3.1. Measurement Data
 HONO concentration during the OASIS campaign varied between ≤ 0.4 and ∼500 pptv (see Figure 1) with an average of 27 pptv. While the average value is still on the upper end of the range of concentrations reported for other polar regions [Li, 1994; Zhou et al., 2001; Beine et al., 2001, 2002; Dibb et al., 2002; Honrath et al., 2002] and also for remote mountain sites [Huang et al., 2002; Kleffmann et al., 2002; Beine et al., 2005; Kleffmann and Wiesen, 2008], the high concentrations observed during several days were out of the range for remote conditions. Similarly, also very high NOx (NO+NO2) levels were observed for these days (see Figure 1). The measured NO, NO2 and NOy levels were in the range 0–20 ppbv, 0–15 ppbv and 0.3–32 ppbv, with average concentrations of 0.41 ppbv, 0.51 ppbv and 1.45 ppbv, respectively, which were also much higher, compared to typical remote conditions.
Figure 1. HONO, NOx and CO concentrations (10 min averages) during the OASIS campaign in Barrow, Alaska, spring 2009.
Download figure to PowerPoint
 To understand the origin of the high HONO levels observed, the polluted days were investigated in more detail (see Figure 2a). High temporal correlation of HONO with NOx and CO was observed for these “polluted” days, whereas no significant correlation with the actinic flux was evident. Since NOx and HONO formation in polar regions are proposed to originate from photolytic sources, the high correlation of HONO with NOx and CO and the missing correlation with the actinic flux is explained by local direct anthropogenic emissions from combustion processes. There are several emission sources nearby, both BARC and the town of Barrow, 0.2 and 5 km away from the measurement site, respectively. For the two most polluted days shown in Figure 2a, a high correlation of HONO and NOx (R2 = 0.52) with a slope of 1.2% was observed. Such a HONO/NOx ratio is close to the ratio of 0.8% quantified for direct emissions from a vehicle fleet [Kurtenbach et al., 2001] and indicates that HONO was mainly caused by fresh local emissions for the polluted days. Since the aim of this study was the investigation of remote polar atmospheric chemistry and not of local anthropogenic emissions, these polluted days were not further considered.
Figure 2. Examples of HONO, NO, NO2, and CO data (a) from two anthropogenically influenced days (24–25 March) and (b) from a “clean day” (26 March) during the OASIS campaign in Barrow, Alaska, spring 2009.
Download figure to PowerPoint
 Only seven “clean days” from the whole campaign – identified by the absence of CO emission peaks – were used in most of the following data evaluation (26–30 March and 7–8 April, for the 8 April, the period 9:30–13:30 was excluded caused by local emissions). One of these days is shown in Figure 2b as an example. HONO concentrations were typically well-correlated with the actinic flux with the highest HONO values of ∼10 pptv occurring around noon Alaska standard time. This observation is in good agreement with other measurements in polar regions and on remote mountain sites [Zhou et al., 2001; Dibb et al., 2002; Kleffmann et al., 2002; Huang et al., 2002; Kleffmann and Wiesen, 2008] and is explained by photochemical HONO sources. From all seven “clean days” a single diurnal average was calculated from the 10-min mean values of the instruments used. This resulted in one “average clean day,” which is shown in Figure 3 for HONO, HONO/NOx and HONO/NOy.
Figure 3. “Average clean day” of the OASIS campaign for HONO, HONO/NOx and HONO/NOy. The HONO error bars reflect the 1σ day-to-day variability.
Download figure to PowerPoint
 For these selected “clean days,” much lower average concentrations of HONO, NO, NO2 and NOy of 4.6 ± 2.5 pptv, 44 ± 28 pptv, 38 ± 21 pptv and 490 ± 90 pptv were determined, respectively. The average HONO/NOx and HONO/NOy ratios of 6.2 ± 4.8% and 1.0 ± 0.6% agreed well with values measured by LOPAP instruments at remote mountain sites (Zugspitze: 2.5% and 1.0%; Jungfraujoch: 4.6% and 1.1% [see Kleffmann and Wiesen, 2008]). However, the HONO/NOx ratio observed was much lower compared to other polar studies or remote mountain sites in which wet chemical instruments were used, having HONO/NOx ratios in the range ∼20–100% [Beine et al., 2001; Huang et al., 2002; Jacobi et al., 2004; Zhou et al., 2007; Jones et al., 2007; Amoroso et al., 2010].
 One possible explanation of the different results could be chemical interferences of wet chemical techniques in agreement with conclusions from other studies [Chen et al., 2004; Grannas et al., 2007; Mauldin et al., 2010]. Chemical interferences are measured in the second channel of the LOPAP instrument and are corrected for in the data evaluation (for details to this correction, see Kleffmann and Wiesen ). For example, when only the HONO data < 10 pptv from the present study were considered, the average interference was 97%. Thus, without correction, the low concentration data would have been overestimated on average by a factor of two. In addition, caused by the different sampling conditions applied, i.e., pH = 0 for the LOPAP and pH ≥ 6 for other wet chemical instruments, other instruments are expected to show even higher interferences (for details to possible interfering reactions at high pH, see Kleffmann and Wiesen ). Finally, heterogeneous formation of HONO in sampling lines up to 30 m length [e.g., Zhou et al., 2001] used in other polar HONO studies may have caused artificial overestimation of HONO. In contrast, caused by the external sampling unit of the LOPAP instrument no sampling lines were used in the present study. Alternatively the much lower HONO/NOx ratios observed by LOPAP instruments under remote conditions compared to other wet chemical instruments may also be caused by different snow properties. For example, alkaline snow surfaces will lead to lower HONO emissions as demonstrated for a high mountain site [Beine et al., 2005]. However, it seems unrealistic that only during all the remote LOPAP measurements alkaline snow surfaces were prevailing, whereas in all studies of other wet chemical instruments acidic snow surfaces were studied, especially when also coastal polar regions were investigated [Beine et al., 2001; Jones et al., 2007; Amoroso et al., 2010].
 The factor 3–15 lower HONO/NOx ratio observed in the present study compared to other polar studies is in good agreement with the factor of 7 lower HONO concentrations observed during the only polar HONO intercomparison exercise using a LIF compared to a wet chemical mist chamber instrument at South Pole [Liao et al., 2006]. Unfortunately, no simultaneous HONO and NOx concentrations were published for the ANTCI 2003 campaign [Liao et al., 2006; Eisele et al., 2008] to calculate HONO/NOx ratios for comparison. From the results of the intercomparison study (factor 7) and from the present study (factor 3–15), it is concluded that chemical interferences of wet chemical instruments should be studied in detail and further HONO intercomparison studies in polar regions, including previously used HONO instruments, the LOPAP technique and spectroscopic techniques, are recommended.
3.2. Is HONO Still a Net Source of OH Radicals?
 For almost all days a high temporal correlation of HONO and OH was observed (e.g., Figure 4a), i.e., each sharp HONO peak observed during daytime was in coincidence with a similar peak in the OH concentration. When all daytime (10–19 h) HONO data < 100 pptv from the OASIS campaign were considered, the correlation coefficient for a plot of OH against HONO was R2 = 0.30. The high variability and coincidence observed for single days could not simply be explained by the variability of the actinic flux (see Figure 4a), although HONO and OH are both proposed to have photochemical sources. This is also confirmed by a smaller correlation (R2 = 0.10) of OH with for all daytime HONO data < 100 pptv, which is in contrast to studies under rural conditions [see Rohrer and Berresheim, 2006, and references therein]. The correlation was improved when OH and the gross rate of OH radicals formed by HONO photolysis (JHONO × HONO) were considered (Figure 4b), leading to a correlation coefficient of R2 = 0.48 for all daytime HONO data < 100 pptv. Considering this coincidence and correlation, it is proposed here that HONO is a major precursor of OH radicals during the OASIS campaign. Alternatively, the high coincidence observed may also be accidentally caused by a high correlation of HONO with any other precursors of the OH radical, e. g., ozone or HCHO. Interestingly, no coincidences of sharp OH radical peaks (see Figure 4a) with the gas phase concentrations of O3 or HCHO were observed, indicating that these two radical sources are of smaller importance for the OH radical levels compared to the HONO photolysis.
 To investigate which of the two above explanations is more reasonable, the net rate of OH radical initiation by HONO photolysis was calculated and compared to that from ozone photolysis, which is typically proposed as the major OH radical source in the atmosphere where water vapor is not limiting. To calculate the OH radical source by HONO photolysis, besides the OH source reaction (1), the gas phase losses of OH radicals by reaction with NO:
and with HONO:
must also be considered. Assuming that the OH source (1) is equal to the losses (6) and (7), a theoretical photo-stationary-state daytime concentration, HONOPSS, can be calculated, if the photolysis frequency of HONO, JHONO, and the NO and OH radical concentrations are known:
 Only when the measured concentration is higher than HONOPSS, should HONO be considered as a net source of OH radicals. Using the NO, OH and JHONO data from the “average clean day” and the temperature dependent rate coefficients for reactions (6) and (7) from the IUPAC database [Atkinson et al., 2004], HONOPSS was calculated for the OASIS campaign. As expected, HONOPSS was much lower compared to the measured HONO concentration (see Figure 5). Thus, an additional HONO source must prevail during daytime. A well known source, still not considered in equation (8), is the heterogeneous conversion of NO2 on humid surfaces in the dark, which was proposed also for snow surfaces [Zhou et al., 2001]. Although the detailed mechanism, either via a slow reaction of NO2 with water (5) [Finlayson-Pitts et al., 2003] or a much faster dark reaction:
Figure 5. “Average clean day” of measured HONO and theoretical HONO concentrations considering only known gas chemistry (HONOPSS) and additionally slow heterogeneous nighttime sources
Download figure to PowerPoint
 (see, e.g., Gutzwiller et al.  and Ammann et al. , and dark experiments from Stemmler et al. ) is still under discussion, both reactions were found to be first order in the NO2 concentration. Accordingly, a first order rate coefficient of khet = 2.7 ± 1.7 × 10−6 s−1 was determined for the heterogeneous NO2 conversion into HONO in the dark using the nighttime increase of the HONO/NO2 ratio from the “average clean day” (for details, see Alicke et al. ). Assuming that this nighttime source is active also during daytime, the theoretical HONO concentration, was not significantly higher than HONOPSS (Figure 5) and thus, the dark conversion of NO2 could not explain the high HONO daytime concentrations observed.
 From the difference between the measured and theoretical HONO concentrations an additional daytime HONO source, d[HONO]/dtextra, was calculated. This source, which is equal to the net OH radical production rate by HONO photolysis, d[OH]/dtHONO net, is shown in Figure 6 together with the rate of OH production by ozone photolysis. It should be pointed out here, that the calculated extra source of HONO can only be applied for the measurement site, since it is based on the source and sinks of OH by reactions (1), (6) and (7), which are calculated by the measured parameters HONO, NO, OH and JHONO. However, in the case of any spatial variations of the involved parameters, like for example vertical gradients, the calculated extra source will be different at other locations. Three important conclusions can be drawn from the figure:
 1. The net OH initiation rate by HONO photolysis is significantly larger compared to that from the ozone photolysis. An OH source by HONO photolysis of up to 90 pptv/h could be determined, whereas that by the O3 photolysis reached only 2.7 pptv/h during daytime. When considering only these two radical initiation sources, the average relative contributions of HONO and O3 were 98% and 2%, respectively. The small contribution by O3 photolysis is caused by the low average humidity (∼0.06%) and O3 levels (5–14 ppbv) for the “average clean day.” Since other radical initiation sources, like alkene ozonolysis, are not expected to be of significant importance for this polar site, it is concluded that HONO is the major OH radical initiation source in the lower polar boundary layer of Barrow, which is in excellent agreement with the high coincidence and correlation of OH radical concentrations in the gas phase and JHONO × HONO, shown in Figure 4b.
 HCHO photolysis was not considered as an OH radical initiation source here, since HO2 radicals are formed by the radical channel of the HCHO photolysis. These HO2 radicals will not be quantitatively converted into OH radicals under the low NOx conditions during the “clean days” caused by peroxyradical cross reactions. In addition, the fraction of HCHO which is formed by the OH-initiated degradation of VOCs in the gas phase, like e.g., methane, will not lead to radical initiation, but is part of the radical propagation cycle, which leads to a net loss of radicals under the prevailing low NOx conditions. And finally, even if all the HCHO were caused by fresh emissions from the snowpack, the total rate of HO2 radical formation by the HCHO photolysis would be on average four times lower compared to the net OH-initiation rate by HONO photolysis. Thus, HCHO is not expected to be a significant source of new OH radicals in Barrow. This conclusion is in excellent agreement with the completely missing correlation of sharp OH radical peaks (see Figure 4) with the concentration of HCHO (not shown).
 Another OH radical source which was not considered here is the reaction HO2+NO, which is not a radical initiation source but part of the propagation cycle. Even for urban high NOx conditions, it was demonstrated that this reaction is not a source of new OH radicals and is balanced by the loss of OH radicals by its reaction with VOCs/CO [Elshorbany et al., 2010]. For the low NOx conditions of the “clean days” from the present study, losses of OH radicals by reaction with VOCs/CO will even exceed the OH source by HO2+OH, caused by peroxyradical cross reactions and thus, should not be considered here as an initiation source.
 In conclusion, although the HONO/NOx ratio was much lower compared to other studies (see section 3.1), HONO is still a major OH radical initiation source in the polar atmosphere of Barrow.
 2. Since OH and NO are formed by the photolysis of HONO, also a net source of NO of up to 90 pptv/h can be derived during daytime from the results of the present study. Thus, HONO photolysis may be also a significant source of NOx in Barrow.
 3. The extra daytime source of HONO necessary to explain the measured HONO concentrations shows the same diurnal variation as the actinic flux indicating a photochemical origin, which is in agreement with most HONO source reactions proposed recently based on laboratory and field results (see section 1).
3.3. HOx and NOx Consistency Tests
 Recently, high importance of HONO as a major source of OH radicals in the polar atmosphere was questioned in model studies based on HOx and NOx balances [Chen et al., 2004; Sjostedt et al., 2005; Bloss et al., 2006, 2010; Grannas et al., 2007; Mauldin et al., 2010]. This was mainly explained by chemical interferences of the instruments used. To verify whether the significant net daytime source of HONO is reasonable for the HONO/NOx data of the present study, HOx and NOx consistency tests were also performed here.
 For the HOx balance, the theoretical OH concentration, OHPSS, was compared with the observed OH. For the steady state calculation of OHPSS, it was assumed that the sources of OH radicals are equal to their sinks. For the gas phase OH sources, only the main sources of OH by HONO and O3 photolysis and by the reaction HO2+NO were considered. For the OH radical losses the measured OH reactivity, kOH, was used. OHPSS was calculated by the following equation (10):
 Unfortunately only a scarce data set of the OH reactivity was available during OASIS. In addition, caused by the low precision of the reactivity measurements, only reactivity data higher than the precision of 1.3 s−1 were used here. For the remaining data, reasonable agreement between OHPSS and OHmeas. was observed for the “average clean day” (see Figure 7). However, since the OH-radical propagation source by the reaction HO2 + NO was on average 10 times higher than the OH-radical initiation source by the photolysis of HONO, OHPSS was only slightly reduced when HONO photolysis was not considered (Figure 7). The difference was in between the error limit for OHPSS, which depended almost exclusively on the uncertainties in the HO2 and OH reactivity data. In conclusion, the HONO concentrations of the “average clean day” were not in contradiction with the available data on the OH radical balance, in contrast to other studies [Chen et al., 2004; Bloss et al., 2010; Mauldin et al., 2010].
Figure 7. Measured OH radical concentration (OHmeas.) and for the “average clean day” during OASIS 2009. In addition, the theoretical OH concentrations calculated by equation (10) considering all measured OH sources (OHPSS) and excluding OH initiation by HONO photolysis (OHPSS w/o HONO) are shown.
Download figure to PowerPoint
 Another consistency test for the HONO data is the measured HONO/NOx ratio. Assuming a steady state of the sources and sinks of HONO and NOx, and assuming that all NOx is generated by the photolysis of HONO as a lower limit, the HONO/NOx ratio is similar to the ratio of their lifetimes in the atmosphere τHONO/τNOx. The lifetime of HONO is almost completely controlled by its photolysis, which was 27 min for the “average clean day” during OASIS. Since the average HONO/NOx ratio was 6% (see above), this translates into an average lifetime of NOx of 7.5 h, which seems to be reasonable in view of potential NOx sinks, e.g., NO2+OH, NO2+O3, NO2+HO2, NOx deposition, etc.. For example, assuming an OH radical concentration of 1 × 106 cm−3 (see Figure 7), the lifetime of NO2 due only to its reaction with OH is 19 h, which will be further reduced by the other NOx sinks and is thus not expected to be far from the estimated value of 7.5 h based on the HONO/NOx ratio. The estimated lifetime of NOx is in good agreement with NOx lifetime calculations of 6 h under polar conditions [Bauguitte et al., 2009].
 In conclusion, since the LOPAP instrument was validated against the DOAS technique both, in the atmosphere and in a simulation chamber even under very complex conditions [Kleffmann et al., 2006], and since the consistency tests using HOx and NOx balances during OASIS do not indicate the opposite, there is no reason to doubt the reliability of the LOPAP HONO data.
3.4. Characterization of the Daytime Source of HONO
 For deeper investigation of the light dependency, the extra daytime source of HONO was plotted against and for the “average clean day” in Figure 8. Both photolysis frequencies reflect a different wavelength range and thus, the correlations may help to answer which of the two main proposed source reactions on snow surfaces is of greater importance for the polar atmosphere. Whereas the photolysis of nitrate, reaction (2) [e.g., Zhou et al., 2001] with a corresponding nitrate absorption maximum at ∼300 nm, correlates well with photosensitized conversion of NO2 on humic acids, reaction (4), is active already in the visible and near UV [see Stemmler et al., 2006] and correlates better with Unfortunately, both plots show the same high correlation, which is in contrast to a recent study at lower latitude [Elshorbany et al., 2009]. Caused by the lower diurnal variability of the solar zenith angle (SZA) during polar spring and show a more linear correlation compared to lower latitude measurements. Thus, by simply using the “average clean day” data, the most probable source reaction could not be identified.
Figure 8. Correlation of the extra daytime HONO source with (a) and (b) for the “average clean day” during OASIS 2009.
Download figure to PowerPoint
 For deeper investigation, similar data evaluation than that shown in Figure 8 was also done for individual days using the daytime (10:00–19:00 h) data. However, besides the seven “clean days,” six additional days were investigated, for which the HONO levels were < 100 pptv (NOx < 1.4 ppbv) for most of the day, to increase the number of data points for the correlations, but still to exclude the very extreme periods with NOx concentrations up to 30 ppbv, for which mainly fresh emissions controlled the observed HONO levels (see above). The slopes of all plots of the extra HONO source against and showed high day-to-day variability. This is explained, at least in part, by the variability of the precursors involved. Thus, to differentiate between the two proposed source reactions, i.e., photosensitized NO2 conversion on humic acids (4) and nitrate photolysis (2), was plotted against the corresponding daily averaged NO2 gas phase concentration (see Figure 9a) and against the corresponding daily averaged snow nitrate concentration (see Figure 9b). For the correlation plots using NO2, it was assumed that the gas phase NO2 concentration is correlated to the NO2 concentration in the snow interstitial air, where reaction (4) is expected to take place. Clearly, the correlation shown for all HONO data < 100 pptv in Figure 9a was much higher compared to Figure 9b, indicating that nitrate photolysis (2) is less likely compared to a NO2 dependent long wavelength range process, like the photosensitized conversion of NO2 on humic acids (4). In contrast, when only the seven “clean days” were considered, a positive correlation between and the snow nitrate concentration could be inferred (see Figure 9b). However, the very significant negative intercept of this correlation is not in line with the photochemical HONO source observed (see Figure 6 and Figure 8) and with the proposed nitrate photolysis mechanism (2), which could be inferred from this plot. In contrast, for the correlation of against NO2, the slopes and the negligible intercepts are not depending on the choice of the data in between the experimental errors (see Figure 9a). Thus, a long wavelength range process, like the photosensitized conversion of NO2 on humic acids (4) is proposed to explain the daytime source of HONO in Barrow.
 Interestingly, there was no correlation between the concentration of HONO and NO2 for the “average clean day” (R2 = 0.0035). From similar observations of previous rural studies [e.g., Zhou et al., 2007] reaction (4) was excluded. Thus, the present study demonstrates that correlation studies using HONO concentrations instead of net HONO production rates should not be used to prove any HONO formation mechanism in the atmosphere during daytime. The argument against the use of concentrations is that a strong HONO source in the atmosphere will be masked by the short photolytic lifetime of HONO only during daytime.
 Unfortunately, only a scarce data of the humic acid concentration in snow were available and thus, a correlation of against the product NO2 × humic acids was not possible. Light absorbing aromatics, fulvic and humic acids and soot have been identified to photosensitize heterogeneous conversion of NO2 into HONO [George et al., 2005; Stemmler et al., 2006, 2007; Monge et al., 2010]. Since many organic carbon (OC) species will either not absorb in the actinic spectral range or do not form excited triplet states, which were postulated to photosensitize the NO2 conversion [George et al., 2005], the measured OC in snow is not a suitable marker for NO2 photosensitizers. Besides OC, the integrated absorption coefficient of melted snow in the UV/VIS corrected for the H2O2 and nitrate absorption (Σαλ(residual)300–450 nm) was also measured during OASIS, which may be a marker for the photosensitizer species mentioned above. However, the day-to-day variability of the daily averaged Σαλ(residual)300–450 nm was very low during OASIS (<factor of two). Thus, there was no further improvement of the correlation of against the product NO2 × Σαλ(residual)300–450 nm compared to the one shown in Figure 9a.
 In conclusion, while the nitrate photolysis (2) could be excluded and the contribution of NO2 to a long wavelength photochemical HONO source is very likely (see Figure 9), the contribution of humic acids in snow cannot be verified based on the existing data set. However, since no other strong photochemical HONO sources besides the excluded nitrate photolysis are yet known, photosensitized conversion of NO2 on humic acids (4) is proposed to explain daytime formation of HONO in the polar atmosphere of Barrow.
 A significant uncertainty in the calculations presented here is the fact that no surface HONO fluxes were measured during OASIS. In contrast, an extra HONO source strength was calculated, in which this source was mathematically treated as a gas phase source. However, in the case of an expected surface source, the calculated source strength would inversely depend on the turbulent vertical mixing of the atmosphere, which is induced, at least in part, by wind shearing. Thus, when plotting against the product of [NO2] × 1/(wind speed) an even higher correlation (R2 = 0.88) compared to the one shown in Figure 9a was observed and thus confirming a ground surface source, in good agreement with previous HONO studies over snow surfaces [Zhou et al., 2001; Honrath et al., 2002, Amoroso et al., 2006; Beine et al., 2008]. However, if only the turbulent mixing were the reason for the correlation with NO2 shown in Figure 9a, the extra daytime source would also correlate with any other ground emitted species. This phenomenon was already discussed in the past to understand nighttime formation of HONO in the urban atmosphere and to explain high correlation of HONO with Radon [Febo et al., 1996], which is definitely not a precursor of HONO. However, when plotting against any other ground emitted species like NO, NOx, NOy, CO significantly lower correlations compared to NO2 were observed (NO: R2 = 0.42; NOx: R2 = 0.61; NOy: R2 = 0.41; CO: R2 = 0.41). Thus, it is concluded that NO2 is a precursor of the extra daytime source of HONO in Barrow and that the correlations observed are not simply caused by the boundary layer dynamics.
 In addition, one may argue, that the high correlation of the extra HONO source with as shown in Figure 8, is simply caused by the excellent correlation of JHONO with (R2 = 1.0) [see also Kraus and Hofzumahaus, 1998] and does not reflect any variability of the source process itself. Caused by the low PSS values observed (see Figure 5), JHONO and the HONO concentration are the most important factors in the calculation of the extra daytime source strength. However, it can be expected that the rate coefficient for the production of HONO by any daytime source is much smaller compared to the high photolysis frequency of HONO, JHONO. Thus, in a simplified consecutive process:
 the formation of HONO will be the rate limiting step for the net OH radical production. Accordingly, the diurnal profile of the net OH source by HONO photolysis, as shown in Figure 6, is reflecting the diurnal variation of the source of HONO and not necessarily that of JHONO. For example, if HONO were formed by a light-independent process, it would not accumulate as strongly as observed and no maximum of the HONO/NOx ratio (see Figure 3) would appear during daytime. Only the diurnal variation of the source caused these observations. Thus, the high correlation of the HONO source with see Figure 6 and Figure 8 and with the NO2 concentration, see Figure 9a, strongly supports a daytime HONO source by photosensitized conversion of NO2, for example on humic acids, as proposed from laboratory studies [George et al., 2005; Stemmler et al., 2006, 2007] and recently also confirmed in studies on humic acid doped ice and snow substrates [Beine et al., 2008; Bartels-Rausch et al., 2010].
 Another important question which should be raised here is whether the conditions studied, even for the considered “clean days,” are representative for the remote polar atmosphere. If a day before a “clean day” NOx levels reach 30 ppbv (see for example Figures 1 and 2), one cannot expect a pristine snow surface and thus, the HONO source determined in the present study may be much stronger compared to the “real” remote polar atmosphere. Thus, further reliable HONO measurements under more pristine conditions are necessary.