Satellite observations have shown large areas of elevated bromine monoxide (BrO) covering several thousand square kilometres over the Arctic and Antarctic sea ice regions in polar spring. These enhancements of total BrO columns result from increases in stratospheric or tropospheric bromine amounts or both, and their occurrence may be related to local meteorological conditions. In this study, the spatial distribution of the occurrence of total BrO column enhancements and the associated changes in meteorological parameters are investigated in both the Arctic and Antarctic regions using 10 years of Global Ozone Monitoring Experiment-2 (GOME-2) measurements and meteorological model data. Statistical analysis of the data presents clear differences in the meteorological conditions between the 10-year mean and episodes of enhanced total BrO columns in both polar sea ice regions. These differences show pronounced spatial patterns. In general, atmospheric low pressure, cold surface air temperature, high surface-level wind speed, and low tropopause heights were found during periods of enhanced total BrO columns. In addition, spatial patterns of prevailing wind directions related to the BrO enhancements are identified in both the Arctic and Antarctic sea ice regions. The relevance of the different meteorological parameters on the total BrO column is evaluated based on a Spearman rank correlation analysis, finding that tropopause height and surface air temperature have the largest correlations with the total BrO vertical column density. Our results demonstrate that specific meteorological parameters can have a major impact on the BrO enhancement in some areas, but in general, multiple meteorological parameters interact with each other in their influence on BrO columns.
Bromine (Br) compounds play an important role in atmospheric chemistry in
particular with respect to removal of ozone. It has been estimated that
bromine contributes about 25 % to the global destruction of stratospheric
ozone and up to 50 % to polar stratospheric O
Stratospheric bromine is predominantly present in its inorganic form,
originating from both natural and anthropogenic organic sources. Man-made
brominated hydrocarbons (halons) are long lived and are transported to the
stratosphere where they release bromine atoms, e.g. Br and bromine monoxide
(BrO), through UV photolysis and oxidation. Methyl bromide (CH
Tropospheric ozone depletions linked to bromine chemistry were first
discovered in the Arctic
BrO is observed by both ground-based and satellite measurements using the differential optical absorption spectroscopy (DOAS) technique (Platt and Stutz, 2008). Ground-based measurements such as long-path DOAS (Hönninger et al., 2004; Stutz et al., 2011), multi-axis DOAS (Hönninger et al., 2004; Frieß et al., 2011), and chemical ionization mass spectrometry (Liao et al., 2011; Choi et al., 2012) provide good temporal coverage and in some cases the vertical profile of BrO, while UV–visible nadir satellite measurements allow us to study the global distribution of BrO columns with good spatial coverage. In particular, since observing BrO by satellites became possible, studies on the mechanism for large-scale release of bromine over both polar sea ice regions, where a management of ground-based instruments is difficult, have been carried out.
The first satellite observations of polar BrO events were performed using
the Global Ozone Monitoring Experiment (GOME) instrument (Wagner and Platt, 1998; Richter et al., 1998; Chance,
1998). A long time series of total BrO columns and the spatial distribution
of regions with enhanced BrO in both hemispheres were investigated using 6 years of GOME data (Hollwedel et al., 2004), finding that general features
of the BrO distribution are similar between years but the strength of BrO
explosion events varies from year to year with an increasing trend in both
polar regions. SCIAMACHY, which was launched after GOME, can measure
scattered and reflected solar radiation in limb and nadir geometry. Thus,
both BrO vertical profiles and BrO column densities could be retrieved from
two different observation modes of SCIAMACHY and an averaged global
background of tropospheric BrO was estimated by comparing the integrated
stratospheric BrO profile with the simultaneously measured total BrO column
(Sinnhuber et al., 2005; Rozanov et al., 2011). OMI and GOME-2, which were
launched in 2004 and 2006, respectively, provide data with improved spatial
resolution and signal-to-noise ratio, enabling us to answer additional
scientific questions related to polar springtime BrO explosion events.
Salawitch et al. (2010) showed that the locations of total BrO column
hotspots during Arctic spring observed by OMI are coincident with high total
O
Meteorological conditions for the occurrence of bromine explosion events have also been studied. Jones et al. (2006) and Jacobi et al. (2006) found that ozone depletion events and elevated BrO concentrations are related to a stable and shallow boundary layer occurring under temperature inversions and low wind speeds. Such conditions allow chemical reactions to proceed efficiently, with the boundary layer acting as a closed reaction chamber. However, enhanced BrO events were also detected during episodes of high wind speed. Jones et al. (2010) found that BrO explosion events can occur under environmental conditions consisting of high wind speeds and the presence of blowing snow. Choi et al. (2018) also showed a significant association between the temporal and spatial extent of tropospheric BrO explosions observed from OMI and GEOS-5 simulated sea salt aerosol emissions generated by blowing snow. They demonstrated that saline aerosol particles generated during blowing snow events serve as a source of reactive bromine in the bromine explosion mechanism. The role of wind speed and atmospheric stability in determining the lower tropospheric BrO vertical distribution was reported by Peterson et al. (2015) using MAX-DOAS observations. In that study, high wind speeds were linked to some of the high columns of BrO, but based on the low frequency of these cases, they argued that high wind speeds and blowing snow are not the sole driver of halogen activation. Sihler (2012) investigated the relationship between tropospheric BrO vertical columns retrieved from GOME-2 measurements and environmental parameters including meteorological parameters from the European Centre for Medium-Range Forecasts (ECMWF) weather model and cloud properties measured by CALIPSO for selected cases. It was found that the occurrence of enhanced tropospheric BrO vertical columns is related to boundary layer meteorology, although the causality is not always clear. A bromine explosion event linked to cyclone development in the Arctic was investigated by Zhao et al. (2016) and Blechshmidt et al. (2016). The vertical lifting and high wind speeds driven by the front of the polar cyclone can transport cold brine-coated snow and salt aerosols into the free troposphere, resulting in bromine explosion events and extending BrO plume lifetime through continuous supply of reactive bromine.
As mentioned above, many studies on the possible sources of BrO enhancements and the driving meteorological conditions in polar regions have been conducted using ground-based and satellite measurements. Study results clearly indicate that meteorological conditions affect the processes of BrO enhancements in several ways. These previous studies mainly focus on specific case studies or analysis using relatively short-term datasets. This study aims at adding to this body of knowledge and obtaining a more general and comprehensive understanding of the enhancements of total BrO columns using a consistent long-term dataset. Therefore, we statistically analyse the spatial distribution of occurrence frequency of enhanced total BrO column and its relationship to various meteorological parameters in the Arctic and Antarctic sea ice regions by using a 10-year long-term dataset. In particular, the relationships between total BrO vertical columns retrieved from GOME-2A/2B and meteorological fields including sea level pressure, surface-level wind speed and direction, surface air temperature, and tropopause height were investigated. The reason for using total BrO columns instead of tropospheric and stratospheric columns separately to examine the relationship with the meteorological fields is that existing separation methods for satellite BrO data are difficult to apply to a long-term dataset in both hemispheres. They also have large uncertainties in connection with low pressure systems and large tropopause height changes which affect both stratospheric and tropospheric columns. This study aims to investigate how the meteorological system generally affects the total BrO column, rather than separate the effects on the enhancement of BrO in each atmospheric layer. Differences in meteorological conditions and their regional characteristics between high BrO situations and the mean field were investigated in order to better understand meteorological effects on processes involved in BrO enhancements. Finally, based on Spearman rank correlation analysis, the degree of influence of different meteorological parameters on total BrO columns was evaluated and the most important meteorological parameters, influencing BrO, and their regional patterns were identified.
GOME-2 is a series of three identical instruments operating aboard the
MetOp satellites. They were launched sequentially to enable continuous
long-term monitoring of atmospheric composition with the same instrument
specification (Callies et al., 2000; Munro et al., 2016). The first GOME-2
aboard MetOp-A was launched in October 2006, and the second and third
aboard MetOp-B and MetOp-C were launched in September 2012 and November 2018, respectively (hereafter referred to as GOME-2A, GOME-2B, and GOME-2C, respectively). GOME-2 is a nadir-viewing scanning UV–visible spectrometer with four channels covering the spectral range between 240 and 790 nm at a spectral resolution of 0.26–0.51 nm (Munro et al., 2016). The spatial resolution of GOME-2 data is typically
In this study, we use data from GOME-2A from 2008 to 2011 and data from
GOME-2B from 2013 to 2018. The first year of data after the launch of
GOME-2A and GOME-2B is not used to avoid sampling bias from incomplete
coverage during this time period. Also, as the swath width of GOME-2A was
reduced from 1920 to 960 km in July 2013 resulting in a change in ground
pixel size to
Summary of DOAS settings used for the GOME-2 BrO slant column retrievals.
When using a Pacific background spectrum, the retrieved differential slant
columns (DSCDs) need to be corrected by adding the BrO slant column over that
region. Here, we follow earlier studies (Richter et al., 2002; Sihler et al.,
2012; Seo et al., 2019) and assume a BrO vertical column of
Sea ice concentrations (Spreen et al., 2008), i.e. the percentage of a
given area covered with sea ice relative to the total, retrieved from AMSR-E
(Advanced Microwave Scanning Radiometer-EOS) and AMSR2 (Advanced Microwave
Scanning Radiometer 2) satellite measurements were used to identify the sea
ice domain for the 2008–2011 and 2013–2018 periods, respectively. The reason for the missing of analysis for 2012 in this study is that there was a gap between AMSR-E and AMSR2. AMSR-E stopped producing data in October 2011, and calibrated brightness temperature data from the AMSR2, the successor of
AMSR-E, have been released in January 2013. Sea ice concentrations are
retrieved by the ARTIST Sea Ice (ASI) algorithm based on the polarization
difference of brightness temperature at the 89 GHz channel and weather
filtering using other channels (Spreen et al., 2008). Sea ice concentration
data are provided on a daily basis with high spatial resolution of
To explore meteorological conditions in polar regions, ECMWF ERA-Interim reanalysis data (Dee et al.,
2011) (
To investigate the relationship between BrO and various meteorological factors over the polar sea ice regions, a spatiotemporal matching of the different datasets is required. According to previous studies, enhanced BrO columns are observed over the continent near the coast in the presence of blowing snow, and even over the interior of the continent by long-range transport of air masses from the sea ice regions (Choi et al., 2018). However, development and maintenance of large-scale enhanced BrO plumes require a continuous supply of reactive bromine over a large area, and the origins of these sources are typically located on the polar sea ice. Thus, we limited the study domain for long-term analysis to the region where the sea ice remains. Since the spatial resolution of daily sea ice concentration data from AMSR-E and AMSR2 is higher than that of GOME-2 BrO columns, GOME-2 BrO data were selected if an average of sea ice concentration located within the GOME-2 satellite pixel is higher than 5 %. The selected GOME-2 BrO data over the sea ice region are also matched with ECMWF meteorological datasets in both time and space. First, the meteorological datasets were linearly interpolated with respect to the observation time of GOME-2 BrO and then spatially matched by interpolation with an inverse weighting proportional to the distance of the nearest four pixels from the GOME-2 BrO pixel. The temporally and spatially matched datasets were used to explore the spatial behaviour of the relationship between BrO columns and meteorological parameters in the next sections.
To investigate the characteristics of occurrences of enhanced BrO, we first
need to establish a reasonable detection criterion for BrO enhancements. In
previous studies, satellite observations of enhanced BrO columns during
spring, termed “BrO hotspots”, were defined as the region where the
total column BrO was elevated by at least
Monthly histograms of the total BrO VCD from GOME-2 measurements
during the study period of 2008–2018 (March and April for the Arctic;
September and October for the Antarctic). The solid blue line is a fitted
Gaussian distribution and the dashed blue line indicates a range of
mode
Figure 1 shows the monthly histogram of GOME-2 total BrO VCDs over the
Arctic (March and April) and Antarctic sea ice regions (September and
October) for the study period from 2008 to 2018. Average total BrO VCDs are
in general normally distributed representing the mean background BrO.
However, the right side of the BrO column distribution has higher
frequencies of occurrence compared to the left side and is long tailed due
to BrO enhancements during polar springtime. We defined a threshold for BrO
hotspot detection as the mode of the BrO column distribution
One issue to note is that monthly thresholds for identification of BrO hotspots in each hemisphere determined by this statistical method have a limitation in discerning a clear source of enhanced total BrO column. In principle, there are three conceivable explanations for enhanced total BrO columns observed by satellite: (1) the enhancement results from descending BrO-enriched stratospheric air when the tropopause is low, (2) the increase results from bromine explosion events occurring in the troposphere, (3) the enhancements are due to a combination of both stratospheric and tropospheric contributions. Thus, to assess whether total column BrO enhancements, detected by using the BrO VCD threshold criteria defined above, are of stratospheric or tropospheric origin, we will investigate related meteorological factors in the next sections.
Monthly spatial distribution of the occurrence frequency of
enhanced total BrO columns over the Arctic (
To identify areas where BrO hotspots occur frequently, the spatial
distribution of the frequency of enhanced BrO occurrences was investigated
over a long time period. To map the spatial distribution of frequency, a
reference grid is required because the positions of the satellite pixels are
not constant. In this study, a reference grid with
Frequency distribution of sea level pressure for all data (black
line) and situations with enhanced BrO columns (red line) in
Figure 2 shows how frequently enhanced total BrO columns were observed over
the Arctic and Antarctic sea ice regions in spring. As can be seen in Fig. 2, patterns and magnitudes of
In this section, we examine the relationship between the occurrence of enhanced total BrO columns and various meteorological conditions, focusing on the magnitude and spatial distribution of changes in meteorological conditions when comparing situations with BrO enhancements to the mean.
The first parameter investigated is sea level pressure. Figure 3 shows
histograms of ERA-Interim sea level pressure data from 2008 to 2018 for the
Arctic and Antarctic sea ice regions during spring (March and April for the
Arctic; September and October for the Antarctic). The red line shows the
frequency distribution of sea level pressure for cases with enhanced total
BrO, whereas the black line represents the frequency distribution of the
mean field using all sea level pressure data for the study period. As can be
seen in Fig. 3, the frequency distribution of sea level pressure is shifted
towards lower sea level pressure during BrO enhancements in both polar
regions, indicating that enhancement in total BrO vertical column is related
to lower sea level pressure. This decrease of sea level pressure can also be
clearly seen from a comparison of the mean sea level pressure map for all
measurements with the mean sea level pressure map for the enhanced BrO cases
(see Fig. 4). When total BrO columns are enhanced, sea level pressure is
generally decreased by up to
Monthly sea level pressure for the mean field
As Fig. 4 but for the Antarctic in September
Our analysis using long-term datasets demonstrates that enhancements in total BrO columns are associated with negative sea level pressure anomalies in both the Arctic and Antarctic sea ice regions. The Arctic and Antarctic have fundamentally different geographical features. The Arctic is a frozen ocean surrounded by land, whereas Antarctica is a frozen continent surrounded by ocean, which leads to differences in the major synoptic pressure systems. The lower tropospheric circulation over the frozen Arctic ocean is dominated by high pressure systems over the continents, while that of the sea ice zone around Antarctica is driven by a strong low pressure belt (Jones et al., 2010; Screen et al., 2018). Although the atmospheric dynamic systems are different between the Arctic and Antarctic sea ice regions, sea level pressure is generally lower during the enhancement of BrO columns in both polar regions, which indicates that atmospheric depressions have an influence on generating enhanced total BrO.
Frequency distribution of surface air temperature for all data
(black line) and the enhanced BrO case (red line) in
Previous studies support the association between atmospheric low pressure systems and enhanced BrO columns. Jones et al. (2010) found that large-scale tropospheric ozone depletion events and enhanced BrO columns appear around large low pressure systems in the Antarctic region using data from tethersondes, free-flying ozonesondes, and satellites. Blechschmidt et al. (2016) revealed a link between polar cyclones and bromine explosion events development for a case over the Arctic. These studies argue that vertical lifting and high wind speeds in synoptic-scale atmospheric low pressure systems result in development of blowing snow, so that the BrO explosion reaction cycle occurs around the wind-blown brine-coated snow particles in tropospheric air. Convergence and ascent of air occur along fronts within low pressure systems. This convective process facilitates air masses potentially having reactive bromine source conditions at the ground to be transported to higher altitudes and cooled. Theys et al. (2009b) and Salawitch et al. (2010) showed that some of the enhancements in total BrO columns are associated with increases in stratospheric BrO due to a decrease of the tropopause height, coincident with low pressure systems. The details of how surface-level wind and tropopause height, which can be affected by changes in atmospheric pressure, are associated with the enhancement of the total BrO column will be discussed in later sections.
The relationship between air temperature at 2 m and the enhancements of
total BrO columns was investigated in the same way as that for pressure in the previous section. From Fig. 6, it can be seen that the surface-level air
temperature is low during enhancements of total BrO column in both the
Arctic and Antarctic sea ice regions. The air temperature frequency
distribution of the Antarctic is shifted more clearly towards lower
temperatures than that of the Arctic. The highest frequencies of 27 % in
the mean field and 32 % in the enhanced BrO field are detected in the
same temperature range of
Monthly surface air temperature for the mean field
As Fig. 7 but for the Antarctic in September
Temperature effects in the chemical mechanism of bromine release were
discussed in several previous studies. Sander et al. (2006) demonstrated why
bromine release is accelerated on cold saline surfaces using a one-dimensional
atmospheric chemistry model. They found that the acid-catalysed atmospheric
bromine explosion cycle is triggered in their simulations by precipitation
of carbonates at a temperature below 263 K, leading to reduced buffering
capacity of the alkaline sea water and facilitating its acidification. Model
calculations identified the strong temperature dependency of the equilibrium
reactions (Reactions R8 and R9) releasing Br
Frequency distribution of wind speed at 10 m for all data (black
line) and the enhanced BrO case (red line) in
Monthly wind speed at 10 m for the mean field
As Fig. 10 but for the Antarctic in September
Next, surface-level wind speed is investigated to evaluate how this may
affect the occurrence of total BrO column enhancements. Figure 9 shows the
frequency distribution of wind speed at 10 m for the average field and for
enhanced BrO cases of the 10 years of measurements in the Arctic and
Antarctic sea ice regions. The distribution is shifted towards high wind
speeds in both polar regions for enhanced total BrO vertical columns, with the
increase in wind speed being more pronounced in the Antarctic region. The
difference in wind speeds is also confirmed by the spatial distribution maps
(Figs. 10 and 11). Higher wind speeds are observed in most Arctic and
Antarctic regions for situations with enhanced total BrO columns compared to
the mean field. In particular, differences in wind speed of more than 5 m s
The relationship between surface wind speed and enhanced BrO column has been
discussed in previous studies and is still under debate. Some of the
related studies showed tropospheric ozone depletion events and bromine
explosion events at lower wind speeds of less than 8 m s
Spatial distributions of the relative frequency of high wind speeds during the enhanced BrO occurrences in the Arctic (
The spatial distribution map of surface wind speed anomalies derived in this
study shows that during the enhancement of total BrO vertical columns, wind
speeds are generally enhanced. However, the average wind speed field during
the high BrO cases shows values of 6–8 m s
In order to investigate if not only the wind speed but also the wind
direction affects the occurrence of BrO enhancement in terms of regions, the
relative frequency of wind direction for data with enhanced BrO columns was
mapped. The relative frequency was calculated for the case that the number
of data collected within the reference grid is greater than 20 to avoid
errors from using too small sample size. The wind direction was divided into
eight groups at intervals of 45
Relative frequency maps of surface-level wind direction for data with BrO enhancements over the Arctic during spring (March to April in 2008–2018). The frequency was calculated for the cases where the number of data points collected within the reference grid is greater than 20. The results are shown separately for different wind directions.
Spatial distributions of the differences in relative frequencies of wind direction between enhanced BrO cases and the mean field for the Arctic in spring (March to April in 2008–2018). The results are shown separately for different wind directions.
As Fig. 13 but for the Antarctic during spring (September to October in 2008–2018).
The spatial distribution of the wind direction frequency during the occurrence of enhanced total BrO columns is clearer in the Antarctic compared to the Arctic. As shown in Fig. 15, the northern winds show low relative frequency overall in the Antarctic sea ice region for BrO enhancement cases, which indicates that the northerly winds have a low association with BrO enhancements. This is consistent with our understanding of the conditions required for BrO enhancements because the northerly winds blowing from the open water to Antarctica are usually relatively warm, and thus sea salt aerosols are not cold enough to trigger the bromine explosion mechanism. Another feature of wind direction distribution in the Antarctic is that easterly winds are strongly related to enhanced BrO columns along the coast of Antarctica, whereas westerly winds are prevailing over the sea ice region except for the Antarctic coastlines. Basically, the main winds of the Antarctic can be divided into two types: (1) large-scale circulations composed of westerly winds and (2) local katabatic winds deflected in the cross-slope direction with an eastward component due to the Coriolis force. Consequently, the predominant wind direction during the BrO enhancement also follows the Antarctic large-scale circulation. Depending on the large-scale atmospheric system, easterly winds in the coastal regions and westerly winds in the sea ice regions are dominant, but the influence of the south wind increases in the presence of enhanced BrO columns. Maps of differences in the frequency of the wind direction between the high BrO situation and the mean field (see Fig. 16) clearly show that occurrence frequencies of wind directions from southerly and westerly directions increase during the occurrence of enhanced BrO columns. In particular, the southwestern winds prevail for situations with enhanced BrO columns, as frequencies increase by more than 20 % over a large area of Antarctic sea ice including the marginal ice zones.
As Fig. 14 but for the Antarctic during spring (September to October in 2008–2018).
We investigated further the impact of high wind speed on the total BrO
column enhancement at each wind direction in the Arctic and Antarctic
regions during spring. Table 2 summarizes the relative frequency of wind
direction for the mean field, for the enhanced BrO cases and for enhanced
BrO cases accompanied by high wind speeds (
Relative frequency (expressed in %) of surface-level wind directions for all data, the enhanced BrO case, and the enhanced BrO with high wind speeds for the Arctic and Antarctic in spring.
The last factor investigated in connection with the enhancement of total BrO
columns is the tropopause height. The relationship between the tropopause
height and BrO hotspots observed from satellites was discussed in previous
studies using data for periods from several days up to 2 years. Salawitch
et al. (2010) found that enhanced total BrO columns over the Hudson Bay
observed by OMI are coincident with a low tropopause of
In this study, the relationship between tropopause height and total BrO column enhancements is investigated in both the Arctic and Antarctic sea ice regions using 10 years of long-term data in terms of magnitude, region, and time, and the results are compared with those from previous studies. The relative frequency distributions of tropopause height for the mean field and the cases with enhanced total BrO columns in the Arctic and Antarctic are presented in Fig. 17. In both polar regions, the frequency distribution is shifted towards lower tropopause height during the occurrence of enhanced total BrO columns, and this effect is larger in the Antarctic than in the Arctic. This indicates that the decrease of tropopause heights is associated with the enhancement of total BrO columns, which is consistent with previous study results. Our results also show that the enhancement of total BrO columns due to tropopause descends is more prominent in the Antarctic than in the Arctic. This conclusion is different from the finding of Theys et al. (2011), who reported that the effect of low tropopause height on the increase of BrO column seems to be more important in the Arctic than it is in the Antarctic region. One possible explanation for the difference apart from the different time periods investigated is that here the frequency distribution of the tropopause height was investigated for situations with enhanced total BrO column, whereas Theys et al. (2011) investigated the impact of tropopause height on the tropospheric BrO column.
Frequency distribution of tropopause height for all data (black
line) and the enhanced BrO case (red line) in
Monthly tropopause height for the mean field
As Fig. 18 but for the Antarctic in September
In addition to the frequency distribution of tropopause height, the
differences in maps of tropopause height between situations with enhanced
total BrO column and the mean were investigated (Figs. 18 and 19). The mean
field of tropopause height shows generally a tropopause height of 9.0–9.5 km range over the Arctic except for the north coast of Canada (e.g. Canadian archipelago), Fram Strait, and the Barents Sea where slightly lower
tropopause heights of 8.0–8.5 km appear. For the BrO enhancement cases,
tropopauses are lower in the 7.5–9.0 km range in most areas. In particular,
large differences in tropopause height of
To assess statistically the dependence between total BrO vertical column and multiple meteorological parameters investigated in the previous sections, we performed a Spearman rank correlation analysis. The Spearman correlation determines the strength and direction of the monotonic relationship between two variables. Thus, this method is less sensitive to strong outliers and distribution type than the Pearson correlation, which is the reason for using the Spearman rank correlation analysis in this study.
Monthly Spearman correlation coefficients (
Spearman rank correlation coefficients between total BrO VCDs and
four meteorological parameters. The results are shown separately for
different months in spring in the Arctic and Antarctic. Note that all
Spearman rank correlation coefficients are significant (
The regional differences of the statistical dependence between total BrO
vertical column and meteorological factors were also investigated by
performing the Spearman correlation analysis for each cell of the reference
grid. Figure 20 displays the spatial distribution of Spearman correlation
coefficients between total BrO vertical column and each meteorological
factor where the
Spatial distributions of the Spearman correlation coefficients between total BrO VCD and four meteorological parameters (sea level pressure, surface air temperature, wind speed at 10 m, and tropopause height) for the Arctic and Antarctic in spring. Only Spearman correlation coefficients with
Surface-level air temperature also shows different Spearman correlation
patterns depending on the region. The correlation between total BrO vertical
column and temperature is negative at relatively lower latitudes of the
Arctic and then turns to positive values over the central Arctic region.
This indicates that an increase of surface air temperature is related to the
enhancement of total BrO vertical column over the central Arctic, while a
decrease of temperature is associated with the increase of total BrO column
in most areas except for the central Arctic. In contrast to the Arctic,
opposite correlation patterns between total BrO column and surface air
temperature as a function of latitude are not detected in the Antarctic sea
ice region. Correlations are not very significant around the Antarctic
coastal region and negative correlation coefficients of
Spatially, sea level pressure is negatively correlated with total BrO
vertical column in most regions of the Arctic and Antarctic. However, sea
level pressure has weak negative correlations with total BrO column compared
to the values found for tropopause height and surface air temperature. Wind
speed is positively correlated with total BrO vertical column in most areas
of both polar sea ice, but the correlation is very weak with
Cross correlations of meteorological parameters for the enhanced total BrO cases in the Arctic and Antarctic sea ice regions.
The correlation analysis using the long-term dataset improves our understanding of the influence of meteorological factors on total BrO vertical columns in terms of the region and the time of year (month). The strongest influence of the different parameters on total BrO column, having a large correlation coefficient, is tropopause height, which indicates that the stratospheric contribution is significant in the total BrO column density variations. This is presumably because low tropopause heights result in a larger contribution of stratospheric BrO column to the total column. Consequently, accurate stratospheric correction is important in studying the tropospheric bromine explosion events and estimating tropospheric BrO content from the measured total BrO columns. Among the surface-level meteorological parameters, air temperature impacts on the total BrO column density, arguably because temperature is an important factor in the chemical mechanism of reactive bromine release in the lower atmospheric layer. Sea level pressure and surface-level wind speed are negatively and positively correlated with the total BrO vertical column density, but correlations are lower than for surface air temperature, which indicates that their influence on total BrO column is not as large as that of surface air temperature.
The relationship between individual meteorological parameters and the total BrO vertical column investigated above illustrates how each meteorological parameter is linked to BrO variations in terms of temporal and spatial distribution. However, since meteorological parameters are not independent of each other and vary systematically in general, cross relationships between meteorological parameters affecting directly or indirectly BrO variations should also be considered. For example, Yang et al. (2019) showed that the sea salt aerosol (SSA) production affecting the enhancement of BrO at the tropospheric level is proportional to the sublimation flux of blowing snow which is a complex function of various meteorological parameters including surface wind speed, temperature and relative humidity. Also, Zhao et al. (2016) and Blechschmidt et al. (2016) showed that large-scale enhanced BrO plumes over the Beaufort Sea are associated with weather systems which change the various relevant meteorological parameters together. They also demonstrated that the size and lifetime of BrO plumes depend on the development stage of the weather system. Therefore, cross correlations between meteorological parameters for those data having enhanced total BrO columns were investigated (see Table 4). During the occurrence of enhanced total BrO, sea level pressure has a negative correlation with surface-level temperature and wind speed, while it has a positive correlation with tropopause height. For example, the development of a low pressure system during the enhancement of BrO columns may correlate with a decrease in tropopause height as well as an increases in surface-level air temperature and wind speed. Although the correlation coefficients found are not large, results show that sea level pressure is linked with both surface-level meteorological conditions and the tropopause height which can account for stratospheric dynamics. Indeed, pressure systems which usually evolve due to interactions of temperature differences in the atmosphere derive directly the airflow motion within the troposphere and also may affect the tropopause height in relation to the convergence or divergence of air masses. It is also interesting to note from Table 4 that the tropopause height has insignificant correlations with surface-level meteorological parameters during the BrO enhancements, except for the air temperature in the Arctic, which is predictable since the tropopause height is a factor more closely related to stratospheric dynamics compared to the surface-level weather system.
All analyses so far were correlating meteorological parameters and BrO
enhancements for the same time step. In order to investigate possible
time-lagging effects of meteorological conditions on total BrO VCDs,
correlations between total BrO VCD and meteorological parameters were
performed with several days' lag for each grid cell. In general, signs of
correlations between meteorological parameters and total BrO VCDs are not
changed with
Box and whisker plots of the time-lagged correlation between the total BrO VCD and each meteorological parameter
As Fig. 21 but for the Antarctic sea ice region.
Bromine monoxide is located in both the stratosphere and the troposphere, and large-scale BrO column enhancement is frequently observed in the Arctic and Antarctic sea ice regions during polar springtime by satellites. In this study, we analysed 10 years of GOME-2 total BrO columns and various meteorological parameters to establish statistical connections between where the enhanced BrO columns mainly appear and the underlying meteorological conditions. The occurrence of enhanced total BrO columns showed regional characteristics. Relatively high occurrence frequencies are detected over the north Canadian coast, the Hudson Bay, and the East Siberian Sea in the Arctic, while in the Antarctic, enhanced BrO columns are often observed across the Weddell and Ross seas, especially in September. The occurrence frequency of enhanced total BrO columns showed more spatial variation in the Arctic, whereas it varied more temporally in the Antarctic region.
Several meteorological parameters such as sea level pressure, surface-level
air temperature, wind speed and direction, and tropopause height were
investigated to assess any significant relationships with the occurrence of
enhanced total BrO columns. If the mean meteorological conditions are
compared with those during enhanced BrO events, the latter are associated
with low pressure systems, cold air temperature, high surface wind speed, and
a decrease of tropopause height in both the Arctic and Antarctic sea ice
regions. Low pressure systems can drive vertical uplifting and high wind
speeds, lifting considerable amounts of saline snow or aerosols acting as a
source of reactive bromine in the troposphere. In the case of temperature,
surface air temperature is clearly lower during high BrO events in most of
both polar regions, but it is slightly higher over the central Arctic and
the Antarctic coastal regions. The slight positive temperature anomalies in
the central Arctic region during BrO events may be influenced by transport
of BrO-rich air with the southern wind blowing from the sea ice margins
where temperatures are relatively higher. Surface wind speed is generally
higher during BrO column enhancements, and in particular, high surface wind
speed above 12 m s
We also performed a Spearman rank correlation analysis between total BrO vertical column density and meteorological factors to assess the relevance of these factors for total BrO column enhancements. Total BrO vertical column density has the strongest negative correlation with tropopause height in both the Arctic and Antarctic regions, reflecting both the importance of contributions in the stratospheric BrO column for the total BrO column and of the link between low tropopause height and tropospheric conditions required for bromine explosions. The next most statistically significant factor is surface air temperature. One remarkable point is that the temperature is negatively correlated with total BrO column in most sea ice regions but has a positive correlation over the central Arctic, which is the same as the result of the temperature anomaly pattern discussed above. Sea level pressure and surface-level wind speed are negatively and positively correlated with the total BrO vertical column density, respectively, but their correlation coefficients are low and the strengths of the relationships are weak. Regarding the time-lagging effects of meteorological factors on total BrO VCDs, the strongest correlation for the tropopause height and surface air temperature is found without the time lag, whereas sea level pressure and surface wind speed have larger correlations with negative time lags. These negative time-lag correlations between total BrO VCD and sea level pressure as well as surface wind speed indicate that the enhanced BrO plumes in polar regions are likely transported and are clearly synoptic system related.
This study has focused on a statistical analysis of a large set of GOME-2 BrO total columns to derive spatial and temporal patterns of links between meteorological conditions and the occurrence of enhanced BrO columns. The results show systematic connections between all of the parameters studied and BrO enhancements. However, such links do not necessarily constitute a cause-and-effect relationship, in particular as quantities such as surface pressure, wind speed, temperature, and tropopause height are closely linked to each other. Another important aspect not covered by the approach of this study is the transport of air masses with enhanced BrO levels away from the region of initial bromine activation – in such cases, the correlation between meteorological parameters such as high wind speed and elevated BrO is not linked to the initial bromine release mechanism. In future studies, other parameters such as sea ice type, snow cover, or the presence of polynyas should also be included and ideally an optimal stratospheric correction applied to the BrO columns to better focus on tropospheric bromine explosions.
The GOME-2 BrO retrieval data are available upon request (contact persons are Sora Seo and Andreas Richter). The ECMWF ERA-Interim reanalysis data were provided by the ECMWF through their website (
SS carried out the GOME-2 BrO retrievals, collected the model meteorological data, performed the analysis, and wrote the paper. AR, JPB, AMB, and IB provided helpful ideas and feedback in designing the study. AR developed the DOAS retrieval code and supported the satellite retrieval. JPB, AMB, and IB supported data interpretation. All co-authors contributed to the writing of the paper.
The authors declare that they have no conflict of interest.
We gratefully acknowledge the funding by the Deutsche Forschungsgemeinschaft
(DFG, German Research Foundation) – project no. 268020496 – TRR 172,
within the Transregional Collaborative Research Center “ArctiC Amplification: Climate Relevant Atmospheric and SurfaCe Processes, and
Feedback Mechanisms (AC)
The article processing charges for this open-access publication were covered by the University of Bremen.
This paper was edited by Michel Van Roozendael and reviewed by three anonymous referees.