Tropospheric distribution of sulphate aerosols mass and number concentration during INDOEX-IFP and its transport over the Indian Ocean: A GCM study

. The sulphate aerosols mass and number concentration during the Indian Ocean Experiment (INDOEX) Intensive Field Phase-1999 (INDOEX-IFP) has been simulated using an interactive chemistry GCM. The model considers an interactive scheme for feedback from chemistry to meteorology with internally resolving microphysical properties of aerosols. In particular, the interactive scheme has the ability to predict both particle mass and number concentration for the Aitken and accumulation modes as prognostic variables. On the basis of size distribution retrieved from the observations made along the cruise route during IFP-1999, the model successfully simulates the order of magnitude of aerosol number concentration. The results show the southward migration of minimum concentrations, which follows ITCZ (Inter Tropical Convergence Zone) migration. Sulphate surface concentration during INDOEX-IFP at Kaashidhoo (73.46 ◦ E, 4.96 ◦ N) gives an agreement within a factor of 2 to 3. The measured aerosol optical depth (AOD) from all aerosol species at KCO was 0 . 37 ± 0 . 11 while the model simulated sulphate AOD ranged from 0.05 to 0.11. As sulphate constitutes 29 % of the observed AOD, the model predicted values of sulphate AOD are hence fairly close to the measured values. The model thus has capability to predict the vertically integrated column sulphate burden. Furthermore, the model results indicate that Indian contribution to the estimated sulphate burden over India is more than 60 % with values upto 40 % over the Arabian Sea.


Introduction
The aerosol effects on climate can be large and complex due to the fact that aerosols chemical composition, abundance and size distribution are highly variable, both spatially and temporally. Most of the earliest investigations on direct aerosol forcing have focused on sulphate aerosols because of their importance as an anthropogenic aerosol component (Charlson et al., 1992;Kiehl and Briegleb, 1993). Sulphate particle is an important component of atmospheric aerosols. Sulphate make up a substantial fraction of tropospheric aerosol in both urban and remote/rural areas (Barth and Church, 1999;Ramanathan et al., 2001;Mallet et al., 2004). Sulphate aerosols are capable of modifying the climate not only by scattering incoming sunlight back to space (direct effect) but also by altering the properties of clouds (indirect effect). In order to further understand sulphate aerosolclimate interactions, the present study is carried out with the Laboratoire de Météorologie Dynamique interactive chemistry global model (Verma et al., 2007) to produce the simulation of INDOEX-IFP. The three fundamental objectives of this study are 1. to simulate sulphate aerosol dynamics as prognostic variables, 2. to compare and evaluate model-derived physical, optical, and radiative properties of sulphate aerosols with observations during INDOEX-IFP, 3. to examine sulphate aerosol transport and radiative forcing, and the contribution of Indian emissions to the regional burden using the emission inventory developed by Reddy and Venkataraman (2002a, b).
INDOEX was conducted during the winter monsoon season, when pristine air-mass from the Southern Hemisphere and not-so-clean air from the Indian subcontinent meet over the tropical Indian Ocean and provide a natural laboratory for studying aerosol influences. Primarily, its goal was to assess the natural and anthropogenic climate forcing due to aerosols and to examine the underlying feedback mechanisms on scales relevant to regional and global climate. Ramanathan et al. (2001) have presented the INDOEX results for the oceanic regions adjacent to the Indian subcontinent, which reveal the presence of a dense anthropogenic aerosol layer over China, South and Southeast Asia. While INDOEX suggests south/southeast Asia as the regions of major sources of aerosols for the Indian Ocean, the controversy still persists regarding the formation, movement, persistence and effects of this regional haze on the regional and global climate, e.g. the South Asian Atmospheric Brown Cloud (UNEP and C 4 , 2002). In order to address such high priority issues of concern to climate, this study, in logical continuation of earlier studies (Krishnamurti et al. 1998;Rajeev et al., 2000;Satheesh and Ramanathan, 2000;Verver et al., 2001;Rasch et al. 2001;Ramanathan et al. 2001;Andronache et al. 2002;Lawrence and Lelieveld 2010 and references therein), deals with the INDOEX region. The main motivation of this study is to reproduce the observed variations in the sulphate aerosol mass and number concentration during the Indian Ocean Experiment from January to March 1999 using LMD interactive chemisry GCM. The present model uses a comprehensive numerical scheme to deal with the gas-phase and aqueous-phase reactions in a GCM, where the concentrations of radicals like OH, HO 2 and gases like O 3 , H 2 O 2 , NH 3 and NO x are computed within the zoom version of LMD-ZT GCM. The sulphate aerosol/chemistry is interactively included in the model. The model considers both sulphur mass and number concentrations as a truly interactive part and not merely a passive component in atmospheric chemistry (Verma et al., 2007). The knowledge of the sulphate aerosol number concentration is important to understand the indirect radiative forcing of the aerosols and to refine estimates of the direct radiative forcing.
The model design therefore differs from the earlier studies of Lawrence et al. (1999), Rasch et al. (2000), Boucher et al. (2002), Hauglustaine et al. (2004), Vignati et al. (2004) and Stier et al. (2005) which employ analyzed meteorology/observations or prescribed three-dimensional short lived radical fields to drive the CTM. The experiments are de-signed to assess impacts of sulphate aerosols on INDOEX region.

The atmospheric global model
The LMDZ is a development of previous LMD model described by Sadourny and Laval (1984), which was utilised along with other global models of major meteorological centres, to achieve the AMIP I objective of atmospheric model intercomparison (Boer, 1992;Cess et al., 1989) and also to carry out long-term simulations for certain climate and aerosol studies (Le Treut et al., 1994;Boucher and Lohmann, 1995;Sharma et al., 1998;Boucher et al., 1998;Boucher et al., 2002 etc). The LMDZ has the capability to zoom with a finer resolution over a specific region of interest over the globe and relatively a coarser resolution over rest of the globe, which gives a unique opportunity to study the regional aspects of aerosols. The model solves the primitive equations of meteorology and mass continuity for sixteen tracers. The discrete analogues of the dynamical equations are obtained using finite difference formulations on Arakawa-C grid and the physical package of the model includes a set of parameterisations for convection, radiative transfer, planetary boundary layer (PBL), gravity wave drag and ground hydrology with no vegetation. For a complete description of the atmospheric model dynamics and physical parameterisation of physical processes one may refer to Sadourny and Laval (1984), Le Treut et al. (1994) and Hourdin and Armengaud (1999). The climatology of this model has been presented by Li (1999) and Lott (1999). An important addition to LMDZ in the version used here is the inclusion of a prognostic sulphur-cycle scheme (Verma et al., 2007). In particular, the model considers an interactive scheme for feedback from chemistry to meteorology with internally resolving microphysical properties of aerosols for radiative transfer computations. Thus, it is a step towards correcting the deficiencies where the microphysics and feedbacks are run offline. Except for some parameterisations, it should be noted that the chemistry module has been developed independently from that of Boucher et al. (2002) and Hauglustaine et al. (2004).

Chemistry and aerosol modules
The chemistry module includes natural and anthropogenic emissions, treatment of sulphur and nitrogen sources, their chemical transformations and deposition processes (Verma et al., 2007). The basic chemistry module consists of over 50 gas/aqueous phase chemical reactions (33 gas phase and 18 aqueous phase) along with the sources and sinks of chemical species. The prognostic chemical species in the LMDZ model are water vapour, liquid water, dimethylsulphide (DMS), hydrogensulphide (H 2 S), dimethylsulphoxide  (Schwartz, 1986). The model does not explicitly take into account mass transfer rates between the gas and the aqueous phase in view of an examination on mass transfer effect by Schwartz (1986) and Venkataraman et al. (2001) who have shown that the mass transfer effect does not limit the rate of sulphate formation. The present approach utilises sulphuric acid produced in the gas phase as an input to the two-mode, two-moment aerosol module (Binkowski and Shankar, 1995) which calculates aerosol mass, number concentration and particle surface area for the Aitken and accumulation modes. The vapour sulphuric acid produced by gas phase oxidation is split into two parts: one going into Aitken and the other one condensing onto the existing particles (accumulation mode). The sulphate produced by aqueous phase oxidation remains dissolved in cloud droplets. However, when the clouds disappear, the sulphate in the dissolved state is added to the accumulation mode mass. Therefore all new sulphate mass produced by aqueous production is added to the accumulation mode but the number of accumulation mode particles remains unchanged (Fig. 1). The model represents a simplified version of the aqueous model of Walcek and Taylor (1986) and is similar to an earlier equilibrium model of Ohta et al. (1981). The equations for chemical equilibrium of the SO 2 -NH 3 -CO 2 -HNO 3 -H 2 O system are those from Chen and Crutzen (1994).
The mass fluxes, simulated by the Tiedtke (1989) scheme, are used to parameterise convective transport of gases and aerosols. Vertical transport of trace species is incorporated in updrafts and downdrafts, accounting for entrainment and detrainment. All chemical species along with aerosol mass and number concentration in the two modes are advected in the model. These species evolve consistently with dynamical, physical, cloud and chemical processes as the model state integrates forward with each time step. Convective transport is performed after wet scavenging calculation, in order to avoid upward transport of material that is scavenged by precipitation.
Dry deposition is parameterised through deposition velocities, which are prescribed for each chemical species and surface types. The model uses a simple parameterisation scheme, which assumes that the rate of deposition at the surface is directly proportional to the mixing ratio (i.e. concentration) of the respective species in the lowest model layer.
To account for the effect of clouds in the grid box, the clear sky photolytic rates are multiplied by a correction factor. The wet removal scheme considers both in-cloud and belowcloud scavenging in the model. Rainfall is computed at every time step in the GCM. Removal of gases and aerosols by rain was parameterised using the model generated precipitation formation rate following Giorgi and Chameides (1986).
The CO 2 concentration is held constant (370 ppm) throughout the model integration. Mie theory is employed to compute the optical properties with prescribed size distribution and refractive indices. Optical properties are computed over the entire shortwave spectrum (0.25-4.0 µm) at 24 wavelengths and grouped into the two model wavebands as weighted averages with a typical spectral distribution of the incoming solar radiation flux at the surface (Verma et al., 2006). The physical, cloud and chemical processes considered in the present model are described in details in (Verma et al., 2007). The global sulphur emissions in GEIA database include emissions from fossil fuels combustion and industrial processes. The biomass burning sulphur emissions are from Pham et al. (1995). A fixed 5 % of sulphur from combustion sources is assumed to be emitted directly as sulfuric acid gas and condenses immediately on ambient aerosol particles. The annual anthropogenic SO 2 emissions (manmade) from fossil fuel combustion and industrial activities are 66.31 Tg S yr −1 globally with a contribution from India of 1.44 Tg S yr −1 which increases to 2.38 Tg S yr −1 if RV inventory is used (Table 1). The simulated SO 2 emissions in the RV and GEIA inventories averaged for the INDOEX-IFP period are shown in Fig. 2. The RV emission inventory is constructed with very high source resolution. The sulphur emissions are estimated using plant specific fuel combustion (Reddy and Venkataraman, 2002b). For biomass burning, the emission inventory for India is constructed from spatially resolved data sets, which results in 0.33 Tg S yr −1 as projection of SO 2 emission for the INDOEX period (1998)(1999).  (Krishnamurti et al., 1998).

Numerical simulations
Here we compare the simulated sulphate concentration with corresponding measured concentration at KCO during INDOEX-IFP (Fig. 3). Modelled surface sulphate concentrations agree with the measured values within a factor of 2-3. The inventory developed by Reddy and Venkataraman (RV) has fine resolution (0.25 × 0.25) with specific emission factors for Indian regions. The identification of large point sources in the RV inventory is an advantage over the global inventories, where the emissions are estimated at country level. The aerosol concentrations downwind of Indian subcontinent are sensitive to the geographical location of the emission sources over the sub-continent. Model simulation shows an increase in concentration in the month of February with RV inventory. This result is in general agreement with the meteorological conditions at KCO as most of the air mass during February originated from India in contrast to March when the air mass originated from East Asia (Verver et al., 2001) flows towards it. The resolution of RV inventory thus gives a better agreement of sulphate concentrations with measurements at KCO.
In contrast to surface concentrations, aerosol optical depth (AOD) is a measure of the entire aerosol column. The measured AOD from all aerosol species averaged over IFP at KCO was 0.37 ± 0.11. Model simulated sulphate AOD at KCO ranged from 0.05 to 0.11 during the IFP period. The KCO aerosol measurements reveal that sulphate is responsible for 29 % Satheesh et al., 1999)  this fact into account, the model simulates the observed sulphate aerosol optical depth at KCO (Fig. 4) well. Further, to see the agreement of model simulated trace gases concentration to observations, we plot in Fig. 5 (upper panel) the CO surface concentration. The simulated CO concentration remarkably agrees with observations between day-62 and day-82 while the model underestimates the observations by a factor of 2 between day-42 and day-62. Next, we plot simulated latitudinal ozone mixing ratios from 0-3 km altitude range. It is evident from Fig. 5 (lower panel) that model is able to simulate the general north-south gradient in the ozone concentration in lower troposphere . The model simulated ozone concentrations of 8 to 31 ppbv agree well with the observed ozone profile (5 to 25 ppbv) obtained after averaging in the region lying between 8 • S to 8 • N.    Kamra et al. (2003) and are decreased by 30 % for the sake of comparing with sulphate only modelled number concentrations. the continent consists of sulphate aerosols and for this reason we report 70 % of their concentrations for the sake of comparing with the sulphate only modelled concentrations. In general, a strong northeasterly flow has been observed over the Indian continent in February and March 1999 leading to a greater transport of atmospheric pollutants over oceanic regions downwind of source regions (Verver et al., 2001;Rasch et al., 2001). The simulated wind pattern in Fig. 13 depicts the above fact very well with a strong northeasterly flow and anti-cyclonic motion near the oceanic regions resulting in higher number concentrations in northern cruise track. The model simulates the order of magnitude and the general north-south gradient feature in aerosol number concentrations very well. The feature is thus in general agreement with the transport pattern prevalent over the region (Fig. 13).

Latitudinal variations
The latitudinal variation of sulphate number concentration and AOD in the Arabian Sea and Bay of Bengal for each month is discussed in this section. The latitudinal variations are obtained by separately averaging the values of a variable in the longitude ranges 60-80 • E (Arabian Sea -Indian Ocean sector) and 80-100 • E (Bay of Bengal -Indian Ocean sector).  (Fig. 7a) show a peak around 10-15 • N region with a value of 8 × 10 3 particles cm −3 while in the Bay of Bengal sector the peak occurs around 25-30 • N region with a maximum of 6.5 × 10 3 particles cm −3 , which decreases to the order of about 3 × 10 3 particles cm −3 upon reaching 5 • S. This result is consistent with the finding of Lelieveld et al. (2001) who also reported a drastic decrease in sulphate aerosols on crossing the ITCZ from north to south. The latitudinal variations of the accumulation mode concentration (Fig. 7b) show a larger magnitude over the Arabian Sea sector than the Bay of Bengal sector with elevated values in the interval 10-25 • N. This indicates a strong influence of continental aerosols in the marine boundary layer over the Arabian Sea. The model predicts large aerosol concentrations ranging between 10 2 and 10 3 particles cm −3 over the North Indian Ocean during IFP and the contribution of Aitken mode concentration along north of ITCZ is significantly higher than that of the accumulation mode. These results generally agree with those of Lelieveld et al. (2001) and Kamra et al. (2003) who showed that the large number concentration during the IFP mostly consists of particles in the Aitken mode. The simulated aerosol number concentration very well illustrates that the average aerosol concentration over the Northern Hemisphere is much larger than the southern Indian ocean (Kamra et al., 2003).

Sulphate aerosol number concentration
An observation on the number concentration and mass in the two modes is interesting. For example, the number concentration in the Aitken mode over the Arabian sea reaches a peak value of 8 × 10 3 particles cm −3 with the corresponding mass of 0.6 µg m −3 . On the contrary the number Atmos. Chem. Phys., 12, 6185-6196, 2012 www.atmos-chem-phys.net/12/6185/2012/ concentration in the accumulation mode reaches a maximum value of 1800 particles cm −3 and the corresponding mass there is about 4.6 µg m −3 . This result is consistent with other studies, which have mentioned that most of the sulphate mass resides in the accumulation mode (Binkowski and Roselle, 2003). Figure 8 shows the latitudinal variations of aerosol optical depth during January, February and March 1999. The AOD is significantly larger in the Northern Hemisphere compared to the Southern Hemisphere during all three months, with maximum AOD near the continents. The latitudinal variation of sulphate AOD over the Arabian Sea sector (Fig. 8a) with Reddy and Venkataraman (RV) inventory shows a highest value (0.12) at around 25 • N. A secondary peak is also observed at 10 • N during the months of February and March. While in GEIA (Fig. 8c), latitudinal sulphate aerosol AOD is larger over (10-15 • N), minimum in ITCZ and noticeably rises over the southern Indian Ocean (0-5 • S) which then decreases linearly further southwards. The highest AOD is simulated in the month of February with a value of 0.08 close to Arabia and the west coast of India. The plume extends from 15 • N to 5 • S in the month of January and March with highest values of 0.06 and 0.08 respectively. Over the Bay of Bengal sector (Fig. 8b) AOD is generally comparable to that over the Arabian Sea with superior values during January. For all three months, a plume lies in the region 20-25 • N while another plume is also visible at 0-5 • S region. The general features revealed by Fig. 8 are as follows: (i) the results show a southward migration of the AOD minimum in consonance with the migration of the ITCZ; (ii) there is a visible rise in sulphate AOD in the Arabian Sea sector over 0-10 • S region with the Indian inventory (RV) suggesting the role of Indian emissions in these concentrations; (iii) the AOD in the Northern Hemisphere is larger than that of Southern Hemisphere by about 30-50 %. These patterns derived from model-simulated values are mostly confirmed by satellite observed patterns especially the maxima over 10-25 • N regions (Chowdhury et al., 2001;Quinn et al., 2002;Reddy et al., 2004).

Regional distribution for INDOEX-IFP
The equatorial Indian Ocean region provides a unique opportunity to observe anthropogenic sulphate effects. From January to April, the predominant circulation in this region consists of a low-level flow from the northeast (Krishnamurti et al., 1998). This NE monsoon (Asian winter monsoon) should facilitate the formation and transport of new sulphate particles to oceanic regions far away from urban centres. The above feature is very crucial as atmospheric circulation and aerosol lifetimes are the dominant factors controlling both   spatial distribution and transport of aerosols that could be evaluated through a chemistry-transport model.

SO 2 concentration
The model simulated spatial distribution of SO 2 concentration over surface and column burden are shown (Fig. 9) over the INDOEX domain averaged for January, February and March 1999. The largest values at surface are found over southeastern China, western and eastern parts of India (8-10 µg m −3 ). Concentrations of SO 2 over ocean are low indicating the short chemical lifetime.

Sulphate mass and number concentrations
The average surface concentrations (Fig. 10a)   advection of sulphate from continents to ocean. The sulphate column burden (Fig. 10b) show a high over south and east India (8-10 mg SO 2− 4 m −2 ). The average vertical profile of sulphate aerosol loading (Fig. 11) shows a constant concentration up to a level of about 900 hPa and then a substantial decrease by at least a factor of 2 at about 3 km (700 hPa). Notably, the observations reported by Rasch et al. (2001) also support the simulated average sulphate concentration for the entire INDOEX region during IFP and are consistent with earlier modelling studies. The Aitken mode average concentration (4×10 3 -6×10 3 particles cm −3 ) over India (Fig. 12) is more than that of the accumulation mode (1×10 3 -2×10 3 particles cm −3 ) with a maximum over Atmos. Chem. Phys., 12, 6185-6196, 2012 www.atmos-chem-phys.net/12/6185/2012/ western India (3 × 10 3 -4 × 10 3 particles cm −3 ) extending to the Arabian Sea region in diminishing magnitudes. Since the precursor gases for the formation of these sub-micron particles come mostly from the continents, the model reveals SE China and Indian region as important contributors to anthropogenic sulphate aerosol production. The observed decrease in the concentration of particles in the accumulation mode from north to south is due to the removal of these particles by dry and wet deposition (Kamra et al., 2003). During winter, the near-surface flow is mostly northeasterly, and the regional pollution forms a thick haze layer in the lower troposphere which spreads out over millions of square km between southern Asia and the Intertropical Convergence Zone (ITCZ), located several degrees south of the equator over the Indian Ocean during this period (Lawrence and Lelieveld, 2010). Wind vectors at the surface (Fig. 13a) also show strong outflows from north and east India into the Arabian Sea and Northern Indian Ocean. These strong flows explain the plausible cause of advection of SO 2 and sulphate from continents and contribute to sulphate column burden over oceanic regions.

Sulphate AOD and its radiative properties
In Fig. 14a, the geographical distribution of the sulphate aerosol optical depth during the INDOEX period is presented. The largest values in AOD lie over SE China where the anthropogenic source of sulphur is very intense. An area of large AOD (0.1-0.14) and radiative forcing at top of atmosphere (−1.25 to −2.0 W m −2 ) is observed (Fig. 14b) from this region covering India: AOD over northern India (0.04-0.1) extends deep into the Arabian Sea and the Indian Ocean region (down to 10 • S). Higher AOD values are also noted over southeastern part of India with a maximum reaching 0.12. From coast to the central Bay of Bengal lower values of AOD occur where aerosol transport, both from India and South-East Asia, is insignificant and reflects the impact of anticyclonic circulation near South India , which does not allow the offshore flow from eastern India  to penetrate deep into the Bay of Bengal. Smaller magnitudes of radiative forcing (< 0.75 W m −2 ) and AOD (< 0.04) are present southwards to 5 • S, near the ITCZ (Quinn et al., 2002). In order to assess the role of Indian emissions in increased pollution levels, we emphasize and show the differences obtained in the radiative forcing and AOD (Fig. 15) by the two inventories. Our result indicates that Indian emission does contribute significantly to the AOD and radiative forcing over the Arabian Sea while there is meagre to very small contribution over the Bay of Bengal.

Indian source contribution to sulphate burden
The Indian emissions are carried deep down into the Arabian Sea sector and to the coastal Bay of Bengal region. Thus the air from India contributes more strongly to Arabian Sea and to the near coastal region of Bay of Bengal. The large aerosol plume off the southwest coast is particularly noticeable (Fig. 16). There is a visible Asian plume entering at about 15-25 • N, 90 • E, contributing significantly to sulphate burden. Atmospheric circulation and trajectory analysis for INDOEX region has shown large southwestward movement of air from peninsular India at lower tropospheric levels. As a consequence, a large amount of sulphate off the west coast of India has been transported over to oceanic regions. Over the Bay of Bengal region however, there is considerable movement of air from southeast Asia due to prevailing easterly and northeasterly winds (Krishnamurti et al., 1998;Rajeev et al., 2000;Rasch et al., 2001).

Conclusions
A fully interactive sulphur chemistry module has been used to simulate sulphate mass as well as number distribution as prognostic variables. The model simulates the sulphate AOD, mass and number concentration to acceptable levels for the entire INDOEX region during IFP. Modelled surface and vertical concentration fields show encouraging agreement in spatial patterns and temporal episodicity with comparable magnitudes of observed quantities. At Kaashidhoo, the model simulated sulphate values agree with observations within a factor of 2-3. The measured aerosol optical depth (AOD) from all aerosol species (of which sulphate constitutes 29 %) at KCO was 0.37 ± 0.1, while the model predicted sulphate AOD is in the range of 0.05 to 0.11. Hence, the predicted values of sulphate AOD are fairly close to measured values, indicating the capability of the model to predict the vertically integrated column sulphate burden. In general, the plume of sulphate mass, number concentration and aerosol optical depth in the Northern Hemisphere is larger than that of Southern Hemisphere by ∼ 30-50 %, which fol-lows the migration of ITCZ with a minimum centred around this region.
In conclusion, Indian emissions are not the sole cause for the higher sulphate concentrations and AOD over the Indian Ocean. The Indian contribution is more than 60 % over India and 30-50 % over the Arabian Sea. The aerosols transported from the East Asia contribute significantly to sulphate burden over the Bay of Bengal. Transport of aerosol by northwesterly winds from Arabia and northeasterly winds from the Indian subcontinent are the major sources of higher concentrations over the Arabian Sea; whereas over the Bay of Bengal, the transport is mainly from Southeast Asia.