Interactive comment on “ Evolution of aerosol chemistry in Xi ’ an , inland China during the dust storm period of 2013 – Part 1 : Sources , chemical forms and formation mechanisms of nitrate and sulfate ”

Anonymous Referee #2 General comments This study focused on aerosol chemistry in inland China and discussed the possible chemical mechanisms of nitrate and sulfate. The authors designed the research during one Asian dust storm. The hourly dust samples were collected to understand the chemical changes from early stage of dust storm to end. Through comparisons chemical compositions of different dust samples at different stages of dust storm and size resolved aerosol sampling, the authors obtained

Due to rapid urbanization and industrialization in China, annual consumption of coal has increased to 3.61 billion tons in the country recently (China Statistic Press, 2010), along with a sharp increase in vehicle numbers, resulting in high burdens of SO 2 , NO x , elemental carbon (EC), organic matter, nitrate, sulfate and other pollutants in the atmosphere over China (Geng et al., 2014;He, 2014;Huang et al., 2014;van Donkelaar et al., 2010;Wang et al., 2014).During longrange transport, dust can become coated with nitrate, sulfate, ammonium and other pollutants, leading to a series of changes in dust behavior in the atmosphere such as water uptake, deposition and scattering sunlight (Boreddy et al., 2014;Creamean et al., 2013;Formenti et al., 2011;Grassian, 2001;Seinfeld et al., 2004;Sullivan et al., 2009).Studies found that SO 2 level in the atmosphere of China has become stable and even decreased since 2006 because the Chinese government has promulgated a strict standard on SO 2 emission (Lu et al., 2010;Menon et al., 2008;Wang et al., 2013).As a result, atmospheric sulfate aerosol loading has been decreasing in many Chinese urban regions since 2006, whereas NO x and particulate nitrate levels have kept stable and even increased (Wang et al., 2011(Wang et al., , 2012(Wang et al., , 2013;;Zhang et al., 2009).
Due to the chemical affinity of nitrogen oxides and nitric acid with dust particles (Hanisch and Crowley, 2001;Mogili et al., 2006;Saliba and Chamseddine, 2012;Song et al., 2013;Usher et al., 2003), such a change in the atmospheric environment of China suggests that dust particle behavior might also be changing compared to the situations 10 years ago.Therefore, it is necessary to investigate the present physicochemical properties of airborne particles in the country during dusty periods.
In the past decades, numerous observations have been conducted along the Asian dust transport pathways -including eastern coastal China, Korea, Japan, the Pacific Ocean and western North America -to investigate the ageing process of East Asian dust during long-range transport (Huebert et al., 2003;Kim et al., 2004).In contrast, only a few field measurements have been done in the upstream regions, especially inland cities of China (Arimoto et al., 2004;Huang et al., 2010), where the dust particle ageing could be in the infant state due to the proximity of the source regions, including deserts and the Loess Plateau.Xi'an is a megacity in inland China, which is located at the south edge of Loess Plateau.A high level of particle pollution has been a persistent problem in the city (Shen et al., 2008).In comparison with that (55 µg m −3 , unpublished data) in 1997, the annual level (27 µg m −3 , unpublished data) of sulfate of PM 2.5 of the city in 2012 has decreased by a factor of around 2 due to the SO 2 emission control, while NO x and nitrate have increased about 25 %.Acidic gas concentration is a key factor affecting heterogeneous reaction rates of dust particles with SO 2 , NO x and HNO 3 (g), apart from relative humidity, temperature and dust mineralogy and morphology.For example, calcite (CaCO 3 ), which is a common component of East Asian dust and accounts for 3.6−21 % of the dust mass (Liu,l985), can rapidly convert to Ca(NO 3 ) 2 in less than 3 min under 1 ppbv HNO 3 (g) but longer than 4 h under a low-HNO 3 (g) mixing ratio (10 pptv) (Sullivan et al., 2009).Since Ca(NO 3 ) 2 is highly hygroscopic, the chemically modified dust can absorb water vapor and form a liquid phase on the surface (Li et al., 2014).Therefore, changes in physicochemical properties of dust particles occur and are probably much more significant than before due to the recent increase in NO x emission.In order to investigate the impact of dust storms on the downwind aerosol chemistry under the current high level of NO x and relatively low level of SO 2 conditions, we performed an intensive filter-based sampling with a 1 h time resolution to investigate chemical evolution of urban airborne particles in Xi'an in the period of 9−12 March 2013, during which the annual heaviest dust storm passed through the city with an hourly maximum of total suspended particulate (TSP) of more than 7000 µg m −3 .In the present work, we focus on the changes in aerosol chemistry specifically nitrate and sulfate during the event.We first investigated the composition and size distribution of airborne particles in the event and compared with those in the non-event to discuss the chemical evolution of dust particles.Then we identified the chemical forms of nitrates and sulfates existing in the dust to explore their sources and formation mechanisms.We found for the first time an enrichment of ammonium nitrate in the dust particles, which is different from a fine-mode accumulation of ammonium sulfate in Asia continental outflow regions such as the western North Pacific and western North America.Our results further revealed that such a phenomenon is relevant to the water-soluble components of mineral dusts, which consist of hygroscopic salts (e.g., NaCl and Na 2 SO 4 ) and originate from dried salt lakes in western China and the Gobi desert.

Collections of TSP and size-resolved samples
TSP samples were collected hourly at an airflow rate of 1.0 m 3 min −1 from 9 March at 18:00 to 12 March at 10:00 LT by using two TCH-1000 air samplers (Tianhong Company, China) on the rooftop of a three-story building on the campus of the Institute of Earth Environment, CAS, which is located in the downtown area of Xi'an.Simultaneously, sizesegregated samples were also collected, with each set lasting for 3 h during the dust storm period and 12 h during the nondust period, by using two size-resolved samplers (Series 20-800, Thermo Electron Corporation USA).The cutoff points of the size-segregated samples are 0.43, 0.65, 1.1, 2.1, 3.3, 4.7, 5.8 and 9.0 µm at an airflow rate of 28.3 L min −1 .All the samples were collected onto pre-combusted quartz filters (450 • C for 6 h).Field blank samplers were also collected at the beginning and the end of the sampling campaign by mounting pre-baked blank filters onto the samplers for about 10 min without sucking any air.After sampling, all the filters were individually sealed in aluminum foil bags and stored at −20 • C prior to analysis.A total of 65 TSP samples and six sets of size-segregated samples were collected.In addition to the above filter-based sampling, online measurement of PM 2.5 was also conducted by using a E-BAM-9800 analyzer (Met One, USA).

Inorganic ions, water-soluble organic (WSOC)
and inorganic carbon (WSIC), and water-soluble organic nitrogen (WSON) Aliquot (size: 12.56 cm 2 ) of the filter was cut into pieces and extracted three times with Milli-Q pure water under sonication.One part of the combined water extracts was determined for inorganic ions using Dionex-6000 ion chromatography.
Another part of the water extracts was determined for watersoluble organic carbon (WSOC), water-soluble inorganic carbon (WSIC) and water-soluble total nitrogen (WSTN) using a Shimadzu 5000 TOC/N Analyzer.The detailed analysis methods for inorganic ions, WSOC and WSTN can be found elsewhere (Wang et al., 2010).NO − 3 and NH + 4 are the major water-soluble inorganic nitrogen (WSIN) species in airborne particles; thus the difference between WSTN and WSIN is defined as water-soluble organic nitrogen (WSON).

Organic carbon (OC) and elemental carbon (EC)
OC and EC in the TSP samples were measured by a DRI Model 2001 Carbon Analyzer using the Interagency Monitoring of Protected Visual Environments (IMPROVE) thermal/optical reflectance (TOR) protocol.Briefly, a 0.53 cm 2 filter was put in a quartz boat inside the analyzer and progressively heated to temperatures of 120, 250, 450 and 550 • C in a non-oxidizing helium (He) atmosphere, and 550, 700 and 800 • C in an oxidizing atmosphere containing 2 % oxygen in helium.

Elements
Elements of the TSP samples were determined by energydispersive X-ray fluorescence (ED-XRF) spectrometry (Epsilon 5 ED-XRF, PANalytical B. V., Netherlands).The X-ray source is a side window X-ray tube with a gadolinium anode and operated at an accelerating voltage of 25e100 kV and a current of 0.5e24 mA (maximum power: 600 W).The characteristic X-ray radiation is detected by a germanium detector (PAN 32).Each sample was analyzed for 30 min to obtain a spectrum of X-ray counts versus photon energy, with the individual peak energies matching specific elements and peak areas corresponding to elemental concentrations.The spectrometer was calibrated with thin-film standards obtained from MicroMatter Co. (Arlington, WA, USA).In the current study, 14 elements (i.e., S, Cl, K, Ca, Ti, Mn, Fe, Zn, Cr, Ni, As, Br, Mo, Pb) were determined.The element concentrations for blank quartz fiber filter are 0.00−0.66µg cm −2 , lower than 10 % of those in the TSP samples.

Results and discussion
During the sampling period a massive dust storm event originating from the Mongolian Gobi desert arrived over Xi'an on 9 March with a highest TSP level of 7527 µg m −3 in the beginning hour (18:00 LT) and a second peak on 10 March at 13:00.From 9 March at 18:00 to 10 March at 21:00 LT, the TSP level decreased from the highest to less than 1000 µg m −3 , which is a typical level of TSP in Xi'an during spring.A 48 h backward-trajectory analysis indicated a three-phase pattern for the movements of air masses arrived in Xi'an during the sampling period.As seen in Fig. 1, the 48 h backward trajectories showed that from 9 March at 18:00 to 10 March at 21:00 air parcels originated from the Gobi desert of Mongolia and northern China and directly moved to Xi'an at 100, 300 and 500 m levels above the ground along the same tracks.TSP within this period ranged from 774 to 7527 µg m −3 with an average of 2109 ± 1360 µg m −3 .We classified this period as the dust storm event (Phase I; see Fig. 1a and b for examples).From 10 March at 21:00 to 11 March at 12:00 LT, air parcels at the 100 and 300 m levels still originated from the Gobi desert but moved to eastern coastal China first and then returned to Xi'an, while the 500 m air parcel originated from southern China and moved slowly to Xi'an after spanning over the Qinling Mountains (Fig. 1c and e).During this period the TSP level ranged from 412 to 1037 µg m −3 with an average of 630 ± 155 µg m −3 (Table 1, hereinafter).We classified this period as the transition period (Phase II; see Fig. 1c for example).After 11 March at 12:00, the three levels of air parcels originated from the North China Plain and moved to Xi'an from the Qinling Mountains (about 1500 m above the ground level, Fig. 1d and e).During this period the TSP varied from 476 to 1399 µg m −3 with an average of 687 ± 194 µg m −3 .We classified this period as the non-dust-storm period (Phase III; see Fig. 1d for example).As seen in Fig. 1e, Xi'an is  located in the Guanzhong Basin and very close to the Qinling Mountains.From the transport tracks it can be seen that air parcels within Phase III moved much slower in comparison with those in Phase I and Phase II.Thus, we believe that aerosols in the non-dust-storm period are mostly derived from the local sources rather than from long-range transport.
In the following sections we will discuss the aerosol chemistry evolution based on the three classified categories.

Hourly changes in chemical compositions of ambient particles
As shown in Fig. 2a, during the dust storm period TSP showed two maxima, with the largest peak at the first hour (i.e., 9 March at 18:00 LT) and a second peak at noontime (11:00−13:00 LT) on 10 March.At the same time PM 2.5 concentration was 152 ± 127 µg m −3 with a maximum of 621 µg m −3 occurring in the first hour (Fig. 2a).
As shown in Table 1, relative abundances of PM 2.5 / TSP were 7.4 ± 3.4 %, 14 ± 2.3 % and 23 ± 4.8 % during the dust storm event, transition period and non-dust-storm event, respectively, suggesting that surface soils in the Gobi desert and Loess Plateau consist of a certain amount of fine particles.From the backward trajectories we found that air parcels in the first 10 h directly moved across the Guanzhong Basin from the north to the south (as exemplified in Fig. 1a), sweeping pollutants out of Xi'an.Then the air mass moved to Xi'an from the east on 10 March at 10:00 LT (see Fig. 1b for an example), with pollutants originating from the upwind cities  such as Weinan and Huayin.As a result, EC was almost undetectable in the beginning 10 h but sharply increased to more than 30 µg m −3 at noontime on 10 March during the dust storm period (Fig. 2b).Total inorganic ions in the dust storm, transition and nondust-storm periods accounted for 4 %, 8 % and 12 % of the TSP mass, respectively (Fig. 3).In contrast, concentration ratios of WSOC to OC at the three periods gradually decreased from 0.4 in the dust period to 0.3 in the non-dust period (Fig. 3), which is lower than that (0.5 ± 0.1) observed for dust storm events in the spring of 2009 (Wang et al., 2013).Our previous studies found that dusts from the Gobi desert contain significant amounts of water-soluble organic compounds, e.g., trehalose (Wang et al., 2011(Wang et al., , 2012)).In addition, heterogeneous formation of secondary organic aerosols (SOAs) on dust surface is another important contributor to WSOC of dust (Sullivan and Prather, 2007;Wang et al., 2013).Therefore, in comparison with that in spring of 2009 the lower ratio of WSOC / OC is most likely due to the differences in the dust source regions and/or SOA formation on the dust surface.The OC / EC ratio presented similar values during the transition and non-dust periods, with an average value around 4.0, which is higher than that for PM 2.5 in the city since coarse particles contain a lower amount of elemental carbon.Our previous investigation on the impact of Asian dust storms on Xi'an aerosols in the spring of 2009 found that 88 % (in mass) of airborne particulate sulfate originated from Gobi desert soil in the presence of the dust storm (Wang et al., 2011(Wang et al., , 2013)).A recent study on the atmospheric aerosols collected in the Taklamakan desert also reported that airborne particles in the desert are abundant in sulfate, which accounts for about 4 % of the particle mass with no significant difference for particles with different sizes (Wu et al., 2012).WSOC / WSON ratios are higher in the dust storm and transition periods and lower in the non-dust period, which can be ascribed to more WSON species emitted from anthropogenic sources such as agricultural fertilizer and livestock dejecta (Cape et al., 2011;Chen et al., 2010;Wang et al., 2010).3.5 ± 1.0 (2.4−5.9)7.6 ± 0.9 (5.6−8.7) 10 ± 5.9 (1.7−28) 6.6 ± 1.9 (2.8−9.8)11 ± 2.5 (6.9−15)As shown in Fig. 4a, extremely high levels of sulfate were observed during the dust storm period (Phase I), with a peak of 180 µg m −3 in the first hour, accounting for 2.3 % of the TSP mass (Table 1), which falls in the range reported for the airborne dust in the Taklamakan desert by Wu et al (2012).Due to the second peak of the dust storm that arrived in Xi'an at noon on 10 March, SO 2− 4 concentration showed a moderate peak as did TSP (see the inserted figure in Fig. 4a).During the whole sampling period, NO  which are different from other ions and are almost continuously increasing from the dust storm period to the nondust-storm period (Fig. 4b).The temporal variation pattern of Cl − is similar to that of SO 2− 4 , while the variation pattern of F − is similar to EC, indicating that Cl − and SO 2− 4 are of similar natural origins but F − and EC are of common anthropogenic sources, e.g., coal combustion (Wang et al., 2010).There are many dried salt lakes in the northern and western parts of China and the Gobi area of Mongolia, in which halite (NaCl), gypsum (CaSO 4 × 2H 2 O), mirabilite (Na 2 SO 4 × 10H 2 O) and other salts are common components of the surface soil (Zheng, 1991).Mineral species containing calcium and magnesium often coexist in desert regions.For example, dolomite (CaMg(CO 3 ) 2 ) is a common mineral salt in surface soil in the Taklamakan desert, Gobi desert and Loess Plateau of China (Li et al., 2007;Maher et al., 2009).Thus, both presented the same temporal pattern during the dust and transition periods when dust particles from the above regions are dominant (Fig. 4d).However, Ca 2+ and Mg 2+ displayed divergent patterns in the non-dust-storm period when the aerosols are dominated by local sources, indicating both ions are of different origins.
As seen in Fig. 5a, relative abundance of elemental calcium to TSP (Ca / TSP) during the whole sampling period was nearly constant, which slightly increased from 6.9 ± 1.0 % in the event to 8.2 ± 1.4 % in the non-event.
In contrast, the ratio of Ca 2+ / Ca kept increasing from  7.1 ± 2.8 % in the event, to 16 ± 6.0 % in the transition period to 22 ± 5.7 % in the non-event, indicating an enrichment of local soil that consists of more water-soluble calcium and/or a continuous conversion of elemental calcium into calcium cation.Although the mass ratio of SO 2− 4 / TSP displayed an increasing trend from about 1.0 % during the dust storm event to about 5.0 % in the non-event, the ratios of SO 2− 4 -S / total-S in the TSP samples during the three periods are almost the same (Fig. 5b): 62 ± 13 % in the dust storm event, 61 ± 6.1 % in the transition time and 54 ± 5.2%, in the nonevent period, suggesting that sulfate in the samples is largely derived from dust/soil and photochemical production of sulfate was minor even in the non-dust period.Such a result is consistent with the observation for the dust storm during the spring of 2009, by which we found only 12 % of particulate sulfate in Xi'an at that time was formed by secondary oxidation and 88 % of the sulfate was transported from the desert region (Wang et al., 2013).

Size distributions
Size is an important parameter of an aerosol, which is related to its origin, formation pathway and composition (Hinds, 1999).To further investigate the chemical evolution process of the urban aerosols, size-resolved chemical compositions were analyzed for the dust storm and non-dust-storm periods.The size-segregated samples collected in the transition period were not used in this study because sampling duration of these samples overlapped somewhat with the non-dust period.As shown in Fig. 6a, Cl − presented a bimodal pattern in the dust storm period with a large peak in the coarse mode (> 2.1 µm) and a minor peak in the fine mode (< 2.1 µm), in contrast to the case of the non-dust-storm period, which is characterized by two equivalent peaks in both the fine and the coarse modes.The pronounced coarse-mode peak in the dust storm period further suggests the origin of NaCl from dried salt lakes in the Gobi desert region.NO − 3 and SO 2− 4 dominated in the coarse mode when dust was present, but in the non-dust-storm period nitrate in the fine mode was much more abundant than in the coarse mode, while sulfate displayed two equivalent peaks in both modes (Fig. 6b and c).The SO 2− 4 distribution patterns in the dust storm and nondust-storm periods are similar to those of Cl − , while NO − 3 displayed similar patterns to those of NH + 4 in both periods (Fig. 6b-d).During the non-dust-storm period Cl − , NO − 3 and SO 2− 4 in the fine mode (< 2.1 µm) accounted for 55, 58 and 54 % of the mass in the whole size range, respectively (Table 2); whereas in the dust event the fine modes of Cl − , NO − 3 and SO 2− 4 decreased significantly, accounting for 40 ± 4 %, 31 ± 6 % and 27 ± 7 % of the total, respectively (Table 2).
Size distribution patterns of Na + are almost identical with those of Cl − in both the dust and non-dust-storm periods (Fig. 6a and e), probably indicating that both ions have the same sources even in the non-dust-storm period.Apart from dried salt lakes, soil and seal salt, Na + and Cl − can also originate from biomass burning.For example, Andreae et al (1998) measured aerosol emissions from savanna fires in southern Africa and found that Na + , Cl − and K + were abundant in the smoke, with 40-90 % of the mass distributing in particles with a diameter less than 1.2 µm.K + is generally considered a tracer of biomass burning smoke and enriched in fine particles (Andreae et al., 1998;Shen et al., 2007;Wang et al., 2012); thus the fine mode (< 2.1 µm) of Table 2. Accumulative percentages (%) of mass concentrations of major ions on the nine-stage filters collected in Xi'an during the spring dust storm and non-dust-storm periods.

Chemical forms, sources and productions of nitrate and sulfate
Figure 7 shows the linear fit regressions for nitrate, sulfate, ammonium and other major cations in the TSP samples during the dust storm and transition periods.Nitrate showed a robust correlation with ammonium (r 2 = 0.76, Fig. 7a) but presented no correlation with the sum of Na + , K + , Mg 2+ plus Ca 2+ (Fig. 7b).However, sulfate during the dust storm and transition periods were not correlated with ammonium but were strongly correlated with the sum of the above cations (Fig. 7c and d).Interestingly, robust linear correlations (r 2 > 0.99) were found for the samples collected at the beginning 3 h (red dots in Fig. 7a−d), suggesting that dust particles in the city within this period are of the same chem-   ical compositions and are not chemically modified.Nitrate and ammonium mass concentrations not only displayed the similar size distribution patterns during the whole sampling period but were also strongly linearly correlated with each other during the dust storm and transition periods with a slope of 0.28, which is equal to the 1 : 1 molar ratio of NH + 4 to NO − 3 (Fig. 7a).Therefore, we assumed that NH 4 NO 3 is the major chemical form of both ions in the airborne particles, especially in the dust storm and transition periods.Equivalent percentages showed that Na + and Ca 2+ are the two major cations during the dust storm and transition periods, accounting for 77 ± 5.0 % (range, 65−87 %) of the total cation equivalent (Fig. 8a).Nitrate and sulfate are the two major anions, accounting for 86 ± 2.1 % (range, 80−90 %) of the total anion equivalent (Fig. 8b).As discussed above, NaCl and NH 4 NO 3 are the major chemical forms of Cl − and NO − 3 in the TSP samples, respectively.Thus, it is reasonably expected that the remaining Na + and other major cations in the samples exist as sulfate salts (i.e., Na 2 SO 4 , CaSO 4 , MgSO 4 and K 2 SO 4 ) during the dust storm and transition periods.Concentrations of these sulfate salts during the dust and transition periods and their relative abundances to the total sulfate in the water-soluble fraction are summarized in Table 3 and shown in Fig. 9. Within these periods Na 2 SO 4 and CaSO 4 are the major sulfate salts, which originate from dust source regions and account for more than 90 % of the total water-soluble sulfate salts.18:00 20:00 22:00 0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 0:00 2:00 4:00 6:00 8:00 10:00 Mass ratio faster than sulfate.In the dust storm period SO 2− 4 almost entirely originated from surface soil in the Gobi desert, and no significant amount of SO 2− 4 in the dust storm event was secondarily produced (Fig. 11).
Hygroscopicity of the ambient aerosols during the whole campaign was also investigated by determining the hygroscopic growth factor of the water-soluble fraction of the TSP samples (Huang and Wang, 2014).The results showed that the κ value of the water-soluble fraction of dust particles ranged from 0.20 to 0.38 (0.30 ± 0.04), indicating a wettable nature of the dust particles (Andreae and Rosenfeld, 2008).Here, we propose a three-step mechanism of heterogeneous formation of nitrate on dust surface to explain the above secondary formation of NH 4 NO 3 (see Fig. 12).As discussed above, the airborne particles during the dust storm and transition period consist of significant amounts of water-soluble NaCl and Na 2 SO 4 .These compounds are very hygroscopic and thus may take up water vapor, forming a liquid phase on the dust surface even under the low-RH conditions of the dust storm period (RH = 22 ± 3.5%, Table 1).The aqueous phase is favorable for the formation of NH 4 NO 3 , which is probably formed via a gas-phase homogeneous reaction of nitric acid with ammonia and a subsequent partitioning into the liquid phase (Nie et al., 2012;Pathak et al., 2011).In addition, NO − 3 can also be produced in the liquid phase via heterogeneous reactions of gaseous HNO 3 , N 2 O 5 and NO x with dust particles (Finlayson-Pitts et al., 2003;Hanisch and Crowley, 2001;Laskin et al., 2005;Usher et al., 2003), www.atmos-chem-phys.net/14/11571/2014/

Figure 1 .
Figure 1.Backward trajectories of air masses arriving in Xi'an during the campaign(a-d) and the topography of the Guanzhong Basin and its surrounding areas (e) (duration 48 h; air parcels of 100, 300 and 500 m above ground level are in red, blue and green, respectively).

Figure 3 .
Figure 3.Comparison of relative abundance of ions, OC,EC, WSOC and WSON in the

Figure 3 .
Figure 3.Comparison of relative abundance of ions, OC, EC, WSOC and WSON in the TSP samples during the dust storm event, transition time and non-dust-storm period.

Figure 4 .Figure 4 .
Figure 4. Temporal variations of inorganic ions during the campaign

Figure 5 .Figure 5 .
Figure 5.Relative abundances of Ca2+/Ca, Ca/TSP, SO4-S/Total S and SO42-of the TSP samples collected during the sampling periods

Figure 6
Figure 6 Size distributions of major ions during the dust storm and non-dust storm periods.

Figure 6 .
Figure 6.Size distributions of major ions during the dust and non-dust-storm periods.

Figure 7 .Figure 7 .
Figure 7. Linear fit regressions for nitrate and sulfate with ammonium and other cations in the TSP samples collected during the dust storm and transition periods (red dots are the three samples collected at the earliest three hours) Figure 7. Linear fit regressions for nitrate and sulfate with ammonium and other cations in the TSP samples collected during the dust storm and transition periods (red dots are the three samples collected at the earliest 3 h).

Figure 9 .Figure 9 .
Figure 9.Chemical forms of water-souble sulfate salts in the TSP samples during the dust storm and transiti

Figure 11 .
Figure11.Mass ratios of nitrate, sulfate and ammonium produced by heterogeneous reactions to the total in the TSP samples , sulfate and ammonium produced by heterogeneous reactions to the total in the TSP samples Date (mm/dd) and time (hh:mm) roduction speed of nitrate during the sampling period ((a) molar ratio of nitrate/sulfate in the TSP samples and (b) linear fit regression for the nighttime molar ratio of nitrate/sulfate with observation duration).

Figure 11 .
Figure 11.Mass ratios of nitrate, sulfate and ammonium produced by heterogeneous reactions to the totals in the TSP samples.

Table 1 .
Meteorological parameters and hourly concentrations of inorganic ions, elements, EC, OC, water-soluble organic (WSOC) and inorganic carbon (WSIC), and water-soluble organic (WSON) and inorganic nitrogen (WSIN) in total suspended particles (TSP) during the dust storm, transition and non-dust periods.

Table 3 .
Concentrations (µg m −3 ) of different sulfates in the water-soluble fraction of TSP samples and their relative abundances (%) to the total water-soluble sulfate during the dust storm and transition periods.
Production speed of nitrate during the sampling period, (a) molar ratio of nitrate / sulfate in the TSP samples and (b) linear fit regression for the nighttime molar ratio of nitrate / sulfate with observation duration.