Observations and the source investigations of boundary layer BrO in Ny-Ålesund Arctic

Bromine monoxide is a reactive halogen species which has crucial impact on the chemistry of the tropospheric polar boundary layer. During polar spring, BrO enhancement can be detected in both northern and southern Polar Regions, while the boundary layer ozone depletion events occur. A considerable challenge for 10 understanding enhanced BrO and the associated ODEs is the difficulty of real-time observations. In this study, a typical process of enhanced bromine and depleted ozone in late April, 2015 at Ny-Ålesund boundary layer was observed using ground-based Multi Axis-Differential Optical Absorption Spectroscopy (MAX-DOAS) technique. The results showed that there were as high as 8×10 14 molecular cm -2 BrO slant columns above the Kings Bay area in 26 April. Considering meteorology, sea ice distribution and air mass history, the floating sea ice in the Kings Bay area was considered as the 15 major source of this bromine enhancement event. During this period, the boundary layer ozone and gaseous elemental mercury (GEM) was synchronously reduced by 85% and 90% separately. The kinetic calculation showed that the ozone loss rate is 10.3 ppbv h -1 , which is extremely high compared to other area. The GEM loss rate is about 0.25 ng m -3 h -1 . The oxidized GEM may directly deposit to snow/ice and thereby influence the polar ecosystem.

Bromide from sea salt aerosol is a sufficiently strong bromine source to explain the rate of ozone destruction or observed BrO enhancements.Sea-ice (first year) surfaces, brine, and frost flowers have been considered as possible former of bromide aerosols (Kaleschke et al., 2004).The reactive bromine species (Br and BrO radicals) are released via a series of photochemical and heterogeneous reactions, known as the "bromine explosion" (Platt and Hönninger, 2003).Moreover, bromine-rich air masses can also be transported over land by monsoon or air turbulence.However, since it is difficult to make detailed chemical observations in source area, direct evidence of bromine source 5 involvement is sparse.One recent study has shown rapid ozone depletion (~7h) in the marginal sea ice zone, which has been interpreted as due to in-situ chemical loss (Jacobi et al., 2006).
Briefly, the chemical ozone depletion process is driven by the BrO cycling reaction in the PBL (Platt and Hönninger, 2003).A typical heterogeneous reaction model between gaseous and condensed phases was shown in Fig. 1.Bromine is released from salty ice surfaces to the atmosphere in an autocatalytic chemical mechanism that oxidizes bromide to 10 reactive bromine.The reaction of HOBr in aerosol is proposed to be the pivot to explain the recycling reaction, which is an acid-catalyzed reaction (Simpson et al., 2007b).
To understanding these processes，optical remote sensing techniques are employed to detect BrO in the PBL relying on their characteristic narrow band absorption structures at UV and visible wavelengths.Ground-based multi-axis DOAS (Differential Optical Absorption Spectrometer) technique has the advantage of being able to separate clearly the 15 tropospheric and stratospheric portions of the atmospheric column, and even derive a crude vertical profile (Frieß et al., 2011).When pointing to a direction slightly above the horizon, high sensitivities for the trace gases close to the ground can be obtained due to the long light path through the trace gas layers.It is also an important calibration of satellite observations, which usually underestimates the tropospheric trace gases concentrations (Platt and Wagner, 1998).
In the Arctic area, ground-based observations have been made at Barrow, Alaska (71°N, 157°W), Alert, northern 20 Canada (82.5°N, 62.3°W) and Ny-Ålesund, Svalbard (Tab.1).Long-path DOAS measurement provides regional determination of BrO in PBL, but few ground-based MAX-DOAS measurement of BrO has been performed in Ny-Ålesund.One of the reasons is that influenced by the North Atlantic Warm Current (NAWC), the near surface air temperature and sea surface temperature (SST) of East Greenland and North of Europe are relatively high (Fig. 2).
The sea ice concentration has dropped severely during Arctic spring.Some areas even have no sea ice at all.In recent 25 years, Kings Bay of Ny-Ålesund has ice-free open water all year round, which is an unique character comparing with other parts at the same latitude in Arctic.The modeling study indicated that sea salt surface (brine, dry salt, or possibly frost flowers) on sea ice is one of the most important prerequisites for a polar bromine explosion (Lehrer et al., 2004).
Therefore, high level of troposphere BrO can be detected much more frequently in East Arctic (coastal area of north Asia and North America) because of widely distributed sea ice.30 In this study, we have discovered a typical process of enhanced bromine and depleted ozone in Ny-Ålesund in late April.The key role of bromine was confirmed by ground-based MAX-DOAS measurements of BrO.This event Furthermore, the mechanisms and environment implications of ozone depletion and gaseous mercury deposition are discussed.

Instrument setup 5
The MAX-DOAS measurement site is located at Yellow River Station (78°55´30"N, 11°55´20"E) at Ny-Ålesund, west coast of Spitsbergen.The observation position is shown in Fig. 3. To have a rough idea of the climate condition, monthly mean sea ice concentrations anomalies and air temperature anomalies in April 2015 are demonstrated in Fig. 2.
The observations were carried out from 24 April to 15 May 2015.Due to the wavelength adjustment, no data is available from 29 to 30 April. 10 The MAX-DOAS instrument operated at Ny-Ålesund consists of indoor and outdoor parts.The telescope receiving scattered sunlight from multi angles is controlled by a stepper motor to adjust elevation angles from horizon (0°) to zenith (90°).The field of view of the telescope is about 1°.The scattered sun light is imported through the quartz fiber with numerical aperture of 0.22 into the indoor spectrograph (Ocean Optics MAYA pro) with a one dimensional CCD (ILX511 linear array CCD) containing 2048 pixels.The wavelength range of the spectrograph is from 290 nm to 420 15 nm, thus enabling the analysis of trace gases including O 3 , NO 2 , BrO, HCHO, and O 4 .The spectral resolution is about 0.5nm (FWHM).The CCD detector is cooled at -30°C while the whole spectrometer is thermally stabilized at +20°C using a thermal controller.A computer sets the configuration of the system and controls the automatic measurements.
The integration time of each measurement depends on the intensity of scattered light which can be influenced by cloud and visibility.The standard mercury lamp is used for spectra calibration.Calibration measurements of dark current and 20 offset are performed after each measurement.
Noontime zenith sky measurements from 26 April is chosen as Fraunhofer reference for the analysis.The O 4 retrieval is performed using the same set of cross sections as for BrO but in the wavelength interval at 340-370 nm.The high resolution cross sections were convoluted with the instrument slit function determined by measuring the emission line 5 of a mercury lamp.A fifth order of polynomial was applied to eliminate the broad band structures in the spectra caused by Rayleigh and Mie scattering.Furthermore, a nonlinear intensity offset was included in the fit to account for possible instrumental stray light.A wavelength shift and stretch of the spectra was allowed in the fit in order to compensate for small changes in the spectral adjustment of the spectrograph.
The fit procedure yields differential slant column densities (dSCD) for the target gas.An example of the fit result of 10 BrO and O 4 is shown in Fig. 4. The spectrum was recorded on 26 April, 2015 19:59 UTC (SZA=86°) at the elevation angle of 2°.The BrO dSCD is 5.10×10 14 molecular cm -2 .The residual root mean square is 4.59×10 -4 , resulting in a statistical BrO dSCD error of 1.63×10 13 molecular cm -2 .
Since SCD is dependent on the light path, wavelength and observation geometry, SCD is then converted to vertical column density (VCD) by dividing the air mass factor (AMF), which is the averaged light path enhancement for solar 15 light traveling through the atmosphere compared to a straight vertical path.
According to Sinreich et al (Sinreich et al., 2013) The near-surface O 4 concentration (c O4 ) is calculated to be 2.8×10 37 molecules 2 cm −6 using local atmospheric temperature and air pressure, which is proportional to the square of the O 2 concentration.
Considering the wavelength dependence of Rayleigh and aerosol scattering, the effective light path systematically varies with the wavelength.Since the light path length based on O 4 dSCD is obtained at 360 nm, extrapolating L to other wavelength ranges for retrieving trace gases such as BrO (350 nm) is an important aspect.The relationships 25 between the light path lengths at 350 nm to those at 360 nm were studied using SCIATRAN for a variety of aerosol scenarios.The effective light path (L eff ) can be approximately calculated as: Thereby BrO SCDs can be converted to near-surface volume mixing ratios (VMR) by dividing effective light path Surface ozone was measured by UV photometry.Gaseous mercury in air was measured using Tekran mercury detector.
Hourly Surface Ozone and gaseous mercury data are downloaded at EBAS database (Tørseth et al., 2012).

Results
Time series of BrO dSCDs at 2°, surface ozone concentrations (from GAW station, Zeppelin), solar zenith angle (SZA), air pressure, air temperature, relative humidity, wind velocity and wind direction from 24 April to 15 May are 20 presented in Fig. 5. Owing to the increase in stratospheric air mass factor at twilight, U-shaped diurnal variation of the BrO dSCDs are derived when BrO is mostly located in the stratosphere.From 26 to 28 April (high light blue area in Fig. 5), BrO dSCDs clearly exceeded the background levels and peaked at 8×10 14 molecular cm -2 in the evening of 26 April.At the same period, severe ODEs occurred with surface ozone sharply decreasing to several ppb and not recovered until BrO had gradually dropped to background levels in 28 April.The occurrence of depleted troposphere ozone and enhanced BrO appears to be unpredictable in May.Partial ozone (not to near zero level) was depleted in the absence of BrO.km under cloud-free conditions, much longer than LP-DOAS (1-3 km) in previous researches (Frieß et al., 2011).The light path increases to more than 22 km in 27 April midnight due to the presence of cloud, which is consistent with the global radiation.The BrO VMR is about 15 pptv during the ODE (Fig. 6e).
BrO VCD from GOME-2 overpass data (Fig. 6f) shows that no obvious BrO enhancement has been detected, which indicates that BrO at Arctic BL is underestimated by satellite-based measurements.Thereby, ground-based 5 observations in this study are crucial complements, which provide calibration basis for the satellite observations.In order to investigate the dynamical and chemical processes affecting bromine chemistry, BrO slant columns at different elevation angles are demonstrated in Fig. 7, which have contributed to better understanding of the vertical distribution of reactive bromine at Arctic boundary layer.The retrieval presents that BrO dSCDs are higher at low elevation angles than those at high elevation angles.Furthermore, results of different elevation angles distinguish 10 obviously, especially during the BrO enhancement period, which implies that BrO is generated from the sea surface.
As a consequence, there is sufficient reactive bromine present locally in the boundary layer to explain the extremely fast O 3 depletion and mercury precipitation simultaneously (Fig. 6h, 6i).

Discussions
In this research, high concentration of troposphere BrO has been detected using the ground-based MAX-DOAS 15 technique.As high as 8×10 14 molecular cm -2 BrO column has been detected above Kings Bay, Ny-Ålesund.The retrieval shows that the enhancement occurred accompanied with severe ozone depletion and mercury precipitations.
The maximum BrO is located at sea surface during the ODE.This enhancement event is a good opportunity to investigate the source of BrO and the impact on the environment of Arctic boundary layer.The following parts are discussed in detail from air mass history, sea ice distribution, and ozone loss and mercury precipitation.20

History of air masses
From the map of GOME-2 measurements from 22 April to 27 April 2015 (Fig. 8), several troposphere BrO clouds existed at coastal North America and Chukchi Sea on 22 April 2015 and gradually diffused.But from GOME2 overpass data at Ny-Ålesund, no apparent enhancement of BrO has been detected on 26 April.Moreover, long-range transport of air masses (Fig. 9) did not show significant change during the fast ODE period.72-hour backward 25 trajectories at Ny-Ålesund 10 and 3000 meters a.s.l.from 26 April (0600 UTC) to 27 April (1800 UTC) were analyzed.
The trajectories followed similar pathways, which indicate a stable circulation pattern for the period before, during, and after the BrO enhancement.Therefore it means that the observed BrO enhancement event is not due to the long-range transport of BrO precursors.

Sea ice distribution
Sea water is the primary source of the boundary layer halogens (Mcconnell et al., 1992).The sea ices provide highly concentrated saline surfaces, which have been reported to play a pivotal role in for bromine activation (Saiz-Lopez and von Glasow, 2012).According to the observation of sea ice concentration from AMSR-E and Zeppelin webcam, Kings Bay is ice-free water area during the measurement period (Fig. 2).However, large amount of sea ice appeared at Kings 5 Bay on 26 April (Fig. 10), floating from the open water by both wind and tidal effect.
The efficient ozone loss is consistent with the temperature decline (Fig. 11).The meteorology data shows that on 26 April air temperature continually goes down and reaches bottom of -11.4 o C at 22:00.According to the precipitation curve of calcium carbonate, large amount of carbonate precipitates below 265K.This process will provide acid aerosol from alkaline sea water, which triggers the transformation of inert sea-salt bromide to reactive bromine (Sander et al., 10 2006).Although the sun radiation intensity is not strong at that time, the heterogeneous reactions can still happen under the twilight.
Thereby, this BrO enhancement event is a local process, mainly influenced by underlying surface change and local environment.The surface ozone concentrations increased along with the melting of sea ice, which indicated that the life span of BrO radicals are very short.When sea ice disappeared, the reaction immediately ended and reactive 15 bromine radicals gradually transformed to soluble bromide (e.g.HOBr), which explained the sink of it (Fan and Jacob, 1992).

Kinetic analysis of O 3 loss
What makes this case very special is that the increasing rate of BrO and the depletion rate of boundary layer ozone are really fast.The surface ozone reduced by 85% within 4 hours.The ozone loss rate is as high as 10.3 ppbv h -1 or 248 20 ppbv d -1 , which is extremely high compared with previous studies in Polar Regions (Tab.2).
The chemical kinetics of bromine enhancement and ozone decay are analyzed assuming that the catalytic reactions are dominated by reactions showed in Fig. 1.A first-order loss of ozone is due to reaction Br+O 3 BrO+O 2 resulting in the rate law: (Eq. 4) 25 [ 3 ] 0 is the ozone concentration at the beginning of decay determined from the measured mixing ratio of 74.72 ppbv.According to the method by Jacobi et al. (Jacobi et al., 2006), the first order rate constant k 1 can be determined as follows: d(ln The measured decrease of ln versus time was fitted by: ) versus time are plotted as black dots in Fig. 12a.The coefficients a and b are obtained from the linear fit in plot.10 The ozone loss begins relatively slow and accelerates with time, which is consistent with the process of bromine explosion.
Assuming that the first order decay is dominated by the reaction Br+O 3 BrO+O 2 , we are able to calculate the Br concentrations as follows: (Eq. 12) 15 is a constant depending on temperature (Fig. 12b).Thereby, the calculated Br concentration increases from 1.1×10 7 to about 1.2×10 9 atoms cm -3 (corresponding to 44.8 pptv) (Fig. 12c).Considering the assumption that the halogens are homogenously distributed in the PBL, the concentrations of Br at sea surface layer in the bromine explosion could be even higher.20

Gaseous mercury precipitation
During ODEs, halogens efficiently oxidize gas-phase mercury and cause it to be transferred from the atmosphere to the snow, called "atmospheric mercury depletion events (AMDEs)".Since 1995, year-round monitoring of atmospheric Hg has been performed at Alert, Canada (Schroeder et al., 1998).Dynamic studies demonstrated that bromine is the major oxidant depleting Hg in the atmosphere (Ariya et al., 2002;Ariya et al., 2004;Lindberg et al., 2002;Lu et al., 25 2001;Steffen et al., 2008).
In this case, the precipitation of gaseous mercury occurred concurrently with tropospheric ozone depletion, as well as the enhancement of BrO (Fig. 13), which indicated that the oxidation of GEM by reactive halogen species (Br atoms and BrO radicals) is considered to be the key process of mercury depletion.The GEM decreases from ~2 ng m -3 to lower than 0.3 ng m -3 during the BrO enhancement event.The mercury loss rate is about ~0.25 ng m -3 h -1 or 6 ng m -3 d -1 .The oxidized GEM may directly deposit to snow/ice or associate with particles in the air that can subsequently deposit onto the snow and ice surfaces, and thereby threaten polar ecosystems and human health.

5
Typical process of enhanced bromine and depleted ozone in Ny-Ålesund boundary layer was observed using ground based MAX-DOAS techniques in late April, 2015.As high as 8×10 14 molecular cm -2 BrO slant columns were detected on 26-27 April.Meanwhile, severe ozone depletion and mercury precipitation occurred under BrO VMR of 15 pptv.
The distribution showed higher BrO concentrations near sea surface.By analyzing the air mass history and sea ice conditions, this BrO enhancement is a local process.The underlying sea ice and low temperature provide acid aerosols, 10 which are prerequisites for the formation of BrO radicals.The kinetic analysis shows that the ozone loss begins relatively slow and accelerates with time, which is consistent with the process of bromine explosion.The ozone loss rate is as high as 10.3 ppbv h -1 , which is much higher than previous studies in Polar Regions.GEM loss rate is about Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2017-546Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 23 August 2017 c Author(s) 2017.CC BY 4.0 License.provides a rare opportunity to investigate the source of bromine and process of ozone depletion at this area.
, the effective light path length (L) in the boundary layer can be determined by the O 4 dSCD for low elevation angle.The diurnal variations of the clear-sky AMFs for BrO are in accordance with O 4 in the boundary layer.
(L eff ): 30  =    (Eq. 3) Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2017-546Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 23 August 2017 c Author(s) 2017.CC BY 4.0 License.2.3 Complementary data Ny-Ålesund is a science community hosting over fifteen permanent research stations.Atmospheric measurements have been measured continuously at Zeppelin Station, Ny-Ålesund since 1990.Located on Zeppelin Mountain with altitude 474 m a.s.l., it is a background atmosphere observatory operated by NPI (Norwegian Polar Institute) and NILU (Norwegian Institute for Air Research) , which are part of the Global Atmosphere Watch (GAW) Framework.5 Meteorology data including temperature, air pressure, relative humidity, wind direction and velocity, and global radiation are recorded by AWIPEV Atmospheric Observatory in Ny-Ålesund.Webcam on the 474m Zeppelin Mountain records the sea ice change of Kings Bay and the cloud situation of 10 Ny-Ålesund.(https://data.npolar.no/_file/zeppelin/camera/)In order to get a rough idea of BrO distribution, station overpass BrO vertical column densities for MetOp-A (GOME-2A) and MetOp-B (GOME-2B) in Ny-Ålesund, Arctic are downloaded from https://avdc.gsfc.nasa.gov.Using Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model via the NASA ARL READY website (http://www.ready.noaa.gov/ready/open/hysplit4.html)(Draxler and Rolph, 2013;Stein et al., 2015), back 15 trajectory analyses were carried out to find the history of air masses.72-hours ensemble back trajectories were driven by meteorological fields from the NCEP Global Data Assimilation System (GDAS) model output.

Fig. 13
Fig.13 Time series of dBrO/dt, dO3/dt and dHg/dt during the BrO enhancement event