Measurements of OH and RO 2 radicals at Dome C , East Antarctica

Introduction Conclusions References


Introduction
Atmospheric chemistry in polar regions has gained growing interest over the two last decades due to the discovery of surprisingly high photochemical activity in both Antarctic and Arctic, i.e. at the South Pole (Mauldin et al., 2001;Davis et Published by Copernicus Publications on behalf of the European Geosciences Union.al., 2001) and at Summit, Greenland (Honrath et al., 1999;Sjostedt et al., 2007).The photochemistry of the boundary layer atmosphere (BL) at these snow covered regions is significantly influenced by the emissions of reactive gases from the snowpack.The emissions are produced by the interaction of solar radiation and photochemically active species in the snowpack (e.g.Grannas et al., 2007 for a review).
At the South Pole (SP) unexpectedly large concentrations of OH radicals in the boundary layer, about 2 ⇥ 10 6 molecule cm 3 , were observed for the first time during ISCAT 1998 (Mauldin et al., 2001) and were confirmed in later campaigns (ISCAT 2000, Mauldin et al., 2004;ANTCI 2003, Mauldin et al., 2010).The elevated concentrations of OH radicals were explained by fast recycling of OH radicals from peroxy radicals in presence of large concentrations of NO.The large concentration of NO molecules exceeded free tropospheric concentrations (Davis et al., 2001) and was attributed to the release of NO x from snowpack (following UV photolysis of the nitrate anion (NO 3 ) on/in snow grains; Jones et al., 2001) and its accumulation in a stable and shallow BL at the SP (Davis et al., 2001(Davis et al., , 2004(Davis et al., , 2008)).The build up of large concentrations of NO x at the SP was suggested to be enhanced by the continuous sunlight during summer and the location at the bottom of a large air drainage basin (Davis et al., 2004(Davis et al., , 2008)).
At the SP the major gas-phase net sources of OH and HO 2 radicals were found to be the photolysis of HCHO and H 2 O 2 .The production of OH radical from O 3 photolysis was found to be less important owing to the small concentration of water at such low temperatures.The photochemical box model for the SP tends to slightly overpredict the concentrations of OH compared to observations, although satisfactorily reproduces the dependence of the concentration of OH on the concentration of NO (Chen et al., 2004;Mauldin et al., 2004).It has to be emphasized that when considering the mixing ratio of ⇠ 30 pptv of HONO measured at the SP by using mist chamber/ion chromatography method (Dibb et al., 2004) the model overpredicts concentrations of OH radicals at the SP by a factor of 3 to 5 (Chen et al., 2004).It was proposed that the HONO concentrations derived with the mist chamber/ion chromatography method during these campaigns were probably biased by some systematic error (e.g. chemical interference) (Chen et al., 2004).
Conditions favouring the accumulation of large concentrations of NO x were thought to be specific to the South Pole (24 h sunlight and shallow stable boundary layer), however a larger part of the Antarctic Plateau also experiences elevated concentrations of NO x despite strongly reduced solar radiation at night and the resulting build up of an unstable convective atmospheric boundary layer.This is supported by airborne observations of large concentrations of NO (Davis et al., 2008;Slusher et al., 2010;Wang et al., 2007) showing that a shallow photochemically active layer is common for large part of Antarctic Plateau atmosphere.However, these airborne measurements of NO x remain limited in time and space.In addition, although the elevated concentrations of NO x represent the key factor in the build up of large OH concentrations in the polar BL, the relationship of concentrations of OH with concentrations of NO is non-linear (Mauldin et al., 2004).Thus, an accurate prediction of the concentration of OH radical in other Antarctic regions may be difficult without detailed information on the radical primary production and net losses.These uncertainties in part motivated the OPALE (Oxidant Production in Antarctic Lands and Export) project aiming at a characterization of the oxidative capacity of the BL atmosphere in the region of East Antarctica.
In addition to the monitoring of surface ozone, that was initiated in 2007 at the Concordia station on the East Antarctic plateau (Legrand et al., 2009), two others studies related to the BL photochemistry have been performed at that site prior to the OPALE campaign: the measurements of gas-phase NO concentrations and, for the first time on the Antarctic Plateau, measurements of gas-phase NO 2 concentrations were conducted during summer 2009/2010 by Frey et al. (2013).Frey et al. (2013) revealed that concentrations of NO at Dome C are comparable to those observed at the SP but the ratios [NO 2 ] / [NO] were found to be significantly larger than calculated, assuming photostationary-state conditions with measured concentrations of O 3 and NO.As a possible explanation Frey et al. (2013) suggested significantly larger peroxy radical concentrations at Dome C than at the SP.The study of Frey et al. (2013) also revealed a distinct diurnal cycle of NO x that has been attributed to boundary layer stability and the flux of NO x released from the snowpack both driven by diel changes in solar radiation (France et al., 2011).The second study was dedicated to the measurements of concentrations of HONO during the summer 2010/2011 (Kerbrat et al., 2012) with a long-path absorption photometry (LOPAP) technique.Kerbrat et al. (2012) concluded that the observed large mixing ratio (⇠ 30 pptv) of HONO would require an unexpectedly large photochemical source of HONO from the snowpack, if the measurements of gas-phase HONO were not biased due to an unknown interference.
In the framework of the OPALE project, aimed at characterization of the oxidative capacity of the atmosphere in the region of East Antarctica (Preunkert el al., 2012), a field campaign was carried out at the top of the high Antarctic Plateau, at Concordia station, from December 2011 to January 2012.The first results of this campaign are presented in several publications accompanying this article, including (Frey et al., 2014) for NO and NO 2 , (Legrand et al., 2014) for HONO, and (Preunkert et al., 2014)   surface meteorological parameters, physics of the boundary layer and photolysis rates.

Methods
Measurements of atmospheric concentration of OH and RO 2 (the sum of hydroperoxy HO 2 and organic peroxy radicals) radicals were conducted from 19 December 2011 to 9 January 2012 at Concordia station (Dome C, 75.10 S, 123.31 E, altitude 3233 m).
The weather at Concordia is dominated by weak katabatic winds (⇠ 3 m s 1 ) and clear sky conditions with frequent presence of elevated cirrus clouds.Daylight lasts 24 h with a significant diurnal variation of solar radiation (as shown in Fig. 1a and b for the photolysis rate coefficients for NO 2 and O 3 ) resulting in a strong diurnal cycle in near-surface temperature (Fig. 1c) and wind speed.During the OPALE campaign the median values of the temperature at a height of 1m and the wind speed ranged from 36 C and 2 m s 1 at 03:00 (all times are local times equivalent to UTC+8 h.) to 27 C and 4 m s 1 at 15:00, respectively.The diurnal solar cycle is responsible for a well defined diurnal cycle of boundary layer structure (Fig. 1c).Starting from about 07:00 in the morning the diurnal heating and upward sensible heat flux (King et al., 2006) drive formation of the convective mixing layer reaching a maximum height of 300-600 m at about 17:00.During the night the boundary layer is stably stratified and confined to a height of several metres.The boundary layer height presented in Fig. 1c comes from the Modèle Atmosphérique Régional (MAR) (Gallée et al., 2014) briefly described in Sect.2.3.
The measurement site was located in a designated cleanair sector 0.7 km to the south of the main station buildings.During the measurement period the wind direction was predominantly from the south, median value of 180 , with few pollution events (on 19, 31 December and 1 January) due to the wind coming from the station, wind sector of 10-50 (Fig. 1c).The data corresponding to the pollution events were filtered out.

Radical measurements
Concentrations of OH and RO 2 radicals, as well as sulfuric acid (not shown here), were measured using chemical ionization mass spectrometry (CIMS) (Eisele and Tanner, 1991;Berresheim et al., 2000).Detailed description of the instrument is presented elsewhere (Kukui et al., 2008(Kukui et al., , 2012;;Michoud et al., 2012).Here we briefly present the principle of the method and report details about the setup of the device and working conditions applied during measurements at Dome C.
The concentration of OH radical was measured by titrating atmospherically sampled OH radicals with SO 2 to form H 2 SO 4 in a chemical conversion reactor in the presence of water vapour and oxygen (Eisele and Tanner, 1991;Tanner et al., 1997;Berresheim et al., 2000).H 2 SO 4 was detected by mass spectrometry as the HSO 4 ion.The HSO 4 ion was produced by chemical ionization with NO 3 in an ion-molecule reactor following the chemical conversion reactor.To distinguish for atmospheric sulfuric acid the chemical titration was performed using isotopically labelled 34 SO 2 leading to the formation of H  (Reiner et al., 1997) followed by conversion of OH into sulfuric acid.
During measurements at Dome C and similarly to previous measurements in coastal Antarctica at Dumont d'Urville (DDU) (Kukui et al., 2012) the instrument was installed in a shipping container with the chemical conversion reactor fixed to the roof of the container via an interface cap covered with a PTFE sheet.The sampling aperture of the reactor (3 mm diameter) was positioned 50 cm above the roof and about 3 m above the snow surface.
Ambient air was sampled at a volumetric flow rate of 7.4 standard litres (sL) min 1 creating turbulent flow in the chemical conversion region of the reactor (Reynolds Number, Re = 6400).The turbulent flow conditions minimize possible influence of wind speed on the measurements and ensure fast mixing of reactants.The reactants used for the chemical conversion ( 34 SO 2 and NO) and the radical quencher (NO 2 ) are introduced into the reactor through a set of injectors.NO 2 used as a scavenger removes not only the OH radicals, but also peroxy radicals converting them into HO 2 NO 2 and RO 2 NO 2 nitrates.The input flow rates of the reactants were similar to those previously used with this instrument (e.g.Kukui et al., 2012), while the sampling flow rate was a factor of 2 smaller than usual.These flow rates were necessary to compensate for the reduced pressure at Dome C (620 hPa) compared to sea-level operation of the instrument, and provided comparable reaction times to previous deployments of this instrument.Switching the reactant flows between the different injectors allows measurements in four different modes: the background mode, two different OH radical measurement modes and the RO 2 radical measurement mode.The two OH measurement modes differ by the time used for the chemical conversion, 4 and 20 ms (Kukui et al., 2008(Kukui et al., , 2012)).Ratio of the signals with the short and the long conversion times may be used as an indicator of an artificial OH formation in the reactor (Kukui et al., 2008).
Measurements of OH, RO 2 and H 2 SO 4 were performed by monitoring the peak intensities at m/z = 62 (NO 3 ), m/z = 99 (H 34 SO 4 ) and m/z = 97 (H 32 SO 4 ).The detection of H 34 SO 4 and H 32 SO 4 corresponds to the measurement of the radicals (OH or RO 2 ) and H 2 SO 4 , respectively.The ion peak intensities were measured sequentially resulting in nine measurements of OH and one measurement of RO 2 for every 28 min.Every measurement of OH was derived from 1 min of OH ion signal count and two 30 s background ion signal counts before and after the OH signal measurement.RO 2 was measured at the end of the OH detection sequence by switching on the NO flow to the corresponding injector for a duration of 2 min.To avoid any possible influence of traces of NO on the OH measurements a time delay of 10 min was introduced after switching off the NO flow and before starting the next OH measurement sequence to ensure flushing of the chemical conversion reactor.The measurements of OH and RO 2 radicals were averaged to 15 and 30 min time intervals, respectively.
The concentration of the radicals, [R], is derived from the measured ratio of the H 34 SO 4 and NO 3 ion peak intensities, I 99 / I 62 : [R] = C R ⇥ ln (1 + I 99 / I 62 ), where C R is a calibration coefficient determined in calibration measurements by production of a known concentration of OH or RO 2 radicals in a turbulent flow reactor by photolysis of water vapour at 184.9 nm (Heard and Pilling, 2003 and references therein;Faloona et al., 2004;Dusanter et al., 2008).Except for additional thermo-stabilization of a Pen-Ray mercury lamp and a VUV phototube the construction of the calibration cell was the same as previously described in Kukui et al. (2008).The concentration of OH and HO 2 radicals generated in the turbulent flow was calculated from the monitored photon flux and H 2 O concentration.For precise control of the concentrations of H 2 O at low temperatures the gas manipulation system for the introduction of the mixture of pure air and water vapour into the photolysis reactor was modified by replacing the previously used water trap by a liquid flow controller (Bronkhorst, µ-FLOW series L01, 1.4 g h allowing simulation of typical atmospheric humidities encountered at Dome C (e.g.50-70 % at 30 C).During the calibrations at Dome C the previously employed cooled mirror dew-point transmitter used for the humidity measurements in the photolysis reactor was found to be unreliable at low temperatures and was replaced by a capacitive humidity sensor (Vaisala, HMP155) providing more accurate humidity data.The water vapour/air mixture was generated inside the container and introduced to the calibration cell positioned on the roof of the container through a 10 m Teflon tube placed outside of the container, providing the cooling of the calibration gas mixture to the ambient temperature.Despite these modifications frequent difficulties were encountered maintaining stable humidity under conditions of low temperatures at Dome C, limiting calibration accuracy.
The atmospheric concentration of total peroxy radicals RO 2 was measured assuming that the HO 2 and CH 3 O 2 radicals represent the major part of all RO 2 radicals at Dome C, i.e. [RO 2 ] = [HO 2 ] + [CH 3 O 2 ], with a ratio of [HO 2 ] / [RO 2 ] of about 0.7, as informed from the model calculations (see below).The calibration of HO 2 and CH 3 O 2 was performed by adding into the calibration cell photolysis reactor either CO or CH 4 converting any OH radical to HO 2 or CH 3 O 2 , respectively (Hanke et al., 2002;Fuchs et al., 2008).The sensitivity to HO 2 was found to be (15 ± 2) % higher than that for the CH 3 O 2 .
The overall accuracy of the calibration coefficients was estimated taking into account uncertainties of all parameters used for calculation of the radical concentrations in the photolysis reactor and the precision of the measurements of the ratio I 99 / I 62 .The main sources of the calibration uncertainty at Dome C were the estimation of the photon flux (±15 %) and the uncertainty of the measurement of humid-ity (±30 %).The overall estimated calibration accuracy (2 ) was of 36 % for OH and of 40 % for HO 2 .The uncertainty of the measurements of RO 2 , accounting for possible variation of the RO 2 composition, was estimated to be of 45 % at Dome C. Accounting for the calibration uncertainties and measurement precision, the overall 2 uncertainty of the 15 min averaged measurements of OH and RO 2 was estimated to be 40 and 47 % during the daytime and 60 and 50 % during the night time, respectively.The lower limits of detection for OH and RO 2 radicals at signal-to-noise ratio of 3 and a 2 min integration time were 5 ⇥ 10 5 molecule cm 3 and 2 ⇥ 10 6 molecule cm 3 , respectively.
To avoid possible contamination of ambient air by the SO 2 , NO and NO 2 reactants added to the chemical conversion reactor, a trap was set up at the pumps exhaust by using two 100 L cylinders containing zeolites.The cylinders were refilled several times during measurements.Flexible exhaust tube of 30 m length was always placed downwind from the container.When the exhaust tube was intentionally placed upwind and close to the radicals sampling point no effect on radical measurements was detected.Also, no influence of the exhaust on the measurements of NO x and HONO could be noticed.

Other measurements
Other chemical measurements conducted at Dome C during the campaign that are relevant to the discussion of sources and sinks of OH and RO 2 radicals include O 3 , HCHO, H 2 O 2 , HONO, NO, NO 2 and photolysis coefficient values.Their median values and ranges are summarized in Table 1, while the detailed discussion of these data and description of the instruments and applied working conditions during the campaign are presented in the corresponding companion papers.
NO x was measured with a two-channel chemiluminescence detector with one channel used for NO detection and the other for the sum of NO and NO originating from the quantitative photolytic conversion of NO 2 (Bauguitte et al., 2012;Frey et al., 2013Frey et al., , 2014)).The NO 2 and NO measurements are discussed in Frey et al. (this issue).The ratios of NO 2 to NO observed at Dome C during 2011-2012 campaign were up to 3 times larger than in 2009-2010 (Frey et al., 2013) and significantly larger, up to seven times, than the ratios estimated assuming photochemical steady-state (PSS) conditions for NO x .It is suggested that some part of this inconsistency may be due to unknown interference leading to an overestimation of the NO 2 concentrations (Frey et al., 2014).
The photolysis rate constants, J , were calculated from measurements of "actinic flux" (Madronich, 1987) measured by a Met-Con 2⇡ spectral radiometer with a CCD detector and a spectral range from 285 to 700 nm.The Met-Con Spectral radiometer was calibrated before and after the campaign using a NIST traceable standard 1000 W tungsten halogen lamp.No significant changes were observed in the performance of the spectral radiometer.The spectral radiometer was mounted on a mast of 1m height on the roof of the container used for the radical measurements.A shadow band was fitted to provide a horizontal horizon.Downwelling unweighted radiance over a complete hemisphere, was recorded as a 5 s average for each of the 532 pixels and internally interpolated to a 1 nm resolution over 285-700 nm.The atmospheric photolysis coefficients, J , were calculated using the measured actinic flux along with quantum yields and absorption cross sections from Sander et al. (2011).The principle of using a 2⇡ CCD Met-Con spectral radiometer to accurately determine photodissociation rate constants, J , has been previously demonstrated by Jäkel et al. (2007).Total 4⇡ steradian radiance was calculated by multiplying the downwelling 2⇡ steradian radiance by a value of 1.9.The value of 1.9 is based on measurements of downwelling and upwelling radiance by inverting the spectral radiometer.
HONO was measured with a LOPAP (Heland et al., 2001;Kleffmann et al., 2002) at height of 1m above the snow surface.In spite of the use of the LOPAP, thought to be free of measurement artefacts, mixing ratios of HONO observed at Dome C in December 2011/January 2012 (hourly means of 35 ± 5 pptv, Legrand et al., 2014) are in the same range as previously observed in December 2010/January 2011 (hourly means of 30.4 ± 3.5 pptv) by Kerbrat et al. (2012).As discussed by Legrand et al. (2014), laboratory experiments with irradiated surface snows collected at Concordia reveal that the snowpack may be a very significant source of HONO.It is shown that this source only accounts for a third of the observed HONO mixing ratios at 1 m height at Concordia.Legrand et al. (2014) report tests done both in the field and in the lab that tend to suggest an overestimation of HONO measurements in the range of 10 to 20 pptv due to the presence of HO 2 NO 2 in the range of 50-100 pptv in the cold atmosphere at Dome C.This range of HO 2 NO 2 mixing ratios is in agreement with the median [HO 2 NO 2 ] of 80 pptv estimated from RO 2 and NO 2 levels measured at Dome C (see Sect. 3.2).Also, as discussed by Legrand et al. (2014), similar levels of HO 2 NO 2 were previously observed in Antarctica.

Model calculations
Observed concentrations of OH and RO 2 were compared with those calculated using a 0-D box model.Photostationary concentrations of OH and RO 2 were calculated by performing numerical integration using a subset from the Master The vertical distribution of HONO in the boundary layer was modelled using a 1-D chemistry-transport box model with a vertical distribution of turbulent diffusivity and boundary layer heights calculated by the regional atmospheric MAR model (Modèle Atmosphérique Régional).A detailed description of the model and its validation with respect to observations from the Automatic Weather Station at Dome C is given in Gallée and Gorodetskaya (2008), Gallée et al. (2014) and references therein.The turbulence scheme is based on an E-e scheme and on the Monin-Obukhov Similarity theory (MOST), outside and inside the lowest model layer of MAR, respectively.Similar to calculations performed by Legrand et al. (2014), we used the MAR data obtained with a horizontal resolution of 20 km centred at Dome C; a vertical resolution of 0.9 m up to a height of 23 m above the surface increasing upward to about 50 m at the height of 500 m; 100 vertical levels with a top level at 1 hPa.For

Presentation of the measurements of OH and RO 2 radicals
Measurements of the concentration of OH and RO 2 radicals are presented in Figs. 1 and 2. The radical concentrations exhibited clear diurnal profiles driven by the solar radiation cycle and ranged from 3 ⇥ 10 5 to 7.5 ⇥ 10 6 and from 1 ⇥ 10 7 to 2 ⇥ 10 8 for OH and RO 2 , respectively, in units of molecule cm 3 .As shown in Fig. 2, the diurnal profiles of OH and RO 2 radicals follow the diurnal variation of the NO 2 photolysis coefficient, J (NO 2 ).At the same time, one can notice some deviation of the diurnal concentration profiles of OH and RO 2 from the J (NO 2 ) profile with larger concentrations for OH and lower for RO 2 in the afternoon.This is reflected in the diurnal profile of the ratio of [OH] to [RO 2 ] (Fig. 2c) which shows a diurnal profile with lower values of [OH] / [RO 2 ] during the day and factor of ⇠ 2 larger in the afternoon.Notably, the [OH] / [RO 2 ] diurnal profile correlates with the daily profile of the concentration of NO (Fig. 2c).
The median concentrations of RO 2 and OH radicals measured at Dome C are compared to those observed at the South Pole in Table 1.At Dome C the J (O( 1 D)) daily median values at 05:00 and 19:00 (9 ⇥ 10 6 s 1 ) are similar to median J (O( 1 D)) observed at the SP ((8.5-9.0)⇥ 10 6 s 1 ).Comparing the corresponding concentrations of OH radical one can see that the median values at Dome C are close to that observed at the SP for 05:00 and somewhat higher at 19:00, 2.5 ⇥ 10 6 and 3.8 ⇥ 10 6 , respectively, compared to 2.3 ⇥ 10 6 at the SP during ISCAT 2000.The concentrations of NO at the SP were similar to the observed at Dome C, 88 and 77 pptv, respectively.
As seen in Fig. 3, the concentrations of both OH and RO 2 correlate linearly with J (NO 2 ) with about 60 and 80 % of the variability of OH and RO 2 , respectively, explained by the variability of J (NO 2 ).The relationship of [OH] with J (O( 1 D)), although with a large scattering as in the case with J (NO 2 ), was close to a power-law dependence with an exponent of ⇠ 0.5 in accordance with a typical close to quadratic dependence of J (O( 1 D)) on J (NO 2 ) observed at Dome C. As seen from the colour coding in Fig. 3 Mauldin et al., 2004) and can be explained in a similar way: when the concentration of NO is small, the concentration of OH increases due to the enhanced recycling from RO 2 until the losses of RO 2 and OH in reactions with NO 2 become important compared to other loss processes and therefore compensate the enhanced OH formation (see below for further discussion).
Owing to an approximate linear correlation of [HONO] with [NO] observed at Dome C (Legrand et al., 2014) together with a approximately linear relation between J (HONO) and J (NO 2 ) the dependence of the concentrations of OH and RO 2 on J (HONO) and HONO is very similar to that on J (NO 2 ) and [NO] presented in Figs. 3 and 4, implying the concentration of HONO also could be an important parameter controlling the radical variability.

Radical sources and sinks
The sources and sinks of OH and RO 2 radicals at Dome C were calculated using the radical field measurements and other available relevant observations presented in Table 1.Some of the key species involved in the radical production and losses, namely CO, CH 4 and H 2 , were not measured.For CO, a mixing ratio of 40 ± 4 ppbv was assumed referring to the value observed at the South Pole (Novelli and Masarie, 2013).Similar mixing ratios of CO for the period of December-January have been measured during ISCAT 2000 at the SP (Davis et al., 2004) and at DDU (Preunkert et al., 2012).
For methane and H 2 mixing ratios the values of 1.8 ppmv (Steele et al., 2002) and 520 ppbv (Steele et al., 2003) are adopted, respectively, based on measurements made at other Antarctic sites.
Measurements of acetaldehyde are sparse in Antarctic regions.As discussed by Legrand et al. (2012), at the coastal site of DDU an acetaldehyde background mixing ratio close to 80 pptv can be assumed, though this value may be highly variable over the Southern ocean (5 to 50 pptv).Hamer et al. (2007) reported a mixing ratio of CH 3 CHO of 60 pptv at the SP.In our calculations we used a value of 60 pptv with an uncertainty of 20 %.Methyl hydroperoxide (CH 3 OOH), acetic acid (CH 3 COOH) and acetone (CH 3 COCH 3 ) were also considered in our calculations.CH 3 OOH was estimated from observations of H 2 O 2 made at Dome C (mean value of 166 pptv, see Table 1) and assuming that it represents 40 ± 10 % of H 2 O 2 as reported by Frey et al. (2005) at the SP.For CH 3 COCH 3 a typical mixing ratio of 130 ± 30 pptv was used on the basis of observations made at the SP (Hamer et al., 2007).For CH 3 COOH, Legrand et al. (2012) measured 60-70 pptv at Dome C between November and February.
The calculated rates of radical sources and sinks are presented in Fig. 5 and Table 2. Based on available measurements the major primary radical source is the photolysis of HONO (1), with other net sources, namely, photolysis of HCHO ( 7), H 2 O 2 (2), O 3 (3) and CH 3 CHO (8) being less important and contributing altogether less than 30 % of the net radical production (the numbering of the reactions is according to the Table 2).It is important to emphasize here that, as detailed in Sect.2.2, the HONO data used in these calculations are thought to be biased by the presence of HO 2 NO 2 (Legrand et al., 2014).Since the raw HONO concentration data cannot be accurately corrected for this measurement artefact, we consider in the following discussion (Sects.3.4 and 3.5) several scenarios for HONO mixing ratios.
The recycling of OH and RO 2 radicals proceeds mainly via the reactions of HO 2 with NO (5) and of OH with CO (10) and CH 4 (11), with the production rate of OH from HO 2 + NO being about half of the generation rate of OH via the photolysis of HONO.Although being less important, additional regeneration of RO 2 by the reactions of OH with HCHO, CH 3 CHO, O 3 and H 2 (12-15) remain significant.Net loss of the radicals proceeds mainly via the reactions OH+NO 2 (18), RO 2 + NO 2 (24) and radical cross reactions OH + RO 2 (20) and RO 2 + RO 2 (25).Note that radical cross reactions become particularly important during the day when concentration of RO 2 is large (Table 2).Reactions of OH with NO, HONO, RO 2 NO 2 and HNO 3 contribute together about 16 % to the radical net losses.The contribution of the reaction of HO 2 (H 2 O) with NO 2 (Li et al., 2014) to the RO 2 losses is estimated with the rate constants given in Legrand et al. (2014) to be less than 1 %.
As the losses of OH and RO 2 via the reactions with NO 2 may be overestimated due to unknown interference in the NO 2 measurements (Frey et al., 2014) we also present in the Table 2 (values in parenthesis) the radical losses for [NO 2 ] estimated assuming PSS conditions for NO x .In this case the net daytime radical losses are dominated by the radical cross reactions RO 2 + RO 2 (25) and OH + RO 2 (20).
According to the above mechanism derived from the field observations, the sum of peroxy radicals RO 2 is composed predominantly of HO 2 and CH which was required for calculating the radical sources/sinks was estimated using steady-state calculations accounting for all identified sources and sinks of HO 2 and CH 3 O 2 .The resulting value of [HO 2 ] / [RO 2 ] is 0.67 ± 0.05(1 ) and is close to the ratio of 0.73 estimated by only accounting for the production of RO 2 in the reactions of OH radical with CO and CH 4 and the loss of RO 2 via reactions of RO 2 with NO.
Concentrations of peroxy nitrates (CH 3 O 2 NO 2 and HO 2 NO 2 ) involved in radical loss reactions (see Table 2) were not measured at Dome C but were estimated assuming steady-state conditions and accounting for their losses via photolysis, the reaction with OH, thermal decomposition and a surface deposition of 7 ⇥ 10 5 s 1 (Slusher et al., 2002).The calculated mixing ratios of HO 2 NO 2 and CH 3 O 2 NO 2 were ⇠ 80 and ⇠ 20 pptv, respectively.These mixing ratios of HO 2 NO 2 are somewhat higher than [HO 2 NO 2 ] of 40-60 pptv observed at the South Pole (Slusher et al., 2010) under conditions of significantly lower [NO 2 ] (estimated at SP from NO observations assuming steady-state).Assuming the steady-state NO 2 concentrations at Dome C the estimated mixing ratios of HO 2 NO 2 and CH 3 O 2 NO 2 , are significantly lower: ⇠ 10 and ⇠ 3 pptv, respectively.In the work presented here the calculated steady-state concentrations of peroxy nitrates were used to estimate the effective loss rates of RO 2 radicals in the reactions with NO 2 by accounting for the regeneration of RO 2 via RO 2 NO 2 thermal decomposition.
The mechanism presented above of the radical production and loss is supported by examination of specific correlations.The major sources of OH radical, both the photolysis of HONO and the recycling from HO 2 , are expected to correlate with photolysis rates.Hence, the observed close to linear correlation between [OH] and J (NO 2 ) or J (HONO) (Fig. 3) tends to support the importance of these two mechanisms of the production of OH radical.However, the correlation of [OH] with J (NO 2 ) explains only about 60 % of [OH] variability, while an additional source of [OH] variability may come from the variability of [HONO] and/or [NO] as it can be seen from Fig. 3. To account for the variability of [HONO] and [NO], in Fig. 6 we plotted [OH] against production rates of OH radical via photolysis of HONO, (P HONO ), and via reaction of HO 2 with NO, (P HO 2 + NO ).The variability of P HO 2 + NO and P HONO then explain ⇠ 80 % of the variability of OH. ⇤ Values in parentheses correspond to NO 2 and HONO estimated assuming PSS; a HNO 3 of (100 ± 30) ppt was assumed (Slusher et al., 2010); b 1 uncertainty estimated with accounting for measurement uncertainties.(e, f).Solid red and black lines represent the total production and loss rates, respectively.Dashed lines correspond to corresponding quartiles.

Photochemical budget of radicals
According to the calculated sinks of OH and RO 2 radicals their lifetimes were less than 3 and 100 s, respectively.The timescales for the variability of atmospheric parameters at Dome C (boundary layer height and vertical diffusivity, solar radiation, NO x concentration etc.) were typically significantly larger, more than 10 min.Hence, assuming validity of a steady-state approximation for OH and RO 2 , their sources and sinks should be balanced.However, as shown in Fig. 5 and Table 2, on the basis of observations the sum of the radical production rates exceeds by about (40-90) and 25 % the sum of loss rates for OH and RO 2 radicals, respectively.Considering the primary sources and net sinks of the sum of OH and peroxy radicals, the measured primary radical production is larger than their measured total removal rate by a factor of 2 and 3 for midnight and noon times, respectively.
A similar result of the budget analysis follows from the comparison of the net OH and RO 2 production rates, i.e. from the difference of the sum of their primary sources and the sum of their net removal, and the total rate of the radical recycling.Under steady-state conditions the net production (or removal) rate of OH, P net pr (OH), should be compensated by the net removal (or production) rate of RO 2 , P net pr (RO 2 ), and equal to the net recycling rate of OH to RO 2 (or RO 2 to OH), R net (OH!RO 2 ), (see Table 2): However, as seen in Table 2, according to the field measurements the net primary production rate of OH significantly exceeds both the OH to RO 2 recycling and the net RO 2 removal rates.At the same time, the net conversion rate of OH to RO 2 is not compensated by the measured net removal rate of RO 2 .This interpretation shows in a different way the same result as Fig. 5: for both OH and RO 2 radicals the measured radical production rate exceeds their loss rate, although for RO 2 the difference is less significant.The observed imbalance between radical production and loss could result from underestimated or unaccounted radical losses, as well as from overestimated radical sources.As the main radical loss processes identified at Dome C were the reactions of radicals with NO 2 and radical cross reactions the underestimation of the radical losses could be related to an underestimation of measured concentration of NO 2 and/or concentrations of OH and RO 2 radicals.The underestimation of the radical concentrations by a factor of 2 is well outside the estimated uncertainties for the measurements of OH and RO 2 , although we can not completely exclude some unknown error in the radical calibration procedure.
Considering the possibility of the underestimation of the concentration of NO 2 , sensitivity analysis shows that increasing of the concentrations of NO 2 by a factor of ⇠ 3 would result in a fair agreement of the model with the observations.However, as the observed [NO 2 ] / [NO] ratios at Dome C are already very high it is very unlikely that measured NO 2 was underestimated.As discussed in Frey et al. (2013, this issue) the ratios of [NO 2 ] / [NO] observed at Dome C are not consistent with other observations available at Dome C. Accounting for the photolysis of NO 2 and conversion of NO to NO 2 in the reactions with measured [RO 2 ] and [O 3 ] results in the ratio of [NO 2 ] to [NO] about a factor of 7 lower compared to the measured values suggesting that there is a missing mechanism of the conversion of NO to NO 2 .An explanation evoking an underestimation of the measured RO 2 can be excluded, because the concentrations of RO 2 radicals needed to explain the observed ratio of [NO 2 ] / [NO] would be a factor of 20 larger than measured.Another possibility would be an additional conversion of NO to NO 2 in presence of halogens (e.g.ClO, BrO, IO) as was found at other polar sites, e.g at the coastal site of Halley (Bauguitte et al., 2012).However, to explain the observed ratio of [NO 2 ] / [NO] at Dome C a very large concentration of halogen species would be required, about 60 pptv in case of BrO and even higher for ClO.Such a large halogen mixing ratio is contradicted by measurements of BrO at Dome C, i.e.Frey et al. (2014) report the BL mixing ratios of BrO of ⇠ 2 pptv.Hence, we conclude that either there is some unknown mechanism of NO to NO 2 conversion, or there was some interference affecting the NO x measurements.
The analysis performed here of the radical chemistry at Dome C could be affected in both cases: the missing NO x chemistry could interfere with chemistry involving OH and RO 2 , i.e. by shifting the ratio [RO 2 ] / [OH] towards OH, like in the case of efficient halogen chemistry at Summit, Greenland (Liao et al., 2011), while biased measurements of NO x could affect the calculated radical losses and interconversion rates.Further speculations about possible impact of an unknown NO x regulating mechanism on the OH and RO 2 chemistry are not feasible at this stage.Concerning possible interference in NO x measurements, any correction consisting in a reduction of NO 2 would result in weaker radical losses and, hence, even larger overestimation of the radical sources.As shown in Table 2 the assumption of steady-state NO 2 concentrations lead to a significant overestimation of the net radical production for RO 2 and (RO 2 + OH) even when neglecting net OH production by the photolysis of HONO.For OH budget, neglecting the net OH production by the HONO photolysis would lead to an underestimation of the OH production.
Considering the possibility of some overestimation of the OH radical production rate and independent of discussions carried out elsewhere (Legrand et al., 2014), one can see from Fig. 5 and Table 2 that among all the measured radical sources only the photolysis of HONO is large enough to explain, in case of its overestimation, the observed imbalance.In the following section we compare the observed concentrations of OH and RO 2 with model predictions to test the sensitivity of the modelled [OH] and [RO 2 ] to the concentration of HONO.

Comparison with 0-D model: sensitivity to HONO
The simulations of concentration profiles of OH and RO 2 were conducted with different constraints on the concentration of HONO: taken as measured by LOPAP, the LOPAP measurements reduced by factors 2, 4, and estimated accounting for HONO production in HO+NO and HONO removal by photolysis assuming PSS conditions.The calculations were constrained with measured concentrations of NO 2 unless specifically stated that PSS-derived NO 2 was used.The modelled concentrations of OH and RO 2 radicals are presented in Figs.7 and 8. Accounting for all radical sources including the photolysis of HONO leads to about a factor of 2 overestimation for modelled [RO 2 ] and [OH] with moderate correlation of the calculated and observed concentrations of OH and RO 2 radicals.Assuming HONO at PSS leads to an underestimation of [OH] and [RO 2 ] with a ratio of modelled to observed concentrations (M / O) of about 0.5 and a distinct difference between modelled and observed diurnal profiles (Fig. 8), with better agreement during the daytime.As shown in Fig. 8 the M / O for OH, RO 2 and their ratio become close to 1 by decreasing [HONO] from the measured to concentrations reduced by a factor of 4.
Similar to the approach adopted at the SP (e.g.Davis et al., 2004) we have tested the model assuming PSS HONO concentrations and using concentrations of NO 2 calculated by assuming photostationary ratio of [NO 2 ] / [NO] calculated from the observed concentrations of NO, O 3 and RO 2 .In this case, similar to the SP results, we obtain quite a good description for [OH], although with up to a factor of 1.5 overestimation of [RO 2 ] and [RO 2 ] / [OH] ratio at noon (Fig. 8).
Another insight on the relationship between the calculated concentrations of radicals and [HONO] can be gained by comparing the dependencies of observed and calculated [OH] and [RO 2 ] normalized by J (NO 2 ) on [NO] (Fig. 4).The model clearly shows that neglecting HONO photolysis and using constraints by measured concentrations of NO x leads to a clear disagreement with observations.In this case, the calculated roll-off of normalized [OH] starts at a mixing ratio of NO of about 50 pptv (Fig. 4a) as a consequence of fast losses due to large [NO 2 ] and the lack of additional OH production at higher concentrations of NO x .When it is assumed that measurements of HONO are correct, [OH]  whole range of observed [NO].A better description by the model is achieved either when reducing the HONO concentrations by a factor of 4 or by neglecting HONO and using [NO 2 ] calculated assuming PSS for NO x (see also Fig. 8).In the latter case, the model underestimates the concentrations of radicals by 30-40 % at large [NO], although the number of observations at large concentrations of NO (>200 pptv) is limited.
Considering diurnal profiles of the M / O ratio presented in Fig. 8 one can see that although reducing the production rate of OH from the photolysis of HONO by reducing the concentration of HONO improves the agreement between the model and the observations, the diurnal profiles of the M / O ratio become more distinct indicating that a simple reduction of [HONO] cannot provide a satisfactory description for the diurnal concentration profiles of OH and RO 2 .Thus, either some chemical mechanism is missing in the model and its effect on the M / O ratio becomes stronger at reduced rates of OH production or, the overestimation of the OH production rate is dependent on time of day (see discussion of effects related to BL diffusivity in Sect.3.5).In fact, median values of the time-dependent correction of [HONO] necessary to provide agreement between calculated and observed radical concentrations can be estimated by solving radical sources and losses balance equations for the concentration of HONO.As shown in Fig. 9, for the OH radical budget the balance is achieved by reducing P HONO by a factor of 3 during the day with a smaller correction required during the night.Balancing the budget of primary production and net removal rates of radicals requires neglecting the photolysis of HONO during the day and reducing it to a quarter of its original value during the night time.According to a sensitivity analysis to different measured parameters the calculated to observed [HONO] ratios, [HONO] calc / [HONO] obs , are most sensitive to the concentrations of NO 2 and OH, RO 2 radicals.By assuming the PSS-derived NO 2 concentrations, the balance for the OH radical budget is achieved by reducing P HONO by a factor of 5, while for RO 2 and the sum of RO 2 and OH the radical production is overestimated even if the net source from HONO photolysis is neglected.

Comparison with 1-D model
It follows from the analysis presented above in Sect.3.4 that the concentrations of HONO observed at Dome C were probably too large and not compatible with the measurements of OH and RO 2 .Legrand et al. (2014)  As an alternative approach, we have compared the concentrations of HONO derived by examining the radical budget (Fig. 9) with the calculations assuming HONO snow emissions at Dome C estimated from the laboratory studies of Legrand et al. (2014).By irradiating the surface of snow collected at Concordia and simultaneously measuring the concentrations of HONO and NO x produced from the snow the authors found that HONO mixing ratios in the experiment chamber outflow were roughly 30-50 % of the mixing ra-  2014) estimated a HONO flux at Dome C ranging from 0.2 ⇥ 10 9 molecules cm 2 s 1 at night to 1.5 ⇥ 10 9 molecules cm 2 s 1 at noon.These fluxes of HONO were used in the work described here to estimate the concentrations of HONO at different heights above the snow using the 1-D chemistry-transport box model (Sect.2.3) with turbulent vertical transport and a simplified chemistry scheme including HONO production via the snow emissions only and HONO removal via photolysis.Formation of HONO in the reaction OH + NO contributing less than 10 % to the concentration of HONO resulting from the snow emissions was neglected in the present 1-D calculations.Another gas-phase source of HONO via reaction of HO 2 (H 2 O) complex with NO 2 was recently proposed by Li et al. (2014).Owing to low temperatures at Dome C the contribution of this source to [HONO] is estimated to be less than 2-3 % (Legrand et al., 2014) and it was also neglected in the 1-D model.
The results of the 1-D calculations are shown in Fig. 10 where the median diurnal profiles of emission fluxes, boundary layer heights and simulated mixing ratios of HONO at 1 and 3 m are shown for the period from 19 December to 5 January.Figure 10 shows that the difference of HONO mixing ratios at 1m and 3 m does not exceed 30 % and, hence, The HONO mixing ratio-time profiles calculated with the 1-D model are compared in Fig. 10 with the HONO profiles resulting from analysis of the radical budgets.The levels of HONO derived from the OH budget with measured NO 2 are about 10 pptv higher than the HONO values obtained using PSS NO 2 concentrations.In both cases the HONO mixing ratios derived from the OH budget are in reasonable agreement with [HONO] predicted by the 1-D model (within 5 pptv).The 1-D model reproduces also the diurnal HONO concentration profile with a minimum during the day and larger concentration of HONO in the evening resulting from interplay between emission rate and BL height variability (Legrand et al., 2014).For the concentrations of HONO derived from the OH + RO 2 budget with measured NO 2 the model significantly underestimates that derived from the budget daytime [HONO] but provides better agreement for the night and reproduces the general temporal trend of HONO variability.The concentrations of HONO derived from the OH + RO 2 budget with steady-state [NO 2 ] are negative (about 8 ppt during the day) and not presented in Fig. 10.

Ozone production
The RO 2 and NO measurements made at Dome C were used to estimate local boundary layer ozone production.As seen in Fig. 11, the peak calculated ozone production rate (P (O 3 )) is about 0.3 ppbv h 1 during daytime (using the measurements of RO 2 at 3 m, NO at 4 m above the snowpack and assuming P (O 3 ) equal to NO 2 production rate in the reaction of RO 2 with NO).The production rate is in fairly good agreement with the previous estimation of 0.2 ppbv h 1 derived from examination of diurnal changes in the concentration of ozone (Legrand et al., 2009).The integrated 24 h production of ozone reaches 4.7 ppbv d 1 at Dome C and is similar to the one calculated for the SP where the daily production of ozone was estimated from a model using measured concentrations of NO and OH to be of 2.2-3.6 ppbv d 1 for ISCAT 1998 (Crawford et al., 2001) and 3.2-4.8ppbv d 1 for ISCAT 2000 (Chen et al., 2004).
It has to be emphasized with respect to the previous discussion of the ratio [NO 2 ] / [NO] observed at Dome C that if the concentration of RO 2 radical was as large as needed to explain the observed [NO 2 ] / [NO] ratio, the corresponding production rate of ozone would be as large as 100 ppbv d 1 and would strongly conflict with observed diurnal change of ozone (see Fig. 3 in Legrand et al. (2014).

Conclusions
Concentrations of hydroxyl radicals and the sum of peroxy radicals, RO 2 , have been measured for the first time on the East Antarctic Plateau at the Concordia station (Dome C).The concentrations of OH and RO 2 radicals were found to be comparable to those observed previously at the South Pole (Mauldin et al., 2001(Mauldin et al., , 2004(Mauldin et al., , 2010) ) confirming that the elevated oxidative capacity found at the SP is not unique but a common characteristic of near-surface atmospheric layer for a large part of the high Antarctic plateau.
Similar to the findings at the SP the major explanation for the large concentrations of OH radical at Dome C was found to be the large concentrations of NO (Frey et al., 2013(Frey et al., , 2014) ) leading to fast recycling of peroxy radicals to OH radicals.Also, similarly to the SP, the variability of the concentration of NO x plays a major role in controlling the variability of radicals (OH and RO 2 ) at Dome C. In contrast to the SP, where there is no diurnal variation of solar radiation, and where radical levels are controlled by the concentrations of NO mostly via changing boundary layer properties (Neff et al., 2008), [OH] and [RO 2 ] at Dome C show strong diurnal variability correlating with the solar cycle, which in turn controls the rate of radical production and the rate of interconversion of OH and RO 2 radicals.
The large concentrations of radicals and NO result in ozone production in the BL at Dome C with a production rate of 0.1-0.3ppb h 1 , similar to that observed at the SP.
The major primary sources of radicals in the atmosphere at Dome C are represented by the photolysis of HONO, HCHO, CH 3 CHO and H 2 O 2 , (in order of decreasing significance), with photolysis of HONO deduced from measurements of HONO contributing for about 75 % of total primary production of radicals.The main net losses of radicals are repre-sented by their reactions with NO 2 and cross radical reactions.However, it is found, that these results are inconsistent with observations of radicals leading to about a factor of 2 overestimation of the concentrations of RO 2 and OH radicals.At the same time, neglecting the production of OH radical from the photolysis of HONO results in about a factor of 2 underestimation of the measured concentrations of radicals.Based on 0-D modelling, to explain the observations of OH and RO 2 radicals, in this case an additional source of OH equivalent to about 25 % of measured photolysis of HONO is required.Similar result follows from analysis of the photochemical budget of the OH radicals for which the balance is achieved by reducing P HONO by a factor of 3.
The conclusions based on the radical budget analysis and 0-D modelling using the measured concentrations of NO and NO 2 may be significantly biased because the chemical mechanism derived from the available field observations at Dome C is inconsistent with observed large ratios of [NO 2 ] to [NO].Assuming that measured NO 2 mixing ratios were overestimated due to unknown interference and using instead [NO 2 ] estimated assuming steady-state results in lower radical losses and, hence, stronger overestimation of the radical production.In this case, based on the analysis of the radical budgets the observed concentrations of OH radicals are consistent with the levels of HONO corresponding to about (15-20) % of the measured values, while for the sum of the radicals the radical production is overestimated even neglecting the net OH source from the photolysis of HONO.Based on 0-D modelling steady-state derived NO 2 , the measured OH concentrations are in agreement with steady-state HONO mixing ratios (about 1-2 pptv), while the concentrations of RO 2 radicals are overestimated by about 50 % even neglecting the net radical production by the photolysis of HONO.
Hence, in both cases corresponding to the measured or the PSS-derived concentrations of NO 2 the calculations, 0-D modelling or budget analysis, overestimate the OH and RO 2 concentrations.If this inconsistency is due to an overestimation of the concentrations of HONO, the degree of the overestimation depends on the concentrations of NO 2 used in the calculations.Using the measured NO 2 results in an overestimation of HONO by a factor of 3-4.If the concentrations of NO 2 are estimated assuming steady-state conditions the net radical production from the HONO photolysis should be reduced by a factor of 5 or completely removed based on the budget of OH or 0-D modelling, respectively.
Assuming that HONO at Dome C originates from snow emissions with the emission strength evaluated by Legrand et al. (2014) we obtain from a 1-D model the concentrations of HONO corresponding to about 20-30 % of measured [HONO] and the diurnal concentration profiles of HONO consistent with those calculated from the budget analysis of OH radicals with the concentrations of NO 2 either calculated assuming PSS or taken from the measurements.We suggest that an explanation for the overestimation of radical production could be an overestimation of measured concentration of HONO, which may originate from the measurement interference from HO 2 NO 2 affecting measurements of HONO by LOPAP (Legrand et al., 2014).Even with a factor of 4 reduction in the concentrations of HONO, the photolysis of HONO represents the major primary radical source at Dome C accounting for about 45 % of primary radical production.
Considering the observed uncertainties in HONO and NO x measurements we suggest that further studies of NO x , HONO, peroxy nitrates (RO 2 NO 2 ) and radical chemistry at Antarctic Plateau are required with specific efforts dedicated to increase the reliability of measurements (especially HONO and NO 2 ) under polar conditions.

12374A.
Kukui et al.: Measurements of OH and RO 2 radicals at Dome C, East Antarctica

Figure 2 .
Figure 2. Median diurnal concentration profiles of OH (a), RO 2 (b), [OH] / [RO 2 ] ratio (c).Solid lines represent the median and quartile levels.Points -15 min averaged measured values with standard deviations.Blue line in (a) and (b) represents J (NO 2 ) in arbitrary units.Red line in (c) represents median NO.
by concentration of NO of the graph points, there is an obvious correlation with [NO], positive for [OH] and negative for [RO 2 ].In fact, about 80 % of the [OH] / [RO 2 ] variability may be explained by the linear correlation with [NO] (Fig. 3c).The important role of NO in controlling the concentrations of OH and RO 2 is seen in Fig. 4 where [OH] and [RO 2 ] are normalized by J (NO 2 ) to remove the dependence on solar radiation and plotted against [NO].For concentrations of NO of up to ⇠ 150 pptv the concentration of OH rises and then levels off, while [RO 2 ] exhibits the reverse dependence on [NO].These dependences are very similar to those observed at the SP (Figs. 2 and 7 in

Figure 4 .
Figure 4. Normalized concentrations of OH (a) and RO 2 (b) versus measured concentrations of NO.Grey and black points represent observations; coloured lines -MCM calculations with [HONO] multiplied by a factor presented in figure legends (x0 corresponds to estimated steady-state HONO concentrations).Magenta -PSS [HONO] and [NO 2 ] is calculated by assuming PSS [NO 2 ] / [NO] ratio.

Figure 5 .
Median diurnal production and loss rates of OH (a, b), RO 2 (c, d); net production and loss rates of radicals

Figure 6 .
Figure 6.Correlation of measured [OH] with production rates of OH radical by photolysis of HONO, P HONO (a), and via reaction HO 2 +NO, P HO 2 +NO (b).

Figure 9 .
Figure 9. Ratio of calculated and measured [HONO]: (a) from OH budget; (b) from net radical (OH and RO 2 ) sources and sinks.Solid lines -median values; dashed lines -upper and lower quartiles of the mean values with added and subtracted standard deviation, respectively.Error bars correspond to standard deviation.

Figure 10 .
Figure 10.Comparison of median diurnal profiles of [HONO] derived from OH (black) and sum of radicals (OH and RO 2 ) (blue) budgets with [HONO] calculated with 1-D model at the heights of 1m (red) and 3 m (magenta) for the days 2-9 of January.Black circles and squares correspond to [HONO] derived using measured and calculated steady-state [NO 2 ], respectively.Grey and green lines represent modelled median boundary layer height and HONO emission rates, respectively.

Figure 11 .
Figure 11.Ozone production calculated from measured [RO 2 ] and [NO]: red line represents hourly median values.
for HCHO.Here, presented for the first time, are measurements of concentrations of OH and RO 2 radicals conducted on the East Antarctic Plateau at the Concordia station (Dome C).Sources and sinks of these radicals are discussed in the light of concurrent chemical observations made during the campaign, including HONO, NO, NO 2 , H 2 O 2 , HCHO, and O 3 , as well as

12376 A. Kukui et al.: Measurements of OH and RO 2 radicals at Dome C, East Antarctica via
342 SO 4 .The concentration of total peroxy radicals RO 2 was measured by converting RO 2 into OH radicals reactions with NO injected in the chemical conversion reactor

2014 12378 A. Kukui et al.: Measurements of OH and RO 2 radicals at Dome C, East AntarcticaTable 1 .
Comparison of measurements at Dome C and at the South Pole.

2014 12382 A. Kukui et al.: Measurements of OH and RO 2 radicals at Dome C, East AntarcticaTable 2 .
Sources and sinks of OH and RO 2 radicals estimated from the measurements available at Dome C ⇤ .