Interactive comment on “ Impact of HONO on global atmospheric chemistry calculated with an empirical parameterization in the EMAC model ” by Y

In this manuscript, the authors examine the impact of HONO on global atmospheric chemistry with a simple parameterization for HONO, i.e., taking a HONO/NOx ratio of 0.02, in the EMAC model. The agreement is surprisingly good between the predicted and the measured HONO concentrations with this simple parameterization, especially in the urban environments. The predicted effects on atmospheric chemistry (OH and O3) are reasonable in the urban regions. I consider this work to be a good modeling effort to understand the potential effects of HONO on a large scale, with interesting results, and thus recommend it for publication after addressing the concerns that I have below:


Introduction
Since three decades, nitrous acid (HONO) photolysis has been shown to be an important source of hydroxyl (OH) radicals, especially during the early morning when other sources are of minor importance (Perner and Platt, 1979;Harris et al., 1982).Recently, HONO photolysis was reported to contribute more strongly to daytime primary OH production than O 3 photolysis, both under urban and rural conditions (Elshorbany et al., 2009a;Sörgel et al., 2011a, respectively).It has been demonstrated that HONO photolysis can be a major OH initiation source under high NO x (e.g., Kleffmann et al., 2007;Elshorbany et al., 2009aElshorbany et al., , 2010a;;Ren et al., 2006;Dusanter et al., 2009) as well as low NO x conditions (e.g., Kleffmann et al., 2005;Elshorbany et al., 2012, Sörgel et al., 2011a).OH radicals constitute the major oxidant of the atmosphere, initiating the removal of most reactive gases (i.e.regulating the self-cleaning capacity of the atmosphere), though also lead to the formation of secondary products such as O 3 and PAN, which can be harmful for human health.The OH oxidation of volatile organic compound (VOC) contributes to the formation of aerosol particles, affecting air quality and climate.HONO is also an important ingredient of the strongly anthropogenically perturbed global nitrogen cycle, which indirectly influences climate change (Kumala and Petäjä, 2011).
Owing to the thus far incomplete knowledge of HONO sources, in particular during daytime, it was not yet possible to simulate realistic HONO levels using global models.Li et al. (2010), Zhang et al. (2011) and Czader et al. (2012) Published by Copernicus Publications on behalf of the European Geosciences Union.

Y. F. Elshorbany et al.: Impact of HONO on global atmospheric chemistry
have shown that additional HONO sources are required to match measured HONO mixing ratios; for some regions tenfold mismatches have been found based on the known gas phase HONO formation only (i.e., OH + NO → HONO).Li et al. (2011) and Goncalves et al. (2012) showed that modelled HONO was consistently lower than observations, even when the most effective recently suggested formation mechanisms were considered.This underestimation of HONO by models may be expected to impact the simulated HO x and O 3 budgets as well as other secondary products.
Several HONO sources have been identified in the laboratory (e.g., Zhou et al., 2003;George et al., 2005;Stemmler et al., 2006Stemmler et al., , 2007;;Bejan et al., 2006;Li et al., 2008, Gustafsson et al., 2006;Ndour et al., 2008;Su et al., 2011), yet these sources cannot account for the HONO levels observed during daytime (Elshorbany et al., 2010b;Su et al., 2011;Czader et al., 2012).While the heterogeneous conversion of NO 2 on humid surfaces (Finlayson-Pitts et al., 2003) in the dark is commonly accepted as the dominant HONO source during the night (Alicke et al., 2002), the exact mechanism is still unclear.For daytime HONO sources, five photochemical mechanisms were recently identified.Three of them dominate under high-NO x urban conditions, i.e., heterogeneous conversion of gaseous NO 2 on photosensitized solid surface organic compounds (George et al., 2005;Stemmler et al., 2006), photocatalytic conversion of NO 2 on TiO 2 (Gustafsson et al., 2006;Ndour et al., 2008) and the photolysis of gaseous nitro-aromatic compounds (Bejan et al., 2006), which are expected to correlate well with j (NO 2 ).Under low-NO x rural conditions, the photolysis of nitric acid (Zhou et al., 2003(Zhou et al., , 2011) ) adsorbed on solid surfaces (including vegetation) may dominate and is expected to primarily correlate with j (O 1 D), related to the much lower wavelength range of nitric acid photolysis than j (NO 2 ).Su et al. (2011) showed that HONO can also be emitted from soils, being a function of temperature (i.e, light independent).This source may be important in tropical forested regions.Sörgel et al. (2011a) showed that unknown HONO daytime sources, normalized with NO 2 mixing ratios have a clear dependency on j (NO 2 ).Recently, Czader et al. (2012) also showed that photo-induced conversion of NO 2 on organic surfaces is a strong daytime HONO source.
In addition to the incomplete knowledge about daytime HONO chemistry, most of the thus far identified HONO sources are generally also associated with large uncertainty as the mechanisms and controlling parameters are only partly understood (Kleffmann, 2007;Li et al., 2010;Czader et a., 2012;Goncalves et al., 2012).Consequently, several recent attempts to simulate HONO formation were based on the modulation of the model parameters (i.e., via several assumptions and simplifications) to match the measurements in individual studies (e.g., Li et al., 2010;Czader et al., 2012;Goncalves et al., 2012).For example, Czader et al. (2012) simulated HONO using an approach adopted by Li et al. (2010), for which a daytime HONO source with a relatively large uptake coefficient was employed based on variable threshold values of light intensity.However, Czader et al. (2012) obtained unrealistically high HONO concentrations and therefore used the same uptake coefficient but scaled by a different and constant factor to match the measured HONO.In contrast, Goncalves et al. (2012) obtained about three orders of magnitude lower production rates of HONO when using the approach adopted by Li et al. (2010) and reverted to another approach, more suitable to the studied area.Unfortunately, this approach to match measurements cannot be implemented in global models, as each region would need its specific parameterization.Thus, a more general approach is needed to calculate more realistic HONO levels on a global scale.
A common factor between the known major day-and nighttime sources is that they are generally surface-based (except the gas phase photolysis of nitro-aromatic compounds).Indeed, such HONO sources contribute a major fraction to the measured HONO levels (Harrison and Kitto, 1994;Stutz et al., 2002;Veitel, 2002;Kleffmann et al., 2003;Zhang et al., 2009;Sörgel et al., 2011b;Wong et al., 2011) and consequently the dispersion into the atmosphere is a function of turbulent mixing (Sörgel et al. 2011b).This was first proposed by Febo et al. (1996) based on Radon measurements, a species emitted almost exclusively from the ground.Veitel (2002) also showed that HONO correlates linearly with Radon with highest values reported in winter.In addition, both, ground-and aircraft-based gradient measurements showed that HONO mixing ratios near to the ground follow a steep gradient during stable conditions (Veitel, 2002;Zhang et al., 2009, respectively).Thus, HONO mixing ratios appear to be controlled by the surface area in a mixed volume of air (s/v), and the changes can often be accounted for by scaling HONO to NO 2 or NO x (Sörgel et al. 2011b, and references therein).Similarly, Zhang et al. (2011) showed that the spatial distribution of simulated HONO during day-and nighttime is generally consistent with the NO x emission pattern.Kleffmann (2007) concluded that the HONO/NO x ratio is a reasonable measure of HONO as it takes into account the dilution of trace gases during transport, after being emitted or formed at the ground.Several previous studies indicated the HONO/NO x ratio as a proxy for HONO formation under rural (e.g.Acker et al., 2006;Elshorbany et al., 2012;Sörgel et al., 2011b), remote (e.g.Kleffmann and Wiesen, 2008) and urban conditions (e.g.Elshorbany et al., 2009aElshorbany et al., , 2010a)).Furthermore, Villena et al. (2011a) showed that the HONO/NO x ratio and its daytime maximum are independent of altitude.Thus, the HONO/NO x ratio may be a practicable metric to help predict HONO in the atmosphere.
In this study, data sets from 15 different field measurement campaigns around the globe are investigated, confirming that the HONO/NO x ratio is a good proxy predictor of the HONO mixing ratio for use in global models.The impacts of simulated HONO levels on the levels of oxidants (OH, O 3 , PAN,  2012) * no reference could be found and data were adopted from Li et al. (2012).
etc.) are determined using an atmospheric chemistry-climate model.

Model description
The applied modelling system is based on the ECHAM5 general circulation model (Roeckner et al., 2006) and the Modular Earth Submodel System (MESSy, Jöckel et al., 2005) to simulate the meteorology and atmospheric chemistry.The ECHAM5/MESSy Atmospheric Chemistry (EMAC) system is a coupled lower-middle atmospheric chemistry general circulation model (AC-GCM), which has been extensively evaluated (e.g., Jöckel et al., 2006;Lelieveld et al., 2007;Tost et al., 2007;Pozzer et al., 2008;2010, 2012;Brühl et al., 2012).The model structure and setup have been described by Jöckel et al. (2006) and only a brief description is given here.Atmospheric chemical reactions are incorporated in the model through the module MECCA1 (Sander et al., 2005), including the Mainz Isoprene Mechanism, version 2 (MIM2, Pöschl et al., 2000;von Kuhlmann et al., 2004;Taraborrelli et al., 2009).To study the effects of changes in the chemistry while avoiding possible feedbacks of radiatively active gases and aerosols through the meteorology, the radiation scheme has been decoupled, and the model is used in the atmospheric chemistry-transport mode.The radiation code in EMAC therefore uses an ozone climatology (Fortuin and Kelder, 1998), fixed vertical profiles for CH 4 , N 2 O and CFCs and constant mixing ratios of CO 2 .The reference run (Base) has been performed for the year 2005 in T42L31 resolution (i.e., with a triangular truncation at wave number 42 for the spectral core of ECHAM5, and with 31 levels on a hybridpressure grid in the vertical, reaching up to 10 hPa).The T42 resolution corresponds to a quadratic Gaussian grid of approximately 2.8 • ×2.8 • in latitude and longitude.To simulate realistic synoptic conditions, we applied a weak "nudging" towards actual meteorology by the assimilation of analysis data from the European Centre for Medium-range Weather Forecasting (ECMWF) through the Newtonian relaxation of four prognostic model variables: temperature, divergence, vorticity and the logarithm of surface pressure (van Aalst et al., 2004).The first 5 months of the simulation are regarded as spin up period and these results are not considered in our analysis.
Output has been archived as 1-hourly instantaneous fields to capture an hourly resolved diurnal cycle.In the sensitivity runs S1, S2 and S3, HONO levels were parameterized such that the HONO/NO x ratio is fixed at 0.01, 0.02 and 0.04, respectively (see Sect. 3.3).This was achieved by an iterative correction of HONO and NO x (i.e., without disturbing the reactive nitrogen budget), which was applied every minute, thus limiting deviations in the HONO/NO x ratio to within ± 0.005.

Field measurements
Average diurnally resolved data sets from eight different field measurement campaigns (see Table 1) were used to analyse the day-and nighttime HONO chemistry.HOxComp took place at the Jülich Research Centre, Germany during 9-11 July 2005 (we used only the diurnal profile of the cloud free day, 10 July, Elshorbany et al., 2012); the Santiago de Chile campaigns took place during 8-20 March 2005 (Santiago S, Elshorbany et al., 2009a) and during 25 May to June 2005 (Santiago W, diurnal average of cloud free days only, Elshorbany et al., 2010a); the PMTACS-NY campaign took place from 10 July to 2 August 2001 in New York (New York, Ren et al., 2003); the Mexico City Metropolitan Area campaign took place during 14-31 March (MCMA-2006, Dusanter et al., 2009); the DOMINO campaign took place in southwestern Spain during November-December, 2008 (DOMINO, diurnal average of cloud free days only, Sörgel et al., 2011a)   campaign at a 55 m tall building in Santiago de Chile from 17-29 November 2009 at the altitudes of 6 m (3rd floor, Santiago 3) and 53 m (21st floor, Santiago 21) above the ground (Villena et al., 2011a).In addition, measured average concentrations of HONO and NO x from 7 other field measurement campaigns, adopted from Li et al. (2012), were also investigated (see Table 1).These measurement campaigns were performed in three different continents (from 52 • N to 33 • S and from 116 • E to 99 • W) under a range of meteorological conditions.These measurements also encompass a wide range of different NO x conditions, typically from NO sensitive conditions (e.g., HOxComp) to VOC sensitive conditions (e.g., Santiago and New York).Furthermore, the results obtained from the analysis of these data are compared with other field campaigns reported in the literature.The range of HONO mixing ratios obtained from these data sets represents the minimum and maximum HONO levels measured during the last decade and therefore, represent an optimal global overview of HONO measurements.

Intercomparison of field measurements
The average measured diurnal profiles of HONO/NO x , HONO and j (NO 2 ) for the first 8 data sets in It is worth mentioning that measured HONO/NO x ratios, using the LOPAP technique at polar and high altitude mountain sites were also within the above-mentioned range though at lower measured HONO and NO x mixing ratios.For example, diurnal average HONO/NO x ratios of 0.046 and 0.025 were reported at high altitude mountain sites at Jungfraujoch and Zugspitze, respectively (Kleffmann and Wiesen et al., 2008).Similarly, the diurnal average measured HONO/NO x ratio at a polar site in Barrow, Alaska was 0.062 ± 0.048 (Villena et al., 2011b).
Average diurnal profiles of gradient HONO measurements in Santiago de Chile at 6 and 53 m altitude are presented in Fig. 2. The diurnal profiles show a clear vertical gradient with HONO mixing ratios at 53 m of about 60 % of those at 6 m during daytime.Similar HONO gradient were observed previously in the Mexico City metropolitan area during the MCMA-2003 at 70 and 16 m altitude (Volkamer et al., 2010).Both NO and NO 2 show similar gradients while O 3 increases with height owing to the inter-conversion of NO/NO 2 through reactions of NO (emitted at ground level) with RO 2 and HO 2 (Villena et al., 2011a).Due to the stronger absolute gradient of NO 2 , HONO/NO 2 ratios show an opposite strong gradient during daytime while HONO/NO x ratios appear to be almost unaffected.Similarly, Kleffmann (2007) also showed that the HONO/NO x ratio decreases only little with altitude from about 2.5 % at 10 m to about 2 % at 190 m.The photolysis of HONO releasing OH and NO, enhances photochemical oxidation processes, which subsequently catalytically convert NO into NO 2 which in turn photolyses to form O 3 .Consequently, the relative change of HONO as a result of its photolysis is also reflected in the O 3 and NO x levels resulting in a nearly constant HONO/NO x ratio.Therefore, the HONO/NO 2 ratio may be considered a good measure of HONO during the night (see also Sörgel et al., 2011b) while during daytime the HONO/NO x ratio seems to be more suited (see Sect. 3.3).

Average HONO/NO x ratio
To obtain an overview of measured HONO/NO x ratios, daytime and 24 h average mixing ratios of HONO and NO x , based on all available data sets, are plotted in Fig. 3.A high correlation (R 2 = 0.87) is obtained between HONO and NO x during daytime with a linear regression slope of 0.02 ± 0.002 (Fig. 3, upper panel).Similarly, a high correlation (R 2 =0.90) is obtained for the 24 h average values with a linear regression slope of 0.02 ± 0.001.However, for the DOMINO, BERLIOZ, HOxCOMP, PRIDE-PRD2006 (Back Garden), Yufa and Kathmandu campaigns a slightly different regression slope of 0.05 ( ± 0.004) was found (Fig. 3, lower panel).This is likely due to the relatively high NO 2 /NO x ratio during these campaigns (see Table 1), especially dur- mixing ratios observed in different field measurement campaigns.For deta references, see Table 1 and the text.

Fig. 3.
Correlation between daytime (upper panel) and 24 h (lower panel) average HONO and NO x mixing ratios observed in different field measurement campaigns.For details and references, see Table 1 and the text.
ing the night when the influence of direct NO emissions in these regions is small.These measurement campaigns were characterised by a high NO morning peak, decreasing to less than 1 ppbv during the rest of the day due to a change in wind direction as mentioned in the respective references (Sörgel et al., 2011a;Alicke et al., 2003;Elshorbany et al., 2012;Li et al, 2012;Yu et al., 2009, respectively, except for the Yufa campaign, for which the data were obtained from Li et al., 2012).The higher NO 2 fraction during the night leads to higher HONO via its heterogeneous formation in the dark, resulting in higher HONO/NO x ratios compared to other measurement campaigns.This influence is less pronounced during daytime as the NO 2 /NO x ratios for all campaigns are much closer (see Table 1).Since HONO photolysis only influences HO x , O 3 and other oxidants during daytime, the HONO/NO x regression slope of 0.02 ± 0.001 determined above appears to be representative and may be used to investigate the impact on HO x and oxidant chemistry.Detailed analyses of the factors controlling the HONO diurnal profile as a function of NO x are presented in the next section.

HONO diurnal profile
To determine the factors controlling the HONO mixing ratio as a function of NO x , the aforementioned 8 field measurement data sets (HOxComp, Santiago S, Santiago W, Santiago 3, Santiago 21, DOMINO, MCMA-2006 and New York) have been investigated.During the first 6 measurement campaigns, HONO was measured by the sensitive LOPAP (Long Path Absorption Photometer) technique (Heland et al., 2001;Kleffmann et al., 2002), which showed excellent agreement with the (absolute) DOAS measurements, both in smog chamber and in urban atmospheric conditions.The good performance of the LOPAP instrument is related to the efficient correction of chemical interferences (Kleffmann et al., 2006).During MCMA-2006, HONO was measured using LP-DOAS (Dusanter et al., 2009 and references therein), which selectively detects HONO based on it specific UV absorption, thus free of sampling artefacts and chemical interferences (Platt and Perner, 1980).During the New York campaign (Ren et al., 2003) HONO was measured by aqueousphase scrubbing and high performance liquid chromatography (HPLC) analysis (Huang et al., 2002), which has been used (also in a modified form) in several field measurement campaigns (e.g., Zhou et al., 2002;Ren et al., 2010;Zhang et al., 2012).It is worth mentioning that HONO measurements under very clean conditions, using wet-chemical techniques (other than LOPAP) have been reported to be biased by uncorrected interferences leading to artificially high HONO/NO x ratios (Kleffmann et al., 2006;Kleffmann and Wiesen, 2008;Su et al., 2008;Lu et al., 2010;Villena et al., 2011b;Zhang et al., 2012).Recent studies acknowledged this problem and corrected for these interferences (e.g., Su et al., 2008) or modified the technique to minimize them (e.g., Ren et al., 2010;Zhang et al., 2012).Under high NO x polluted conditions this problem is less pronounced (e.g., Kleffmann et al., 2006;Kleffmann and Wiesen, 2008).In the current study, only HONO measurements by LOPAP are considered for low NO x conditions while for high NO x conditions, LOPAP, DOAS and aqueous-phase scrubbing and HPLC or ion chromatography (IC) analysis techniques are used.To analyse the factors influencing the HONO diurnal profile, we distinguish 4 time sectors, midnight to sunrise (sector A), sunrise to mid-noon (sector B), mid-noon to sunset (sector C) and from sunset to midnight (sector D), grouped into two main sectors, nighttime (A and D) and daytime (B and C).The onset of sunrise and sunset is defined at 30 % of the maximum measured j (NO 2 ) values.The duration of each sector period varies based on the j (NO 2 ) values measured in each study.The obtained linear regression expression is then used to calculate HONO mixing ratios in each sector in terms of NO x and the other related parameters.

Nighttime HONO levels
In contrast to daytime HONO levels, which are generally higher than expected based on model calculations (e.g., Kleffmann et al., 2005;Acker et al., 2006;Li et al. (2010); Sörgel et al., 2011a;Su et al., 2011;Elshorbany et al., 2012), HONO concentrations during nighttime can be largely explained by known sources, i.e., the heterogeneous reaction on humid surfaces (Alicke et al., 2002) and direct emissions (Vogel et al., 2003).All measurement campaigns are included in this analysis except MCMA-2006, for which no nighttime HONO data are available.Though the correlation between HONO and NO x during both night sectors, A and D, reveal a similar slope of about 0.02 (see Fig. 4), the HONO/NO x ratio itself is found to be logarithmically related to NO x , which can be explained by the nighttime heterogeneous formation of HONO (Kleffmann et al., 2003).This is confirmed by the HONO/NO x dependency on wind speed (ws) (see Fig. 4).Similarly, Elshorbany et al. (2009a) showed that the HONO heterogeneous rate constant in the dark (calculated using the HONO/NO 2 ratio) correlates inversely with the wind speed.The positive correlation between HONO/NO x and the wind speed in Fig. 4 is related to the anti-correlation typically observed between HONO and the NO 2 /NO x ratio for all investigated campaigns (see Table 1).In addition, the slope of the HONO/NO x ratio versus NO x was found to be about two times higher in sector A compared to sector D (see Fig. 4), resulting in the following two regression equations: Similarly, an anti-correlation between HONO/NO x ratio and NO x was previously reported (Sörgel et al., 2011b).Again, for the correlation between HONO/NO x ratio and ws, the slope in sector A is higher than that in sector D, though with a higher correlation coefficient in sector D (see Fig. 4).Nevertheless, as shown in Fig. 5, simulated HONO using equations (Eq. 1) and (Eq.2) (sector A and D, respectively) is only improved for measurement campaigns with a nighttime HONO/NO x ratio > 0.02, i.e., HOxComp, Santiago S and DOMINO, compared to that using the mean HONO/NO x ratio of 0.02, while for other campaigns measured HONO was reasonably well simulated based on the mean HONO/NO x ratio of 0.02.

Daytime HONO mixing ratio
During daytime, a similar regression slope of 0.02 ± 0.002 between measured HONO and NO x mixing ratio was obtained for all measurement campaigns (see Fig. 6), though with a lower correlation coefficient in comparison to  nighttime values, which is mainly due to the low correlation between HONO and NO x during the afternoon (sector C, see below).However, in contrast to nighttime conditions, the HONO/NO x ratio during daytime is independent of HONO and NO x mixing ratios while it has a light dependency (given by j (NO 2 ), though with a low correlation coefficient (R 2 = 0.23).These results are in agreement with Zhang et al. (2012), who showed that the daytime HONO flux does not correlate significantly with measured NO x mixing ratios, suggesting that under these conditions NO x is not an important precursor of HONO daytime formation.In addition, using the Community Multiscale Air Quality (CMAQ) model, Czader et al. (2012) demonstrated that photochemical HONO formation can be a strong source during daytime, which directly impacts HO x and ozone levels.
During daytime, HONO mixing ratios are controlled by the photostationary state concentration of HONO, [HONO]pss, calculated from the known gas phase chemistry by the following equation: in addition to other sources that are not known or not accounted for, henceforth denoted as "unidentified".Owing to their negligible contributions during daytime, heterogeneous formation of HONO and its deposition on the ground (e.g., Elshorbany et al, 2009;Sörgel et al., 2011a) are ignored.
Given the reported low HONO/NO x ratio of 0.3-0.8% for directly emitted HONO estimated at a NO/NO x ratio of > 0.9 (Kurtenbach et al., 2001)  (1) to Eq. ( 6) for the different sectors (simulated) and those based on the mean HONO/NO x ratio of 0.02 (global mean), compared to measurements in different campaigns.Small spikes appear between sectors due to the different parameterizations applied.
NO/NO x ratios (0.2 to 0.6) and the much higher measured HONO/NO x ratios for the investigated campaigns, emitted HONO contributions are not expected to contribute significantly to measured HONO levels (Su et al., 2008) during the investigated campaigns and are therefore not considered in the following calculation.For the estimation of [HONO]pss the OH concentration and the HONO photolysis frequency (j (HONO)) in addition to measurements of NO and HONO mixing ratios are required, which were available only for four measurement campaigns, HOxComp, Santiago S, Santiago W and DOMINO.For all campaigns [HONO]pss was  mixing ratios versus NO x levels for the investigated measurement campaigns during both, sector B and daytime (sectors B + C).However, the regression slopes for the measured HONO and unidentified HONO levels versus NO x are higher during daytime in comparison to that of sector B, showing that unidentified sources contribute most during the afternoon, again in good agreement with previous studies, which showed that the high afternoon HONO/NO x ratio points towards unknown HONO daytime sources (e.g., Kleffmann et al, 2005;Elshorbany et al., 2009aElshorbany et al., , 2010bElshorbany et al., , 2012)).Consequently, a higher correlation coefficient is obtained for the relationship between HONO, [HONO]pss and NO x during sector B in comparison to the average daytime values.In contrast, for [HONO]pss, similar slopes are obtained during daytime as well as sector B, implying similar average contributions (i.e., of [HONO]pss to HONO measured mixing ratios) during both sectors, B and C. For sector B, the following linear regression expressions are derived: [HONO]pss = (0.014 ± 0.001) [HONO]B = (0.021 ± 0.001) As shown in Fig. 5, HONO mixing ratios in sector B calculated by equation (Eq.5) and by the global mean of 0.02 are similar and in very good agreement with measurements, related to the similar slope of HONO/NO x in sector B and the mean HONO/NO x ratio of 0.02, except for Santiago 3, San-tiago 21 and MCMA-2006.In fact, measured HONO during sector B for Santiago 3 and Santiago 21 could have been accurately simulated based on (Eq.4) alone, which shows that measured HONO during sector B in these data sets can be largely explained by [HONO]pss.However, for consistency we used (Eq.5) for sector B in all data sets.For MCMA-2006, the overestimation of HONO mixing ratios in sector B is due to the very high measured NO x mixing ratio, which is not related to that of HONO (see below).The noon to sunset time period (sector C) is the most important one in the HONO diurnal profile, during which unidentified daytime HONO sources contribute most significantly.The linear regression analysis for this time period was found to result in two different slopes as shown in Fig. 8 (middle panel).For measurement campaigns with a HONO/NO x ratio of ≤ 0.02, i.e., DOMINO, Santiago 3, Santiago 21, New York and MCMA-2006, a high correlation is obtained with a regression slope of 0.017 ± 0.003, while measurement campaigns with a HONO/NO x ratio > 0.02, i.e., HOxComp, Santiago S, Santiago W indicate a regression slope of 0.042 ± 0.01 (Fig. 8, middle panel).The regression slope for measurement campaigns with HONO/NO x ratios ≤ 0.02 is quite comparable to that of [HONO]pss vs NO x (see Fig. 7), which shows that HONO mixing ratios during this time period for these measurement campaigns can be described accurately by [HONO]  small contribution from unidentified sources.Interestingly, a very high correlation (R 2 =0.99) between the HONO/NO x ratio and j (NO 2 ) has been obtained only for measurement campaigns with HONO/NO x ratios > 0.02 (see Fig. 8), from which HONO mixing ratios during sector C can be calculated as: For those with a HONO/NO x ratio ≤ 0.02, a much worse correlation (R 2 = 0.44), even with a negative slope was obtained.Consequently, for measurement campaigns with HONO/NO x ratios > 0.02, calculated HONO mixing ratios based on the above equation (Eq.6) match the measurements well (see Fig. 5), while for those with a HONO/NO x ratio ≤ 0.02, measured HONO mixing ratios are better simulated using the average of 0.02.One potential explanation for this different j (NO 2 ) dependency is that for measurement campaigns with a HONO/NO x ratio ≤ 0.02, unknown daytime HONO sources that are j (NO 2 ) dependent do not con-tribute significantly to the measured HONO mixing ratios.Indeed, for two of these measurement campaigns, DOMINO and New York, unidentified daytime HONO source strengths of about 100 and 150 pptv h −1 , respectively were reported, being very low compared to that of about 0.33, 1.7 and 2.5 ppbv h −1 for HOxComp, Santiago S and Santiago W, respectively.Similarly, the unknown source strengths for other campaigns reported in the literature are also in the range of 200-500 pptv h −1 , 500 pptv h −1 and up to 2 ppbvh −1 calculated for rural, semi-urban and urban conditions, respectively (Kleffmann et al., 2007).For MCMA-2006, an additional daytime HONO source was reported (Dusanter et al., 2009) despite the low average daytime HONO/NO x ratio of about 0.01.However, as mentioned above (Sect.2.2), this is due to the different spatial account of HONO and NO x (i.e., measuring different air masses of HONO and NO x ).During MCMA-2006 HONO was measured using an LP-DOAS system over an optical path length of 5285 m while NO x was measured using a local commercial standard instrument,  which is known to be affected by interferences from other NO y components (e.g., HNO 3 , PAN, RNO 3 , etc).Therefore, the low HONO/NO x ratio of 0.01 is likely to be caused by higher measured NO x levels in different air masses.Unfortunately, no strict method or correlation could be found to predict the HONO/NO x ratio (i.e., > 0.02 or ≤ 0.02).Therefore the different dependency of HONO/NO x on j (NO 2 ) can only be used as an indicator of the presence of significant unknown HONO sources, rather than to predict the HONO concentration.

Unidentified HONO sources
The contribution of unidentified HONO sources to measured HONO during the afternoon (sector C, from noon to sunset) was largest during HOxComp (∼91 %) followed by Santiago W (∼69 %) and Santiago S (∼25 %).As shown in Fig. 9, high correlations are obtained between unidentified HONO mixing ratios and their derived source strength (ppb h −1 ) with j (NO 2 ), j (O 1 D) and temperature for sector C. The good agreement for both j (NO 2 ) and j (O 1 D) relates to the onsets of sunrise and sunset being defined at 30 % of the maximum j (NO 2 ) values, when the j (O 1 D) values correspond to about 10 % of their daytime maximum.This method allows investigating the dependency of unidentified daytime HONO sources on both, j (NO 2 ) and j (O 1 D).In previous studies (e.g., Elshorbany et al., 2009aElshorbany et al., , 2012)), better agreement was obtained for the correlation between unidentified HONO sources and j (NO 2 ) in comparison to j (O 1 D) by considering the full photolysis frequency range of these photolysis rates, and the latter produces an intercept due to the smaller span of j (O 1 D).However, the good correlation between unidentified HONO sources (see Fig. 9) and temperature (in addition to j (NO 2 ) and j (O 1 D)) implies that photolytic sources active at longer wavelengths (e.g., photosensitized conversion of NO 2 on organic surfaces (Stemmler et al., 2006) or photolysis of nitro aromatic compounds (Bejan et al. ( 2006)) or those that are j (O 1 D) dependent, e.g., the photolysis of HNO 3 (Zhou et al., 2003(Zhou et al., , 2011) ) or those that are temperature dependent (Su et al., 2011), are all of significance.Despite that these results leave the question of unknown HONO daytime sources open, they offer linear regression equations that can be used as an indicator for unknown HONO production within the range of conditions investigated (i.e., at a HONO/NO x ratio > 0.02).
Interestingly, the unidentified HONO sources were found to correlate inversely with j (NO 2 ), j (O 1 D) and temperature (see Fig. 9).The highest absolute values of the unidentified sources are estimated during the winter season in Santiago W, under highly polluted conditions, coinciding with lower actinic fluxes and temperatures, compared to Santiago S, followed by HOxComp.Veitel (2002) showed that nocturnal HONO mixing ratios directly correlate to Radon, especially in winter.He also measured the highest HONO mixing ratios of 2.4 ppb at night in January (1999) coinciding with very high Radon levels.Thus, Veitel (2002) concluded that surfaces at or near the ground form the most prominent HONO source both by day and night.Similarly, Febo et al. (1999) stated that especially trace gases emitted or formed near the ground show a direct and strong correlation with Radon.Their results agree with our findings that unidentified sources are mainly ground-based with highest contributions in the winter season.The negative slopes obtained in Fig. 9 should not be confused with the positive slopes obtained in Elshorbany et al. (2009aElshorbany et al. ( , 2012) )   conditions) and therefore may only relate to the similar diurnal cycle of j (NO 2 ) and the unidentified HONO sources (i.e., may not necessarily follow any physical dependency).The correlations in Fig. 9 are also quite interesting because they show that unidentified HONO sources during the afternoon can be directly calculated as a function of j (NO 2 ) or j (O 1 D) (also vs. temperature but with less accuracy) as follows: Unidentified HONO(ppbv h −1 ) = (7) (−1.78 ± 0.94) × 10 5 j (O 1 D) + (3.65 ± 1.12) Indeed, considering an estimated daytime average j (NO 2 ) for Xinken, China (PRIDE PRD2004) and Back Garden, China (PRIDE-PRD2006) of about 0.005 and 0.007 s −1 , the approximate unidentified HONO source strength based on the relationship (Eq.7) would be about 2.5 ± 1.1 and 0.78 ± 0.34 ppb h −1 , respectively which is in good agreement with those reported by Li et al. (2012).Nevertheless, since the above correlations were derived based on a limited number of measurement campaigns, further evaluation is required.

Measured vs simulated HONO levels
Figure 10 shows the correlations between measured and simulated HONO mixing ratios based on equations (Eq. 1) to (Eq. 6) and the mean HONO/NO x ratio of 0.02 for all investigated measurement campaigns.It appears that measured HONO diurnal profiles for all campaigns can be simulated quite well for both 24 h and daytime averages with  a linear measured/simulated regression slope of 1.10 ± 0.05 and 0.91 ± 0.07, respectively.Similar agreement is also obtained between measured and calculated HONO mixing ratios using the mean HONO/NO x ratios of 0.02 with slopes of 1.02 ± 0.05 and 1.06 ± 0.13, respectively, albeit with lower correlation coefficients, which is due to the underestimation of measured HONO mixing ratios during the afternoon for measurement campaigns with HONO/NO x ratios > 0.02 (see Sect. 3.2.2).This can only be accounted for by equation (Eq.6).However, the calculation of afternoon HONO mixing ratios using (Eq.6) requires the prior knowledge of the HONO/NO x ratio (see Sect. 3.2.2),which is not feasible for a prognostic method.In addition, during daytime the slope of the relationship between HONO and NO x for all measurement campaigns is 0.022 ± 0.002, which is quite similar to the overall mean HONO/NO x ratio.Therefore, the overall mean HONO/NO x ratio of 0.02 is considered to be most suitable to approximate the HONO mixing ratios and investigate the global impact of HONO photolysis during daytime on HO x and O 3 chemistry.

Global impact
In global models HONO is typically simulated based on the known gas phase chemistry (i.e., only accounting for [HONO]pss, see Sect.3.2.2),thus significantly underestimating ambient HONO concentrations.Consequently, HO x , O 3 and other secondary oxidation products from HONO photolysis during daytime are also expected to be significantly underestimated.Since most polluted regions with high NO x levels are located in the Northern Hemisphere, the impact of using more realistic HONO levels should be most pronounced in this part of the world, where NO levels are often sufficient for efficient radical recycling (Lelieveld et al., 2002;Elshorbany et al., 2010b).To investigate the impacts by HONO photolysis, including the seasonal dependency, we focus on model output for the months June and December.
The global mean HONO/NO x ratio of 0.02, empirically derived from the correlation between HONO and NO x has been implemented in the EMAC model.In addition, the sensitivity runs S1 and S3 apply differences by a factor of two, thus using HONO/NO x ratios of 0.01 and 0.04, respectively.As shown in Fig. 11, simulated HONO levels based on a HONO/NO x ratio of 0.02 (S2) are typically more than order of magnitude higher than those simulated by the base run, which considers only gas phase HONO formation (i.e., OH + NO → HONO).These results are in agreement with Li et al. (2010), Zhang et al. (2011) and Czader et al. (2012) who also showed that measured HONO mixing ratios are about tenfold of those calculated when this reaction is the www.atmos-chem-phys.net/12/9977/2012/only formation mechanism.The simulated HONO levels in the S2 run in December are higher than those during June, which is also in good agreement with previous studies, showing that HONO levels in winter are higher than in summer (e.g., Elshorbany et al., 2010a).The simulated HONO levels in June in South Korea and in December in east China compare well to those measured during May-July 2005 in Seoul (Korea) and during Oct-Jan 2004/2005 in Shanghai (China) (Li et al., 2012, and references therein).Furthermore, the day (24 h) average HONO mixing ratios, averaged over the Northern Hemisphere (from 30 • N to 50 • N) show a clear gradient within the first 1200 m above the ground (i.e., the lowest model layer is centred at ∼30 m), both in summer (on 29 June, 2005) and winter (on 30 December, 2005), being most apparent in highly polluted regions, especially in summer (see Fig. 12).These results are in agreement with HONO gradient measurements in previous studies (Volkamer et al., 2010;Vellina et al., 2011;Wong et al., 2011Wong et al., , 2012)).
Figure 13 shows that the shape of the simulated diurnal HONO profile on 10 July 2005, in Northwest Germany (50.93 • N, 6.36 • E) using the base model, differs strongly from the observations during the HOxComp and  other campaigns presented in Fig. 1.Owing to the coarse spatial resolution of the model the profiles in Fig. 13 represent larger areas than the measurements, though the agreement is rather good, probably indicating that HOxComp is representative of this area.In contrast to the base scenario, the shape of the simulated HONO diurnal profiles, S1, S2 and S3 based on the HONO/NO x ratios of 0.01, 0.02 and 0.04, respectively match the measurements much better with maximum HONO mixing ratios during the early morning and minimum values during afternoon.The simulated HONO by the S2 run (based on a HONO/NO x ratio of 0.02) agrees best with the measurements.These results corroborate our earlier conclusion that simulating HONO based on a mean HONO/NO x ratio of 0.02 provides a realistic representation of HONO, rather than considering only the gas phase formation of HONO.Therefore, the S2 run is used hereafter to investigate the impact of implementing this more realistic HONO parameterization.As also shown in Fig. 13, enhancements in OH, HO 2 , O 3 , PAN, H 2 O 2 , NO x and HNO 3 are calculated only at NO levels of about 1 ppbv (marked by the dashed line in the diurnal NO profile).When NO levels are below 1 ppbv, enhancements are typically small or negligible.This is due to the low recycling efficiency (under NO sensitive conditions), under which any increase in the OH levels as a result of HONO photolysis does not lead to a significant increase in the HO 2 levels from reaction (RO 2 +NO → HO 2 ), H 2 O 2 from the reaction of (HO 2 +HO 2 →H 2 O 2 ) or O 3 from the reaction of (RO 2 /HO 2 +NO → NO 2 →O 3 ).These results show that an increase in the HO x levels as well as in the secondary oxidation products only occurs under conditions of efficient recycling of OH through RO 2 /HO 2 , in agreement with the sensitivity analysis performed for HOxComp (Elshorbany et    Li et al., 2011).
To further investigate the impact of the more realistic HONO levels in relatively polluted areas (high-NO x , VOClimited conditions), Fig. 14 compares the simulated diurnal profiles in the eastern USA and downwind on 1 June 2005.It appears that the shape of the simulated diurnal HONO profiles from the runs S1, S2 and S3 match well to those measured in field campaigns (see Fig. 1), in contrast to the base scenario.As shown in Fig. 14, significant increases in HO x and the secondary oxidation products result from the more realistic HONO levels (simulated by S2).Unfortunately, no HONO field measurements are available for comparison in this area for 2005.The simulated maximum O 3 levels (S2) agree with the reported levels in New York State in 2005 (New York State Department of Environmental Conservation).Similar to the HOxComp case, NO and NO 2 decrease while HO x , O 3 , PAN, H 2 O 2 and HNO 3 increase, with higher relative enhancements compared to HOxComp, owing to the relatively high NO x conditions during daytime (see Fig. 14).These results underscore that the more realistic representation of HONO in a global model leads to an enhancement in simulated HO x and oxidation products, predominantly under high NO x conditions.Figures 15 and 16 show the absolute and relative enhancements, respectively, in OH and O 3 together with simulated NO levels.As shown in Fig. 15, in the relatively polluted Northern Hemisphere average monthly NO mixing ratios in December are typically two times higher than in June.In addition, in June NO levels are relatively high in the northeastern USA and in northern Europe, e.g., in comparison to eastern China.Consequently, the enhancements of OH and O 3 levels are relatively strong in these high-NO x regions.Also, over eastern China in December where NO levels are higher than in the USA and Europe, the enhancements of OH and O 3 levels are larger.In addition, the relative enhancements of OH and O 3 are significantly higher in the winter hemisphere compared to summer (see Fig. 16).Similarly, previous studies showed that the impact of HONO photolysis on OH is higher in winter than in summer (Aumont, 2003;Elshorbany et al., 2010a), owing to the relatively minor importance of the other photochemical sources such as O 3 and HCHO (Elshorbany et al., 2010b), in addition to the higher NO levels in winter.Thus, the increased HONO levels in the model significantly enhance the oxidation capacity in polluted regions, especially in winter, when other photolytic OH sources are of minor importance.
Figures 17 and 18 show some details of the modelcalculated enhancement in OH, O 3 , NO x , HNO 3 and PAN over the eastern USA in June and eastern China in December, together with the NO mixing ratios (top panels in Fig. 17).It is evident that the relative enhancements in these species follow the relative changes in NO.In these high-NO x regions, monthly average OH enhancements up to about 1.5×10 6 and 0.6×10 6 molecules cm −3 are calculated for June and December 2005, respectively, while O 3 is calculated to increase up  to 12 and 8.5 ppbv, respectively.These increases are highly significant, corresponding to relative OH and O 3 enhancements of about 80 and 30 % in June and about 90 and 40 % in December, respectively.These results are quite comparable to those in previous studies for specific locations applying small-scale models.For example, model simulations of daytime HONO in Mexico City, which could only account for about 30-50 % of the observed HONO, indicate an average midday O 3 enhancement of 6 ppbv (Li et al., 2010).Similarly, Li et al. (2011) reported an average O 3 enhancement of 12 ppbv (∼50 %) compared to the base case considering only gas phase HONO formation.These results are also in good agreement with Elshorbany et al. (2009b) who compared box model calculations constrained by measured HONO to those only considering OH + NO → HONO.In the latter case, O 3 levels decreased on average by 30 % during daytime.Similarly, Harris et al. (1982) and Clapp and Jenkin (2001), employing photochemical trajectory models, showed that HONO photolysis may have an important impact on the level of oxidants mostly under polluted high-NO x conditions.Jorba et al. ( 2012) also showed that O 3 levels are enhanced in response to additional OH production only in high-NO x regions though can be decreased in regions under low-NO x conditions.
These results thus illustrate that the HONO/NO x ratio is a reasonable proxy to simulate HONO levels in global models, thus avoiding the strong underestimation of HONO concentrations, with consequences for simulated HO x and sec-ondary oxidation products.In view of the many unknowns and uncertainties about daytime HONO sources, global models could apply this method until the chemistry of HONO is resolved in greater detail.

Conclusions
HONO, NO x and auxiliary atmospheric chemistry parameters have been investigated using data from 15 field measurement campaigns around the globe (from 52 • N to 33 • S and from 116 • E to 99 • W).The high correlation between HONO and NO x in all data sets reveals a robust and consistent linear regression slope of 0.02 ± 0.002.Our analysis indicates that, given the ambient NO x mixing ratio, the HONO/NO x ratio is a suitable proxy of HONO mixing ratios both during night and daytime.It should be mentioned, however, that owing to the limited number of field measurements and the large uncertainties associated with HONO measurements under very low NO x and HONO conditions, i.e., remote from pollution sources such as high altitude and polar regions, the obtained average HONO/NO x ratio of 0.02 should be applied with caution.Additional HONO and NO x measurements with improved measurement techniques (i.e, avoiding or correcting for interferences) under these conditions are needed to further evaluate the findings presented in this paper For nighttime conditions, we derive a linear regression slope of 0.02 ± 0.001 based on all investigated campaigns.The HONO/NO x ratio itself appears to follow a logarithmic relationship with NO x , which can be explained by the nighttime heterogeneous formation of HONO.This is corroborated by the HONO/NO x dependence on wind speed.In addition, the regression slope of the HONO/NO x ratio versus NO x is different during both nighttime sectors A (midnight to sunrise) and D (sunset to midnight), from which the HONO concentration can be calculated as a function of NO x levels.
During daytime, a similar regression slope of 0.02 ± 0.002 between measured HONO and NO x mixing ratios is obtained for all investigated measurement campaigns.However, in contrast to nighttime conditions, the HONO/NO x ratio is independent of HONO and NO x mixing ratios, in agreement with previous studies.The correlations between [HONO]pss and NO x during time sector B (sunrise to noon) as well as during daytime (time sectors B + C) reveal a similar regression slope of 0.014, implying comparable contributions dur-ing both time sectors B and C. In contrast, the correlations between unidentified HONO and NO x mixing ratios during the time sector B reveal a regression slope of 0.007, much lower than the 0.011 obtained for daytime, implying higher contributions by unidentified sources during the afternoon, also in agreement with previous studies.In contrast to time sector B, the correlation between HONO and NO x during sector C (noon to sunset) results in two different correlation slopes.For measurement campaigns with HONO/NO x ratios ≤ 0.02, a regression slope of 0.017 ± 0.003 is obtained, while measurement campaigns with HONO/NO x ratios > 0.02 follow a regression slope of 0.042 ± 0.01.These results imply that for conditions with HONO/NO x ratios ≤ 0.02, HONO mixing ratios during time sector C are largely represented by [HONO]pss, with small additional contributions by unidentified sources.A very high correlation (R 2 = 0.99) between the HONO/NO x ratio and j (NO 2 ) is obtained www.atmos-chem-phys.net/12/9977/2012/only for conditions with a HONO/NO x ratio > 0.02, while for those with HONO/NO x ratios ≤ 0.02, a much worse correlation (R 2 =0.44), even a negative slope is obtained.A potential explanation for this different j (NO 2 ) dependency is that during measurement campaigns with HONO/NO x ratios ≤ 0.02, unknown daytime HONO sources that are j (NO 2 ) dependent do not contribute significantly.
High correlations are also obtained between unidentified HONO and its source strength (ppb h −1 ), and j (NO 2 ), j (O 1 D) and temperature for time sector C (noon to sunset).These results imply that photolytic sources active at longer wavelengths or those that are j (O 1 D) dependent are of significance.In fact these results suggest that sources that are solely temperature dependent may also be important, as predicted by Su et al. (2011).Interestingly, we also find that unidentified HONO sources correlate inversely with j (NO 2 ), j (O 1 D) and temperature, from which they can be estimated.
The mean HONO/NO x ratio of 0.02, derived from all measurement campaigns investigated, has been implemented into the EMAC chemistry-climate model to calculate the global HONO distribution.Strong HONO enhancements are predicted relative to the base model, which only accounts for the gas phase reaction of OH with NO.Reasonable agreement is obtained between simulated and measured HONO mixing ratios during both summer and winter.The simulated HONO mixing ratios appear to have a significant impact on HO x , O 3 and other oxidants, however, predominantly under polluted high-NO x conditions.Furthermore, the relative enhancements in OH and secondary products are higher in winter than in summer, thus enhancing the oxidation capacity in polluted regions, especially in winter, when the other photolytic OH sources are of minor importance.The results of the current study suggest that realistic HONO levels should be accounted for by global models and that simulated HO x , O 3 and secondary oxidation products should be revised accordingly.

Fig. 1 .
Fig. 1.Diurnal cycles of the HONO/NO x ratio, HONO and j (NO 2 ) observed in different field measurement campaigns.For abbreviations and references, see the text.

Figure 3 :
Figure 3: Correlation between daytime (upper panel) and 24 h average HONO a

Figure 4 :
Figure 4: Correlation of nighttime average HONO and NO x mixing ratios as well as wind speed (ws) for sector A (right panel, from midnight to sunrise) and sector D (left

Fig. 4 .
Fig. 4. Correlation of nighttime average HONO and NO x mixing ratios as well as wind speed (ws) for sector A (right panel, from midnight to sunrise) and sector D (left panel, from sunset to midnight).Error bars represent 1 σ of the mean.

Figure 5 :
Figure 5: Simulated diurnal HONO cycles based on the equations E1 to E6 for the different sectors (simulated) and those based on the mean HONO/NO x ratio of 0.02 (global mean), compared to measurements in different campaigns.Small spikes appear between sectors due to the different parameterizations applied.

Fig. 5 .
Fig. 5. Simulated diurnal HONO cycles based on the Eq.(1) to Eq. (6) for the different sectors (simulated) and those based on the mean HONO/NO x ratio of 0.02 (global mean), compared to measurements in different campaigns.Small spikes appear between sectors due to the different parameterizations applied.

Figure 6 :
Figure 6: Correlations between HONO, NO x and j(NO 2 ) daytime average values for different measurement campaigns.Error bars represent 1 σ of the mean.

Fig. 6 .
Fig. 6.Correlations between HONO, NO x and j (NO 2 ) daytime average values for different measurement campaigns.Error bars represent 1 σ of the mean.

Figure 8 :
Figure 8: Correlations between average afternoon (sector C, mid-noon to sunset) HONO mixing ratios versus the corresponding NO x and j(NO 2 ) values observed in different measurement campaigns.Error bars represent 1 σ of the mean.

Fig. 8 .
Fig. 8. Correlations between average afternoon (sector C, mid-noon to sunset) HONO mixing ratios versus the corresponding NO x and j (NO 2 ) values observed in different measurement campaigns.Error bars represent 1 σ of the mean.

Figure 9 :
Figure 9: Correlation between unidentified HONO mixing ratios and their source strength versus j(NO 2 ), j(O 1 D) and temperature observed in field measurement campaigns with a HONO/NO x ratio of >0.02.Error bars represent 1σ of the mean.

Fig. 9 .
Fig. 9.Correlation between unidentified HONO mixing ratios and their source strength versus j (NO 2 ), j (O 1 D) and temperature observed in field measurement campaigns with a HONO/NO x ratio of > 0.02.Error bars represent 1σ of the mean.

Figure 10 :
Figure 10: Correlations between measured and calculated HONO mixing ratios for 24 h (upper panel) and daytime averages (lower panel) based on the HONO/NO x dependency (simulated, left panel, see text for details) and that calculated using the mean HONO/NO x ratio of 0.02 (global mean, right panel) for the different measurement campaigns.Error bars represent 1 σ of the mean.

Fig. 10 .
Fig.10.Correlations between measured and calculated HONO mixing ratios for 24 h (upper panel) and daytime averages (lower panel) based on the HONO/NO x dependency (simulated, left panel, see text for details) and that calculated using the mean HONO/NO x ratio of 0.02 (global mean, right panel) for the different measurement campaigns.Error bars represent 1 σ of the mean.

Figure 11 :
Figure 11: Simulated monthly average HONO mixing ratio near the surface in ppbv for the

Fig. 11 .
Fig. 11.Simulated monthly average HONO mixing ratio near the surface in ppbv for the base (upper panel) and S2 (lower panel) scenarios in June (left) and December (right).

Figure 12 :
Figure 12: Simulated one day average HONO mixing ratios (ppbv), averaged over the

Fig. 12 .
Fig. 12. Simulated one day average HONO mixing ratios (ppbv), averaged over the Northern Hemisphere (30 • N to 50 • N) during summer (on 29 June 2005, left) and winter (on 30 December, 2005, right) within the first 1200 m above the ground.

Figure 15 :
Figure 15: Simulated monthly average enhancement of OH (middle panel, 10 6 molecules cm - 3 ) and ozone (lower panel, ppbv) near the surface, indicating the difference between

Fig. 15 .
Fig. 15.Simulated monthly average enhancement of OH (middle panel, 10 6 molecules cm −3 ) and ozone (lower panel, ppbv) near the surface, indicating the difference between the S2 and base scenarios as a function of NO (base scenario, upper panel, ppbv) in June (left) and December (right).

Figure 16 :
Figure 16: Monthly average relative enhancements (in %) in OH (upper panel) and ozone

Fig. 16 .
Fig. 16.Monthly average relative enhancements (in %) in OH (upper panel) and ozone (lower panel) near the surface, indicating the difference between the S2 and base scenarios (relative to the latter) in June (left) and December (right).

Figure 17 :
Figure 17: Simulated monthly average enhancement of OH (middle panel, 10 6 molecules cm - 3 ) and O 3 (lower panel, ppbv) near the surface as a function of NO (base scenario,

Fig. 17 .
Fig. 17.Simulated monthly average enhancement of OH (middle panel, 10 6 molecules cm −3 ) and O 3 (lower panel, ppbv) near the surface as a function of NO (base scenario, upper panel, ppbv) in June over the eastern USA (left) and December over eastern China (right).

Fig. 18 .
Fig. 18.Simulated monthly average enhancement of NO x (upper panel), HNO 3 (middle panel) and PAN (lower panel) in ppbv near the surface in June over the eastern USA (left) and December over eastern China (right).

Table 1 .
Summary of the field measurement data sets used.

field measurement 9980 Y. F. Elshorbany et al.: Impact of HONO on global atmospheric chemistry
and we used two gradient measurement data sets obtained during a