Springtime boundary layer ozone depletion at Barrow, Alaska: Meteorological influence, year-to-year variation, and long-term change



[1] In April 2008 and March–April 2009 near daily ozonesonde measurements were made over a several week period to study springtime Arctic boundary layer ozone loss in the vicinity of Barrow, Alaska. A detailed picture of the vertical structure of the depletion events from the soundings was obtained showing that the depletion was confined to approximately the lowest 1000 m with an average height of the top of the layer at ∼500 m. The two years were strongly contrasting in the frequency of ozone depletion events providing an opportunity for investigating the differing conditions under which these events develop. Short-term variability of the ozone depletion events is closely tied to the frequency of airflow that is primarily Arctic Ocean in origin (more depletion) or originates at lower latitudes (less depletion). The ubiquitous depletion events are interrupted by periodic mixing of ozone rich air into the boundary layer with the onset of synoptic scale weather changes that interrupt flow from off the Arctic Ocean. A 38-year record of surface ozone measurements at Barrow provides a unique time series that reveals the strong year-to-year variability of ozone depletion event occurrence. During March, but not April or May, there has been a significant increase in the frequency of ozone depletion events. This long-term increase in March depletion events appears to follow the decline in multiyear sea ice in the Arctic Ocean and its replacement by first-year ice. This significant change in the occurrence of boundary layer ozone events in March may signal a change in the oxidative chemistry in the Arctic that is related to climate change in this sensitive region.

1. Introduction

[2] Identification of near zero ozone mixing ratios in the springtime Arctic boundary layer [Oltmans, 1981; Bottenheim et al., 1986] and their association with halogen chemistry [Barrie et al., 1988; Oltmans et al., 1989] has spawned a number of campaigns (e.g., Polar Sunrise Experiments [Bottenheim et al., 1990], AGASP [Schnell et al., 1984], ALERT2000 [Bottenheim et al., 2002], TOPSE [Ridley et al., 2003], and Circumpolar Flaw Lead System Study [Seabrook et al., 2011; Nghiem et al., 2012]) to investigate these dramatic ozone depletion events. Two recent campaigns in spring 2008, the Arctic Research of the Composition of the Troposphere from Aircraft and Satellites (ARCTAS) and Aerosol, Radiation, and Cloud Processes affecting Arctic Climate (ARCPAC) [Jacob et al., 2010; Neuman et al., 2010] and the Ocean-Atmosphere-Sea Ice-Snowpack (OASIS) campaign in spring 2009 [Shepson et al., 2003] have reinvigorated investigation into the springtime Arctic boundary layer ozone loss. Simpson et al. [2007a] present a comprehensive review of the current state of knowledge related to polar boundary layer ozone depletion and related phenomena. This review provides a systematic overview of the many factors, meteorological, chemical, biogeochemical, microphysical, geographical, and more that play a role in Arctic springtime boundary layer ozone depletion.

[3] While detailed studies of the meteorology and chemistry related to the spring boundary layer ozone depletion events have been carried out at other Arctic locations such as Alert [e.g., Bottenheim et al., 2002; Morin et al., 2005] a more limited set of studies have focused on Barrow, Alaska [Tackett et al., 2007]. This despite the high frequency of depletion events at Barrow compared to other Arctic locations [Bottenheim and Chan, 2006]. In this study the intensive ozone profile measurements using ozonesondes made in conjunction with the 2008 and 2009 campaigns are used to provide a high vertical and temporal resolution picture of the structure of the depletion events at Barrow over an extended period of several weeks. The two years were strongly contrasting in the frequency of ozone depletion events providing an opportunity to investigate the differing conditions under which these events develop. The transport of ozone depleted air to Barrow and the interruption of the depletion events are investigated using trajectory analysis that is carried out for a number of years. In addition the influence of synoptic scale weather systems in transitioning from low ozone amounts at the surface to values representative of the free troposphere is studied during the two ozonesonde campaign periods.

[4] A 38-year record of surface ozone measurements at Barrow provides a unique time series that reveals the strong year-to-year variability of ozone depletion event occurrence. This long surface ozone record also presents a timely opportunity to explore possible longer term changes and in particular possible relationships with the significant changes that have taken place in the Arctic related to changing sea ice characteristics. The multiple meteorological and ozone data sets available for Barrow are used to describe the vertical structure, temporal evolution, and air mass origin of springtime ozone behavior.

2. Observation and Data Sources

[5] The continuously updated National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalysis data set [Kalnay et al., 1996] is used to compute 10-day back trajectories [Harris and Kahl, 1994] from Barrow. The kinematic three dimensional trajectories [Harris et al., 2005] determine the vertical position explicitly from the vertical wind field. The trajectories use the 6-h analysis fields and are calculated in one hour time steps. The starting altitude for all of the trajectories was 300 m. Both individual trajectories associated with specific profile and surface observations and monthly clustered trajectories [Harris and Kahl, 1994] are employed to trace air mass origins and to differentiate year-to-year patterns. The meteorological analysis database used for the trajectory calculations has relatively low vertical and horizontal resolution. The information provided by the trajectories gives a time dependent broad scale view of the synoptic meteorological conditions influencing Barrow, but does not diagnose the specific local meteorological conditions that also play an important role in what is observed in the ozone surface and profile measurements. In addition to the trajectories, synoptic maps showing the heights of the 1000 hPa, 925 hPa, and 850 hPa surface for the region in the vicinity of Barrow were produced from the National Oceanic and Atmospheric Administration (NOAA) Earth System Research Laboratory (ESRL), Physical Sciences Division web site: (http://www.esrl.noaa.gov/psd/cgi-bin/db_search/DBSearch.pl?Dataset = NCEP+Reanalysis+Daily+Averages+Pressure+Level&Variable = Geopotential+ Height&group = 0&submit = Search). The daily maps during the period ozonesondes were flown in 2008 and 2009 were used to determine when airflow patterns to Barrow underwent weather system related changes. Continuous surface meteorological data are available from the Barrow Observatory where both the surface ozone and ozonesonde observations were made. The NOAA operated Barrow Observatory (71.3°N, 156.6°W) is located 8 km east of the city of Barrow and 3 km southeast of the Arctic Ocean at an altitude of 8 m above sea level.

[6] The surface ozone measurements have been made with several varieties of instrumentation. From 1976 to 2010 ultraviolet absorption analyzers from Dasibi and Thermo Environmental Instruments spanning a number of instrument versions were used. In the initial period of observations from 1973 to 1975 an electrochemical instrument using the same methodology employed in the ozonesonde was used. Several years of overlap observations have been employed in transition from one instrument to another [Oltmans and Komhyr, 1986]. Through the entire measurement record the observations have been traced to the U.S. National Institute of Standards and Technology (NIST) ozone reference [Norris et al., 2004] through comparison to a NOAA/ESRL Global Monitoring Division (GMD) network standard, which in turn has been periodically compared with the NIST standard.

[7] Ozone profiles were obtained using balloon-borne Electrochemical Concentration Cell (ECC) ozonesondes [Komhyr, 1969] coupled to a Vaisala RS-80 radiosonde. The procedures are those used in the NOAA/ESRL/GMD ozonesonde network and are based on a number of recent studies of ozonesonde performance [Johnson et al., 2002; Smit et al., 2007; Deshler et al., 2008]. Because of the near zero ozone mixing ratios measured during the spring depletion episodes, particular attention has been paid to the background current [Vömel and Diaz, 2010], which is the signal measured by the ozonesonde for zero ozone air. The agreement between the surface ozone amount measured by the ozonesonde and the value measured by the surface based analyzer under near zero ozone conditions confirms that the zero ozone signal, which is very small in recent ozonesonde versions, is properly accounted for in the ozonesonde measurement. In order to provide higher altitude resolution data in the relatively shallow boundary layer depletion region, several techniques were employed to slow the balloon ascent rate from the nominal 5 m/sec to about 2–3 m/sec. Data transmission to the ground station from the radiosonde was 1 Hz.

[8] Data on multiyear sea ice extent during March was extracted from work by Nghiem et al. [2007] and updated through 2009 (S. V. Nghiem, personal communication, 2011). The Arctic Oscillation Index for the month of March (http://www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/ao.shtml) was obtained from the NOAA/NCEP Climate Prediction Center (CPC). It is used to indicate possible longer term circulation changes that could influence transport characteristics in the Arctic with implications for transport to Barrow.

3. Description of the Observational Results

[9] Both in the stratosphere and the troposphere outside the boundary layer ozone reaches its seasonal maximum in the spring [Oltmans, 1991]. Near the surface on the other hand (Figure 1) a spring minimum is present at Barrow. This seasonal pattern is similar to that presented by Bottenheim and Chan [2006]for Barrow, but differs from that seen at two other Arctic coastal locations with many more occurrences of mixing ratios <10 ppb in March and April at Barrow. Both the highest and lowest boundary layer ozone mixing ratios are observed during the spring at Barrow. A high-frequency picture of the variations at the surface is seen in the hourly ozone measurements from the surface observations during 2008 and 2009 for the month of April (Figure 2). April is the month with the greatest occurrence of spring boundary layer ozone depletion events defined as ozone mixing ratios less than 10 ppb. In addition to the highly variable nature of ozone at the surface it is clear that the two years are also quite different in the frequency of occurrence of low ozone mixing ratios (<10 ppb). These year-to-year differences will be discussed further when looking at the long-term variability of springtime ozone depletion occurrence. The strong variability between periods of higher and lower ozone is closely linked to the airflow reaching the Barrow location (Figure 2). Low ozone events are associated with flow from off the Arctic Ocean while higher ozone values are associated with flow from westerly or more southerly directions where the air parcels do not spend significant time over the Arctic Ocean. This is born out when considering the flow patterns for two contrasting months, April 2007 and April 2008 (Figure 3). Back trajectories computed for all days show that in April 2008 a large number of the trajectories originate to the south and west of Barrow traveling primarily over land. In April 2007 on the other hand less than 20% of the trajectories have a southerly origin. Also in 2007 there was a significant portion of trajectories that came from the Russian Arctic Ocean sector or the Arctic Ocean near Barrow that have been identified as particularly prominent sectors for the source of ozone depleted air [Bottenheim and Chan, 2006]. In April 2008 there were almost no trajectories originating from these regions. The wind roses that depict the ozone mixing ratios as a function of local wind direction (Figure 3) also show that in 2007 low ozone amounts are associated with local wind from off the Arctic Ocean with a high number of values of 0–20 ppb in the northeast sector. In 2008 local wind directions from the south dominate with ozone amounts >20 ppb.

Figure 1.

Seasonal behavior of surface ozone at Barrow, Alaska (71 N). The dot is the mean for each month, the horizontal bar within the box is the median, the box encompasses the 25th to 75th percentiles of the data, and the whiskers are the 5th and 95 percentiles. Data are for the period 1999–2008.

Figure 2.

The hourly average ozone mixing ratios at Barrow for April 2008 and 2009. Accompanying the ozone data are a 10-day back trajectory for a high and low ozone day during each year.

Figure 3.

Back trajectories and accompanying wind roses for April 2007 and 2008. The wind roses show the ozone mixing ratios as a function of the local wind direction. The concentric circles show the mixing ratio scale in ppb with 0 ppb at the center and 40 ppb for the outer circle.

[10] In the spring of both 2008 and 2009 near daily ozonesonde observations were carried out at Barrow (Figure 4a and 4c). Although the time periods covered by the ozonesondes in the two years do not overlap completely, the sharp difference between the two years is readily apparent as was also noted for the surface observations. In the first half of April when there were ozone profiles in both 2008 and 2009, there are only a couple of days with relatively weak depletion events (mixing ratios <20 ppb) in 2008. In contrast there are extended periods below 20 ppb and a number of days with near zero ozone levels in 2009. In the lower troposphere above the surface layer (∼500–2000 m) the differences are also large with ozone amounts over 50 ppb for much of this region in 2008 and only occasional profiles with regions exceeding 50 ppb in 2009. In April 2008 tropospheric transport was particularly vigorous from Eurasia. In addition, in the latter half of April biomass burning in Eurasia contributed to greatly enhanced ozone in the lower troposphere and episodically at the surface at Barrow based on trajectory analysis and observations of co-transported biomass burning related atmospheric constituents [Oltmans et al., 2010]. The near daily ozone profiles capture almost all of the ozone depletion events seen in the continuous surface record (Figure 4b and 4d). Exceptions were March 20, 21, and 29, 2009 when there were no soundings. On March 29 the interruption of a depletion event by higher ozone mixing ratios that is seen in the surface observations was not covered by a sounding. The soundings in most cases captured the full range of the ozone mixing ratios between the high and low events. The surface value measured by the ozonesonde matches the corresponding surface mixing ratio extremely well. The ozone depletion events vary in vertical extent from a relatively restricted height of about 200 m to a maximum depth of ∼1000 m (Figures 4 and 5). Typical depth was ∼500 m. In all cases the events were connected to the surface as was also noted by Seabrook et al. [2011] from shipboard profiles obtained by LIDAR in the Amundsen Gulf off the south coast of Banks Island. The typical depths and maximum vertical extent of the depletion layers was similar at Barrow to those seen in the LIDAR profiles. A similar vertical structure has also been observed at Alert, Canada [Bottenheim et al., 2002]. There were no cases at Barrow where depleted air was found in isolated elevated layers above the surface. Whenever a depletion event occurred, that air must have been transported through the depth of the boundary layer when it reached Barrow. However, the ozone profile within the boundary layer was often variable in structure (Figure 5). In some cases near zero ozone was found through the entire layer with a sharp boundary at the top of the layer while in other cases there was a more gradual increase from the low values at the surface to the top of the boundary layer.

Figure 4.

(a) Vertical cross-section of ozone mixing ratio at Barrow during ARCTAS/ARCPAC in April 2008. (b) The ozonesonde surface value is overlaid on the hourly average surface ozone mixing ratios during the time period covered by the soundings. (c) Vertical cross-section of ozone mixing ratio at Barrow during OASIS in March–April 2009. (d) The ozonesonde surface value is overlaid on the hourly average surface ozone mixing ratios during the time period covered by the soundings.

Figure 4.


Figure 5.

Individual ozone profiles from April 2008 and 2009 showing the detailed vertical structure through the boundary layer.

4. Transport Patterns

[11] There are significant differences between years in the occurrence of low ozone events (defined as ozone mixing ratios <10 ppb) during the spring at Barrow (Figure 6a). The year-to-year variability is closely tied to the persistence of various transport patterns during a particular year. Spring 1981 stands out with less than 10% of the time with ozone mixing ratios less than 10 ppb while in spring 1984 more than half the time ozone values were below this threshold. The years 2008 and 2009 were also contrasting, not only during the time of the ozonesonde measurements but also during the entire spring season with each of the spring months having a greater frequency of depletion events in 2009 than in 2008. As noted in the examples for April 2008 and 2009 ozone mixing ratios in the spring at the surface at Barrow are strongly related to the source of air reaching the site. To investigate this dependence on transport characteristics 5-day back trajectories at 6-h intervals were segregated for cases when the surface ozone mixing ratio in the spring was less than 10 ppb versus when it was greater than 35 ppb (Figure 6b6e). This was done for several contrasting years in the frequency of ozone depletion events. In years such as 2009 (Figure 6b) and 1984 (Figure 6d) a large fraction of the trajectories show up in the <10 ppb panel and indicate air parcel paths that spent almost all of their time over the Arctic Ocean. In 2008 (Figure 6c) and 1981 (Figure 6e), on the other hand, a large number of trajectories are associated with ozone amounts >35 ppb with a significant number coming from the south or west with little Arctic Ocean contact. There are some trajectories with ozone mixing ratios >35 ppb that do spend significant time over the Arctic Ocean but a large number of them are confined to near the Canadian Arctic coast, which is not a region where significant depletion takes place [Bottenheim and Chan, 2006]. This is particularly evident in 2008. In 1984, the spring with the highest frequency of depletion, the large number of trajectories for cases of ozone mixing ratio <10 ppb are strongly confined to the Arctic Ocean. These trajectories are largely from April and May and are clustered toward the Russian sector, the region where low ozone amounts have been found to arise most frequently [Bottenheim and Chan, 2006]. Other years such as 1994 with frequent low ozone events and 1998 with relatively few events show similar trajectory patterns (not shown) with the preponderance of air transported over the Arctic Ocean in 1994 and much more southerly flow in 1998. The strong associations of both the ozone depleted air masses that originate over the Arctic Ocean and higher ozone amounts that primarily come with air parcels from the Pacific sector are indicative of the strong role of synoptic weather changes [Jacobi et al., 2010]. Mixing of ozone rich air to the surface that ends a depletion event is related to a change in air mass at Barrow. The weakening of the boundary layer that restricts mixing from aloft during the depletion events occurs with a shift in air mass, not a localized weakening of the boundary layer. This can be seen in the contours of temperature (Figure 7) for the period of the ozonesonde measurements. The transition from low ozone episodes in the boundary layer to higher ozone values is accompanied by a significant change through the lower troposphere in temperature that results from synoptic weather system changes on a time scale of several days. To further investigate the link between the cessation of a depletion event with the synoptic weather pattern, maps of the geopotential height contours on several pressure surfaces were constructed for the days surrounding depletion events where ozone values dipped below 10 ppb during the ozonesonde campaigns. Three events were identified in 2008 and 10 in 2009. During April 2008 the depletion events were separated by at least a day so that the daily maps of the height of the pressure surfaces distinctly show the transition from one pressure pattern and the related flow to Barrow to another regime. For example, on April 16, 2008 at 925 hPa (Figure 8a) there was low pressure over the Canadian Archipelago to the east of Barrow with northerly flow off the Arctic Ocean. By April 18 (Figure 8b) a strong low pressure system has developed to the west of Alaska with strong flow out of the south to Barrow. In 2009 several of the depletion events only occur over a portion of the day so that daily maps are not as conclusive in showing the change in the meteorological situation (for example March 18). However, for most of the events the transition is clearly reflected in the change in the meteorology. This is illustrated in the transition from the extended depletion event on March 26–28 using the maps for March 27 (Figure 8c) and March 29 (Figure 8d). On March 27 a low pressure system to the south and east of Barrow brought air from the northeast to Barrow. On March 29 a strong low pressure center now located over the Aleutian Islands brought air from the south over interior Alaska to Barrow. The airflowing off the Arctic Ocean to Barrow will be routinely depleted of ozone given the ubiquitous nature of such depletion [Bottenheim et al., 2009]. The interruption of the prevailing flow from off the Arctic Ocean by migrating synoptic weather systems temporarily restores surface ozone amounts to levels near those seen just above the boundary layer (Figures 4). This link to synoptic conditions at Barrow is consistent with that seen in the study by Jacobi et al. [2010].

Figure 6.

(a) The normalized frequency of hourly average ozone values <10 ppb for the spring of each year from 1973 to 2010. The normalized frequency is the number of hours with ozone values <10 ppb divided by the total number of hourly measurements. Individual trajectories for selected years with (b and d) high and (c and e) low frequency of depletion events. Trajectories are shown for cases with high ozone (>35 ppb) and low ozone (<10 ppb) amounts.

Figure 6.


Figure 7.

Vertical cross-section of temperature at Barrow during (top) ARCTAS/ARCPAC in April 2008 and (bottom) OASIS in March and April 2009.

Figure 8.

Geopotential height contours in meters on the 925 hPa pressure surface for (a) April 16, 2008, (b) April 19, 2008, (c) March 27, 2009, and (d) March 29, 2009. Contour intervals are 25 m. Purple and blue contours are lower pressure while red and orange are higher pressure. Data courtesy of the NOAA Earth System Research Laboratory, Physical Sciences Division (see text for the URL).

5. Longer-Term Changes

[12] The significant relationship between annual (first year) sea ice and bromine activation [Simpson et al., 2007a] suggests that the there could be an increase in the frequency of springtime boundary layer ozone depletion events, given the reported dramatic decline in perennial or multiyear sea ice [Maslanik et al., 2007, 2011; Nghiem et al., 2007; Perovich et al., 2011]. First-year or annual sea ice has replaced multiyear sea ice in much of the Arctic [Kwok and Untersteiner, 2011; Perovich et al., 2011]. As discussed above, year-to-year variability in the frequency of depletion events at Barrow is related to synoptic weather patterns that control airflow characteristics to the observing site. This is a general characteristic throughout the ozone depletion season of March–April–May. A significant increase as shown below in the normalized frequency of boundary layer ozone depletion events is, however, confined to the month of March based on the long record of surface ozone observations at Barrow (Figure 9). The normalized frequency is the number of observed hours less than 10 ppb divided by the number of hours in the month with observations. Although April dominates the total frequency of low ozone events, there is no apparent long-term change in the number of hours with ozone <10 ppb. May also shows no significant change. The average normalized frequency of hours with ozone mixing ratios less than 10 ppb for March during the second half of the record, 1993–2010 (18 years) is 85% greater than for the first half of the record, 1973–1992 (18 years with data - 2 years have missing data). In the first half of the record depletion events with ozone <10 ppb occurred on average less than 15% of the time. There were several years with either no or very few occurrences. In the second half of the record depletion events with <10 ppb of ozone occurred on average more than 25% of the time and there were no years with less than 10%. To test the significance of these differences a test for the difference of the means was carried out. The difference (0.14 versus 0.26 frequency units) was significant at the 99% level. A linear fit to the data also shows a significant increase of 0.0054 ± 0.0034 normalized frequency units per year (slope ± 2 standard errors). This represents a change of ∼4 h/year in March over the 38 year period of observation. There is also a significant increase of 0.0039 ± 0.0026 when including only the strongest depletion events (<5 ppb) or 0.0061 ± 0.0046 when including modest depletion events (<20 ppb) as well (Figure 10). Similar tests for April and May did not show a significant change. By the mid 1990s there appears to have been a consistent shift to more hours with ozone <10 ppb in March that is absent in April and May.

Figure 9.

The normalized frequency of hourly average ozone values <10 ppb for the months of March, April, and May for each year from 1973 to 2010. The normalized frequency is the number of hours with ozone values <10 ppb divided by the total number of hourly measurements.

Figure 10.

The normalized frequency of hourly average ozone values (top) <10 ppb, (bottom left) <5 ppb, and (bottom right) <20 ppb for the month of March of each year from 1973 to 2010. The normalized frequency is the number of hours with ozone values less than the threshold value divided by the total number of hourly measurements. The linear trend and two standard error limits for the normalized frequency for the 5 ppb, 10 ppb, and 20 ppb thresholds are 0.0039 ± 0.0026, 0.0054 ± 0.0026, and 0.0061 ± 0.0046 respectively.

6. Discussion

[13] In order to link the increase in March surface ozone depletion at Barrow with possible changes in Arctic Ocean sea ice conditions use is made of several studies detailing the March changes in age of Arctic Ocean sea ice. There has been a dramatic decline in overall Arctic sea ice extent (see, e.g., http://nsidc.org/arcticseaicenews/index.html and Kwok and Untersteiner [2011]). In the spring while the overall ice cover has declined, the more dramatic change has been a shift from perennial or multiyear ice to annual or first year ice [Nghiem et al., 2007; Maslanik et al., 2007, 2011; Perovich et al., 2011] that may have important implications for the occurrence of spring boundary layer ozone depletion [Simpson et al., 2007a, 2007b]. Of particular relevance is the decline in March perennial ice cover (Figure 11) [Nghiem et al., 2007; Maslanik et al., 2007, 2011; Perovich et al., 2011] for spring boundary layer ozone depletion at Barrow and its replacement by younger ice, and in particular first year ice [Perovich et al., 2011]. It has been shown [Simpson et al., 2007b] that air parcels arriving at Barrow that have contact with first year sea ice are well correlated with measurements of BrO. The work of Bottenheim and Chan [2006] showed that air masses depleted in ozone arriving at Barrow were correlated with first year ice and in particular areas of polynya formation. The higher salinity of first year ice compared to multiyear ice [Wagner et al., 2001] appears to be an important factor in BrO activation [Simpson et al., 2007b]. Ice melting in April and May provides conditions where openings in the ice (leads, polynyas) are more consistently present [Eicken et al., 2006]. In March, however, a shift to more openings in the ice that are present in annual ice compared to perennial ice may provide enhanced conditions for halogen activation and subsequent ozone depletion. It now appears that even with year-to-year variations in transport to a particular location such as Barrow there will likely be conditions in March over the sea ice that promote ozone depletion, whereas in the past these conditions were less frequent.

Figure 11.

March Arctic Ocean perennial sea ice extent in millions of square kilometers. The blue bars are values based on a drift-age model. The red bars are QuikSCAT observations. The smooth curve is a 5th order polynomial fit to the drift-age model data (adapted fromNghiem et al. [2007]).

[14] Because of the prominent role played by variations in transport to the Barrow location in the frequency of ozone depletion events, a long-term change in the transport pattern to this site could also influence the occurrence of these events. The Arctic Oscillation is a prominent and well studied atmospheric mode related to circulation patterns over the Arctic region [Thompson and Wallace, 1998]. Long-term changes in the Arctic Oscillation (AO) Index in the spring and in March in particular could suggest a shift that could result in an altered airflow regime to Barrow. While there was a more positive phase of the AO Index in the early 1990s, the difference in the average AO Index value for March in the period 1973–1992 compared to 1993–2010 is not significant and no significant trend was found for the period covered by the ozone observations. In order to look more specifically at the transport characteristics to the Barrow site, trajectories were computed four times each day for every day in March for the period 1973–2010 and then clustered for the two periods 1973–1992 and 1993–2010 (Figure 12). The clusters were determined for the entire 38 year time period. For each of the two partial time periods the trajectories were clustered according to the pattern determined for the complete record (Figure 12). In examining the percentage of trajectories in each cluster for the two periods only small differences were noted (Figure 12). The individual trajectories in each cluster for each time period were also examined to determine if the dispersion of the individual trajectories within the cluster was also similar. Again the small differences are suggestive that changing transport to the Barrow site in March was not responsible for the long-term change in frequency of the boundary layer ozone depletion events. However, the average ozone amount associated with each cluster for the later period was significantly lower for all clusters where the average air parcel path was primarily over the Arctic Ocean (4 of the 6 clusters) with clusters 3 and 4 showing the largest difference (Table 1). The region along the Canada and Alaska coasts represented by clusters 3 and 4 that in the past was covered with multiyear ice that extended well beyond the coast but is now confined to a narrow region along the coast, is now dominated by first year ice [see Perovich et al., 2011, Figure 5.12; Maslanik et al., 2007]. Clusters for March representing air parcel paths from the south (2 of 6 clusters) showed no differences between the earlier and later periods (Table 1). Figure 13 shows the cluster members for two of the clusters; one from over the Arctic Ocean (cluster 4) where ozone changes have been significant and one from the south (cluster 5) with essentially no change. The lack of discernable changes in the transport pattern as a likely cause for the increase in the occurrence of ozone depletion events in March at Barrow leaves the changing sea ice characteristics (reduced multiyear ice and more first year ice) as the prime reason for the increase.

Figure 12.

Back trajectory clusters for Barrow for March for the period 1973–2010. The percentages of the trajectories in each cluster for the periods March 1973–1992 and March 1993–2010 are given.

Table 1. Mean Ozone Mixing Ratio (ppb) for the Six Clusters Shown in Figure 12 for the Two Time Periods March 1973–1992 and March 1993–2010a
Cluster Number1973–1992 (ppb (Number))1993–2010 (ppb (Number))Difference/Significance (ppb (p-Value))
  • a

    The difference for each period and the significance of the difference based on a t-test are also given (NS = not significant).

Cluster 123.1 (551)20.4 (640)−2.7 (0.01)
Cluster 224.3 (466)21.0 (444)−3.3 (0.01)
Cluster 325.9 (481)19.6 (383)−6.3 (0.01)
Cluster 422.2 (240)17.5 (269)−4.7 (0.01)
Cluster 532.3 (135)32.8 (172)+0.5 (NS)
Cluster 630.3 (108)30.4 (139)+0.1 (NS)
Figure 13.

Individual cluster members (trajectories) for clusters 4 and 5 (numbers as shown in Figure 12) for the periods 1973–1992 and 1993–2010. One trajectory (00Z) for each day is plotted instead of all four daily trajectories for clarity. The average ozone for each of the clusters is shown below each plot.

7. Summary

[15] The presence at Barrow of springtime boundary layer air containing very low mixing ratios of ozone is a defining characteristic of this site. During the spring season (March–April–May) air with ozone mixing ratios < 10 ppb can vary in occurrence from less than 10% of the time to 50% depending on the year. Some depletion (mixing ratios < 20 ppb) occurs on average almost half the time in the spring and 70% of the time in April, the month with the most frequent depletion events. In general the site at Barrow is dominated by flow from off the Arctic Ocean but there is large year-to-year variability in the frequency based on both analysis of trajectories and local wind conditions. Overwhelmingly, air depleted in ozone arrives at Barrow from over the Arctic Ocean where such conditions are ubiquitous [Bottenheim et al., 2009]. Abrupt shifts from ozone depleted conditions in the boundary layer to ozone levels seen just above the boundary layer result from the passage of synoptic scale weather systems.

[16] Near daily ozonesonde measurements for three weeks in April 2008 during ARCTAS and ARCPAC and for five weeks in March and April 2009 during OASIS provide valuable information for the first time on the vertical structure of the ozone depletion events at Barrow. Since these were strongly contrasting years in the frequency of depletion events, they provide insight on the varying meteorological conditions that drive year-to-year differences. During the overlapping time period of soundings in the first half of April during the two years, there was only one weak depletion event in 2008. In 2008 the lower troposphere above the boundary layer also had consistently higher ozone mixing ratios than during 2009 (Figures 4a and 4c). Temperatures in the lower troposphere during this time were also warmer in 2008 than in 2009 (Figure 7) consistent with a transport regime in which air was coming from more a southerly sector in 2008. Although March 2008 did not see as dramatic reduction of depletion events as April 2008 (Figure 8), it still had the fewest events for the period since 1993 when depletion events in March have generally been increasing. The contrast persisted into May where in 2008 there were no depletion events after May 5 while in 2009 an event occurred as late as the last three days of the month. The difference in the ozone profiles during the time of the soundings between the two years was reflective of the contrast in the meteorology and transport through the entire season as shown in the trajectories for the entire season (Figure 8).

[17] Tarasick and Bottenheim [2002]have reported on a possible increase in boundary layer ozone depletion events at Resolute, Nunavut based on the 35 year record of ozonesonde measurements. The tendency toward more frequent events appears to be most pronounced during the latter one third of that record. The unique 38-year record of surface ozone observations at Barrow reveals a very significant shift to more frequent depletion events during March at Barrow during the second half of the record compared to the earlier half. This shift is not seen in the months of April and May and is indicative of more conducive conditions for ozone depletion early in the season. The striking shift from multiyear to annual ice during March is consistent with a change to conditions that can result in enhanced ozone depletion in the Arctic Ocean boundary layer [Simpson et al., 2007a]. The increased occurrence of ozone depleted air in the boundary layer in March at Barrow may signal a change in the chemical environment of the Arctic. This change in the oxidative state in the Arctic tied to changing characteristics of sea ice in March that has been linked to a warming climate [Serreze et al., 2007] may have implications for other atmospheric chemical cycles including that of mercury [Simpson et al., 2007a].


[18] Funding for the ozonesonde measurements at Barrow in 2008 was provided under NASA grant NNH08AH471 as part of ARCTAS. Multiple personnel over many years have been responsible for the surface ozone measurements at Barrow and their contribution is appreciated. Of particular note is the stewardship of the NOAA Barrow Observatory by Dan Endres, who guided the operations at the observatory including the surface ozone observations for 25 years. He personally witnessed the changes in the Arctic environment over his historic tenure at Barrow. Comments and suggestions by three reviewers contributed significantly to the improvement of the manuscript.