Atmospheric Measurements at the Foot and the Summit of Mt. Tai-Part I: HONO Formation and Its Role in the Oxidizing Capacity of the Upper Boundary Layer

Chaoyang Xue , Can Ye , Jörg Kleffmann, Chenglong Zhang , Valéry Catoire, Fengxia Bao, Abdelwahid Mellouki , Likun Xue, Jianmin Chen, Keding Lu, Yong Zhao, Hengde Liu, Zhaoxin Guo, Yujing Mu 4* 1 Research Centre for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China 2 Laboratoire de Physique et Chimie de l’Environnement et de l’Espace (LPC2E), CNRS–Université Orléans–CNES, Cedex 2, Orléans 45071, France 3 Physical and Theoretical Chemistry, University of Wuppertal, Gaußstrasse 20, Wuppertal 42119, Germany 4 Centre for Excellence in Regional Atmospheric Environment, Institute of Urban Environment, Chinese Academy of Sciences, Xiamen 361021, China 5 Multiphase Chemistry Department, Max Planck Institute for Chemistry, Mainz 55128, Germany 6 Institut de Combustion Aérothermique, Réactivité et Environnement, Centre National de la Recherche Scientifique (ICARECNRS), Cedex 2, Orléans 45071, France 7 Environmental Research Institute, Shandong University, Qingdao, Shandong 266237, China 8 Shanghai Key Laboratory of Atmospheric Particle Pollution and Prevention, Department of Environmental Science and Engineering, Institute of Atmospheric Sciences, Fudan University, Shanghai 200438, China 9 State Key Joint Laboratory of Environment Simulation and Pollution Control, College of Environmental Sciences and Engineering, Peking University, Beijing, 100871, China 10 Taishan National Reference Climatological Station, Tai’an, Shandong, 271000, China

The NO2 uptake on aerosol surfaces was proposed to be much less important than that on ground surfaces in previous studies 65 because of the low S/V (surface to volume ratio) of particles compared to ground surfaces and the similar reaction kinetics on the same types of surfaces (Nie et al., 2015;Stemmler et al., 2007). However, the contribution of NO2 uptake on aerosol surfaces to HONO formation in the extremely polluted region is not well constrained. For example, previous studies using box models or regional transport chemistry models found the NO2 uptake on aerosol surfaces lead to a negligible impact on daytime HONO formation in the polluted NCP (Liu et al., 2019b;Xue et al., 2020;Zhang et al., 2019aZhang et al., , 2019b. Nevertheless, a recent 70 chamber study (Ge et al., 2019) found a high dark NO2 uptake coefficient (2.0×10 -5 to 1.7×10 -4 ) on NaCl particles under high RH (90%), NH3 (50-2000 ppbv), and SO2 (600 ppbv) conditions. First, such severe pollution rarely occurred. Second, if such a high NO2 coefficient on the aerosol surface was applied in night-time HONO budget analysis, the dominant role of NO2 uptake on the ground surface in night-time HONO formation, which was already generally accepted, might be challenged (Kleffmann, 2007;Kurtenbach et al., 2001;Stutz et al., 2002;Xue et al., 2020). Besides, recent nocturnal vertical 75 measurements of HONO in Beijing found both ground-based and aerosol-derived sources may play important roles in HONO formation during the clean period and haze period, respectively (Meng et al., 2020). Therefore, the contribution of NO2 uptake on the aerosol surface to HONO formation still needs more field constraints.
The photolysis of particulate nitrate (pNO3) was found to be an important HONO source in low NOx areas such as forest canopy and marine boundary layer. High enhancement factors (EF = J(pNO3)/J(HNO3)), within the range of tens to thousands, 80 were proposed in forest area, marine boundary layer, and polluted areas like the NCP (Bao et al., 2020;Ye et al., 2016Ye et al., , 2017Zhou et al., 2007Zhou et al., , 2011. However, model studies with field constraints (Romer et al., 2018;Xue et al., 2020) found that the EF was moderate (7-30) rather than tens to thousands obtained in laboratory studies (Bao et al., 2020;Ye et al., 2016Ye et al., , 2017Zhou et al., 2007). Moreover, a recent laboratory flow tube study (Wang et al., 2021) revealed that the EF was lower than 1 in the aqueous phase. Another flow tube study (Laufs and Kleffmann, 2016) also reported a slow HONO formation from 85 secondary heterogeneous reactions of NO2 produced during HNO3 photolysis. Besides, a very recent chamber study (Shi et al., 2021) found that the EF values of airborne nitrate were lower than 10 (generally around 1), which also indicates an insignificant contribution of nitrate photolysis to HONO formation. Furthermore, when considering the large variation of EF values (from digits to thousands) in the model, model performance on HONO simulations could be improved but accompanied by large uncertainties (Fu et al., 2019;Liu et al., 2019b). Therefore, HONO formation from nitrate photolysis still needs more field 90 constraints.
In addition, the role of HONO photolysis in the oxidizing capacity of the upper boundary layer remains unclear. As there exists a significant gradient in HONO distribution from the ground level to the upper troposphere, HONO photolysis was accounted https://doi.org/10.5194/acp-2021-529 Preprint. Discussion started: 20 July 2021 c Author(s) 2021. CC BY 4.0 License.
to be much less important compared to O3 photolysis (Ye et al., 2018;Zhang et al., 2009). However, in mountainous regions, mountain winds, including mountain breeze (downslope) and valley breeze (upslope) can accelerate the air mass exchange 95 between the mountain top and the ground levels, which may affect HONO levels and the atmospheric oxidizing capacity at the summit level Schmid et al., 2020;Ye et al., 1987).
Herein, atmospheric measurements at the foot (~150 m a.s.l.) and the summit (~1534 m a.s.l.) of Mt. Tai allow us to understand more about 1) the transport of ground-formed HONO and its role in the upper boundary layer; 2) HONO formation from the aerosol-derived sources as the ground sources might be less effective; 3) the oxidizing capacity of the upper boundary layer 100 and its contributors.

Site Description
HONO was alternately measured at two locations: the foot and the summit of Mt. Tai (Figure 1 and S1). The foot station was inside Shandong College of Electric Power, a typical urban site (36. 18°N, 117.11°E). HONO, VOCs, OVOCs, CO, O3, SO2, 105 NOx, PM2.5, PM10, J(NO2), and meteorological parameters were continuously measured at this station. Details about the foot station and the used instruments can be found in the companion study to be submitted to the same journal (entitled "Atmospheric Measurements at the Foot and the Summit of Mt. Tai (1534 m a.s.l.) Part II: HONO Budget and Radical (ROx + NO3) Chemistry in the Lower Boundary Layer"). The summit station (36.23°N, 117.11°E) is located inside a meteorological observatory at the eastern part of the summit of Mt. Tai, with an altitude of about 1534 m a.s.l. It is in the north part of Tai'an 110 city (altitude: ~150 m, population: ~5.6 million), and about 60 km south of Jinan city (the capital city of Shandong province, altitude: ~20 m, population: ~8.7 million).
Since Mt. Tai is a famous tourist place, most of the tourist activities on the summit happen around the Southern Heavenly Gate, the Bixia Temple, and the Jade Emperor Peak. The most crowded period is around sunrise when visitors come for the view of sunrise. The Southern Heavenly Gate is about 1 km west of and about 100 m lower than our station. There are several small 115 restaurants nearby, but they don't cause significant emissions as they only use electricity for the energy supply. The Bixia Temple is about 200 m west to and about 50 m lower than our station, and small anthropogenic emissions may be produced here because of the incense burning, but the impact on our measurements is expected to be negligible as a result of the fast dilution process at the summit level. The Jade Emperor Peak is about 200 m northwest of and has a similar altitude to our station. Visitors generally stay there for a short time and don't have activities that may produce significant emissions. A detailed 120 discussion about the influence of anthropogenic emissions at the summit level on our measurement is presented in Section 3.2.1.

Instrumentation
During the campaign, HONO was continuously measured by the LOPAP technique (LOng Path Absorption Photometer, Model-03, QUMA GmbH, Germany) with a detection limit of 1.5 pptv for 5 min average (Heland et al., 2001;Kleffmann et 130 al., 2006). The performance of LOPAP was well assessed and recorded in different environmental conditions (Heland et al., 2001), including low-NOx and high-altitude sites (Kleffmann and Wiesen, 2008). The LOPAP instrument was installed at the foot station from 29 th May to 8 th July 2017, and then transported to the summit station with successful measurements from 9 th to 31 st July 2017. At the summit station, NO2 was measured by a Model-T500U-CAPS-NO2-analyzer (Teledyne API, USA) that utilizes a patented Cavity Attenuated Phase Shift (CAPS) technique to measure NO2 in the air directly. NO and NOy were 135 measured by API-T200U-NOy-analyzer (Teledyne API, USA) based on the chemiluminescence principle coupled with a remote NOy converter via umbilical to allow measurements with a lower detectable limit of 50 pptv. PM2.5 was measured by a SHARP 5030 monitor (Thermo Scientific, USA). CO and SO2 were measured by a T300U-CO monitor (Teledyne API, USA) and a Model 43C SO2 monitor (Thermo Scientific, USA), respectively. J(NO2) was measured by a 4-π J(NO2) filter radiometer (Metcon GmbH, Germany). Other J-values used in this study, including J(HONO), J(O( 1 D)), and J(HNO3), are calculated by 140 the trigonometric SZA function (MCM default photolysis frequency calculation, see the companion paper and Jenkin et al. (1997)) and scaled by the measured J(NO2). For instance, J(HONO) = J(HONO)model * J(NO2)measured / J(NO2)model.
Water-soluble ions, including particulate nitrate (pNO3) of PM2.5, were collected by filter method and analyzed by an ion chromatograph  every 2 hours late June and early July, but it suffered a sampling problem after 12 th July. Counter (CPC, Model 3775, TSI Inc., USA). Meteorological parameters (temperature, relative humidity, wind speed, wind direction) were measured by instruments from the Shandong Taishan Meteorological Station simultaneously, and details can be found in previous studies at this station . In this study, 10-min averaged data were used for the following analysis. Details about the instrumentation at the foot station could be found in the companion ACP paper. Measurements at 150 the foot station ended on 16 th July. To compare pollutants between the foot and the summit levels during the same period (Section 3.2.2), measurements (only hourly CO, NO2, PM2.5, PM10, O3, and SO2 were available) from the monitoring station (~200 m east to the foot station) were used. Figure 2 shows the meteorological parameters measured at the summit of Mt. Tai during the campaign. The air temperature (T in °C) was slightly lower (~17 º C) in the first two days compared to the period after 10 th July (~20 º C). As clouds were frequently formed at the summit , the observed relative humidity (RH) commonly reached 100%, with a mean of 96%. Based on the wind measurements, air mass at the summit mainly came from the south (direction of Tai'an city), with a mean wind speed (WS) of 5.1 m s -1 . In particular, during the period of 23 rd to 26 th July, high wind speed (1-min max: 19.5 m 160 s -1 , 10-min max: 18.5 m s -1 ) was observed, accompanied by a relatively low temperature, low pressure (p), low radiation (J(NO2)), and high RH.  ppbv) were observed during daytime on some days (i.e., from 14 th to 26 th July). NO mixing ratios were generally lower than 0.5 ppbv due to significant suppression by high O3 levels of usually higher than 50 ppbv. NO2 was generally lower than 2 ppbv with several events, during which NO2 was relatively higher. Besides, the measured HONO mixing ratio varies from 1.1 pptv (close to the detection limit) to 880 pptv, with a mean of 133 pptv and a median of 101 pptv, respectively (Table 1). For the 170 same sampling site at the summit of Mt. Tai, as listed in Table 2, the observed mean HONO mixing ratios in summer is similar to those observed at the same site in winter (150 pptv, December 2017) and spring (130 pptv, March -April 2018) reported by Jiang et al. (2020), but the variation of HONO mixing ratios in summer was within a much narrower range (1 -880 pptv) than in winter (0 -1140 pptv) and spring (0.5 -3230 pptv).

Overview of the Observations 155
With an exception for relatively lower HONO levels at altitudes higher than 2000 m or in the free troposphere (Ye et al., 2018), 175 HONO mixing ratios are significantly higher at the summit of Mt. Tai than at other mountain sites (   Table 2). For example, mean HONO mixing ratios observed at Mt. Whiteface in the USA (Zhou et al., 2007) and Mt.
Hohenpeissenberg in Germany (Acker et al., 2006) were 46 and 100 (daytime)/30 (night-time) pptv, respectively. This phenomenon could be explained by fewer human activities around these mountains, while Mt. Tai locates in the middle of the NCP with a relatively high pollution level. 180 Note that high HONO mixing ratios were observed during the periods from 14 th to 26 th July, with the co-occurrence of high SO2 (a primary pollutant generally emitted at the ground level with a relatively short lifetime). To better understand HONO formation and its role at different pollution levels, data were classified into two periods: high HONO period (HP, 14 th to 26 th ) and low HONO period (LP) that covers all the other days. Statistics of observations during the two periods are summarized in Table 1. Average HONO, NOy, SO2, and PM2.5 during LP are 76 pptv, 4.7 ppbv, 0.3 ppbv, and 12 µg m -3 , respectively, slightly 185 lower than those during HP (194 pptv, 7.0 ppbv, 0.8 ppbv, and 17 µg m -3 , respectively).  *: near or below the detection limit of the used instrument  Kleffmann and Wiesen (2008) and some unpublished data from the study of Kleffmann et al. (2002). High values of NOx/NOy were expected in a very fresh plume with significant local emissions. Throughout the campaign, the 200 average NOx/NOy ratio was 0.43 ± 0.28, which was much lower than fresh plumes observed in the nearest city of Tai'an with an average of 0.93 ± 0.05 (from the measurement at the foot station), indicating an aged air mass and a general small impact of nearby anthropogenic emissions at the summit level.
However, regular local emissions caused rapid increases of some pollutants. As an example, the most rapid increase of HONO and other pollutants, which was observed between 5:20 and 6:20 on 29 th July 2018 is shown in Figure 4. During this event, 205 HONO rapidly increased from 18 to 700 pptv, in concert with rises in NO, CO, PM2.5, NOx, and NOy but a decrease in O3 (Table 3). The synchronous increase in NO (a primary pollutant of combustion) and the decrease in O3 indicates a relatively fresh plume due to the fast titration reaction, as shown in R-1: As there was no significant anthropogenic emission at the summit level, the polluted plume was expected to originate from the 210 level below rather than long transport. During this period, the ∆HONO/∆NOx was 8%, much larger than that inferred from direct emissions (typically inferred as less than 1%). The ratio could be enhanced by: 1) night-time NO2-to-HONO conversion at the ground level and 2) in-plume NO2-to-HONO conversion along the mountain slope (rock and vegetation surfaces, etc.) and 3) particle surfaces as both the boundary layer height (BLH) elevation and the valley breeze are initialized after sunrise.   , and CO were available but NO and NOy were not available at the foot station during this period. All the data were in the same measurement period from 9 th to 31 st July, except for HONO, HONO/NOx, and J(NO2) for the foot station measured from 29 th May to 8 th July.
In Figure 5 the campaign averaged diurnal data is shown, in which most observed species, including HONO, NO, NO2, NOy, 225 CO, and PM2.5, showed small peaks during 6:00 -6:30. This suggests a regular process responsible for this phenomenon rather than an accidental event. Note that the sun started to rise and to heat the ground surface, as well as the mountain surface, one hour before those peaks, leading to an increasing BLH (Anisimov et al., 2017). On the other hand, sunrise would initiate the https://doi.org/10.5194/acp-2021-529 Preprint. Discussion started: 20 July 2021 c Author(s) 2021. CC BY 4.0 License. daytime upslope valley breeze wind (Kalthoff et al., 2000;Schmid et al., 2020;Ye et al., 1987), which could also be supported by the increasing pressure and temperature (1 hour after sunrise) observed at the summit ( Figure S2). Hence, it can be inferred 230 that the morning peaks resulted from the rising air parcel, within which pollutants accumulated during night-time. Interestingly, similar morning peaks were also observed in winter and spring ( Figure S3A), indicating the persistent impact of this process.

Insight on the Seasonal HONO Variations
In addition to the morning peaks analysis, seasonal HONO variations at the summit were also summarized ( Figure 6 and S3), including measurements in winter, spring, and summer. Distinctly higher PM2.5 and NO2 were observed in winter ( Figure 6B  235 and 6C) than in summer. However, HONO levels in winter/spring/summer were similar ( Figure 6A), indicating that the aerosol-derived sources did not dominate HONO formation at the summit level. In general, HONO levels observed at the ground level of the NCP were significantly higher in winter than in summer Nie et al., 2015;Xue et al., 2020).
A similar HONO level observed in summer was possible because of a more rapid vertical exchange between the ground level and the summit level (see Section 3.2.3). 240

Insight on the Comparison of Pollutants at the Foot and Summit Level 245
Comparison of daytime (5:00 -18:00) average PM2.5, CO, O3, and SO2 observed at the foot and the summit stations are shown in Figure 7. It is apparent that all the average daytime levels of primary pollutants (CO and SO2), partially primary pollutant (PM2.5), and secondary pollutant (O3) show very similar variation trends at both monitoring station, revealing 1) a significant or even dominant impact of pollutants at the foot level on that at the summit level, and 2) the presence of a pathway that enables the vertical air mass exchange between the summit and the foot levels. This was also consistent with the higher daytime 250 HONO ( Figure S3A) observed at the summit station in winter than in summer because the regional pollution was generally much severer in winter than in summer.
Besides, during night-time, the summit (~1500 m altitude) is above the boundary layer (in the residual layer), and similar variation trends of pollutants were also found at the foot and the summit stations ( Figure S4 vertical air mass exchange at night. This could also be inferred from the higher night-time HONO ( Figure S3A) in summer 255 than in winter because 1) more south winds (the direction to Tai'an city) were observed in summer ( Figure S5) and 2) the nocturnal boundary layer height was generally much lower in winter than that in summer.

Impact from Tai'an City (150 m a.s.l.)
Besides the discussion in Section 3.2.1, five arguments point to the potential impact from pollution in the nearest city (Tai'an city, ~150 m a.s.l.) on the summit HONO level: a) the "∩" shape of HONO variation in the daytime was different from that of NO2 (a constant level during the daytime), NOy (which increased in the early morning and then remained stable at noontime, followed by a continuous increase in the late 265 afternoon) and PM2.5 (which also showed a "∩" shape variation but its peak was 3 hours later than the HONO peak). These observations indicate that the observed HONO at the summit was not dominated by in-situ aerosol-derived formation ( Figure   5) but an external HONO source such as transport; b) high-level HONO was frequently observed at the ground level (150 m a.s.l.) in Tai'an city (   Table 2), and almost the same variation trends of HONO/NOx were observed at both the summit and foot stations ( Figure 5G); 270 c) HONO peaks at the summit occurred at noontime when the BLH was high, and valley breeze wind was strong; https://doi.org/10.5194/acp-2021-529 Preprint. Discussion started: 20 July 2021 c Author(s) 2021. CC BY 4.0 License. d) high-level HONO (>200 pptv) observed at the summit mainly appeared when the air mass came from south or southwest (the direction to Tai'an city, see Figure 8); e) HONO peaks occurred synchronously with the peaks of SO2, which is mainly emitted at the ground level, and NO2, which is an important HONO precursor (Figure 3). 275 The impact could be achieved through: 1) the air mass ascending by valley breeze upslope wind (daytime) and by the north wind (daytime and night-time, Section 3.2.2.3), and 2) HONO formation during the air mass ascending process, i.e., HONO formation through the NO2 heterogeneous uptake on the mountain slope surfaces (George et al., 2005;Marion et al., 2021;Stemmler et al., 2006). The HONO production from the above processes was defined as P(HONO)transport and will be discussed in Section 3.6. 280 The length of the hypotenuse from the foot to the summit is about 4.2 km, with an average elevation angle of about 20°. In the daytime, the valley breeze could occur with a upslope wind speed of 2 -5 m s -1 reported in previous measurements (Kalthoff 285 et al., 2000;Schmid et al., 2020;Ye et al., 1987; also see https://glossary.ametsoc.org/wiki/Upvalley_wind), it takes about 14 -35 min (ttransport) for the air mass to be transported from the foot to the summit. The south wind could enhance the upslope valley breeze wind; for example, the mean south winds measured at the ground and summit stations are >2 and >5 m s -1 , respectively. If the integrated wind speed along the mountain slope is 4 -10 m s -1 , the calculated ttransport will be reduced to 7 -17.5 min. 290 The key question is the quantity of HONO that still exists after transport from the foot to the summit levels regarding its photolysis in the daytime. Assuming first-order decay of HONO by photolysis during the transport, the remaining HONO and ratio at the summit can be calculated: where ct, c0, J(HONO), and α represent the remaining HONO after a transport period (ttransport), the initial HONO concentration at the foot, the HONO photolysis frequency, and the remaining proportion of HONO. Figure 9 shows the calculated α with ttransport = 7 or 17.5 min during the daytime. It is apparent that α was larger than 40% with ttransport = 17.5 min and larger than 70% with ttransport = 7 min, providing a theoretical basis for the potential role of vertical HONO transport from the ground to the summit levels. This calculation only included the daytime HONO sink through 300 photolysis, but the sources, such as NO + OH and heterogeneous NO2 reactions, were not considered, and hence, the calculated α represents a lower limit. Thus, the impact of transport was expected to be larger when 1) taking other HONO formation paths (e.g., NO2 heterogeneous reactions on the mountain surfaces and the vegetation surfaces) into account, and 2) vegetation shadows on the mountain surface slow down HONO photolysis during the transport. Therefore, ground level (~150 m a.s.l.) HONO as well as its formation during transport may affect the HONO measurement at the summit significantly. The 305 quantification of the contribution will be discussed in Section 3.6.

Daytime Unknown HONO Source Strength 310
The photo-stationary state (PSS), presented by the following equations, is valid to calculate the unknown HONO source strength (Pun) when local emission was negligible (Crilley et al., 2016;Kleffmann et al., 2005;Michoud et al., 2012).The predicted HONO concentration by PSS ([HONO]PSS) and Pun could be calculated by Eq-3 and Eq-4, respectively. OH measurements were not available during this campaign but were available for the summit station in June 2006 (Kanaya et al., 2009). The significant correlation of OH and J(O 1 D) observed during the former campaign was used here to estimate OH 320 concentrations. Since NO + OH was not the dominant HONO source (8.0%) or HONO + OH not the dominant sink (1.9%), the uncertainty caused by the estimated OH should be not significant.
The diurnal variation of the calculated noontime (10:00 -16:00) Pun is shown in Figure 10. Campaign-averaged Pun was about 290 ± 280 pptv h -1 with a maximum of about 1800 pptv h -1 . The maximum Pun value appeared at midday (13:00), indicating a photo-enhanced HONO source. Similarly, high correlations (r = 0.79, 0.83, or 0.83) were found between Pun and NO2*J(NO2), 325 pNO3*J(HNO3), or NOy*J(HNO3) ( Table 4), suggesting the potential HONO formation from photosensitized NO2 reactions or photolysis of NOz (NOz = NOy -NO -NO2) species such as particulate nitrate (pNO3). Moreover, the relatively poor correlations (r = 0.17 or 0.64) between Pun and NO2*Sa or NO2*Sa*J(NO2) ( Table 4) suggested a minor role of NO2 uptake on the aerosol surface in the HONO formation. Besides, a high correlation between Pun and HONO (r = 0.76) was obtained. A possible reason could be that HONO and other pollutants were not dominated by in situ formation but by transport, as discussed 330 in Section 3.2.2. As correlation analysis is only a preliminary indicator and it might not be instructive for HONO budget analysis when the vertical air mass exchange occurs, further investigation of NO2 uptake on aerosol surface and photolysis of pNO3 is presented in Section 3.4 and Section 3.5, respectively.

Constraint on HONO Formation from NO2 Uptake on the Aerosol Surface 335
During the daytime, the HONO production rate from the NO2 uptake on the aerosol surface (P(HONO)a) with the photoenhanced effect was parameterized by the following equation. where (NO2), Sa, a, J(NO2)measured/0.005 are the molecular speed of NO2 (m s -1 ), aerosol surface density (m -1 ), the photoenhanced uptake coefficient of NO2 on the aerosol surfaces, and the photo-enhancement factor normalized to a J(NO2) = 0.005 340 s -1 . In Eq-5, an upper limit HONO yield for the NO2 conversion of 100% was assumed. Additionally, RH was proposed to significantly influence aerosol surface density, especially at our site, with frequently high RH up to 100%. Then besides calculating the aerosol surface density based on the measured aerosol size distribution (Sa_measured), we estimated the effective aerosol surface density in m -1 with an RH enhancement factor ( ) (Sa = Sa_measured * ( )) using the following equation: where a and b are empirical values of 2.06 and 3.60, respectively (Liu et al., 2008). The average Sa without and with RH enhancement is 3.0×10 -4 and 8.3×10 -4 m -1 , respectively. Note that the uncertainty of Sa is not expected to cause a significant uncertainty on HONO budget analysis as P(HONO)a was not the dominant source (Section 3.6).
As Pun includes all the sources except NO + OH, then P(HONO)a << Pun can always be obtained. Hence, the real a value should be much lower than the inferred ones (a_inferred) from Pun = P(HONO)a. In total, 606 a values were inferred based on 350 the measurements, varying from 1.3×10 -4 to 8.5×10 -3 , with a mean of (8.3 ± 7.5) ×10 -4 . However, the minimum (a_inferred_mini) of 1.3×10 -4 is still very high, compared to the results of most lab studies, in which values of a of typically at few times 10 -5 or even less were observed (Han et al., 2016;Liu et al., 2019a;Ndour et al., 2008;Sosedova et al., 2011;Stemmler et al., 2007).
Hence a popularly used value of a = 2×10 -5 was used to calculate P(HONO)a and a_inferred_mini of 1.3×10 -4 was also used for uncertainty analysis as the upper limit. 355 The calculated P(HONO)a with these a values are shown in Figure 10. It is obvious that P(HONO)a is significantly lower than Pun with either lab-based a = 2×10 -5 or even a_inferred_mini = 1.3×10 -4 , pointing out the minor role of NO2 uptake on the aerosol surface in daytime HONO formation. With the lab-based a = 2×10 -5 , P(HONO)a could only explain 3% of Pun, which is similar to previous model studies (Liu et al., 2019b;Xue et al., 2020;Zhang et al., 2016). The contribution of P(HONO)a to Pun increased when using a_inferred_mini, but resulting from an overestimated a as discussed before. Nevertheless, analysis in this 360 study still could be an important effort in the field constraints on the NO2-to-HONO conversion on the aerosol surface.

Constraint on HONO Formation from the Photolysis of Particulate Nitrate
As one of the important inorganic components of aerosols, particulate nitrate (pNO3) could undergo photolysis, with the production of HONO. This process needs more field constraints as discussed in the Introduction section. During the present campaign, pNO3 at the summit was measured by a filter method every 2 hours , but it suffered a sampling problem after 12 th July. Because NOz (NOz = NOy-NO-NO2) mainly contains pNO3 and its precursors, e.g., HNO3 and N2O5, 370 similar variations were expected between NOz and pNO3. As shown in Figure S6, NOz and pNO3 exhibited a very high correlation (R 2 = 0.895), for which pNO3 makes 44% of the NOz and this fraction was used to estimate pNO3 in the period when it was not measured. The uncertainty of the estimated pNO3 should have no significant impact on daytime HONO formation concerning its small contribution to daytime HONO formation (see Section 3.6).
A high correlation between Pun and pNO3*J(HNO3) was found (Table 4), suggesting a possible impact of pNO3 on HONO 375 formation. But one should bear in mind that the high correlation might also be caused by the remarkable impact on pNO3 formation from HONO-related reactions (e.g., ℎ → 2 → 3 → 3 )  or other photolytic processes. For parameterization, an enhancement factor (EF) was defined as the ratio of photolysis frequencies of pNO3 to gas-phase HNO3. Then HONO production from pNO3 photolysis (P(HONO)nitrate) could be quantified by Eq-7: Similar to NO2 uptake on the aerosol surface, one can always find P(HONO)nitrate << Pun. Hence, it is expected that the real EF should be much lower than the inferred ones (EFinferred) from P(HONO)nitrate = Pun. Therefore, 606 EF values were inferred, in the range of 15.6 to 1072, with a mean of 173 ± 98, which is much higher than those (around 1) determined in recent flow tube or smog chamber studies (Shi et al., 2021;Wang et al., 2021). The minimum (EFinferred_mini = 15.6) is at a similar level to field studies of Romer et al. (2018) and Zhou et al. (2003), and the lower values in the laboratory studies (Bao et al., 2018;Ye et 385 al., 2016Ye et 385 al., , 2017Zhou et al., 2011). To quantify the HONO production from pNO3 photolysis, the EF value of 7 from a recent field study (Romer et al. 2018) was used for P(HONO)nitrate calculation, and EFinferred_mini (15.6) from this study and EF values of ~1 from recent laboratory studies (Shi et al., 2021;Wang et al., 2021) were also used for the uncertainty analysis and comparison.
The calculated P(HONO)nitrate with these EF values is shown in Figure 10. With the EF = 7, P(HONO)nitrate was at a level of 390 half of NO + OH but much lower than Pun, which was also observed at the summit of Mt. Whiteface (Zhou et al., 2007). P(HONO)nitrate could explain 4.3% of the observed Pun. While its contribution varied from 0.6% to 9.6%, depending on EF values varying from 1 to 15.6 in the sensitivity tests. Therefore, with a P(HONO)a (a = 2×10 -5 ) contribution of 3% (Section 3.4) and a P(HONO)nitrate (EF = 7) contribution of 4.3%, the other sources (defined as P(HONO)other = Pun -P(HONO)a -P(HONO)nitrate, mainly transported from the ground level as discussed in Section 3.2.2), made a dominated contribution of 395 92.7% to the observed Pun.
Moreover, significant differences between EF values obtained from field studies and laboratory studies indicate a complex process of pNO3 photolysis that may be influenced by various environmental parameters, e.g., the aerosol pNO3 loading and the aerosol composition (Bao et al., 2018(Bao et al., , 2020Ye et al., 2016Ye et al., , 2017, and experimental laboratory conditions, e.g., collected particles on the filter or generated airborne particles (Shi et al., 2021;Wang et al., 2021). We, therefore, suggest that this 400 process still needs further field or laboratory constraints. To our knowledge, this study provided the first field constraint on the aerosol-derived HONO sources in the NCP region, where the abundance of aerosol was frequently observed, and its role in HONO formation is still highly controversial.
The landscape (e.g., mountains) enhances the vertical air mass exchange, leading to a weak vertical HONO distribution within the boundary layer, which is not yet considered in previous studies. This will underestimate the role of ground-derived sources 405 in HONO formation in the upper boundary layer over mountain regions. Regarding the short enough lifetime of OH (<<1 s), it is expected that OH cannot be transported. However, part of HONO was formed at the ground level through NO + OH, and the same amount of OH consumed at the ground level would be released at the summit level through HONO photolysis, constituting a potential pathway of OH transport through the reservoir of HONO. Moreover, the enhanced vertical air mass exchange could also lead to fast transport of other pollutants (PM2.5, O3, CO, SO2, etc.) from the ground to the summit levels, 410 which is expected to significantly impact the atmospheric composition as well as its chemistry in the upper boundary layer or the residual layer. The discussion and implications in this study will be instructive for further laboratory or model studies.

Role of HONO in the Oxidizing Capacity of the Upper Boundary Layer
O3 was typically the major HOx (HOx = OH + HO2) source at high altitude regions, including the upper boundary layer. Then we compared the HOx production rates from O3 and HONO photolysis to investigate whether HONO could play a significant 415 role in the oxidizing capacity of the atmosphere at this high-altitude site. Photolysis of HONO and O3 with their integrated https://doi.org/10.5194/acp-2021-529 Preprint. Discussion started: 20 July 2021 c Author(s) 2021. CC BY 4.0 License.
HOx production is shown in R-2 and R-5 to R-7, respectively. OH loss through HONO + OH and NO + OH was not subtracted from P(HOx)HONO because of the dominated role of vertical transport in the HONO formation at the summit level (net HOx production at the summit level and comparison with the foot level is calculated and discussed in the companion ACP paper). where the reaction constants were taken from the IUPAC kinetic database (https://iupac-aeris.ipsl.fr). The atmospheric RH and 425 temperature largely influenced the branching ratio of R-6 to R-7. The average OH yield (ϕ) during the campaign of 20% was used for calculating OH production from O3 photolysis. Figure 11 displays the diurnal profiles of HOx production rates from HONO and O3 during the clean and polluted periods. It is apparent that O3 photolysis contributed mostly to HOx production, owing to that P(HOx)O 3 was generally more than two times of P(HOx)HONO during daytime except in the early morning and later afternoon. Variations of P(HOx)HONO in the clean 430 period started to increase after sunrise, followed by a decline after the peak at 11:00. The rapid decline of P(HOx)HONO and P(HOx)O 3 around noon was caused by low solar intensity as a result of frequently (near-) saturated humidity and resulting cloud formation (Figure 2 and S1F). The largest P(HOx)HONO of the diurnal profile in the clean period was 1.9×10 6 molecules cm -3 s -1 , and contributed 21% of total HOx production through the photolysis of HONO and O3 (P(HOx)sum), which was close to the average contribution (24%) in the clean period. During the polluted period, however, the curve of diurnal P(HOx)HONO was 435 broader than that in the clean period, with an unexpectedly sustainable high level (>2.0×10 6 molecules cm -3 s -1 ) from 10:20 to 16:00, increasing the contribution of P(HOx)HONO to P(HOx)sum to 27.5%. At 15:00, the maximum average P(HOx)HONO of 3.3×10 6 molecules cm -3 s -1 was observed in the polluted period, contributing about 34% of P(HOx)sum.
In particular, in the early morning at the ground level, HONO photolysis is found to initialize daytime photochemistry after night-time accumulation Kleffmann, 2007;Platt et al., 1980). If the air mass exchange between the ground 440 level and the summit level is fast enough, HONO photolysis may also initialize daytime photochemistry at the summit level.
Then we compared the relative contribution of HOx production rates from HONO and O3 at different hours. As shown in Figure 12, P(HOx)HONO arose earlier than P(HOx)O 3 and dominated HOx production (>50%) in the early morning (5:00 -7:20), demonstrating that daytime atmospheric photochemistry at the summit level is also initialized by HONO photolysis.
Taken together, the average contribution of P(HOx)HONO to P(HOx)sum was about 26%, more than one-third of P(HOx)O 3 (the 445 contribution is 18% if OH loss through HONO + OH and NO + OH is subtracted, see the companion paper). As discussed before, the transport from the ground level to the summit level contributed to the majority of HONO observed at the summit level, pointing to a new insight that ground-derived HONO played an important role in the oxidizing capacity, not only at the https://doi.org/10.5194/acp-2021-529 Preprint. Discussion started: 20 July 2021 c Author(s) 2021. CC BY 4.0 License. ground level but also in the upper boundary layer (~1500 m) in mountainous regions. Yet this vertical exchange might be only valid in the mountainous areas, and the follow-up regional impact still needs to be quantified by further model studies. 450

Summary
Observations of HONO and related parameters at the summit of Mt. Tai (1534 m a.s.l.) in July 2018 were presented. The average HONO mixing ratio is 133 ± 106 pptv, with a maximum of 880 pptv, significantly higher than observations at other mountain summits worldwide. Along with observations at the ground level (the nearest city, Tai'an city), HONO formation 460 from different paths and its role in the atmospheric oxidizing capacity of the upper boundary layer were explored and discussed.
The main conclusions are listed as follows: 1. Constraints on the kinetics of NO2 uptake coefficient on the aerosol surface and photolysis of pNO3 were obtained based on the assumption that Pun could be solely explained by NO2 uptake on the aerosol surface, P(HONO)a or particulate nitrate photolysis, P(HONO)nitrate. The inferred λa and EF values were much higher than most values 465 https://doi.org/10.5194/acp-2021-529 Preprint. Discussion started: 20 July 2021 c Author(s) 2021. CC BY 4.0 License.
obtained from recent laboratory studies, indicating that the aerosol-derived HONO could not explain the observed Pun. In the NCP region, the abundance of aerosol was frequently observed, but its role in HONO formation is still highly controversial as a result of uncertain kinetics. This study provided the first field constraints on aerosol-derived HONO sources in this region and will be instructive for further laboratory or model studies.
2. With a λa value of 2×10 -5 and an EF value of 7, P(HONO)a and P(HONO)nitrate showed small contributions (3% and 470 4.3%, respectively) to daytime HONO formation at the summit station. Both P(HONO)a and P(HONO)nitrate varied from negligible to moderate levels (similar to NO + OH), depending on a and EF values, suggesting the necessity to further study the related kinetics. Additionally, although high values of a (1.3×10 -4 ) and EF (15.6) compared with recent studies were tested here, both sources were still much lower than the observed Pun. The remaining majority (92.7%) of Pun was dominated by the rapid vertical transport from the ground to the summit levels including 475 heterogeneous HONO formation on surfaces of the mountain slope, which was inferred from comprehensive evidence presented in this study.
3. Photolysis of HONO initialized daytime photochemistry in the early morning. In addition, for the daytime average, it contributed 26% of P(HOx)sum, more than one-third of P(HOx)O 3 , indicating the important role of HONO in the oxidizing capacity of the atmosphere in mountainous areas. HONO formation at the ground level could significantly 480 influence the HONO mixing ratios and the atmospheric oxidizing capacity at the summit level through the vertical air mass exchange. Moreover, the enhanced vertical air mass exchange could also lead to a fast exchange of other pollutants between the ground and the summit levels and significantly impact the atmospheric composition as well as the chemistry in the upper boundary layer or the residential layer. However, those follow-up impacts, by far, are not quantified by the current model studies. 485