The Base Camp is located in the middle of the Rongbuk valley (86.85° E, 28.14° N), situated ∼5200 m above sea level (m a.s.l.) on the north slope of the Mt. Qomolangma (Zou et al., 2008; Zhu et al., 2006). The surrounding surface consists of loosed soil, gravel, broken rocks of various sizes, with sparse vegetation due to the semi-arid status (Ming et al., 2007; Zou et al., 2008). Rongbuk valley is characterized by a depth of ∼1000 m and a floor width of ∼1000 m, with elevations of the surrounding mountains exceeding 6000 m on both sides (Zou et al., 2008). Attributed to the unique topography, the local air circulation is dominated by mountain and valley breezes. The predominant wind regime is the katabatic flow of southerly and southeasterly, which is typically persists from noon to midnight (Zhu et al., 2006; Zou et al., 2008; Zhou et al., 2011). The nearest accessible area for residents and visitors is at least 2 km north of the Base Camp. During the campaign, electricity and natural gases were routinely used for cooking and hot water production. There were intermittent vehicle exhaust emissions around the station during daytime for the daily necessaries supporting, i.e., water and food. To minimize the influence of local anthropogenic activities on sampling, the instruments were set in the southeast (upwind direction, Fig. S1 in the Supplement) and approximately 100 m away from the living space of the Base Camp. The anthropogenic influence on the sampling is expected to be minimal.

From 26 April to 6 May 2022, TSP samples were collected simultaneously with NO₂ using a homemade denuder-filter system (Zhou et al., 2022). A similar system has been widely used to separately collect atmospheric particulate matter and N-containing gases for isotopic analysis (Chai et al., 2019, 2021), suggesting the mutual interference between the particulate and gaseous phases to be minimal. A polytetrafluoroethylene (PTFE) sleeve is used to assemble the homemade denuder with the filter pack, flowmeter, and pump. The filter pack was placed in the front of the denuder. All connections between the various parts of the sampling apparatus are made using 3/8′′ Teflon tubing. A detailed description of the sampling apparatus can be found in our previous report (Zhou et al., 2022). Whatman quartz filter (circles, diam. 47 mm, pre-heated at 400 °C for 3 h before use) was placed into the filter pack to collect TSP sample. In the present study, the collected bulk aerosol can be regarded as TSP since no size-selective inlet was employed. The flow rate was controlled at 30 L min⁻¹ using a flowmeter. Previous reports have indicated that the flow rate has negligible effect on the concentrations of sulphate, nitrate, and ammonium when using quartz filter for sampling (Keck and Wittmaack, 2005). To minimize the potential influence of the loose ground surface on the TSP collection, a mountain tent was used to separate the pump (out the tent) with the denuder-filter system (in the other side of tent), and the inlet Teflon tube was stretched out of the tent for ∼1.5 m height straight.

From 24 April to 6 May 2022, PM_2.5 were sampled using a high-volume aerosol sampler (TH-1000F; Wuhan Tianhong Instruments Co. Ltd., China) equipped with PM_2.5 inlet and Whatman quartz-fiber filters (sheets, 203 mm × 254 mm) at a flow rate of 1.5 m³ min⁻¹. All the quartz filters were pre-heated at 400 °C for 3 h before use. In general, TSP and PM_2.5 samples were collected with diurnal resolution during this campaign, with daytime samples from approximately 09:00–20:00 and nighttime samples from 21:00–08:00 (local time), respectively. From 2 to 4 May, we collected the daytime TSP and PM_2.5 samples in the morning (09:00–14:00) and afternoon (14:00–20:00), respectively. A total of 24 TSP samples (including 2 blanks) and 29 PM_2.5 (including 3 blanks) were collected during this campaign. After each sampling period, quartz filters were removed from the filter pack/PM_2.5 inlet and immediately wrapped in pre-baked aluminum foil and then stored in frozen until analysis to minimize potential loss of nitrite. During the campaign, snowfall events occurred on the night of 29 April and during the daytime of 30 April.

Surface soil samples (0–5 cm depth, n=9) were collected in May 2023 from the east slope, west slope and south sides of the Rongbuk valley. A polytetrafluoroethylene (PTFE) shovel was used to collect soil. The collected soil was immediately transferred to clean plastic bags, sealed and kept frozen. Soil samples were transported into laboratory using a cold chain. Upon arrival at our laboratory, the soil samples were passed through a 60-mesh screen (∼0.25 mm) to remove larger particles and thoroughly homogenized prior to chemical and isotopic analysis.

Water-soluble inorganic ions (WSIs, including Na⁺, NH4+, K⁺, Mg²⁺, Ca²⁺, Cl⁻, NO2-, NO3-, and SO42-) in TSP (entire filter, ∼16.6 cm²) and PM_2.5 (1/32 section, ∼ 13.0 cm²) were extracted using 20 mL Milli-Q ultrapure water (18.2 Ωcm) in an ultrasonic bath at room temperature for 30 min. Note the TSP filters were cut to fit the filter holder from the standard Whatman quartz-fiber filters which were also used as the PM_2.5 filters. After filtration through a 0.22 µm pore size syringe filter which was pre-cleaned with ultrapure water, the filtrate was subjected to inorganic species analysis using ion chromatography (Dionex Aquion) (Zhang et al., 2020). Blank filters were pretreated and measured the same as real samples, and the limits of detection (LOD) was calculated as 3 times of standard deviations of blanks (Fang et al., 2015a). In general, Na⁺ in the blank filters is comparable to samples, a well-known issue for the Whatman quartz filter which is high in Na⁺ blank. Therefore, in this study, we discarded the Na⁺ data. The volatile components (i.e., NH4+, NO2-, NO3-) and K⁺ in blank are low but several times higher than the detection limits; SO42-, Mg²⁺ and Ca²⁺ in blank are significantly higher than the respective LOD but lower than samples by at least five times (the lower end). All reported concentrations of each ion were blank corrected as follows: 1Ci=(ρsample-ρblank)×Vwater×FVair with Ci representing the ambient concentrations of specie i in air (ng m⁻³ or µgm-3), ρsample and ρblank are the concentrations determined by the ion chromatography (ng mL⁻¹), Vwater is the volume of ultrapure water used for extraction (20 mL), Vair is the volume of air sampled for each PM_2.5 or TSP filter, F is the ratio of particulate matter collection area for PM_2.5 or TSP filters to the filter area used for extraction.

The overall uncertainty in ion concentration was estimated according to the law of error propagation, accounting for the sampling air volume (3 % for PM_2.5 samples and 1 % for TSP samples as provided by the manufactures), the extraction of water volume (∼0.3 % for pipetting from the manufacture Eppendorf), the blanks, and the analytical uncertainty from ion chromatography and calibration, assuming that these factors are independent. The analytical uncertainty for water-soluble ions concentration determination using ionic chromatography has been extensively assessed in our laboratory, with values typically <5 % for all inorganic species at concentration of 500 ng mL⁻¹. The combined uncertainty about the ionic concentrations in PM_2.5 and TSP are shown in Table S1 in the Supplement. In general, TSP samples are associated with relatively high overall uncertainty compared to PM_2.5 samples, perhaps due to the relatively high blank variability from the low mass loadings in TSP.

For soil ionic concentration analysis, 4.0 g sieved soil was extracted using 20 mL Milli-Q ultrapure water, shaken for 30 min at room temperature. After centrifugation, the extract was passed through 0.45 mm filters before ions analysis. The concentration determination of water-soluble inorganic ions in soil extract was similar to that of aerosol samples. Detailed procedure for the extraction of soil inorganic ions, especially the NO2- has been descried in previous report (Homyak et al., 2015).

After concentration measurements, isotopic analyses (δ15N, δ18O, and Δ17O) of NO2- in TSP were conducted using the azide method (Casciotti et al., 2007). The azide method can reduce nitrite ion in solution into N₂O in a single step, while nitrate ion remains unchanged, ensuring no interference on the nitrite isotopic analysis. The azide reagent is prepared by mixing 2 M sodium azide with 40 % acetic acid at a 1:1 ratio by volume in our laboratory. NO2- standards and samples were pretreated under identical conditions concerning the total volume, nitrite amount, water isotope, and matrix. The δ15N, δ18O, and δ17O of N₂O reduced from NO2- in the TSP samples and standards were determined using a Finnigan^® MAT253 plus isotope ratio mass spectrometer (IRMS) equipped with a GasBench II and preconcentration system. The data calibration followed the procedures described in Albertin et al. (2021), using three international KNO₂ salt standards (RSIL-N10219, RSIL-N7373, and RSIL-N23 with respective δ15N and δ18O values of 2.8 ‰/88.5 ‰, -79.6 ‰/4.2 ‰, and 3.7 ‰/11.4 ‰). The Δ17O values of the three international references have not been certified. To address this, a series laboratory experiments was conducted to determine the true values of three international references in our laboratory (Zhang et al., 2025). In brief, each nitrite international reference was oxidized into NO3- by O₃ produced from commercial ozone generator. A parallel flow of O₃ was also used to convert a normal KNO₂ salt (Δ17O=0) into NO3- to quantify the Δ17O transfer during O₃ oxidation, following the approach of (Vicars and Savarino, 2014). Based on these experiments, the Δ17O of RSIL-N7373 and RSIL-N23 are determined to be negligible, consistent with previous findings (Albertin et al., 2021), while the Δ17O of RSIL-N10219 is determined to be (-9.3±0.2) ‰ in our laboratory. The ¹⁷O-excess in samples is then calculated as Δ17O=δ18O-0.52×δ17O. The standard deviations for δ15N, δ18O, Δ17O of reference materials (n=10) were determined to be less than 0.1 ‰, 0.6 ‰, and 0.4 ‰, respectively.

For soil NO2- isotopic analysis, ∼30.0 g sieved soil was extracted using 150 mL Milli-Q ultra-pure water. The soil extract was then preconcentrated into 10 mL using ion-exchange resin before isotopic analysis. The preconcentration approach was widely used for nitrate isotopic analysis in snow and ice samples, and the detailed procedures and the performance have been provided in Erbland et al. (2013) and evaluated in our laboratory (Sect. S1 in the Supplement). The concentrated soil extracts (50 nmol NO2-) was then subjected to soil NO2- isotopic analysis by converting into N₂O via the azide method. For soil NO3- isotopic analysis, ∼5.0 g sieved soil was extracted with 15 mL of 2 M KCl solution (Fang et al., 2015b). The remaining procedures of soil NO3- and NO2- analysis were same as those for TSP samples, as aforementioned.

To evaluate the possible impact of biomass burning emissions or other pollution sources from South Asia, the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model (performed using TrajStat plugin of the MeteoInfo software) and archived Global Data Assimilation System (GDAS) of meteorological data were used to model the air mass back trajectories (Wang, 2014). In this study, 3 d air mass backward trajectories with arriving height of 1000 m above ground level were simulated to identify the most likely pathway and potential source regions of the air masses at the Base Camp (Bhattarai et al., 2023; Lin et al., 2021). Moreover, the Fire Information and Resource Management System (FIRMS) developed by Moderate Resolution Imaging Spectrometer (MODIS) (https://worldview.earthdata.nasa.gov, last access: 7 September 2024) was employed to identify the distribution of active fire spots during the sampling period.

Figure 1 displays the chemical compositions of water-soluble inorganic ions, their corresponding time series and fractional contributions in TSP and PM_2.5. Throughout the campaign, substantial variations of total WSIs in PM_2.5 and TSP were observed. For PM_2.5 samples, the mass concentrations of total WSIs before 30 April were higher than that from 1 to 6 May (4.1±1.7 versus 1.7±0.6 µgm-3; p<0.05). The cut-off date of 30 April was selected based on observed shifts in concentrations of water-soluble ions and air mass origins (described in Sect. 4.2). This decline of total WSIs after 30 April was predominately driven by significant reductions in secondary inorganic species, i.e., SO42-, NO3- and NH4+, with the magnitude by more than 60 %. In particular, NH4+ in PM_2.5 was on average (322±243) ng m⁻³ before 30 April whereas NH4+ in PM_2.5 collected during daytime and nighttime of 1 May were 1 and 3 ng m⁻³, respectively, and NH4+ in PM_2.5 extractions from 2 to 6 May was below the detection limit. Therefore, the fractional contribution of secondary inorganic species in PM_2.5 also decreased (Fig. 1d). Similarly, K⁺ in PM_2.5, a good tracer of biomass burning (Ma et al., 2003), also declined significantly after 30 April (269±432 versus 22±12 ng m⁻³; p<0.05). The elevated concentrations of WSIs before 30 April (4.1±1.7 µgm-3) are comparable to previous reports at QOMS station (4.2±2.2 µgm-3) in the spring (Lin et al., 2021). In comparison, concentrations of Ca²⁺ and Mg²⁺, tracers of wind-blown dust (Wang et al., 2002), decreased by less than 20 % after 30 April. In general, SO42-, NO3-, and Ca²⁺ are the most abundant species in PM_2.5, accounting for the majority of the mass of total WSIs. In addition, no clear diurnal variation of water-soluble inorganic ions in PM_2.5 was observed in this study (Fig. S2).

Some water-soluble inorganic ions (e.g., SO42-, NO3- and NH4+) in TSP showed similar variation trends with that in PM_2.5 throughout the campaign, while others, such as K⁺, exhibited divergent behavior (Fig. 1). For example, after 30 April, the secondary inorganic species in TSP declined considerably by over 50 % (i.e., NH4+ in TSP declined by more than tenfold), while Ca²⁺ in TSP decreased with a smaller degree (<15 %) and Mg²⁺ remained stable. In contrast, TSP K⁺ (from both crustal and biomass burning sources) (Hsu et al., 2009) drastically surged on 3 and 4 May. Note that other species in TSP, i.e., SO42- and NO3- also increased to some extent on 3 May. The time series of water-soluble inorganic ions in the coarse-mode particulate, calculated as the differences between TSP and PM_2.5, are presented in Fig. 2. Due to the overall analytical uncertainties, the air concentrations of SO42- in PM_2.5 samples occasionally exceeded that in corresponding TSP samples on some days.

Figure 3 presents the ion balance of measured water-soluble ions in PM_2.5 throughout the campaign as well as in TSP collected before and after 30 April, respectively, to highlight the significant decline in TSP NO2-. There is a strong correlation between cation and anion equivalents in PM_2.5 samples (R2=0.70), whereas the correlations decreased in TSP samples (R2=0.46 before 30 April and 0.49 after that, respectively). The nanogram-equivalent weights of cation were significantly higher than that of anions for all samples, with ratio of cation to anion equivalent of ∼1.5 for PM_2.5, ∼1.9 for TSP collected before 30 April, ∼4.2 for TSP collected after 1 May. Clearly, the declines in TSP NO2- after 30 April resulted in relatively higher cation/anion equivalent ratio. The slopes of the correlation lines exceeded unity for PM_2.5 and TSP samples, indicating the alkaline nature of aerosol. The observed deficiency of anion can be attributed to the presence of carbonates (i.e., CaCO₃), which can dissolve in water during extraction to release CO32- and/or HCO3- despite the relatively low solubility (Zhang et al., 2021a).

The most distinct feature of chemical compositions in TSP was the elevated level and significant variation of NO2-, ranging from 0.2 to 1291 ng m⁻³ in air and with an average of 375±386 ng m⁻³. In comparison, NO2- consistently remained below the detection limit in PM_2.5 samples. Note during the laboratory measurements of ionic concentrations, TSP and PM_2.5 filters were extracted with ultrapure water and it was the extraction analyzed by ion chromatography. To ensure fair comparisons, similar areas of filters were extracted with same volume of ultrapure water, so that the extractions from the PM_2.5 filters should be more concentrated in atmospheric particulate species compared to that from the TSP filters, since PM_2.5 samples were collected at a much faster sampling speed (1.5 m³ min⁻¹ vs. 30 L min⁻¹) over the same sampling duration. Nevertheless, NO2- was detectable only in the extractions of TSP filter. The determined NO2- concentrations in TSP in this study (375±386 ng m⁻³) were higher than previous reports conducted in various remote sites, such as at QOMS station (∼60 ng m⁻³ for TSP) (Bhattarai et al., 2023), at a forest site in the Southeast Tibet Plateau (∼140 ng m⁻³ for TSP) (Bhattarai et al., 2019), in the middle hills of the central Himalayas (∼210 ng m⁻³ for TSP) (Tripathee et al., 2021).

In particular, there was a dramatic decrease in TSP NO2- after 30 April, from a mean of (625±457) to (147±145) ng m⁻³, in line with the declines in other secondary inorganic species. Over the course of the campaign, NO2- comprised approximately 8 % of the total WSIs mass in TSP, while its contribution reached maximum of ∼20 % on 27 and 28 April, being one of the most abundant components on the 2 d. In addition, there was a strong correlation between NO2- and NO3- throughout the campaign (r=0.75, p<0.05, Fig. S3). Meanwhile, the mean mass ratio of NO2- to NO3- was ∼50 % throughout the campaign, but on several days (i.e., night of 27 April) NO2- concentrations significantly exceeded that of NO3- (802 vs. 663 ng m⁻³). Previous study also reported comparable NO2- and NO3- concentrations in at a forest site in the Southeast Tibet (summer: 100 vs. 110 ng m⁻³; winter: 180 vs. 270 ng m⁻³) (Bhattarai et al., 2019).

Figure 4 presents the times series of δ15N, δ18O and Δ17O of NO2- in TSP, along with the NO2- concentrations. Similar to the variation trend of NO2- concentrations, NO2- isotopes varied in a wider range before 30 April, but became more stable afterward. For example, δ15N(NO2-) ranged from -10.9 ‰ to 0.8 ‰, with a relatively large standard deviation before 30 April compared to that after 30 April ((-6.4±4.3) ‰ vs. (-8.0±0.7) ‰). The large variability in δ15N(NO2-) before 1 May is predominately attributed to the two high values observed in daytime of 27 April and night of 28 April. Note that the two high δ15N(NO2-) samples were also associated with relatively high NO2- mass concentrations. In contrast to the declining trend of NO2- concentrations, the mean δ15N(NO2-) values were comparable before and after 30 April. Relatively large variability was observed in TSP δ18O(NO2-), ranging from -9.0 ‰ to 3.9 ‰ and with an average of (-3.4±3.8) ‰. TSP Δ17O(NO2-) varied within a narrow range from -0.2 ‰ to +0.7 ‰ and with a mean of (0.2±0.3) ‰. During the campaign, no significant correlations were observed between δ15N(NO2-) and the NO2- concentrations; while δ18O(NO2-) appeared to be moderately correlated with the NO2- concentrations (Fig. S5).

The concentration of soil NO2- (and NO3-) as well as the corresponding isotopic signatures are displayed in Table 1. The soil NO2- content on the west slope of Rongbuk Valley (on average 124.7 ng g⁻¹) was higher than that observed on the east and south sides (75.3 and 48.3 ng g⁻¹, respectively). The mean surface soil NO2- and NO3- in the Rongbuk Valley were 82.8 and 2740.8 ng g⁻¹, respectively. The soil NO3- concentrations were significantly higher than the NO2- by a factor of ∼8–40. In general, the measured soil NO3- concentrations at the Rongbuk valley were significantly lower than other remote regions of TP (i.e., 23.1 µgg-1 at Naqu, 8.4 µgg-1 at Yangbajing), while soil NO2- concentrations were comparable to these reports (i.e., 90.3 ng g⁻¹ at Naqu, 131.2 ng g⁻¹ at Yangbajing) (Wang et al., 2019). Soil δ15N(NO2-) values ranged from -13.2 ‰ to -5.2 ‰ (on average -10.4 ‰), which are comparable to TSP δ15N(NO2-) (-7.3 ‰). In comparison, we observed positive soil δ18O(NO2-) and Δ17O(NO2-), ranging from 2.7 ‰ to 18.1 ‰ (on average 10.5 ‰) and 1.4 ‰ to 7.3 ‰ (on average 3.8 ‰), respectively, in contrast to the negative δ18O(NO2-) and near-zero Δ17O(NO2-) observed in TSP samples. The determined soil δ18O(NO2-) is comparable to that in laboratory incubated soil (11.8 ‰–12.5 ‰) (Lewicka-Szczebak et al., 2021).

Increasing evidence supports particulate nitrate photolysis as an important source of atmospheric HONO especially in pristine atmosphere, with NO2- serving as the intermediate in the subsequent gas-particle partition process (Andersen et al., 2023; Ye et al., 2016). In theory, particulate NO2- (and HONO) produced from particulate nitrate photolysis might be associated with extremely negative δ15N values, due to the significant nitrogen isotopic fractionation effect during nitrate photolysis (Erbland et al., 2013; Frey et al., 2009). For example, NO2- in water of hypersaline ponds and soil of McMurdo Dry Valleys, Antarctica, produced from the NO3- photolysis, were characterized by significantly negative δ15N values (< -80 ‰) (Peters et al., 2014). However, δ15N(NO2-) in this study was only ∼ 2 ‰ lower than the δ15N(NO3-) in TSP samples collected during this campaign (on average (-5.3±3.3) ‰, Sect. S1) and that at the QOMS stations (annual average of (-5.1±2.3) ‰) (Wang et al., 2020). The similarity in δ15N isotopes between NO2- and NO3- suggests particulate nitrate photolysis is unlikely to be the primary source of the TSP NO2-.

In addition to nitrate photolysis, the absorption of HONO on alkaline aerosols (i.e., lofted dust and road salt particles) can also result in accumulation of NO2- into the particle phase (VandenBoer et al., 2014a; Chen et al., 2019). For example, VandenBoer et al. (2014a) observed a synchronous enhancement of fine particle NO2- (as high as 730 ng m⁻³) alongside the buildup of HONO (up to 1.37 ppbv) after sunset in an agricultural site (VandenBoer et al., 2014a). However, the levels of HONO in terrestrial background environments, typically on the order of dozens of pptv (Ye et al., 2023), are obviously too low to support the observed unexpectedly high levels of particulate NO2- (up to 1300 ng m⁻³). What is more, previous studies indicated the HONO uptake predominately occurs on fine particles (Wang et al., 2015; Chen et al., 2019; Shang et al., 2021), while our observations indicated NO2- only exists in coarse particles.

The uptake of NO₂ on mineral dust has also been identified as a significant route for the formation of particulate NO2- or gas-phase HONO (Nie et al., 2012; Ndour et al., 2008). For example, Nie et al. (2012) found a significantly enhanced NO2- in coarse particle during daytime in a dust storm event in Mt. Heng (up to 4.5 µgm-3) (Nie et al., 2012). The proposed mechanism is initiated by photo-enhanced conversion of NO₂ to NO2- via photo-produced electrons on surface of dust. Nevertheless, given the relatively small NO₂ uptake coefficients on mineral dust or salt surface (generally lower than 1×10-6) (Yu et al., 2021; Bao et al., 2022; Xuan et al., 2025) and the low concentration of NO₂ in pristine environment of TP (e.g., ∼140 pptv at Namco (Wang et al., 2023), and would be even lower at Mt. Qomolangma), such high levels of particulate NO2- are beyond the capacity of NO₂ heterogeneous reactions.

Other than the above-mentioned rationales, the atmospheric physicochemical processes leading to NO2- production would influence both fine and coarse-mode particles, and some of the processes (e.g., HONO uptake) preferentially interact with fine-mode particles. However, the observation indicated NO2- only exists in TSP but not PM_2.5, suggesting atmospheric physicochemical processes are unlikely to account for the elevated levels of NO2- in TSP.

There is a growing body of compelling evidence indicating that the elevated aerosol loadings and some chemical species in TP during spring, i.e., black carbon (Cong et al., 2015; Kang et al., 2019) and soluble components (Dasari et al., 2023; Bhattarai et al., 2023; Lin et al., 2021; Wang et al., 2020; Zhao et al., 2020) are significantly linked to biomass burning emissions from South Asia, which can penetrate into TP via long range transport. Recently, Bhattarai et al. (2023) observed synchronously elevated water-soluble nitrogen compounds (i.e., NO2-, NH4+, NO3-), levoglucosan (a molecule marker for biomass burning) and bulk δ15N signatures in TSP at QOMS station, once upon the arrival of biomass burning plumes from South Asia (Bhattarai et al., 2023). Specially, they found that TSP NO2- averaged 60 ng m⁻³ during the spring biomass burning influenced episodes, while NO2- was always below the detection limit in other seasons (Bhattarai et al., 2023). This suggests that biomass burning events in South Asia and the subsequent long-range transport could contribute to the accumulation of NO2- in aerosols in Tibet.

In this study, the mass concentrations and compositions of water-soluble ions, including SO42-, NO3-, NH4+ and TSP NO2-, also varied substantially throughout the springtime campaign (Fig. 1). The potential effect of South Asian pollutants on our observations was explored by analyzing the air masses origins during the sampling period. As shown in Fig. 5, during the first-half of the campaign (i.e., before 30 April), air masses predominately originated from or passed through northern India and Nepal with intensive human activities and numerous fire hotspots (represented by the dense red dots in Fig. 5), indicating the potential impact of South Asia pollutants on aerosol loadings of TP. Correspondingly, elevated concentrations of secondary inorganic ions (i.e., NH4+, NO3- and SO42-) in TSP and PM_2.5 were observed before 30 April, which are comparable to the values of previous reports in QOMS station when arrived air masses experiencing severe biomass burning emissions (Bhattarai et al., 2023; Lin et al., 2021). Meanwhile, elevated levels of TSP NO2- (625±457 ng m⁻³) was observed during this biomass burning-impacted period, and δ15N(NO2-) values exhibited substantial variability. TSP samples collected during daytime of 27 April and night of 28 April were associated with high NO2- concentrations and δ15N(NO2-) values. In comparison, TSP NO2- in other samples were significantly ¹⁵N-depleted. Assuming that biomass burning emission accounted for the two high δ15N(NO2-) samples (Bhattarai et al., 2023), the observed relatively low δ15N(NO2-) in other samples before 30 April likely indicated the potential contribution from additional emission sources.

From 1 to 6 May, air masses originated primarily from the inside of the TP or surroundings and none of the fire hotspot was detected at the whole TP and along the air mass trajectories (Fig. 5), potentially excluding the influence of biomass burning and anthropogenic emissions during this period. Accompanied by this significant shift in air mass origins, the concentrations of NH4+, NO3- and SO42- in both PM_2.5 and TSP, as well as the TSP NO2- apparently decreased. In particular, NH4+ in most PM_2.5 and TSP samples were below the detection limit during this period. Furthermore, K⁺ in PM_2.5, a common tracer for biomass burning, declined from 269±432 ng m⁻³ before 30 April to 22±12 ng m⁻³ after 30 April, with the difference being statistically significant (p<0.05). Meanwhile, the average TSP NO2- also declined significantly from 625±457 ng m⁻³ before 30 April to 147±145 ng m⁻³ after 30 April. Note substantially high levels of TSP NO2- were also observed on several days after 30 April, i.e., during daytime of 3 May (∼400 ng m⁻³). Although the TSP NO2- concentrations varied in wide range after 30 April, the δ15N(NO2-) was relatively stable and comparable to that determined before 30 April (except for the two high δ15N(NO2-) samples).

In short summary, these results likely suggest the significant impact of South Asia pollutants (i.e., biomass burning emissions) through long-range transport on the TSP NO2-, especially for samples collected before 30 April. However, it is difficult for the long-range transport to explain why NO2- is predominately present in coarse particles, as fine mode particle is typically easier to be transported in principle. While further studies need to be conducted to find out the exact reasons, one possibility is that the size partition of fine particle NO2- toward the coarse mode range during transport. Similar size shifts of NO3- have been observed and used to explain the enrichment of NO3- in coarse-mode aerosols in marine environment (Matsumoto et al., 2009). In addition, the coarse mode particles in this study contain more alkaline species (Fig. 2), which makes nitrite more stable in TSP during the transport or more stably exist in TSP after being uptake when the polluted air masses reaching TP. For the period after 30 April, since air masses originated from clean regions with little to no biomass burning sources, other sources of nitrite might be required.

Previous reports have confirmed that wind-blown mineral dust contributed significantly to coarse mode aerosols in TP (Kang et al., 2016; Zhang et al., 2021a). Notably, surface soil collected from the Rongbuk valley is characterized with elevated NO2- concentration (up to 180 ng g⁻¹, Table 1). The surface soil NO2- is expected to mainly reside in the coarse mode after suspended in the atmosphere (Drakaki et al., 2022), consistent with our observations that particle NO2- was predominately confined into coarse particle. The unique environment of Rongbuk valley, characterized by exposed surface soil and strong wind (reaching as high as 9 m s⁻¹ during this campaign) could facilitate the resuspension of soil components into the atmosphere. Furthermore, small localized tornadoes were frequently observed before 30 April, while the snow events occurred in 30 April would reduce the soil-derived dust emission by increasing the snow coverage and enriching the soil moisture. In fact, concentrations of Ca²⁺ in TSP, which predominately originate from dust emission, also declined to some extent after 30 April.

In addition, the comparable δ15N values of NO2- between TSP collected after 1 May (-8.0±0.7 ‰) and the surface soil (-10.3±3.0 ‰) also supports that locally emitted surface soil is an important contributor to the observed high levels of TSP NO2-. But one should note that the oxygen isotopes (δ18O and Δ17O) of TSP NO2- were significantly lower compared to that in soil NO2-, indicating that the original soil NO2- oxygen isotope may have been modified after resuspension. This discrepancy could be explained by the potential oxygen isotope exchanges between TSP NO2- and aerosol liquid water (fractionation effect of 18εeq≈16 ‰ at local temperature, T=270 K, 18εeq=-0.12T+48.79 (Buchwald and Casciotti, 2013)), which tend to deplete both the initial δ18O(NO2-) and Δ17O(NO2-).

We noted the oxygen isotope exchange process between NO2- and H₂O also occurs in surface soil. Previous study indicated that in high-altitude arid regions of TP (i.e., >5000 m), denitrification process dominated the surface soil NO2- production, accounting for ∼75 % (Wang et al., 2019). Soil NO2- generated from denitrification process is expect to inherit the Δ17O signatures of substrate NO3-. In this study, the surface soil Δ17O(NO3-) were positive with average values of 9.6 ‰ (Table 1). The positive soil Δ17O(NO3-) have been observed on arid environments (Wang et al., 2016), such as desert soil, where the low water moisture favors the preservation of atmospherically derived NO3-. One could estimate that soil Δ17O(NO2-) derived from the denitrification would be 9.6 ‰, nitrite from other sources (e.g., nitrification) should possess zero Δ17O, thus in total nitrite in soil should possess Δ17O of ∼7.2 ‰ (estimated as 0.75×Δ17O(NO3-)). However, the determined soil Δ17O(NO2-) (3.8 ‰) is significantly lower compared to the estimated soil Δ17O(NO2-), indicating the occurrence of exchange process between soil water and NO2-, which would reduce the soil Δ17O(NO2-) to some extent. The exchange process between soil water and NO2- is particularly evident in west slope of Rongbuk Valley, where soil Δ17O(NO2-) is as low as 1.5 ‰ while soil Δ17O(NO3-) is on average 10.3 ‰. Soil Δ17O(NO2-) should be erased to near-zero if exchange process between soil water and NO2- was efficient. Therefore, the fact that the observed soil Δ17O(NO2-) remain above 0 ‰ indicates unfavorable conditions for the oxygen isotope exchange process, likely due to the extremely low soil moisture content (∼1 %) in surface soil (Ma et al., 2023).

Upon resuspension into atmosphere, the soil-derived dust aerosols usually exhibited a certain degree of hygroscopicity (Tang et al., 2016; Chen et al., 2020), allowing the absorption of water molecule on dust aerosol. For example, the aerosol water was determined to account for ∼ 20 % of the total PM₁₀ mass during Saharan dust plumes (Cardoso et al., 2018). Based on laboratory experiment, Tang et al. (2019) reported that Asian dust also exhibited substantial hygroscopic property and revealed that the water-soluble inorganic ions, such as Cl⁻, SO42- and NO3- played a critical role in the absorption of water molecules on dust aerosol (Tang et al., 2019). In the present study, the SO42- and NO3- account for ∼30 % of the total mass of water-soluble ions in TSP, implying the potential uptake of water vapor on aerosol surface. In addition to water-soluble ions, the hygroscopicity of mineral dust also depends on the surface areas, and wind-blown dust experiences a substantial increase in surface area after being lifted into the atmosphere, enhancing its capacity for water uptake (Chen et al., 2020; Seisel et al., 2004). The hygroscopicity of dust aerosol is expected to accelerate the oxygen isotope exchanges between NO2- and aerosol liquid water. The atmospheric water vapor δ18O isotope in TP is determined to be significantly negative (approximately -35 ‰ to -15 ‰ at a remote site in TP with altitude of ∼4200 m) (Yu et al., 2015). Similarly, the oxygen isotope exchange between NO2- and H₂O would also homogenize and erase original soil Δ17O(NO2-) signals, because aerosol liquid water is characterized by negligible Δ17O values (Luz and Barkan, 2005). Consequently, isotope exchange with aerosol water would further reduce both δ18O and Δ17O of TSP NO2-, effectively masking the original isotopic signature inherited from surface soil.

We also noted that NO2- in surface soil is significantly lower than NO3-, by on average 35 times (Table 1), contrasting with the chemical compositions in TSP. It is important to note that during the complex dust generation and aerosolization processes, water-soluble ions exhibit significant chemical enrichment relative to that in parent soil (Wu et al., 2022; Gao et al., 2023). For example, Wu et al. (2022) reported that contents of nitrate in sand dust-derived PM₁₀ is higher than the original soil samples by ∼2–80 times, while sulphate can be up to 500 times higher than the original soil. Although the enrichment of nitrite has not been evaluated to the extent of our knowledge, we propose that the observed discrepancy of nitrate / nitrite ratio between TSP and surface soil can be reconciled if the nitrite is enriched more efficiently than nitrate during dust aerosolization, i.e., by a factor of 30. To further assess the potential contribution of resuspended soil to elevated nitrite in TSP, we conducted a rough estimation based on laboratory investigations from Wu et al. (2022). First, the springtime TSP concentrations observed in the nearby QOMS station is used (65.1±50.9 µgm-3; Liu et al., 2017), as TSP mass concentrations surrounding the Base camp were not available. Second, we assume the nitrite content in soil-derived coarse particles to be similar to that of nitrate (0.2 %) in laboratory-generated dust aerosol from natural sandy and gravel soils (Wu et al., 2022), which is comparable to the soil texture of the Rongbuk Valley. This assumption can also explain the comparable amounts of nitrite and nitrate observed in TSP samples in this study. Results shown that the concentrations of nitrite in TSP can be approximately 130±102 ng m⁻³, on the same order with the observed nitrite concentration in TSP after 30 April (147±145 ng m⁻³) but substantially lower than that before 30 April (625±457 ng m⁻³). Therefore, we speculate that the resuspension of surface soil may account for the observed TSP nitrite after 30 April, whereas the biomass burning and soil together co-contributed to the high TSP nitrite before 30 April. This is also consistent with the shift of air mass origins, which clearly indicated that airmasses before 30 April is significantly impacted by the biomass burning emissions, and after 30 April airmasses primarily originated from clean regions. While we acknowledge that this simplistic estimation is subject to substantial uncertainty, it provides a first-order assessment supporting the hypothesis that wind-blown soil dust contributes to coarse-mode particulate nitrite. Previously, wind-blown mineral dust has been verified as a potential source for aerosol water-soluble ions (Engelbrecht et al., 2016; Wu et al., 2022, 2012) and dust aerosol is also recognized as one of the important aerosol types over TP (Pokharel et al., 2019).