Summertime nitrate aerosol in the upper troposphere and lower stratosphere over the Tibetan Plateau and the South Asian summer monsoon region

We use the global three-dimensional Goddard Earth Observing System chemical transport model (GEOSChem) to examine the contribution of nitrate aerosol to aerosol concentrations in the upper troposphere and lower stratosphere (UTLS) over the Tibetan Plateau and the South Asian summer monsoon (TP/SASM) region during summertime of year 2005. Simulated surface-layer aerosol concentrations are compared with ground-based observations, and simulated aerosols in the UTLS are evaluated by using the Stratospheric Aerosol and Gas Experiment II satellite data. Simulations show elevated aerosol concentrations of sulfate, nitrate, ammonium, black carbon, organic carbon, and PM2.5 (particles with diameter equal to or less than 2.5 μm, defined as the sum of sulfate, nitrate, ammonium, black carbon, and organic carbon aerosols in this study) in the UTLS over the TP/SASM region throughout the summer. Nitrate aerosol is simulated to be of secondary importance near the surface but the most dominant aerosol species in the UTLS over the studied region. Averaged over summertime and over the TP/SASM region, CNIT (the ratio of nitrate concentration to PM2.5 concentration) values are 5–35 % at the surface, 25–50 % at 200 hPa, and could exceed 60 % at 100 hPa. The mechanisms for the accumulation of nitrate in the UTLS over the TP/SASM region include vertical transport and the gasto-aerosol conversion of HNO3 to form nitrate. The high relative humidity and low temperature associated with the deep convection over the TP/SASM region are favorable for the gas-to-aerosol conversion of HNO3.


Introduction
Aerosols in the upper troposphere and lower stratosphere (UTLS) have much longer residence times than those in the lower troposphere, influencing atmospheric chemistry and the Earth's climate with large spatial and temporal coverage (Rasch et al., 2008).Aerosols in the UTLS influence the concentrations of chemical species via changes in photolysis rates and heterogeneous reactions (Pitari et al., 2014).For example, heterogeneous reactions on sulfate aerosol can perturb the chemical partitioning in the lower stratosphere, leading to significant O 3 depletion through enhanced chlorine, bromine, and odd-hydrogen catalytic cycle (Zhao et al., 1997;Considine et al., 2001;Talukdar et al., 2012;Tang et al., 2014;Pitari et al., 2014).Aerosols in the UTLS also influence climate by altering properties of cirrus clouds via homogeneous or heterogeneous ice nucleation (Li et al., 2005;Liu et al., 2009;Yin et al., 2012;Fadnavis et al., 2013).Injection of aerosols into the UTLS has been reported to induce complex responses in circulation, temperature, and water vapor Published by Copernicus Publications on behalf of the European Geosciences Union.(Liu et al., 2009;Wu et al., 2011;Su et al., 2011;Fadnavis et al., 2013).
Aerosols over the Tibetan Plateau (TP) and the Asian summer monsoon region are especially important.The TP is surrounded by countries with large anthropogenic emissions (Li et al., 2005;Lau et al., 2006).Aerosols from India, Southeast Asia, and southern China can be transported to the TP by prevailing winds in the pre-monsoon and monsoon seasons (Lawrence and Lelieveld, 2010;Xia et al., 2011).Observational and modeling studies have shown that persistent maxima of atmospheric constituents, such as water vapor (Gettelman et al., 2004;Randel and Park, 2006;Park et al., 2007), CO (Kar et al., 2004;Li et al., 2005;Park et al., 2007Park et al., , 2008Park et al., , 2009)), CH 4 (M.Park et al., 2004;Xiong et al., 2009), NO x (M. Park et al., 2004), HCN (Park et al., 2008;Randel et al., 2010), C 2 H 6 and C 2 H 2 (Park et al., 2008), exist in the UTLS above the TP and the South Asian summer monsoon (SASM) region because of the deep convection during boreal summer.Satellite observations suggested that the convection associated with the SASM is a vital pathway to transport air mass from the lower troposphere into the stratosphere (Chen et al., 2006;Randel and Park, 2006;Randel et al., 2010;Bian et al., 2011a).The heating associated with the persistent deep convection during summertime leads to the formation of the Tibetan anticyclone in the UTLS, which acts to isolate air within the anticyclone and traps the uplifted pollutants at that altitude (Park et al., 2007;Vernier et al., 2011;Bourgeois et al., 2012;Fadnavis et al., 2013;He et al., 2014).The stratosphere-troposphere exchange (STE) over the TP contributes largely to the global STE (Chen et al., 2006).
Previous studies have reported that aerosols exist in the UTLS over the TP/SASM region.Kim et al. (2003) carried out optical measurements with a ground-based lidar in Lhasa from August to October of 1999, and found an enhancement in aerosol concentration near the local tropopause with scattering ratio (SR, the ratio of aerosol plus molecular backscatter to molecular backscatter alone) of 1.1-1.2.Tobo et al. (2007) reported an enhancement of submicron aerosols (effective radius r = 0.15-0.6µm) near the summertime tropopause (about 130 to 70 hPa), on the basis of in situ balloon measurements from an Optical Particle Counter at the same location in August of 1999.Vernier et al. (2009) examined satellite measurements from the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) on board Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) and reported the presence of small depolarizing particles with high SR values (about 1.20 at 532 nm) at 16-17 km altitude over South Asia in July and August of 2007 and 2008.Bourgeois et al. (2012) found that an aerosol layer existed at 16-18 km altitude over the Asian continent and Indian Ocean (20 • S-30 • N, 5-105 • E) on the basis of the CALIOP observations.Recently, He et al. (2014) examined the vertical profiles of aerosol extinction coefficients measured with a Micro Pulse Lidar at Naqu, a meteorological station located in the central part of the TP, and also showed a maximum in aerosol extinction coefficient (∼ 2.10 −3 km −1 ) in the UTLS (18-19 km) during the summer of 2011.
A number of previous studies have attempted to understand the chemical composition of aerosols in the UTLS.Froyd et al. (2009) measured aerosol composition with the National Oceanic and Atmospheric Administration (NOAA) single-particle mass spectrometer aboard the National Aeronautics and Space Administration (NASA) WB-57 highaltitude aircraft platform, and reported that particles in the tropical tropopause layer were rich in nitrogen.Vernier et al. (2011) suggested that aerosol layer at the tropopause of Asia could be sulfur and/or organics, considering that Asian pollutants consisted of black carbon, organic carbon, SO 2 , and NO x (Park et al., 2009;Randel et al., 2010).Weigel et al. (2011) analyzed the volatility of aerosols obtained from in situ airborne measurements and reported that about 75-90 % of the particles in the tropical tropopause layer were volatile, but this study did not give any detailed analyses of chemical composition of aerosols.Bourgeois et al. (2012) showed, by using the ECHAM5.5-HAM2model, that sulfate, water, and organic carbon (OC) contributed, respectively, 53, 29, and 11 % to aerosol extinction in the vicinity of the tropical tropopause layer.The ECHAM5.5-HAM2 model used by Bourgeois et al. (2012) simulated all major aerosol species in the atmosphere except for nitrate.
Few previous studies have examined nitrate aerosol in the UTLS, although nitrate is expected to be important for the following reasons.First, emissions of precursors of nitrate, such as NO x and NH 3 , are high over India, Southeast Asia, and China (Streets et al., 2003;Datta et al., 2012;Huang et al., 2012).Second, simulated nitrate concentrations are high over those regions (Liao and Seinfeld, 2005;Mu and Liao, 2014;Lou et al., 2014).Third, measured concentrations of nitrate are comparable to or larger than those of sulfate at rural and urban sites in the SASM region.Shrestha et al. (2000) carried out measurements of aerosols at Phortse, Nepal, during September 1996-November 1997, and showed that the average concentration of nitrate during the monsoon season (June-September) was 0.34 µg m −3 , higher than that of sulfate (0.17 µg m −3 ).Decesari et al. (2010) reported, on the basis of measurements at the Nepal Climate Observatory-Pyramid from 2006 to 2008, that the concentrations of nitrate and sulfate were 0.37 and 0.50 µg m −3 , respectively, during the monsoon season.Chatterjee et al. (2010) measured aerosols at a high-altitude station in the northeastern Himalayas during January-December 2005.They found that the average concentrations of fine-mode nitrate and sulfate were 3.31 ± 2.25 and 3.80 ± 2.9 µg m −3 , respectively.At Lahore, an urban site in Pakistan, the observed daytime nitrate concentration of 21.8 µg m −3 was also higher than sulfate concentration of 12.6 µg m −3 (Lodhi et al., 2009), as the observations were averaged over November 2005 to March 2006.Fourth, the low temperatures in the UTLS would favor nitrate formation (Seinfeld and Pandis, 2006).Therefore, it is of interest to take nitrate aerosol into consideration when we examine aerosols in the UTLS.
In this work we simulate nitrate aerosol and its contribution to aerosol concentrations in the UTLS over the TP (70-105 • E, 25-40 • N) and the SASM region (70-105 • E, 10-25 • N) by using the global Goddard Earth Observing System chemical transport model (GEOS-Chem) driven by the assimilated meteorological fields.These regions of interest are shown in Fig. 1.Simulated surface-layer aerosol concentrations are compared with ground-based observations, and simulated aerosols in the UTLS are evaluated by using the Stratospheric Aerosol and Gas Experiment II (SAGE II) satellite data.Section 2 is a brief description of the GEOS-Chem model and numerical experiment.Section 3 presents the simulation and evaluation of distributions and concentrations of HNO 3 and O 3 to show the model's capability in simulating the NO x -O 3 -HNO 3 cycle over the studied regions.Section 4 shows simulated aerosols and Sect. 5 presents the simulated contribution of nitrate to aerosol concentrations in the UTLS over the TP and the SASM region.Section 6 discusses the mechanisms for high concentrations of nitrate in the UTLS.Section 7 discusses the impacts of uncertainties in surface-layer aerosol concentrations on simulated nitrate in the UTLS.

GEOS-Chem model
We simulate gas-phase species and aerosols using the global chemical transport model GEOS-Chem (version 9-01-03, http://geos-chem.org)driven by the GEOS-5 assimilated meteorological fields from the Goddard Earth Observing System of the NASA Global Modeling and Assimilation Office.The version of the model used here has a horizontal resolution of 2 • latitude by 2.5 • longitude and 47 vertical layers extending from the surface to 0.01 hPa.Over the TP and the SASM region, the model has about 34 layers in the troposphere and 12 layers in the stratosphere.
The gas-aerosol partitioning of nitric acid and ammonium is calculated using the ISORROPIA II thermodynamic equilibrium module (Fountoukis and Nenes, 2007).In the version of the GEOS-Chem model used in this work, ions considered in ISORROPIA II include H + /Na + /NH + 4 /Cl − /SO 2− 4 /HSO − 4 /NO − 3 /OH − .The two-way coupling between aerosols and gas-phase chemistry provides consistent chemical fields for aerosol simulation and aerosol mass for heterogeneous processes and calculations of gasphase photolysis rates.Heterogeneous reactions include hydrolysis of N 2 O 5 (Evans and Jacob, 2005), irreversible absorption of NO 3 and NO 2 on wet aerosols (Jacob, 2000), and the uptake of HO 2 by aerosols (Liao and Seinfeld, 2005;Thornton et al., 2008).Aerosol species are treated as an external mixture in the calculation of aerosol optical properties.
With respect to chemistry in the stratosphere, stratospheric O 3 concentrations are calculated using the linearized parameterization scheme (McLinden et al., 2000).The monthly mean production rates and loss frequencies of other stratospheric species (including long-lived species such as CFCs and N 2 O) use those from NASA Global Modeling Initiative (GMI) Combo simulations (Considine et al., 2008;Murray et al., 2012).
Convective transport in GEOS-Chem mimics that in the parent GEOS general circulation model (Hack, 1994;Zhang and McFarlane, 1995), which accounts for updraft, downdraft, and entrainment mass fluxes for deep and shallow convection (Wu et al., 2007).The aerosol wet deposition scheme in the GEOS-Chem follows that of Liu et al. (2001).For the scavenging of aerosols, SO 2− 4 , NO − 3 , NH + 4 , and hydrophilic OC and hydrophilic BC aerosols are assumed to be fully soluble.Dry deposition follows the standard resistance-in-series model of Wesely (1989).
Global emissions of aerosols and their precursors in the GEOS-Chem follow R. J. Park et al. (2003Park et al. ( , 2004)), with anthropogenic emissions of NO x , CO, SO 2 , and non-methane volatile organic compounds in Asia overwritten by David Streets' 2006 emission inventory (http://mic.greenresource.cn/intex-b2006).Emissions of NH 3 in Asia are taken from Streets et al. (2003).Since NH 3 emissions in China showed large uncertainties in previous studies (Streets et al., 2003;Kim et al., 2006;Y. Zhang et al., 2010;Huang et al., 2011Huang et al., , 2012)), we use the most recent estimate of NH 3 emissions in Figure 2. Monthly variations in emissions of NO x (Tg N month −1 ), SO 2 (Tg S month −1 ), NH 3 (Tg N month −1 ), OC (Tg C month −1 ), and BC (Tg C month −1 ) over Asia.Values shown are the total emissions (anthropogenic plus natural emissions listed in Table 1).
China by Huang et al. (2012), which is 9.8 Tg yr −1 , instead of 13.5 Tg yr −1 from Streets et al. (2003).Natural NO x emissions from lightning are calculated using the scheme described by Sauvage et al. (2007) and Murray et al. (2012), and those from soil are simulated following Wang et al. (1998).Natural NH 3 emissions from soil, vegetation, and the oceans are taken from the Global Emissions Inventory Activity inventory (Bouwman et al., 1997).Biomass burning emissions are from the monthly Global Fire Emissions Database (GFED v3) driven by satellite observations of fire activity (van der Werf et al., 2010).Biogenic volatile organic compounds emissions are calculated from the Model of Emissions of Gases and Aerosols from Nature (Guenther et al., 2006).
The monthly variations of emissions of SO 2 and NO x follow Wang et al. (2013) and those of BC and OC follow Lou et al. (2014).The monthly scaling factors for NH 3 emissions follow the global inventory described in Fisher et al. (2011).Monthly variations of emissions (anthropogenic plus natural emissions) of NO x SO 2 , NH 3 , OC, and BC over Asia are displayed in Fig. 2. The emissions of NH 3 are the highest in June as a result of the agriculture practice and high temperatures (Wang et al., 2013).

Numerical experiment
To examine the contribution of nitrate to aerosol concentrations in the UTLS over the TP/SASM region, we simulate aerosol concentrations by using the emissions of and meteorological fields of year 2005.Year 2005 is chosen so that we can use the observational data sets for this year from SAGE II and MLS (Microwave Limb Sounder), as described in Sects. 3 and 4. Following Rasch et al. (2008), we perform a 10-year spin-up run to generate the initial conditions (to al-low the stratospheric species to reach quasi-steady-state conditions).We would consider that the tropospheric simulation can be representative of year 2005 but stratosphere simulation should represent a multi-year average, because the production rates and loss frequencies in the stratosphere are the averages over years of 2004-2010 (http://wiki.seas.harvard.edu/geos-chem/index.php/Stratospheric_chemistry).
3 Simulated concentrations of HNO 3 and O 3 and model evaluation Nitrate aerosol forms when nitric acid (HNO 3 ) reacts with alkaline gases (for example, ammonia) in the atmosphere (Seinfeld and Pandis, 2006).HNO 3 , as the important precursor of nitrate, is the major oxidation product of nitrogen oxides (NO x = NO + NO 2 ) (Seinfeld and Pandis, 2006).To show the model's capability in simulating the NO x -O 3 -HNO 3 cycle over the studied regions, we present and evaluate the simulated HNO 3 and O 3 in this section.Simulated mixing ratios of HNO 3 and O 3 in the UTLS are evaluated by using data sets from the limb viewing satellite instrument of Microwave Limb Sounder (MLS, version 3.3, level 2, ftp://acdisc.gsfc.nasa.gov/data/s4pa///Aura_MLS_Level2/).The MLS data sets provide valuable information on atmospheric compositions in the UTLS (Waters et al., 2006).For HNO 3 , the MLS provides scientifically useful data sets for 215 to 1.5 hPa, with a vertical resolution of 3-4 km and a horizontal resolution of 400-500 km.Since further evaluations are needed for data sets at altitudes with pressures higher than 215 hPa (Livesey et al., 2011), we use only data sets for pressures lower than that.For O 3 , the MLS provides scientifically useful data sets for 261 to 0.02 hPa, with a vertical resolution of 2.5-3 km and a horizontal resolution of 300-400 km in the UTLS (Santee et al., 2007;Livesey et al., 2011).The uncertainties of the MLS HNO 3 and O 3 data sets in the UTLS are discussed in Livesey et al. (2011).The MLS data HNO 3 product at 100 hPa represents an average of a 3-4 km layer reaching from below the tropopause layer into the lower stratosphere (Duncan et al., 2007).The data screening is conducted strictly according to the instructions of Livesey et al. (2011).For comparison with the MLS observations, the simulated HNO 3 and O 3 mixing ratios are converted to x, following the method of Livesey et al. (2011): where A is the averaging kernel matrix of the MLS, x is the modeled vertical profile of HNO 3 or O 3 interpolated at the vertical grid of MLS, and x a is the HNO 3 or O 3 prior profile of MLS.trations of HNO 3 exceed 1 ppbv over the industrialized areas such as Europe, North America, central and eastern Asia, and over biomass burning regions in the tropics, in agreement with the distributions and magnitudes reported in Liao et al. (2003).Over South Asia, simulated HNO 3 concentrations are high (0.3-1 ppbv) in the northern Indian subcontinent, because the emissions of NO x and NH 3 are high in this region (Streets et al., 2003;Zhang et al., 2009;Datta et al., 2012).Figure 4a and b show the simulated HNO 3 concentrations in the UTLS averaged over June-August of 2005.Since the tropopause is located at 70-150 hPa (12-15 km) over the TP/SASM region (Li et al., 2005;Bian et al., 2011b;Fadnavis et al., 2015), we choose the vertical layers of 200 and 100 hPa to represent the UTLS.At both 200 and 100 hPa, the highest HNO 3 concentrations are simulated to occur in the high-latitude regions in the Northern Hemisphere (NH) (Fig. 4a and b).Simulated HNO 3 concentrations at 100 hPa are low over the region of 40-100 • E and 10-30 • N, which is part of the anticyclone region defined in Fig. 1. Figure 4c shows the latitude-altitude cross section of simulated seasonal mean HNO 3 mixing ratios averaged over 70-105 • E. In boreal summer, the highest HNO 3 mixing ratios are simulated to occur at 30 hPa over the polar regions in both hemispheres.Over high latitudes, HNO 3 concentrations in the Southern Hemisphere (SH) are simulated to be higher than those in the NH.

HNO 3
To evaluate the simulated HNO 3 , Fig. 4d-f show HNO 3 concentrations in the UTLS from MLS that are averaged over June-August of 2005.At 200 and 100 hPa altitudes, the observed HNO 3 mixing ratios are high in the high latitudes in the NH, which are captured by the GEOS-Chem model.The observed HNO 3 at 100 hPa exhibits low values of less than 400 pptv over 30-100 • E and 10-30 • N in the Asian monsoon anticyclone region (Fig. 4e).At 100 hPa, the observed HNO 3 mixing ratio averaged over the TP/SASM region (70-105 • E, 10-40 • N) is 335.4 pptv, which is lower than the simulated value of 372.6 pptv.Considering all the grid cells with MLS HNO 3 data available, the simulated seasonal mean HNO 3 concentrations show normalized mean bias (NMB) of +11.1 % at 100 hPa over the TP/SASM region in summer of year 2005.The observed pattern of the HNO 3 vertical distribution (Fig. 4f) is also captured by the GEOS-Chem model (Fig. 4c).The distributions of HNO 3 in the UTLS are associated with the Brewer-Dobson circulation proposed by Brewer (1949) and Dobson (1956), traveling upwards across the tropopause to the stratosphere at the equator and downwards to the troposphere near the polar region.in the UTLS over the TP/SASM region are caused by the deep convection that transports O 3 -poor air upward (Fu et al., 2006;Randel and Park, 2006;Park et al., 2007;Bian et al., 2011b).Figure 5c displays the latitude-altitude cross section of seasonal mean O 3 mixing ratios averaged over 70-105 • E. As a result of the Brewer-Dobson circulation, O 3 concentrations in the UTLS are lower over the tropics than in the polar regions, even though the maximum O 3 concentrations are located around 10 hPa over the tropics (Brewer, 1949).Our simulated O 3 concentrations in the UTLS agree well with the measurements from MLS (Fig. 5d-f).At 100 hPa, simulated and MLS observed O 3 mixing ratios averaged over the TP/SASM region (70-105  Stevenson et al., 2006).

O 3
In addition to the comparisons against MLS products, the simulated O 3 profiles are compared with balloon-borne sonde measurements in Fig. 6.The measurements were carried out at Kunming (KM, 102.7 • E, 25.0 • N) in August of 2009 and 2012, and at Lhasa (LH, 91.1 • E, 29.7 • N) in August of 2010 and 2013.The uncertainties of the observed O 3 mixing ratios were estimated to be within 5-10 % (Bian et al., 2012).The comparisons with multi-year observations show that the model can reproduce the vertical distributions of O 3 above 12 km in Kunming and Lhasa.At 100 hPa, the simulated monthly mean O 3 mixing ratio in KM is 112.6 ppbv, and the observed value is 124.2 ppbv in 2009 and 113.5 ppbv in 2012.In LH, the simulated monthly O 3 mixing ratio at 100 hPa is 152.6 ppbv, and the observed O 3 mixing ratio at that altitude is 142.4 ppbv in 2010 and 167.9 ppbv in 2013.The magnitudes of O 3 mixing ratios from these balloonborne sonde measurements support those from MLS; O 3 mixing ratios in the UTLS are less than 200 ppbv over the TP/SASM region.4 Simulated aerosols and model evaluation

Simulated aerosols
Figure 7a shows the simulated surface-layer concentrations of SO 2− 4 , NO − 3 , NH + 4 , OC, BC, and PM 2.5 (the sum of the mass of SO 2− 4 , NO − 3 , NH + 4 , BC, and OC aerosols) averaged over June-August of year 2005.As expected, simulated aerosol concentrations are high over polluted regions such as India and eastern China as a result of the high anthropogenic emissions of aerosol precursors and aerosols (Streets et al., 2003;Huang et al., 2012).Over the TP/SASM region (70-105 • E, 10-40 • N), the average concentrations of SO 2− 4 , NO − 3 , NH + 4 , BC, and OC are 1.70, 0.94, 0.85, 0.30, and 0.94 µg m −3 , respectively.NO − 3 is simulated to be of secondary importance at the surface over the region of our interest.The simulated distributions and magnitudes of these aerosol species are similar to those reported in Wang et al. (2013) and Mu and Liao (2014).
Figure 7b and c also show the simulated concentrations of SO 2− 4 , NO − 3 , NH + 4 , OC, BC, and PM 2.5 in the UTLS.Elevated concentrations of SO 2− 4 , NO − 3 , NH + 4 , OC, BC, and PM 2.5 are simulated over the TP and the plateau south slope at 200 hPa altitude, and extend from the eastern Mediterranean to western China at 100 hPa.The simulated enhanced concentrations of SO 2− 4 , OC, and BC at 100 hPa over the anticyclone region (20-120 • E, 10-40 • N) agree with previous observational and modeling studies (Lelieveld et al., 2001;Li et al., 2005;Fadnavis et al., 2013).Li et al. (2005) reported elevated CO concentrations in the upper troposphere over the TP, on the basis of both MLS measurements and the GEOS-Chem simulation for September 2004.Fadnavis et al. (2013) also simulated maximum concentrations of SO 2− 4 , OC, BC, and mineral dust aerosols in the UTLS during the Asian summer monsoon season owing to convective uplifting of the boundary layer pollutants.With NO − 3 aerosol accounted for in our simulation, NO − 3 is simulated to be the most dominant aerosol species in the UTLS over the TP/SASM region, followed by SO 2− 4 , NH + 4 , OC, and BC.At 100 hPa, the averaged concentrations of SO 2− 4 , NO − 3 , NH + 4 , OC, and BC over the TP/SASM region (70-105 • E, 10-40 • N) region are 0.026, 0.069, 0.014, 0.011, and 0.002 µg m −3 , respectively.

Comparisons of simulated aerosol concentrations with in situ observations
The simulated aerosol concentrations in East Asia in the GEOS-Chem model have been evaluated in previous studies (L.Zhang et al., 2010;Fu et al., 2012;Jeong and Park, 2013;Jiang et al., 2013;Wang et al., 2013;Lou et al., 2014).
Here we are focused on the evaluation of aerosols in the South Asian monsoon region.For lack of publicly accessible in situ measurements of summertime aerosols in the South Asia monsoon area, we compiled monthly or seasonal mean measured concentrations of each aerosol species based on measurements reported in the literature (see Table S1 in the Supplement).These measurements were carried out over the years of 1992-2010.The locations of sites with measurements available are shown in Fig. 8a.Most sites are located in the upwind directions of the TP, with pollutants that can be transported to the UTLS during the South Asian summer monsoon season.The observed PM 10 concentrations listed in Table S1 are multiplied by 0.6 to convert to PM 2.5 for model evaluation, following the suggestions in Zhang et al. (2002) and Chatterjee et al. (2010).Figure 8b-f show the scatter plots of simulated vs. observed seasonal mean aerosol concentrations.Compared with measurements, simulated SO 2− 4 , NO − 3 , NH + 4 , OC, and BC have NMBs of −17.0, +38.8, +42.0, −69.7, and −41.0 %, respectively, as the concentrations of all seasons are considered.The correlations between model results and observations have R values of 0.49-0.85for all aerosol species, indicating that the model is capable of capturing the spatial distributions and seasonal variations of each aerosol species in the South Asian monsoon region despite the biases in concentrations.If we consider simulated and measured concentrations for JJA alone, the simulated concentrations of SO 2− 4 , NO − 3 , NH + 4 , OC, and BC exhibit seasonal NMBs of −14.7, +51.5, +74.9, −57.2, and −32.2 %, respectively, and the values of R are in the range of 0.24-0.85.Note that the measurements of NO − 3 and NH + 4 are quite limited in terms of the number of samples, and the discrepancies between model results and measurements may also arise from the mismatch of the model year 2005 with the years of 1992-2010 with observations available.

Comparisons of simulated aerosol extinction coefficients with SAGE II data sets
Satellite data sets from the Stratospheric Aerosol and Gas Experiment II (SAGE II, https://eosweb.larc.nasa.gov/project/sage2/sage2_v620_table) are used to evaluate the simulated aerosol extinction in the UTLS.The SAGE II instrument was launched in October 1984 aboard the Earth Radiation Budget Satellite (ERBS) and terminated on 8 September 2005 (McCormick, 1987;Chu et al., 1989).The data sets used here are aerosol extinction coefficients at 525 nm from the version 6.20 SAGE retrievals, covering from 0.5 to 40 km with a vertical resolution of 0.5 km.Many validation studies have been conducted on the SAGE II aerosol data (Russell and McCormick, 1989;Oberbeck et al., 1989;Wang et al., 1989), which indicated that extinction coefficients have uncertainties of 20-30 %.The extinction coefficients of aerosols in the GEOS-Chem model are calculated using aerosol mass concentration, extinction efficiency, effective radius, particle mass density, and the assumed aerosol size distribution (Drury et al., 2010).The hygroscopic growth of each aerosol species with relative humidity is accounted for, using the hygroscopic growth factors listed in Martin et al. (2003).Figure 9a presents the simulated monthly mean distribution of aerosol extinction coefficients at 100 hPa for July of 2005.At 100 hPa, the simulated aerosol extinction coefficients are relatively high over the anticyclone region, where anthropogenic aerosol species (Fig. 7) and natural aerosols such as mineral dust and sea salt contribute to aerosol ex-  Accounting for all aerosol species, the GEOS-Chem model reproduces the aerosol extinction coefficients above 10 km well, but the discrepancies are rather large in altitudes less than 10 km.Note that the uncertainties in satellite data sets increase as the altitude decreases (Vanhellemont et al., 2008;Kulkarni and Ramachandran, 2015), and the missing data in the lower troposphere along the satellite trajectories over the region of our interest also contribute to the discrepancies.
Comparisons of profiles of aerosol extinction coefficients with and without nitrate aerosol indicate that the profiles show small differences in altitudes less than 6 km but large discrepancies from 6 km to the tropopause.With nitrate aerosol accounted for, the simulated aerosol extinction coefficients agree closely with SAGE II data sets in the UTLS (averaged over 14-16 km, the simulated value is 8.6 × 10 −4 km −1 , while the observed value is 8.0 × 10 −4 km −1 ).Without nitrate aerosol, the simulated aerosol extinction coefficient at 14-16 km altitude is 1.5 × 10 −4 km −1 , which underestimates the aerosol extinction coefficient by 82.6 % compared to that calculated with all the aerosol species.These comparisons of extinction coefficients with and without nitrate aerosol suggest that nitrate aerosol plays an important role in aerosol extinction in the UTLS over the region of our interest.

Contribution of nitrate to aerosol concentrations in the UTLS
Since nitrate aerosol is simulated to be the most abundant aerosol species in the UTLS over the TP/SASM region, we analyze the contribution of nitrate to PM 2.5 concentration (C NIT = nitrate concentration × 100 % / PM 2.5 concentration) in this section.Figure 10 shows the simulated seasonal mean distributions of C NIT for June-August of year 2005.At the surface layer (Fig. 10a), simulated high C NIT values are located over the areas with high nitrate concentrations (India and eastern China) as well as the oceans where NO − 3 also forms on sea salt and mineral dust particles (Arimoto et al., 1996;Nakamura et al., 2005;George and Nair, 2008).Over the TP/SASM region, the C NIT values in JJA are 5-35 % at the surface, 25-50 % at 200 hPa (Fig. 10b), and could exceed 60 % at 100 hPa (Fig. 10c).The latitudealtitude cross section of C NIT (Fig. 10d) shows that C NIT over 20-40 • N increases with altitude and reaches maximum values around the extratropical tropopause.
Table 2 lists the mean concentrations of SO 2− 4 , NO − 3 , NH + 4 , BC, and OC, and their contributions to PM 2.5 during summertime of 2005 over the TP/SASM, TP, and SASM regions.Over the TP/SASM region SO 2− 4 , NO − 3 , NH + 4 , BC, and OC are simulated to contribute 35.9,19.8,18.1,6.4,and 19.8 %,respectively,to PM 2.5 mass concentration at the surface layer.The contributions increase significantly in the UTLS.The largest C NIT is simulated in the SASM region at 100 hPa, where NO − 3 accounts for 60.5 % of PM 2.5 mass concentration.The high C NIT values indicate that NO − 3 plays an important role in the aerosol layer in the UTLS over the TP/SASM region.
Considering the large uncertainties in simulated sea salt (Jaeglé et al., 2011) and mineral dust (Fairlie et al., 2007) aerosols, we tend to be focused on anthropogenic aerosol species (SO 2− 4 , NO − 3 , NH + 4 , BC, and OC) in this work.In our model, concentrations of sea salt (or mineral dust) are simulated to be 1.0-1.7 ng m −3 (or 5.0-7.0 ng m −3 ) over the studied region in the summer of 2005, which contribute less than 1.2 % (or 5.0 %) to total aerosol mass at 100 hPa.Therefore the consideration of sea salt and mineral dust can slightly reduce C NIT values, but C NIT values at 100 hPa are still as high as 45-65 % over the TP/SASM region in summer.
6 Mechanisms for high nitrate concentrations in the UTLS 6.1 Upward transport of nitrate from the lower troposphere The intense convective transport of chemical species into the UTLS over the TP/SASM region during summertime has been widely discussed in previous studies (Randel et al., 2010;Bian et al., 2011a;Fadnavis et al., 2013Fadnavis et al., , 2015;;Qie et al., 2014;He et al., 2014), evidenced by both the satellite observations (Fu et al., 2006;Luo et al., 2011) and the calculation of the outgoing long-wave radiation which is a convective proxy over the region (Randel and Park, 2006;Park et al., 2007;Fadnavis et al., 2013).Since nitrate aerosol is simulated to be of secondary abundant aerosol species in the surface layer over the TP/SASM region (Fig. 7), the vertical mass transport through the deep convection in this region contributes to the accumulation of NO − 3 in the UTLS. Figure 11 shows the latitude-altitude cross sections of simulated concentrations of SO 2− 4 and NO − 3 averaged over 70-105 • E in June-August of 2005, together with the wind vectors obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-Interim Reanalysis data.Note that the assimilated GEOS-5 meteorological fields do not have vertical winds (http://wiki.seas.harvard.edu/geos-chem/index.php/List_of_GEOS-5_met_fields), so we use the ECMWF reanalysis wind fields to do the analysis here.High values of aerosol concentrations are found on the south slope of the Himalayas, where the deep convection exists.Although both SO 2− 4 and NO − 3 are transported upward to the extratropical tropopause, the details of the vertical distributions are different.At altitudes higher than 8 km, the concentrations of NO − 3 do not decrease with altitude as quickly as those of SO 2− 4 , and the concentrations of NO − 3 over 10-40 • N are higher than those of SO 2− 4 .The chemical mechanisms for the formation of SO 2− 4 NO − 3 , and NH + 4 aerosols in the GEOS-Chem model were described in R. J. Park et al. (2004), which are comprehensive and have been used extensively in previous studies to simulate these three aerosol species (R. J. Park et al., 2004;Pye et al., 2009;L. Zhang et al., 2010;Zhu et al., 2012;Jiang et al., 2013;Lou et al., 2014)   SO 2 by O 3 and H 2 O 2 .Nitrate forms from the partitioning of HNO 3 between gas and aerosol phases, which is calculated by the ISORROPIA II thermodynamic equilibrium module (Fountoukis and Nenes, 2007) in the GEOS-Chem model.HNO 3 is produced by the reaction of NO 2 with OH during daytime and by hydrolysis of N 2 O 5 on aerosol surfaces at night (Table 3).The chemical mechanisms for SO 2− 4 and NO − 3 have different sensitivity to meteorological conditions.During the vertical transport, temperature decreases, which reduces the gas-phase oxidation of SO 2 (Yao et al., 2002;Seinfeld and Pandis, 2006) but promotes the formation of NO − 3 by shifting gas-particle equilibria (Dawson et al., 2007;Liao et al., 2009).Dawson et al. (2007) examined the sensitivities of sulfate and nitrate concentrations to temperature by using the Particulate Matter Comprehensive Air Quality Model with extensions (PMCAMx).The sensitivity test was performed by fixing all meteorological parameters but perturbing temperature.Their sensitivity simulations showed that the increases in temperature led to increases in sulfate concentrations and decreases in nitrate concentrations.Compared to nitrate, sulfate concentrations showed smaller sensitivity to temperature changes (Dawson et al., 2007); as temperature increased, nitrate concentrations decreased by 19 and 17 % K −1 in January and July respectively, while sulfate concentration increased by 0.12 and 1.3 % K −1 in January and July, respectively.Therefore the different chemical mechanisms for SO 2− 4 and NO − 3 formation contribute to the differences in their vertical distributions.

Net chemical production of HNO 3 during the vertical transport
As mentioned above, the formation of gas-phase HNO 3 and the partitioning of HNO 3 between gas and aerosol phases are the two major chemical processes that influence NO − 3 concentrations.The ability of the GEOS-Chem model to simulate gas-phase HNO 3 has been evaluated in Sect.3.1 (by comparisons of our model results with MLS observations and concentrations from previous modeling studies).Major reactions for the production and loss of HNO 3 are listed in Table 3. Figure 12 shows the net chemical production of HNO 3 by gas-phase reactions and heterogeneous reactions (chemical production by Reactions (R1)-(R23) minus chemical loss by Reactions (R24)-(R25) in Table 3) summed over the TP/SASM region.The net chemical production has an overall trend of decreasing with altitude.Since NO x emis-Table 3. List of gas-phase reactions and heterogeneous reactions involve HNO 3 in the GEOS-Chem model (Version 9-01-03).

Reactants Products
Chemical productions sions from aircraft and lightning are located between 200 and 300 hPa over the TP/SASM region during summer (Martin et al., 2007;Murray et al., 2012;Pitari et al., 2015), net chemical production of HNO 3 shows a small peak at those altitudes.The mixing ratio of HNO 3 decreases with altitude between 500 and 200 hPa and increases with altitude above 200 hPa.At 100 hPa, the average HNO 3 mixing ratio over the TP/SASM region agrees with the values of 300-400 pptv over the same region shown in Fig. 4. Figure 12 also shows the transport of HNO 3 over the TP/SASM region, defined as convergence of horizontal flux of HNO 3 (inflow minus outflow) plus convergence of vertical flux of HNO 3 (inflow minus outflow) for a specific vertical model layer.At altitudes below 200 hPa, concentrations of HNO 3 over the TP/SASM region are dependent on the net chemical production by gas-phase reactions and heterogeneous reactions as well as the transport of HNO 3 .The vertical variation of the horizontal transport (the north-south and the east-west transport) follows the development of the anticyclone in the upper troposphere during summertime.As the altitude increases, the westerlies enhance over the TP and the easterlies develop over the SASM region (see Fig. 1 for our definitions of TP and SASM regions).At latitudes between 100 and 500 hPa, vertical transport of HNO 3 occurs over the TP due to the deep convection activities over the region which can arrive at 100-200 hPa (Fig. 11;Fadnavis et al., 2013;Qie et al., 2014).

6.3
The gas-to-aerosol conversion of HNO 3 to form nitrate during the vertical transport NO − 3 formation from gas-to-aerosol conversion of HNO 3 is calculated by using the ISORROPIA II thermodynamic equilibrium module (Fountoukis and Nenes, 2007).As shown in Fig. 12, the nitrate formation from gas-to-aerosol conversion of HNO 3 peaks between 100 and 300 hPa, indicating that the gas-aerosol partitioning plays an important role in the enhancement of nitrate in the UTLS.
The gas-to-aerosol conversion of HNO 3 to form NO − 3 is very sensitive to relative humidity (RH) and temperature (Fountoukis and Nenes, 2007;Dawson et al., 2007).Low temperature and high RH are favorable for NO − 3 formation.Figure 13 shows the seasonal mean horizontal distributions of RH and temperature at 100 hPa and the latitude-altitude cross sections of these two parameters averaged over 70-105 • E. RH exhibits high values in the TP/SASM region, which are consistent with the high H 2 O mixing ratios in this area reported in Gettelman et al. (2004), M. Park et al. (2004), andFu et al. (2006).At 100 hPa, the locations with high RH of exceeding 45 % correspond well with those with high C NIT values (Fig. 10c).The latitude-altitude cross section of RH (Fig. 13c) shows that RH has high values over the places with intense upward transport (Fig. 11).For temperature, as Fig. 13b and d show, summertime temperatures are cold (190-200 K) at 100 hPa in the TP/SASM region, consistent with the distribution and magnitude reported for August 2011, in He et al. (2014) on the basis of the NCEP Reanalysis data.The low temperatures over the TP/SASM region are associated with the adiabatic expansion of ascending air mass of the deep convections (Yanai et al., 1992;Park et al., 2007;He et al., 2014).
Because of the favorable conditions of RH and decreasing temperature, the gas-to-aerosol conversion of HNO 3 to form nitrate can occur during the upward transport and in the UTLS. Figure 14 shows the mass budget for nitrate aerosol within the selected box of 70-105 • E, 10-40 • N, 8-16 km to see the role of nitrate formation over the TP/SASM region.The horizontal mass fluxes have a net negative value of 0.10 Tg season −1 , reducing nitrate aerosol in the selected box.The vertical transport and the gas-to-aerosol conversion of HNO 3 increase nitrate mass in the selected box, with values of 0.09 and 0.11 Tg season −1 , respectively, indicating that the gas-to-aerosol conversion plays an important role in the enhancement of nitrate in the UTLS over the TP/SASM region.Although relatively high RH exists near the tropopause of the TP/SASM region, the air near the tropopause is still drier compared to that in the lower altitudes.Model results show that the gas-to-aerosol partition of HNO 3 decreases with altitude between 100 and 300 hPa, indicating that the gas-to-aerosol conversion contributes to nitrate accumulation in the UTLS mainly during the process of upward transport.
Previous studies have also reported that nitric acid trihydrates (NAT, HNO 3 q (H 2 O) 3 ) could form in the polar and tropical stratosphere at low temperatures through two mechanisms: (1) the homogeneous nucleation out of supercooled ternary solutions, and (2) the heterogeneous formation on ice particles (Hofmann et al., 1989;Carslaw et al., 1998;Voigt et al., 2000;Popp et al., 2006;Kirner et al., 2011).A typical NAT condensation temperature is approximate 193 K (Kirner et al., 2011).As shown in Fig. 13, the temperatures around 100 hPa over the TP/SASM region are in the range of 190-200 K, which are low enough to produce some NAT particles.However, balloon-borne measurements of depolarization ratio and backscattering ratio of aerosols at Lhasa during August-October of 1999 by Kim et al. (2003) and Tobo et al. (2007) suggested that coarse and aspherical particles such as NAT are scarce in the UTLS of the TP/SASM.7 Sensitivities of simulated nitrate in the UTLS to anthropogenic NO x , NH 3 , and SO 2 emissions in Asia Since simulated SO 2− 4 , NO − 3 , and NH + 4 concentrations have NMBs of, respectively, −17.0, +38.8, and +42.0 % on an annual mean basis and of −14.7, +51.5, and +74.9 % in summer (Sect.4.2), we perform four sensitivity simulations to examine the impacts of uncertainties in surface-layer aerosol concentrations on simulated nitrate in the UTLS.In the first three cases, anthropogenic emissions of NO x , NH 3 , and SO 2 in Asia are changed by −50, −50, and +20 %, respectively, relative to those in our standard simulation.In the last case, anthropogenic emissions of all these three species are changed simultaneously, with NO x reduced by 50 %, NH 3 reduced by 50 %, and SO 2 increased by 20 % in Asia relative to the standard case.The purpose of these sensitivity studies is to reduce NMBs of simulated surface-layer concentrations of SO 2− 4 , NO − 3 , and NH + 4 and to see whether NO − 3 is still the most dominant aerosol species in the UTLS.Model results from these sensitivity studies for summer of 2005 are presented in Table 4.
As anthropogenic emissions of SO 2 in Asia are increased by 20 %, the NMB of simulated surface-layer SO 2− 4 concentrations is −4.4 %, which is an improvement compared to the NMB of −14.7 % in the standard simulation.However, the increases in SO 2 emissions lead to larger NMBs of surfacelayer NO − 3 and NH + 4 because of the increased formation of ammonium sulfate or ammonium bisulfate.The percentage contributions of SO 2− 4 to total aerosol mass in the UTLS increase slightly by 2.7 % at 200 hPa and by 1.6 % at 100 hPa, and nitrate in the UTLS also shows small sensitivity to the change in SO 2 emissions.
With anthropogenic emissions of NO x in Asia reduced by 50 %, the NMB of simulated surface-layer NO − 3 concentra- tions changes from +51.5 % in the standard simulation to −11.7 % in this sensitivity run.The contribution of each of SO 2− 4 , NO − 3 , and NH + 4 aerosols to total aerosol mass in the UTLS is not sensitive to this reduction in NO x emissions at the surface; the percentage contribution obtained from this sensitivity run is very close to the value obtained in the standard simulation (Table 4).Similarly, in the sensitivity study with NH 3 emissions reduced by 50 % in Asia, simulated surface-layer concentrations of NO − 3 and NH + 4 are improved in terms of the values of NMBs, but the improvement in simulated aerosol concentrations at the surface layer does not in-fluence our conclusion of high nitrate aerosol concentration in the UTLS.
As shown in Table 4, for the surface layer, simulated nitrate concentration over the TP/SASM region decreases by 46.8 % (from 0.94 to 0.50 µg m −3 ) with a 50 % reduction in anthropogenic NO x emissions in Asia, and it decreases by 22.3 % (from 0.94 to 0.73 µg m −3 ) when anthropogenic NH 3 emissions are reduced by the same percentage, indicating that surface-layer nitrate aerosol is more sensitive to anthropogenic emissions of NO x than to those of NH 3 .Relative to the baseline simulation, simulated nitrate concentrations at 200 and 100 hPa decrease, respectively, by 49.0 % (from 7.57 × 10 −2 to 3.86 × 10 −2 µg m −3 ) and 17.7 % (from 6.90 × 10 −2 to 5.68 × 10 −2 µg m −3 ) with a 50 % reduction in NH 3 emissions, whereas only by 2.1 % (from 7.57 × 10 −2 to 7.41 × 10 −2 µg m −3 ) and 1.3 % (from 6.90 × 10 −2 to 6.81 × 10 −2 µg m −3 ) with a 50 % reduction in NO x emissions.Over the studied region, the role of NH 3 in the sulfate-nitrate-ammonium aerosol system can be quantified by the gas ratio of GR = free ammonia total nitrate = TA−2×TS TN (Ansari and Pandis, 1998), where TA = NH 3 + NH + 4 , TS = SO 2− 4 , and TN = HNO 3 + NO − 3 .Over the TP/SASM region, the GR is generally positive both at the surface and in the UTLS, especially over 20-40 • N where deep convection exits (Fig. 11), indicating that free ammonia is available to react with nitrate (Seinfeld and Pandis, 2006).However, the GR is generally less than 1.0 above 400 hPa in summer over the TP/SASM region, which indicates that nitrate concentrations are most sensitive to changes in NH 3 and explains the small sensitivity of nitrate aerosol to NO x emissions in the UTLS.
In the sensitivity study with emissions of NO x , NH 3 , and SO 2 in Asia changed simultaneously, simulated surfacelayer concentrations of SO    of −8.3, −27.0, and +55.4 %, respectively, which are all improved compared to those in the standard simulation.Even though nitrate aerosol is now underestimated at the surface, it still accounts for 53.3 % of the PM 2.5 concentration at 100 hPa over the TP/SASM region in summer.
It should be noted that the concentrations of OC and BC are also underestimated, with NMBs of −57.2 and −32.2 %, respectively, in summer (Sect.4.2).We have done a simple calculation with the concentrations of OC and BC in the UTLS multiplied by 2.3 and 1.5, respectively, and nitrate is still the most dominant aerosol species in summertime in the UTLS over the TP/SASM region (not shown in Table 4).
Therefore the uncertainties in surface aerosol concentrations do not compromise the conclusion of this study.

Conclusions
In this work we simulate nitrate aerosol and its contribution to aerosol concentrations in the UTLS over the TP/SASM region (70-105 • E, 10-40 • N) for summertime of year 2005, using the global chemical transport model GEOS-Chem driven by the assimilated meteorological fields.
Simulated HNO 3 and O 3 are evaluated to show the model's ability to simulate the NO x -O 3 -HNO 3 cycle over the studied region.In the UTLS, both the horizontal and vertical distributions of simulated HNO 3 and O 3 agree well with the MLS observations.At 100 hPa, simulated seasonal mean HNO 3 and O 3 mixing ratios show NMBs of +11.Averaged over the TP/SASM region, the surface-layer concentrations of SO 2− 4 , NO − 3 , NH + 4 , BC, and OC are simulated to be 1.70, 0.94, 0.85, 0.30, and 0.94 µg m −3 , respectively.Nitrate aerosol is simulated to be of secondary importance near the surface over the region of our interest.Comparisons of simulated aerosol concentrations with groundbased observations show that simulated summertime concentrations of SO 2− 4 , NO − 3 , NH + 4 , OC, and BC have NMBs of −14.7, +51.5, +74.9, −57.2, and −32.2 %, respectively.

FigureFigure 3 .
Figure3ashows the simulated global distribution of HNO 3 concentrations averaged over June-August of 2005.Concen-

Figure 4 .
Figure 4. Comparisons of simulated HNO 3 concentrations (pptv) with observations (pptv) from MLS. Panels (a) and (b) show simulated concentrations at 200 and 100 hPa, respectively.Panel (c) shows the latitude-altitude cross section of simulated HNO 3 concentrations averaged over 70-105 • E. Panels (d)-(f) are the same as (a)-(c), except that (d)-(f) show observations from MLS.The GEOS-Chem HNO 3 simulations are smoothed by the corresponding averaging kernels (AKs).White areas indicate lack of data meeting the retrieval quality criteria.All the data are averaged over June-August of 2005.

FigureFigure 5 .
Figure 3b shows the global distribution of simulated summertime surface-layer O 3 concentrations.Simulated O 3 concentrations are in a range of 40-70 ppbv over Europe, North

Figure 6 .
Figure 6.The simulated and observed vertical profiles of monthly mean O 3 mixing ratios at (a) Kunming and (b) Lhasa in August.The model results are from the simulation of year 2005.The observations in Kunming were conducted during 7-13 August (11 profiles of O 3 collected) in 2009 and during 12-31 August in 2012 (daily observations).The observations in Lhasa were conducted during 22-28 August in 2010 (12 profiles of O 3 collected) and during 4-26 August in 2013 (daily observations).

Figure 7 .
Figure 7. Simulated seasonal mean concentrations (µg m −3 ) of sulfate, nitrate, ammonium, organic carbon, black carbon, and PM 2.5 (defined as the sum of sulfate, nitrate, ammonium, BC, and OC) at (a) the surface layer, (b) 200 hPa, and (c) 100 hPa, during summer (June-August) of year 2005.Note that color bars are different for concentrations at the surface, 200, and 100 hPa.

Figure 8 .×
Figure 8.(a) Locations with measured aerosol concentrations from previous studies.Surface winds during summertime are also shown.Panels (b)-(f) show the comparisons of simulated seasonal mean concentrations of sulfate, nitrate, ammonium, OC, and BC with measured values, respectively.In (b)-(f) the 1 : 1 line (dashed), linear fit (solid line and equation), correlation coefficient between simulated and

Figure 9 .
Figure 9. (a) Monthly mean distribution of aerosol extinction coefficients (km −1 ) at 100 hPa for July of 2005.(b) Monthly mean vertical distributions of aerosol extinction coefficients (at 525 nm for SAGE II and 550 nm for GEOS-Chem, with the contributions of both anthropogenic and natural aerosols considered) (km −1 ) averaged over the Asian monsoon anticyclone region (20-120 • E, 10-40 • N) for July of 2005.The horizontal dashed line represents the tropopause averaged over the Asian monsoon anticyclone region simulated by the GEOS-Chem model.

Figure 11 .Figure 12 .
Figure 11.Latitude-altitude cross sections of simulated concentrations (color shades, µg m −3 ) of SO 2− 4 and NO − 3 averaged over 70-105 • E in June-August of 2005, together with the wind vectors obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-Interim Reanalysis data.The black line denotes the tropopause simulated by the GEOS-Chem model.
, and BC, and their contributions to PM 2.5 (in percentages) during summertime (June-August) of 2005.The mass concentrations are averaged over the TP/SASM region, with a unit of µg m −3 at the surface layer and of 10 −2 µg m −3 at 200 and 100 hPa.The NMBs are also shown, as the simulated surface-layer concentrations are compared with measurements described in Sect. 4

Figure 14 .
Figure 13.(a, b) Distributions of RH (%) and temperature (K) at 100 hPa.(c, d) The latitude-altitude cross sections of RH (%) and temperature (K) averaged over 70-105 • E. RH and temperature are from the GEOS5 assimilated meteorological fields, and all the values are the averages over June-August of year 2005.
1 and +3.5 %, respectively, over the TP/SASM region (70-105 • E, 10-40 • N) in summer of year 2005, and the model biases lie within the confidence range of the MLS instruments.Both simulated and observed O 3 concentrations show relatively low values of less than 200 ppbv at 100 hPa over the TP/SASM region.

Table 2 .
Simulated seasonal mean concentrations of aerosols and their contributions to PM 2.5 (in percentages in parentheses) during summertime (June-August) of 2005 for the TP/SASM, TP, and SASM regions.The unit is µg m −3 for concentrations at the surface, and 10 −2 µg m −3 for concentrations at 200 and 100 hPa.