Observation of strong NOx release over Qiyi Glacier, China

NOx is released from sunlit snowpack surfaces, and this significantly influences the oxidizing capacity of the clean boundary layer atmosphere and the potential interpretation on the historical atmospheric composition recorded in the ice core. The Tibetan Plateau is an important snow-covered region in the northern midlatitudes, with strong solar radiation and relatively high NO3 in snow/ice. Released NOx on the glacier surface of the Tibetan Plateau should have a higher concentration than in 10 Antarctic and Arctic regions. To verify this hypothesis, field observations were carried out at 4600 m asl in Qiyi Glacier in late August 2004. In late August, the surface ultraviolet-B (UVB) radiation level at 4600 m asl in Qiyi Glacier reached >4.5 W/m and was increased by the strong reflection of snow/ice and clouds against the sun, and strengthened by the topographical effect. The concentrations of NO3 and NH4 in water from melting snow were hardly detected, but the average concentration (±1σ) of NO3 in snow samples was 8.7 ± 2.7 μmol/L. Strong correlations were observed between NOx (NO2) mixing ratios 15 and UVB radiation levels in the Tibetan glacier. Vertical experiments revealed a negative gradient of NOx (NO2) mixing ratios from the glacier snow surface to a height of 30 cm. As a result of the high levels of UV radiation and high NO3 concentrations in snow/ice, the mixing ratios of NOx released by fresh snow in Qiyi Glacier in late August reached to several parts per billion (ppbv) and were approximately 1 order of magnitude higher than those observed in polar regions. This observation provides direct evidence to support the research hypothesis and confirms that the release of high concentrations of NOx in the boundary 20 layer of highland glaciers and snow surfaces.


Introduction
NOx and other gases may be released from sunlit snowpack in Antarctic and Arctic regions, significantly influencing the oxidizing capacity of the polar boundary layer atmosphere and the potential interpretation of the historical atmospheric composition recorded in the ice core (Honrath et al., 1999;Summer and Shepson, 1999;Davis et al., 2001;Mauldin et al., 25 2001;Zhou et al., 2001;Dominé and Shepson, 2002;Honrath et al., 2002;Jone and Wollf, 2003;Mauldin et al., 2004;Wang et al., 2011). The ice core record of nitrate was used for the estimation of preindustrial NO emission levels (Preunkert et al., 2003). This promotes scientific awareness and research on the importance of photochemical processes on polar ice surfaces (Bottenheim et al., 2002;Davis et al., 2004;J. Erbland et al., 2013;Van Dam et al., 2015;Bock et al., 2016;Chan et al., 2018).
NOx release in the photolysis experiments of natural snow was observed not only in polar regions but also in midlatitudes

Result 80
It snowed heavily during 22-23 August 2004, and then it was cloudy; the wind speed was strong on 24-25 August.
Measurements were obtained from this fresh snow surface from 26 August, through its melting and spoilage processes, and no new snowfall was noted in the following observation days. After heavy snowfall, 26 August was the first sunny day. The wind was strong around noon and in the afternoon, and relatively low in the morning and evening. The clouds drifted one after another, and the UVB radiation level exhibited large variations during the daytime. Signs of Loess dust falling into glaciers 85 were clearly identified around this time ( Figure S1). Figure 2 shows the diurnal variation in UVB radiation levels measured using the UVB-1 during the observation period. UVB radiation levels could exceed 4.5 W/m 2 , which was much higher than the radiation level measured at noon in June (3.8 W/m 2 , with a lower solar zenith at noon) at an altitude of 5050 m (Lin et al., 2008) and at Zhongshan Station (69°22'S, 76°22'E), 90

Measurement and simulation of UVB radiation
Antarctica (Zheng et al., 2020). The highest UVB value on clear days (27 and 29 August) was lower than that on cloudy days (26,28,and 30 August). The measured UVB values were higher than the simulated values even when the surface albedo value was set at 0.9 (Fig. 2). In the daytime of 28 and 29 August, the relative humidity was 61%-94%, with a water vapor density  (Table 1). The values were lower than ultraviolet albedos (0.70-0.90) over the fresh snow surface (Blumthaler and Ambach, 1988;Chen, 1995;Feister and Grewe, 1995). Table 2 lists the inorganic ion concentrations in snow and water samples. Samples were collected during the daytime. SO4 2− was the most abundant anion, followed by Cl − and NO3 − , and Ca 2+ was the most abundant cation, followed by Na + and Mg 2− 100 ( Table 2). The concentrations of NO3 − and NH4 + in the snow samples were much higher than those in the water samples, in which NO3 − and NH4 + in the water samples was difficult to detect. The average concentration (±1σ) of NO3 − in 10 snow samples was 8.7 ± 2.7 μmol/L (0.44 ± 0.13 ppm (w/w)), which was much higher than that in Antarctic (0.4-2.0 μmol/L) and Arctic (1.9-2.6 μmol/L) snow/ice regions (Li et al., 1995;Xiao et al., 2002). Figure 3 shows variations in the mixing ratios of NO2 and NOx and UVB radiation level on 26 August 2004. Before 16:00, the LMA-3D was used for continuously measurement for 1 or 2 hours in fixed NO2 or NOx mode. At 16:00, the instrument switched to automatic switching mode, and NO2 and NOx were measured in turns for 1 minute. The change in NOx (NO2) concentration corresponded well with the increase and decrease in UVB radiation, and a clear correlation was observed between them ( Fig. 3 and Fig. S2). When the UVB radiation level was high, the concentration of NOx (NO2) could reach 110 magnitudes of several parts per billion, approximately 1 order of magnitude higher than that observed in polar regions (Ridley et al, 2000;Davis et al., 2001;Davis et al., 2004a). After sunset, the concentration of NOx (NO2) gradually decreased to the background level at night. Hence, a sunlit condition was a key factor influencing NOx (NO2) concentrations on the surface of ice and snow. According to the slopes in Fig. S2, with 1 unit change in UVB radiation level, NO2 and NOx changed 0.47 and 1.17 ppb, respectively. NO2 accounted for 40.2% of the total NOx. 115

Relationships between NOx and UVB 105
In the following days, the snow melted further. Strong correlations were observed between NOx (NO2) and UVB (Fig. S3-S5), similar to the observations on 26 August, but with much lower mixing ratios of NOx and NO2. On 30 August, we cleared the remaining surface snow near the observation region, and the sample inlet was deployed near the glacier ice surface. Figure   4 shows variations in NO2, NOx, and UVB on 30 August 2004, and Fig. S6 shows correlations between NOx (NO2) and UVB.
On the glacier ice surface, a strong relationship was observed between NOx (NO2) and UVB radiation levels. However, unlike 120 on the snow surface, the ratio of NO2 to NOx on the glacier ice surface was approximately 90%, which means that NO2 rather than NO was directly released from the ice surface. In addition, the levels of NOx (NO2) released on the glacier ice surface were much lower and more fluctuating than those on the fresh snow surface. Wind speed can be an important factor influencing the processes of snowpack release and air dilution. To weaken the influence of these processes on the relationship between NOx and UVB, the corresponding data of NOx, NO2, and UVB radiation at 125 different binned wind speeds (0.5 m/s as an interval) were calculated, and the relationships between the binned NOx (NO2) mixing ratios and UVB radiation levels are shown in Fig. 5. Similar results were obtained from no binned data (Fig. S2), but with a lower ratio (30.8%) of NO2 to NOx. Gaseous NO2 was the main product of NO3 − photolysis in snow or ice (Jones et al., During the night, when the temperature drops, the recrystallization effect causes increases in snow particle size, thereby reducing the albedo. Another important factor was the deposition of sand and dust (Fig. S1), which blow from surrounding 155 barren mountains as a result of strong winds, making the snow dirty and leading to easy absorption solar radiation, causing the snow particles to melt and the water content to increase, thus reducing the UV albedo.
As stated above, the measured UVB values were higher than the simulated values even when the surface albedo value was set at 0.9, which value can be only found in fresh snow. Albedo is not the only factor to enhance the surface UV radiations. In the Tibetan plateau, studies have observed high surface solar radiation in summer at values frequently larger than the solar constant 160 owning to cloud reflection overhead (Kou et al., 1975;Ren et al., 1999;Li et al.,2000). Although multiple reflections on the cloud overhead might lead to an increase in the ground UVB radiation level, the measurement values on clear and cloudless days (27 and 29 August) were significantly higher than simulated values. A complex terrain might be one of the biggest influence factors. Qiyi Glacier is a valley-shaped glacier, and UVB radiation was observed at a gently sloping mountainside.
This might collect strong solar radiation as a result of reflection by the snow covering mountain peaks, resulting in high 165 radiation levels recorded by the instrument. Therefore, model challenges will be faced on simulating the solar radiations over a mountain-valley-shaped glacier surface.
O3 data were not available during observation because of instrument malfunction. However, other observations (Zhu et al., 2006;Lin et al., 2015;Xu et al., 2016) have proven that the background O3 mixing ratio in the Tibetan Plateau is high (Hourly mean can be as high as 100 ppbv and the monthly mean can be as high as >60 ppbv) due to its high altitude. In the daytime of 170 28 and 29 August, the relative humidity measured using a Assmann ventilation dry and wet meter was 61%-94%, with a water vapor density of 2.1-4.6 g/m 3 . Additionally, strong UV radiation and abundant water vapor promote OH radical generation through O3 photolysis, which might indicate that glacier surface air has a high oxidizing capacity.

O ( 1 D) + H2O → 2OH
(2) 175 The concentrations of NO3 − and NH4 + in the snow samples were much higher than those in the water samples, in which NO3 − and NH4 + in the water samples was difficult to detect. This phenomenon might indicate a quick loss of NO3 − during the melting process under sunlight. However, we only collected limited number of water samples, so it was difficult to determine whether this phenomenon was universal.
Strong correlations were observed between the NOx (NO2) mixing ratios and UVB radiation levels obtained from Qiyi Glacier, 180 Tibetan Plateau. The correlations were much more significant than the findings in the North and South pole regions, where no such direct and strong correlations were reported. Moreover, the mixing ratios of NOx released by fresh snow in Qiyi Glacier in late August reached up to several parts per billion (ppbv), approximately 1 order of magnitude higher than those observed in the polar region. This observation provides direct evidence to support the hypothesis that the concentrations of NOx released on the glacier surface of the Tibetan Plateau are higher than those in Antarctic and Arctic regions. Due to the limited covering 185 areas of mountain-valley glaciers which are less than the huge covering areas in Antarctic and Arctic regions, the photochemical generation NOx left from the snowpack in Tibetan plateau might be more easily to be transported to somewhere https://doi.org/10.5194/tc-2021-32 Preprint. Discussion started: 17 February 2021 c Author(s) 2021. CC BY 4.0 License. else by advection, and hardly return to the snow by deposition through the reaction between NOx and HOx to form HNO3 and HO2NO2 (Honrath et al., 2002;Oncley et al., 2004;Munger et al., 1999;Cohen et al., 2007), resulting in the loss of nitrogen in the snow. High levels of NOx release certainly affect atmospheric chemical processes in the boundary layer of highland 190 glaciers and snow surfaces as well as the conversion equation of snow/ice records to real atmospheric concentrations.

Conclusions
In late August, the surface UVB radiation level at 4600 m asl in Qiyi Glacier reached >4.5 W/m 2 and was enhanced by the strong reflection of snow/ice and clouds against the sun, which was strengthened by the topographical effect. At Qiyi Glacier, a clear phenomenon of dust/sand deposition exists, which helps to absorb sunlight, promotes the melting of ice and snow, 195 increases the air humidity, and reduces the surface albedo. The average concentration (±1σ) of NO3 − in 10 snow samples was 8.7 ± 2.7 μmol/L, but NO3 − was barely detected in three melting snow (water) samples.
Very strong correlations were observed between the NOx (NO2) mixing ratios and UVB radiation levels obtained from Qiyi Glacier, Tibetan Plateau. Vertical experiments showed a negative gradient of NOx (NO2) mixing ratios from the glacier snow surface to a height of 30 cm. Moreover, the mixing ratios of NOx released by fresh snow in Qiyi Glacier in late August reached 200 up to several parts per billion (ppbv). This might be due to strong UV radiation and much higher NO3 − concentrations in snow/ice in the Tibetan Plateau comparing with that in Antarctic and Arctic regions. Further in-depth researches needed to access the effect of such high levels of NOx release on the atmospheric chemical processes in the boundary layer of highland glaciers and snow surfaces.