masses (∼300 km) being characterized by the most extreme ozone depletion. Previous measurements of ozone at inland and coastal sites (e.g. Jones et al., 2013; Van Dam et al., 2013) showed that the observation of an ODE at inland sites was coincident with observation of an ODE at coastal sites.

While the role of bromine radicals in ODEs is well-known (Barrie et al., 1988; Simpson et al., 2007b), the differing lifetimes of O₃ (hours to days) and BrO (minutes in the absence of heterogeneous recycling (McConnell et al., 1992; Platt and Hönninger, 2003)) means that the relationship between measured BrO and O₃ at a given location is not always clear. Many studies (e.g. Strong et al., 2002; Bottenheim and Chan, 2006; Jones et al., 2006) have found that rapid changes in observed ozone mixing ratios at coastal sites were often due to transport rather than local halogen chemistry. Indeed, Halfacre et al. (2014) showed, through BrO and O₃ measurements in the remote Arctic, that local BrO concentrations combined with chemical modeling are often insufficient to explain concurrently measured trends in O₃ (mean decrease of 3.5 nmol mol^-1 hr^-1), pointing to the larger role of transport mechanisms in determining levels of near-surface O₃.

The termination of ODEs and the recovery of boundary layer ozone to background levels is attributed to two main mechanisms, both of which rely on mixing rather than in-situ production. Jacobi et al. (2010) used ozone measurements at coastal sites, as well as from multiple research cruises, to argue that widespread boundary layer ozone depletion is typical throughout the Arctic in the polar spring, and transport of ozone rich air masses from lower latitudes causes ozone to recover to background levels. The second mechanism involves the transport of ozone rich air to the surface from aloft (e.g. Strong et al., 2002; Bottenheim et al., 2009). In particular, Moore et al. (2014) showed that open leads enhance convective mixing, causing ozone depleted air masses to recover to near background ozone levels. Presumably, other events with enhanced vertical mixing, such as mesoscale cyclones, would have similar effects.

This work uses ground-based ozone measurements obtained in the spring of 2005 at Barrow and Atqasuk, Alaska, to explore the frequency of differences in observed ozone between the two sites. Additionally, airborne measurements of BrO and O₃ conducted over tundra and sea ice areas around Barrow during the 2012 BRomine, Ozone, and Mercury EXperiment (BROMEX, Nghiem et al., 2013), as well as Moderate Resolution Imaging Spectrometer (MODIS) images of surrounding sea ice conditions in 2005 and 2012, allow exploration of potential causes of these differences.

From March 19 to April 30, 2005, ozone measurements were conducted at both Barrow, Alaska (71.325° N, 156.668° W) and Atqasuk, Alaska (70.469° N, 157.399° W), approximately 90 km to the south (Fig. 1). Atqasuk is 17 m above mean sea level, and there is minimal topography between the two measurement sites. Local wind conditions at Atqasuk were generally correlated with those at Barrow over the course of the 2005 study, while temperatures tended to be lower (Fig. S1). Ozone measurements at the National Oceanic and Atmospheric Administration (NOAA) Barrow Observatory near Barrow, Alaska in 2005 were provided by the NOAA Earth Systems Research Laboratory/Global Monitoring Division (ESRL/GMD) (Oltmans and Levy, 1994). All ground-based ozone measurements are reported as 30 min averages.

As part of the March 2012 BROMEX, the Purdue University Airborne Laboratory for Atmospheric Research (ALAR) enabled flight altitude O₃ measurements, as well as determination of BrO lower tropospheric vertical column densities (LT-VCDs) using differential optical absorption spectroscopy (DOAS) (Platt and Stutz, 2008). Airborne DOAS observations were used instead of satellite based DOAS observations due to the higher spatial resolution (hundreds of m vs. tens of km) and lower detection limits for BrO LT-VCDs of the airborne measurements. These measurements were performed during nine flights of BROMEX; herein, we focus on one of nine flights during which ozone depletion was observed inland in absence of concurrent depletion near the coast. Figure 1 shows the flight track on 28 March 2012 during which the data presented in this paper were acquired. The flight took off at 21:45 on 28 March and landed at 01:08 on 29 March (UTC), which corresponds to mid-afternoon local time.

In-flight DOAS measurements were performed using the Heidelberg Airborne Imaging DOAS Instrument (HAIDI) (General et al., 2014). The DOAS data presented in this work were obtained using measurements of scattered sunlight beneath the aircraft with a scanning nadir viewing telescope. The calculation of BrO LT-VCDs from the acquired spectra is a two-step procedure. First, DOAS fitting of these spectra was performed using DOASIS (Kraus, 2006) and a reference spectrum acquired in flight, over the 336 to 365 nm region to obtain differential slant column densities (dSCDs) of BrO. The calculated dSCDs were then converted to LT-VCDs using a geometric approximation that assumes the aircraft flew above the BrO layer and the stratospheric column of BrO was a constant 5×10¹³ molecules cm^-2. The average limit of detection (2σ) for BrO LT-VCDs during this flight was 4 × 10¹² molecules cm^-2. The full details of the DOAS fitting procedure and subsequent calculation of LT-VCDs can be found in General et al. (2014).

O₃ measurements on board ALAR were completed using a 2B Technologies model 205 dual-beam O₃ monitor. The instrument detection limit is 2 nmol mol^-1 (2σ), and ozone mixing ratios below this level were sometimes reported as negative values. Meteorological measurements were obtained using a “Best Air Turbulence” (BAT) probe (Garman et al., 2006). ALAR acquired vertical profiles of ozone and potential temperature at both Barrow and Atqasuk, during the ascent and descent of the airplane, on 28 March 2012.

To assist in interpretation of ground-based ozone measurements from 2005, as well as the in-flight measurements, we calculated 72 hour backward air mass trajectories with 50 m arrival heights at Barrow and Atqasuk (see Fig. 1) using the HYSPLIT trajectory model (Stein et al., 2015), incorporating meteorological fields derived from the NCEP Global Data Assimilation System (GDAS) modeling. Altitude changes along the modeled backward air mass trajectories were similar for each site, with the average difference in altitude change along a trajectory being less than 100 m over the first 36 h and less than 200 m over the entire trajectory.

Sea ice and lead formation in the Beaufort Sea and the Chukchi Sea were observed with the Moderate Resolution Imaging Spectrometer (MODIS) aboard the NASA Terra and Aqua spacecraft in sun-synchronous orbit around the Earth. For the 2012 BROMEX field campaign, a special MODIS Rapid Response subset was created to provide satellite images for near-real time observations. In particular, daily MODIS images of ice conditions were composed of the 7-2-1 bands for blue, green, and mid-infrared wavelengths (620–670, 841–876, and 2105–2155 nm wavelengths, respectively) and were gridded in a 250-m posting. In such composite images, ice is identified as light blue, while leads appear as a dark color. For the 2005 measurements, the composite MODIS images using the composite of 7-2-1 bands were not readily available at this time. Thus, MODIS true-color images in the same 250-m grid were used to monitor sea ice and leads, which can be also identified in the satellite images. Over the March-April period of the field measurements in both years, MODIS imagery shows a major lead opening event, followed by subsequent events leading to numerous leads in the vicinity of Barrow. While there are differences in details on different days, the general spatial pattern of lead distribution during spring (March-April) is similar in both years, suggesting the recurrent behavior of lead formations in this region where bathymetry may exert some control on sea ice processes (Nghiem et al., 2012).

Examining the correlation between O₃ measurements at Barrow and Atqasuk, shown in Fig. 2, illustrates that O₃ mixing ratios at both sites are correlated (R=0.78), with an orthogonal distance regression yielding a fit line with a slope near unity (1.03) and a near-zero y intercept. However, the root mean square deviation of the fit of 9 nmol mol^-1 suggests that there were times when large differences in ozone mixing ratios existed between the two sites.

To evaluate the frequency of these O₃ differences between Barrow and Atqasuk, we defined an ODE as measured O₃< 10 nmol mol^-1 (Oltmans et al., 2012) and examined the frequency of ODE occurrences at each site, shown in Fig. 3. During March-April of 2005, ODEs occurred at Barrow 41% of the time, while occurring slightly less often (39%) at Atqasuk. Both sites were simultaneously impacted by an ODE 31% of the time, with just one site being depleted in ozone 10% of the time at Barrow, and 7% of the time at Atqasuk. The finding that approximately two thirds of ODEs observed during this study occurred concurrently at both the inland and coastal sites (Fig. 3) is consistent with the advection of large (100s of km (Ridley et al., 2003; Bottenheim et al., 2009; Halfacre et al., 2014)) ozone depletion events to coastal areas from sea-ice areas (Bottenheim and Chan, 2006; Oltmans et al., 2012). The remaining third of ODEs occurred at only a single site, with ODEs being observed at only Barrow or Atqasuk with similar frequencies (21% and 15%, respectively) (Fig. 3). Given the tendency for ODEs occurring at Barrow to originate in the Arctic Ocean (Oltmans et al., 2012), our observations of an ODE occurring more frequently at a coastal site was expected. However, the finding that ODEs occurred at Atqasuk in the absence of an ODE at Barrow presents a contrast to previous comparisons of inland and coastal ozone measurements (Jones et al., 2013; Van Dam et al., 2013), where ODEs were only observed at inland sites when there were corresponding ODEs at coastal sites.

Moore et al. (2014) showed that the interaction between an ozone depleted air mass and an open lead caused a recovery in ozone to near background levels through increased vertical mixing. These lead interactions provide a potential mechanism for ozone differences observed between Barrow and Atqasuk in 2005. Figure 4 shows one instance where an ODE is observed at both sites while the backward air mass trajectories were over consolidated sea ice. As the ozone values measured at the two sites started to diverge on subsequent days, the averaged trajectories for each site also diverged, with trajectories from Barrow passing over leads, while trajectories from Atqasuk remained over coastal tundra and consolidated sea ice. This event coincided with an ozone recovery at Barrow, prior to recovery at Atqasuk which occurred when the trajectories from both sites were passing over newly opened leads. Thus, the observed differences in surface ozone mixing ratios at Barrow and Atqasuk are consistent with the influence of convective mixing associated with upwind sea ice leads and the associated downward mixing of ozone-rich air from aloft. This interpretation is consistent with and assumes a surface-based ODE phenomenon.

The observation of ozone depletion inland in the absence of concurrent depletion near the coast occurred during one of nine BROMEX flights. Flight-based measurements on 28 March 2012 provide a detailed look at the halogen and ozone conditions during times when ozone differences existed between Atqasuk and Barrow. Vertical profiles conducted at Barrow and Atqasuk on 28 March 2012 showed clear differences in ozone vertical profiles between the two sites. Figure 5 (left panel) shows a shallow (∼200 m) ODE at Atqasuk, while ozone surface mixing ratios were much greater (∼25 nmol mol^-1) at Barrow. Since ozone depletion is attributed to reactions with bromine radicals (e.g. Barrie et al., 1988; Fan and Jacob, 1992; Platt and Hönninger, 2003; Thompson et al., 2015), which react with ozone to form BrO, we also examined the spatial distribution of BrO during this flight using imaging DOAS techniques. As seen in Fig. 1, the amount of BrO observed within 50 km of Atqasuk is quite low, with the average observed BrO LT-VCD below the detection limit (∼4 × 10¹² molecules cm^-2). The observed BrO peaks over sea ice areas north of Barrow and 200 km to the southeast of Atqasuk, a finding to be investigated further in a future manuscript. It should be noted that during times when ozone was near-zero, as observed at Atqasuk, the formation of BrO would be inhibited, and the majority of BrO_x (BrO + Br) would consist of bromine atoms (Helmig et al., 2012). However, the absence of BrO, both near and upwind of Atqasuk, combined with backward airmass trajectories originating in the Arctic Ocean (Fig. 1) suggests the O₃-depleted air was transported to Atqasuk after being depleted by halogen chemistry upwind, rather than due to halogen activation chemistry occurring preferentially at Atqasuk.

A closer examination of the backward air mass trajectories and MODIS imagery for the 28 March 2012 flight show that air masses reaching Barrow previously interacted with an open lead just north of Barrow, while the air masses observed at Atqasuk failed to interact with an open lead in the previous 48 hours (Fig. 1). Potential temperature profiles, measured concurrently with ozone (Fig. 5), show steeper boundary layer slopes at Barrow than at Atqasuk, indicating stronger mixing of the boundary layer at Barrow than at Atqasuk. This finding suggests the occurrence of lead-induced vertical mixing, which caused a recovery of ozone at Barrow and not at Atqasuk, where an inversion persisted, leading to the observed differences in the ozone profiles. Thus, the ozone difference observed during this flight can likely be attributed to nearby lead dynamics causing a recovery of ozone at Barrow, while not at Atqasuk. These findings suggest that local sea ice dynamics play a role in determining the spatial heterogeneity of Arctic boundary layer ozone.

Arctic boundary layer ozone depletion is primarily associated with halogen activation in snow-covered sea ice regions. The majority of ozone depletion events observed during this study occurred concurrently at both inland and coastal measurement sites; however, in some cases, variation in upwind sea ice conditions led to ozone depletion events being observed inland while not being observed simultaneously at a coastal location. The differences in ozone mixing ratios observed at the two sites did not show a clear relationship with measured BrO LT-VCDs, likely due to the role of transport mechanisms, as well as upwind halogen chemistry, in determining local ozone levels. Both ground based measurements in 2005 and airborne measurements during the 2012 BROMEX campaign showed that lead interactions can drive the recovery of ozone to background levels at Barrow, while ODEs continue to be observed inland at Atqasuk. These findings suggest variations in boundary layer ozone exist in inland tundra regions despite the absence of significant topography or localized ozone sinks or sources. Given the drastic reduction of the multi-year sea ice pack and increasing prevalence of first year sea ice areas in the Arctic, under stormier conditions, which facilitate ice deformation and lead formation (Vavrus et al., 2012), these findings also suggest in the future, ODEs over the Arctic Ocean may be more frequently terminated by lead activity, increasing the spatial heterogeneity of ozone in the Arctic boundary layer and altering atmospheric composition across the Arctic.

Data from Barrow and Atqasuk are available via the NOAA Barrow Observatory (http://www.esrl.noaa.gov/gmd/obop/brw/) and ACADIS data gateway (http://www.aoncadis.org/), respectively. BROMEX flight data can be obtained by contacting the corresponding author.

© 2016 Peterson et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Chris Moore (Desert Research Institute) is thanked for discussions. John W. Halfacre (Purdue University) is thanked for calibration of the ALAR ozone monitor. Dana Caulton (Purdue University) is thanked for processing the potential temperature data. Ryan Tedd and Jillian Cellini contributed to ALAR data analysis as part of the University of Michigan Undergraduate Research Opportunity and Michigan Space Grant Consortium Fellowship Programs, respectively. The authors gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model used in this publication. Atqasuk meteorological data were provided by Walter Oechel (San Diego State University).