2.1 Satellite and ground-based measurements of AOD

We analyze aerosol retrievals from Multi-Angle Imaging Spectroradiometer (MISR) (level-3 version 4, at 550 nm wavelength during 2000–2016) (Martonchik et al., 2002), The MISR AOD measurements give aerosol properties over the global ocean and land with bright targets such as deserts (Kahn et al., 2001). Aerosol Robotic NETwork (AERONET) sun photometer, level 2.0 version 3 daily AOD observations during 2006–2016 (Holben et al., 1998) were also analyzed at the stations in the Indo-Gangetic Plain, (Bihar: 25.87^∘ N, 84.12^∘ E; Jaipur: 26.90^∘ N, 75.80^∘ E; Kanpur: 80.23^∘ E, 26.51^∘ N; Karachi: 67.13^∘ E, 24.95^∘ N), and China (Xianghe: 39.76^∘ N, 110.00^∘ E; Nghia Do: 21.04^∘ N, 105.80^∘ E).

2.2 SO₂ measurements from the Ozone Monitoring Instrument

The Ozone Monitoring Instrument (OMI) aboard the NASA Aura spacecraft retrieves SO₂ data from Earthshine radiances in the wavelength range of 310.5–340 nm (Levelt et al., 2006). It gives the total number of SO₂ molecules in the entire atmospheric column above a unit area (https://disc.gsfc.nasa.gov/datasets/OMSO2e_V003/, last access: 5 August 2019). Details of the retrieval technique are documented by Li et al. (2017). To understand the impact of SO₂ emission changes over India and China, we estimate a trend in the SO₂ (2007–2017) over the Indian region (8–35^∘ N, 70–95^∘ E) and the Chinese region (20–45^∘ N, 95–130^∘ E) (see Fig. 2e). For this purpose, we used version 1.3, level-2, OMI retrievals that assume all SO₂ is located in the planetary boundary layer. We use a regression model described by Fadnavis and Beig (2006). A model regression equation is given as follows:

where θ(t,z) is the daily mean number of SO₂ molecules averaged over the Indian or Chinese region, with altitude z set to 1 km, as we use column data. The model uses the harmonic expansion to calculate the seasonal coefficient, α, and the trend coefficient, β. The harmonic expansion for α(t) is given as

where $\mathit{\omega }=\mathrm{2}\mathit{\pi }/\mathrm{12}$; A₀, A₁, A₂, …. are constants and t (t=1, 2, …n) is the time index. The estimated trend value for SO₂ is 4.8±3.2 % yr⁻¹ over the Indian region and 7.0±6.3 % yr⁻¹ over the Chinese region (99 % confidence interval). These trend values are used while designing the model sensitivity simulations (discussed in Sect. 2.4).

2.3 CloudSat and Cloud-Aerosol Lidar Infrared Pathfinder Satellite Observations

We use the ice water content (IWC) dataset from a combination of CALIPSO lidar and CloudSat radar data (2C-ICE dataset, version L3_V01) for the period 2007–2010 (Deng et al., 2013). The Cloud Profiling Radar (CPR) onboard the CloudSat satellite is a 94 GHz nadir-looking radar which measures the power backscattered by clouds as a function of distance. It provides information on cloud abundance, distribution, structure and radiative properties. CALIOP is an elastically backscattered active polarization-sensitive lidar instrument onboard CALIPSO. CALIOP transmits laser light simultaneously at 532 and 1064 nm at a pulse repetition rate of 20.16 Hz. The lidar receiver subsystem measures backscatter intensity at 1064 nm and two orthogonally polarized components of 532 nm backscatter signal that provide the information on the vertical distribution of aerosols and clouds, cloud particle phase, and the classification of aerosol size (Winker et al., 2010). The details of the data retrieval method are explained in Li et al. (2012).

2.4 The model simulations

The ECHAM6–HAMMOZ aerosol–chemistry–climate model used in the present study comprises the ECHAM6 global climate model coupled to the two-moment aerosol and cloud microphysics module HAM (Stier et al., 2005; Tegen et al., 2019) and the sub-model for trace gas chemistry MOZ (Kinnison et al., 2007). HAM predicts the nucleation, growth, evolution and sinks of sulfate (${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$), black carbon (BC), particulate organic matter (POM), sea salt (SS) and mineral dust (DU) aerosols. The size distribution of the aerosol population is described by seven log-normal modes with prescribed variance as in the M7 aerosol module (Stier et al., 2005; K. Zhang et al., 2012). Moreover, HAM explicitly simulates the impact of aerosol species on cloud droplet and ice crystal formation. Aerosol particles can act as cloud condensation nuclei or ice-nucleating particles. Other relevant cloud microphysical processes such as evaporation of cloud droplets, sublimation of ice crystals, ice crystal sedimentation and detrainment of ice crystals from convective cloud tops are simulated interactively (Lohmann and Ferrachat, 2010; Neubauer et al., 2014). The anthropogenic and fire emissions of sulfate, black carbon (BC) and organic carbon (OC) are based on the AEROCOM-ACCMIP-II emission inventory for the study period 2010–2011 (Textor et al., 2006). The MOZ sub-model describes the trace gas chemistry from the troposphere up to the lower thermosphere. The species included within the chemical mechanism are contained in the O_X, NO_X, HO_X, ClO_X and BrO_X chemical families, along with CH₄ and its degradation products. Several primary non-methane hydrocarbons (NMHCs) and related oxygenated organic compounds are also included. This mechanism contains 108 species, 71 photolytic processes, 218 gas-phase reactions and 18 heterogeneous reactions on aerosol (Kinnison et al., 2007). Details of anthropogenic, biomass burning, biogenic, emissions fossil fuel sources, etc., are reported by Fadnavis et al. (2017a).

The model simulations are performed at the T63 spectral resolution corresponding to $\mathrm{1.875}{}^{\circ }\times \mathrm{1.875}{}^{\circ }$ in the horizontal dimension, while the vertical resolution is described by 47 hybrid σ−p levels from the surface up to 0.01 hPa. The model has 12 vertical levels in the UTLS (50–300 hPa). The simulations have been carried out at a time step of 20 min. Atmospheric Model Intercomparison Project (AMIP) sea surface temperature (SST) and sea ice cover (SIC) (Taylor et al., 2000) were used as lower boundary conditions. We performed 10-member ensemble runs by varying the initial conditions (both SST and SIC) starting between 1 and 10 January 2010 and ending on 31 December 2011 to obtain statistically significant results. The analysis is performed for the year 2011. The 2011 Indian monsoon was well within the long-term norm, with no strong influences from the Indian Ocean Dipole or El Niño modes of interannual climatic variability. We refer to it as the control simulation (CTRL). In previous work, Fadnavis et al. (2013, 2017b) used the ensemble means from 6 to 10 members to analyze the variability of aerosols and associated impacts during the monsoon season. In two emission sensitivity simulations we applied (1) a flat 48 % increase in anthropogenic SO₂ emissions over India (referred to as Ind48 simulation) and (2) a flat 48 % increase in anthropogenic SO₂ emissions over India and a flat 70 % decrease in anthropogenic SO₂ emissions over China simultaneously (referred to as Ind48Chin70 simulation); the same assumptions were made for simulated years. The simulation design is based on the estimated trend of 4.8 % per year over India and −7.0 % over China, from OMI SO₂ observations during 2007–2017. The Ind48 and Ind48Chin70 simulations are also 10-member ensemble runs for the same period as CTRL and are analyzed for the year 2011 (see Table 1). We compare the CTRL and Ind48 and Ind48Chin70 simulations to understand the seasonal impact of enhanced sulfate aerosol on the UTLS, radiative balance and cirrus clouds. We should mention that our simulations are canonical in design in order to show the impact of Asian sulfate aerosols; they do not include many of the observed complexities, like radiative forcing due to non-sulfate aerosols (e.g., organics, nitrates and dust). The Quasi-biennial Oscillation (QBO) is not internally generated in the model. Notwithstanding this, the present work provides valuable insight into the relevance of the impact of sulfate aerosol originating from India and China on the UTLS.

Table 1Details of model simulations performed.

Download Print Version | Download XLSX

The seasons considered in this study are pre-monsoon (March–May), summer monsoon (June–September), post-monsoon (October–November) and winter (December–February).

2.5 Offline radiative calculations

We use offline radiative calculations to explore the radiative impacts of enhanced sulfate aerosol loadings in the UTLS only (300–50 hPa), compared to the all atmosphere enhancement. Radiative effects associated with the sulfate aerosol enhancement are calculated using the SOCRATES radiative transfer model (Edwards and Slingo, 1996; Rap et al., 2013) with the CLASSIC aerosol scheme (Bellouin et al., 2011). We used the offline version of the model with six shortwave and nine longwave bands and a delta-Eddington two-stream scattering solver at all wavelengths.

Figure 1Seasonal mean distribution (2007–2010) of cloud ice mass mixing ratio (mg kg⁻¹) from CloudSat and CALIPSO combined 2C-ICE L3 averaged for 20–40^∘ N for the (a) pre-monsoon, (b) summer-monsoon, (c) post-monsoon and (d) winter seasons; (e)–(h) same as (a)–(d) but from CTRL simulations.

Download

4.1 A layer of aerosol in the UTLS

The Asian region (8–45^∘ N, 70–130^∘ E) experiences convective instability throughout the year with a peak in the monsoon season (Manohar et al., 1999; Luo, 2013). The distribution of seasonal mean outgoing longwave radiation, simulated ice crystal number concentration and cloud droplet number concentrations representing convection is shown in Fig. S1 in the Supplement. It depicts convection over the Asian region rising to the upper troposphere throughout the year and is widespread during the monsoon season. The summer-monsoon convection lifts the boundary layer aerosols to the upper troposphere, leading to the formation of the Asian Tropopause Aerosol Layer (ATAL) (Fadnavis et al., 2013; Vernier et al., 2015). The CALIPSO lidar and Stratospheric Aerosol and Gas Experiment II (SAGE-II) satellite observations reveal that the ATAL extends over a wider Asian region (15–40^∘ N, 60–120^∘ E) between 12 and 18 km (Vernier et al., 2015; Fadnavis et al., 2013). The ECHAM6–HAMMOZ simulations reproduce the formation of an ATAL (extinction and sulfate aerosol) in the UTLS during the summer-monsoon season (Fig. 3a–b). The aerosol layer in the UTLS is connected to the troposphere during the pre-monsoon, indicating the transport of tropospheric aerosols into the UTLS. From March to November, the altitude of convective outflow propagates deeper into the UTLS. Strong uplift during the summer-monsoon season lifts the mid-tropospheric aerosols and aerosol precursors to the UTLS, generating aerosol minima in the mid-troposphere (Fadnavis et al., 2013). During the summer-monsoon season, the convective transport mostly occurs from the Bay of Bengal, the South China Sea and southern slopes of the Himalayas (Fadnavis et al., 2013; Medina et al., 2010). After the convective uplift, at altitudes above ∼360 K, radiatively driven upward transport in the anticyclonic monsoon circulation occurs at a rate of ∼1 K d⁻¹; this is a slower uplift than convection but faster than outside the anticyclone (Vogel et al., 2019). The simulated distribution of aerosol extinction and sulfate aerosols at 100 hPa from the CTRL simulation shown in Fig. 3c–d indicates maxima in aerosol extinction (Fig. 2c) and sulfate aerosols (Fig. 2d) in the anticyclone region.

Figure 3Monthly vertical variation in (a) extinction (km⁻¹ × 10⁻⁴) averaged for 25–40^∘ N, 70–120^∘ E; (b) same as (a) but for sulfate aerosols (ng m⁻³); (c) distribution aerosol extinction (km⁻¹ × 10⁻⁴) at 100 hPa averaged for the summer-monsoon season; (d) distribution of sulfate aerosol (ng m⁻³) at 100 hPa averaged for the summer-monsoon season. Wind vectors in (d) indicate the extent of the anticyclone. Panels (a)–(d) are obtained from CTRL simulations. The black line in (a) and (b) indicates the tropopause.

The estimated ratio of ECHAM6–HAMMOZ simulated sulfate aerosols in the UTLS to the total aerosol amount is 6:10 pointing to sulfate aerosols as a major ATAL constituent. Balloon sonde observations over south Asia also indicate that large amounts of sulfate aerosols may be present in the ATAL (Vernier et al., 2015). Tropospheric SO₂ and sulfate aerosol transported into the stratosphere during volcanically quiescent periods are potentially large contributors to the stratospheric aerosol burden (SPARC-ASAP, 2006).

4.2 Transport into the upper troposphere and lower stratosphere

We investigate the transport pathways of sulfate aerosol during different seasons from anomalies in sulfate aerosol for (1) Ind48 and (2) Ind48Chin70 simulations. Firstly, we present a vertical distribution of anomalies (relative to CTRL) in sulfate aerosol for Ind48 simulations in Fig. 4a–h. The striking feature is the poleward transport of Indian emissions in the UTLS throughout the year. A layer of sulfate aerosol enhancement extending from India to the Arctic (68–90^∘ N) is seen near the tropopause during pre-monsoon (3–15 ng m⁻³) and in the lowermost stratosphere during summer-monsoon (2–15 ng m⁻³), post-monsoon (2–6 ng m⁻³) and winter (0.5–3 ng m⁻³) seasons. This layer may be due to the transport of Indian sulfate aerosols to the Arctic by the lower branch of the Brewer–Dobson circulation – the AOD in the UTLS is 0.184×10⁴ (i.e., 1.1 %) to $\mathrm{4.15}\times {\mathrm{10}}^{-\mathrm{4}}$ (i.e., 4.17 %) over India and the Arctic (seasonal details in Table 2). Past studies also indicate the transport of pollution from south Asia and east Asia to the Arctic predominantly in the UTLS (Shindell et al., 2008; Fisher et al., 2011). From multi-model simulations, Shindell et al. (2008) show that seasonally varying the transport of south Asian sulfate aerosols to the Arctic maximizes in the pre-monsoon season. This enhancement of sulfate aerosols that maximizes during the pre-monsoon is also illustrated in Fig. 4a.

Table 2Seasonal mean AOD in the UTLS (300–90 hPa) over India (20–35^∘ N, 75–95^∘ E) and the Arctic (65–85^∘ N, 75–97^∘ E) from simulations performed. AOD is calculated at different altitude ranges indicated in brackets for some seasons since the sulfate aerosol layer varies in altitude in the UTLS.

Download Print Version | Download XLSX

Figure 4Vertical cross section of anomalies in sulfate aerosols (ng m⁻³) from Ind48-CTRL simulations for the pre-monsoon season (a) latitude–pressure section and (b) longitude–pressure section; (c)–(d) same as (a)–(b) but for the summer-monsoon season; (e)–(f) same as (a)–(b) but for the post-monsoon season; (g)–(h) same as (a)–(b) but for the winter season. The averages obtained over latitudes or longitudes are indicated in each panel. The black vertical bars indicate topography, and a black line indicates the tropopause.

Download

Figure 4 also shows that during most seasons the vertical transport occurs from the Bay of Bengal, the Arabian Sea and southern slopes of the Himalayas (15–35^∘ N, 60–100^∘ E), except during the post-monsoon season when it occurs from the west Asia and Tibetan Plateau region (20–35^∘ N, 60–95^∘ E). This may be due to the transport of sulfate aerosols from India to these regions, which might have been lifted to the UTLS by post-monsoon convection (see Figs. S1c, h, k, and S2 c). The enhancement of sulfate aerosols in the monsoon anticyclone (an ATAL feature) and the cross-tropopause transport associated with the summer-monsoon convection is evident in Fig. 4c–d (enhancement ∼5–15 ng m⁻³; 10 %–36 %). Past studies show that the aerosols transported into the lower stratosphere by the monsoon convection are recirculated in the stratosphere by the lower branch of the Brewer–Dobson circulation (Randel and Jensen, 2013; Fadnavis et al., 2013, 2017b). Yu et al. (2017) report that ∼15 % of the Northern Hemisphere column stratospheric aerosol originates from the Asian summer-monsoon anticyclone region. Figure 4d shows that aerosols spread to east and west from the anticyclone (20–120^∘ E), likely due to east- or westward eddy shedding from the anticyclone (Fadnavis and Chattopadhyay, 2017; Fadnavis et al., 2018). Eddy shedding is not evident in the seasonal mean distribution (Fig. 3b) due to its short duration (i.e., days) and episodic nature.

Figure 5Vertical cross section of anomalies in sulfate aerosols (ng m⁻³) from the Ind48Chin70-CTRL simulation for the pre-monsoon season (a) latitude–pressure section and (b) longitude–pressure section; (c)–(d) same as (a)–(b) but for the summer-monsoon season; (e)–(f) same as (a)–(b) but for the post-monsoon season; (g)–(h) same as (a)–(b) but for the winter season. The averages obtained over latitudes or longitudes are indicated in each panel. The black vertical bars indicate topography, and a black line indicates the tropopause.

Download

The influence of the Chinese SO₂ emission reduction (Ind48Chin70) on the vertical distribution of sulfate aerosols is shown in Fig. 5a–h. In the pre-monsoon season, the transport pattern is similar to the Ind48 simulations; however, the enhancement of sulfate aerosols at the Arctic tropopause is significantly hindered (1–3 ng m⁻³). The subsidence over north India (20–35^∘ N) has prevented sulfate aerosols from crossing the tropopause (Fig. 9a, e). A feeble plume tilted westward is seen during the monsoon season (Fig. 5c–d), and it is eastward and equatorward during the post-monsoon due to changes in circulations (ascending winds over south India and strong subsidence over north India; Fig. 9f–g). Entrainment into the anticyclone and cross-tropopause transport of the sulfate aerosols, seen in the Ind48 simulation, is inhibited by this subsidence. Interestingly, during summer-monsoon and post-monsoon seasons, poleward transport of south Asian sulfate aerosols has also been cut off due to circulation changes (subsidence over north India; see below in Fig. 9f–g). During winter, vertical winds over ∼20^∘ N lift aerosols from India to the mid-troposphere and further transport them to the Arctic (Figs. 5k–l, 9h). The vertical transport of sulfate aerosols increases AOD in the UTLS over India by $\sim \mathrm{0.32}\times {\mathrm{10}}^{-\mathrm{4}}$ (0.61 %) to $\mathrm{19.20}\times {\mathrm{10}}^{-\mathrm{4}}$ (19.25 %) (except in winter), and it increases over the Arctic by $\mathrm{2.09}\times {\mathrm{10}}^{-\mathrm{4}}$ (16.45 %) during the pre-monsoon season (see Table 2).

Figure 6Seasonal distribution of anomalies in clear-sky direct net radiative forcing (W m⁻²) simulated by ECHAM6–HAMMOZ at the top of the atmosphere, from Ind48-CRTL simulations for the (a) pre-monsoon (b) summer-monsoon, (c) post-monsoon and (d) winter seasons; (e)–(h) same as (a)–(d) but from Ind48Chin70-CTRL simulations; (i)–(l) same as (a)–(d) but from the offline model; (m)–(p) same as (e)–(h) but from the offline model. The black hatched lines in (a)–(h) indicate the 99 % significance level.

5.1 Radiative forcing

The seasonal mean anomalies in net radiative forcing at TOA due to sulfate aerosols from the Ind48 and Ind48Chin70 simulations of the ECHAM6–HAMMOZ model are illustrated in Fig. 6a–h. In general, both simulations show negative forcing over India and the surrounding region where sulfate aerosols are dispersed during that season (−0.2 to −2 W m⁻²). The distribution of anomalies in sulfate aerosols at 850 hPa (Fig. S2a–d) and Fig. 4a–d show that in the Ind48 simulations, during all seasons, sulfate aerosols are transported southwest over the Arabian Sea and partially to the east (during pre-monsoon, monsoon and winter towards Myanmar; during post-monsoon and winter to northeast China). These regions are associated with negative radiative forcing for Ind48 in Figs. 6a–d. This negative radiative forcing extending from north India towards the Arctic during pre-monsoon and summer monsoon is likely due to the poleward transport of south Asian sulfate aerosols in the UTLS (2–10 µg m⁻³) reflecting back solar radiation (see Fig. 4a, c). The poleward extension of negative radiative forcing (RF) is not evident during the post-monsoon and winter seasons (Fig. 6c, d). This may be due to a fine and thinner sulfate aerosol layer (∼1–4 µg m⁻³) in the upper troposphere which partially reflects back solar radiation, leading to weak positive and negative RF (−0.1 to +0.5 W m⁻²) over mid–high latitudes (40–70^∘ N).

The simulated RF at TOA in the Ind48Chin70 simulations is negative over India during all seasons ($\sim -\mathrm{0.6}$ to −2 W m⁻²) (Fig. 6e–h) similar to Ind48 (Figs. 6a–d). In addition, the Chinese SO₂ emission reductions in Ind48Chin70 have produced a significant positive forcing ∼0.6 to 6 W m⁻² over China (100–140^∘ E). The positive RF is also seen over the western Pacific (pre-monsoon, summer monsoon and winter) and the Bay of Bengal (post-monsoon and winter). This is due to the negative anomalies in sulfate aerosols over these regions in Ind48Chin70 (Fig. S2e–h). The southwestward transport of Indian sulfate aerosols to the Arabian Sea in the lower troposphere (Fig. S2e–h) during all seasons producing a negative RF in that region is evident in Fig. 6e–h. During the monsoon season, the narrow localized plume leads to a negative regional forcing (30–40^∘ N, 80–95^∘ E) of $\sim -\mathrm{0.6}$ W m⁻². The negative RF near 40–50^∘ N may be due to sulfate aerosols in the lower troposphere (Fig. 5c). The negative RF values (−0.1 to −0.4 W m⁻²) extending from the Indian region to the Arctic are likely due to the poleward transport in the upper troposphere during the pre-monsoon season and in the lower to mid-troposphere during the winter season (Fig. 6e, h). The seasonal mean net radiative forcing due to sulfate aerosols at the surface and at TOA is similar for both the Ind48 and Ind48Chin70 simulations (Fig. S3a–h), due to the strong scattering properties of the sulfate aerosols (Forster et al., 2007).

The comparison of RF at TOA obtained from ECHAM6–HAMMOZ simulations over the Arabian Sea (0–20^∘ N, 60–75^∘ E) during winter (Ind48: −2.0 W m⁻²; Ind48Chin70: 1.5 W m⁻²) (Fig. 4a) shows reasonable agreement with the INDOEX experiment (−1.25 to −2.0 W m⁻² over north India during January–March 1999 (Verma et al., 2012). Yu et al. (2016) reported that the increase in sulfate AOD (0.06–0.15) over the tropics (30^∘ S–30^∘ N) since the preindustrial period has exerted a forcing of −0.6 to −1.3 W m⁻².

The corresponding distribution of sulfate aerosol DRF at TOA estimated with our offline simulations for the four seasons for Ind48 and Ind48Chin70 is shown in Fig. 6i–p. The results from the offline model are in reasonable agreement with the ECHAM6–HAMMOZ simulations, although their magnitude differs spatially. Both the Ind48 and Ind48Chin70 simulations have produced negative RFs, varying between −0.2 and −2.0 W m⁻² over India. The reduction in SO₂ emission over China leads to an increase in RF of 2–6 W m⁻², comparable with the corresponding values simulated in ECHAM6–HAMMOZ. The differences in estimated RF in the offline calculations and the ECHAM6–HAMMOZ simulations are likely due to the fact that the implicit dynamical responses in ECHAM6–HAMMOZ are not captured in the offline simulations. However, the offline calculations are important insofar as they isolate the direct radiative impact of the simulated changes in aerosol loading.

The offline calculations further allow the specific effect of the enhanced aerosol layer in the UTLS (300–50 hPa) to be distinguished (Fig. 7a–h). Figure 7a–d show the direct radiative forcing at TOA (estimated from our offline simulations) induced by the sulfate aerosol enhancement in the UTLS (300–50 hPa) during the four seasons. The RF values from Ind48 are mostly negative over India and China and extending to the Arctic ($\sim -\mathrm{0.001}$ to −0.015 W m⁻²), due to the presence of the sulfate aerosol plume in the UTLS. Interestingly, the Ind48Chin70 simulation also shows negative RFs in the region colocated with the UTLS plume, e.g., in the summer-monsoon season, the plume over north India leads to negative RF values. Similarly, in the post-monsoon season, the sulfate aerosols plume extends to 15^∘ S and leads to negative RF values ($\sim -\mathrm{0.001}$ to −0.005 W m⁻²) (see Figs. 7g and S4). In the pre-monsoon season, the aerosol plume travels to the Arctic below or near the tropopause; therefore, partial contribution to RF from the UTLS (300 to 50 hPa) might have produced positive anomalies of 0.0001 to 0.0005 W m⁻² in mid–high latitudes. During winter, sulfate aerosols do not reach above the tropopause (Fig. 5g–h) and therefore RF values are positive over India and China. Thus, the radiative forcing caused specifically by UTLS aerosol shows a much clearer signal than the forcing due to the entire aerosol column (compare Figs. 6 and 7a–h). The sulfate aerosol layer, corresponding to the ATAL in the summer-monsoon season, leads to an RF of $\sim -\mathrm{0.011}$ to −0.015 W m⁻² (Fig. 7b) in the Ind48 simulation. It is reduced to −0.001 to −0.003 W m⁻² in the Ind48Chin70 simulations (Fig. 7f) due to a reduction in the transport of sulfate aerosols in the UTLS. The short-term ATAL RF at TOA has previously been estimated as about $\sim -\mathrm{0.1}$ W m⁻² over the Asian region during 1998–2015 (Vernier et al., 2015). The radiative forcing reported here caused solely by the sulfate aerosol particles in the UTLS is lower than the value reported by Vernier et al. (2015), who give an integral value for the ATAL and not only for the sulfate particles.

5.2 Incoming solar radiation, temperature and stability of the troposphere

An important impact of sulfate aerosols in the atmosphere is solar dimming, which counteracts the surface temperature response to the anthropogenic CO₂ increase (Ramanathan et al., 2005). There is observational evidence (1300 sites globally) indicating that one-third of potential continental warming attributable to increased greenhouse gas concentrations has been compensated for by aerosol cooling during 1964–2010 (Storelvmo et al., 2016). Solar radiation measurements over the Indian region (at 12 stations) during 1981–2004 show a declining trend varying between −0.17 to −1.44 W m⁻² yr⁻¹ (Padma Kumari et al., 2007). While not directly comparable to these previous studies, Ramanathan et al. (2005) reported a negative trend in solar flux observations at 10 different Indian stations (−0.42 W m⁻²), and their model simulations show a trend of −0.37 W m⁻² induced by the changes in BC and sulfate aerosols over India (0–30^∘ N and 60–100^∘ E).

Figure 7Simulated clear-sky direct net radiative forcing at TOA (W m⁻²) using the offline model due to sulfate aerosols on the UTLS–only for the (a) pre-monsoon (b) summer-monsoon, (c) post-monsoon and (d) winter seasons for Ind48; (e)–(h) same as (a)–(d) but for Ind48Chin70 simulations. The distribution of anomalies in net solar radiation (SR) (W m⁻²) at the surface from Ind48 for the (i) pre-monsoon (j) summer-monsoon, (k) post-monsoon and (l) winter seasons; (m)–(p) same as (i)–(l) but for Ind48Chin70 simulations.

We estimate the changes in net solar radiation at the surface for four seasons from the Ind48 and Ind48Chin70 simulations. Figure 7i–l show that the Ind48 simulations have produced negative anomalies in net solar radiation (SR) at the surface ($\sim -\mathrm{0.5}$ to −3 W m⁻²) over India and parts of China (where sulfate aerosols are transported) due to the enhanced sulfate aerosol layer reflecting back solar radiation. In general, the seasonal mean distribution of anomalies in net solar radiation at the surface is similar to the distribution of the anomalies in RF at TOA. A reduction in Chinese SO₂ emissions along with an increase in SO₂ emissions over India (Ind48Chin70) has produced a reduction in solar radiation over India, while there is a significant increase over China (1–5 W m⁻²) (see Fig. 7m–p).

Figure 8Vertical cross section of anomalies in temperature (K) from Ind48-CRTL simulations for the pre-monsoon season (a) latitude–pressure section and (b) longitude–pressure section; (c)–(d) same as (a)–(b) but for the summer-monsoon season; (e)–(f) same as (a)–(b) but for the post-monsoon season; (g)–(h) same as (a)–(b) but for the winter season. Figures (i)–(p) same as (a)–(h) but from Ind48Chin70-CRTL simulations. For the vertical cross section, averages obtained over latitudes or longitudes are indicated in each panel. The black hatched lines indicate the 99 % significance level. The black vertical bars indicate topography, and a black line indicates the tropopause.

Download

Sulfate aerosols also absorb infrared radiation thus causing heating locally and producing a cooling in the region below by solar dimming (Niemeier and Schmidt, 2017). Therefore, seasonally varying transport of sulfate aerosol may affect the thermal structure in the receptor region. Figure 8 shows a temperature enhancement near the region of transport of sulfate aerosols in the UTLS and a cooling of the atmosphere below it. For example, in the Ind48 simulations, positive temperature anomalies are seen near the sulfate aerosol layer extending to the Arctic, with negative anomalies below the layer during all seasons (except winter) (Fig. 8a–h). Similarly, a warming of ∼0.1–0.7 K over India is simulated in the Ind48Chin70 simulations in pre-monsoon and post-monsoon (Fig. 8i–j, m–n). During winter, in the Ind48Chin70 simulation, poleward transport occurs from the Indian lower or mid-troposphere to the lower stratosphere of mid–high latitudes. This region shows positive anomalies in temperature of ∼0.2 to 1 K (see Figs. 8o–p and 5g–h).

As shown in Fig. 8 the amplitude of the temperature anomalies in the UTLS varies seasonally and regionally. In general, there is temperature enhancement in the UTLS over north India and south China (20–35^∘ N, 75–130^∘ E) of $\sim \mathrm{0.2}\pm \mathrm{0.15}$ to 0.8±0.72 K in Ind48 (all four season) and $\sim \mathrm{0.1}\pm \mathrm{0.08}$ to 0.5±0.23 K in Ind48Chin70 (pre-monsoon and post-monsoon). Temperature uncertainties in this paragraph are obtained by determining the variability within the 10-member ensemble. After reaching the Arctic, these sulfate aerosols cause substantial warming in the lower stratosphere, i.e., $\sim \mathrm{1}\pm \mathrm{0.62}$ to 1.6±1.07 K in Ind48 during all seasons and 0.7±0.60 to 1.6±1.43 K in Ind48Chin70 in pre-monsoon and winter seasons. Figure 8 also shows a reduction in temperature of $-\mathrm{0.1}\pm \mathrm{0.05}$ to $-\mathrm{0.6}\pm \mathrm{0.4}$ K in the troposphere, below the warming, corresponding to the UTLS sulfate aerosols layer.

Figure 9The distribution of anomalies in vertical velocity (m s⁻¹) from Ind48-CTRL for the (a) pre-monsoon, (b) summer-monsoon, (c) post-monsoon and (d) winter seasons; (e)–(h) same as (a)–(d) but for Ind48Chin70-CTRL simulations. Vertical velocity is scaled by 1000. Seasonal distribution of anomalies in the Brunt–Väisälä frequency (s⁻¹ × 10⁻⁵) from Ind48-CTRL for the (i) pre-monsoon, (j) summer-monsoon, (k) post-monsoon and (l) winter seasons; (m)–(p) same as (i)–(l) but from Ind48Chin70-CTRL simulations. For the vertical cross section, averages obtained over latitudes or longitudes are indicated in each panel. The black vertical bars indicate topography.

Download

The changes in the circulation are illustrated in Figs. 9a–h. These panels show ascending winds in the region of the sulfate aerosol plume. For example the Ind48 simulations show ascending winds over northern India (while there is subsidence in the upper troposphere over 10–30^∘ N) during all seasons and in the Ind48Chin70 simulations during the pre-monsoon season. The reduction in Chinese SO₂ emissions (Ind48Chin70) induces strong descending winds over northern India during the summer monsoon and post-monsoon. It hindered the poleward transport of the plume as discussed in Sect. 4.2.

The sulfate aerosol-induced cooling in the upper troposphere (below the layer of sulfate aerosols) and subsidence in the upper troposphere cause a stabilization of the upper troposphere (Pitari et al., 2016). Figure 9i–p show that anomalies in the Brunt–Väisälä frequency are positive (0.2–3 s⁻¹ × 10⁻⁵) in the upper troposphere (250–150 hPa) over north India and south China (20–35^∘ N, 70–130^∘ E) during all the seasons in Ind48 and for the pre-monsoon and post-monsoon seasons in the Ind48Chin70 simulations. Thus, enhanced Indian sulfate aerosols have increased the stability of the upper troposphere and produce a cooling of ∼0.2–1.2 K (Fig. 8) in the upper troposphere. They induced upper-tropospheric subsidence (10–30^∘ N) in Ind48 and ind48Chin70 simulations (except in winter in Ind48Chin70). Upper-tropospheric temperature and stability play important roles in rainfall suppression (Wu and Zhang, 1998; Fadnavis and Chattopadhyay, 2017). Thus, upper-tropospheric cooling and enhanced stability may suppress the rainfall over India in all seasons in Ind48 and in the pre-monsoon and post-monsoon seasons in the Ind48Chin70 simulations. However, a complete analysis of the impact of the enhanced surface aerosols on rainfall is beyond the scope of this study.

5.3 Cirrus clouds

Cirrus clouds cover at least about 30 % of the Earth's area on annual average (Stubenrauch et al., 2013; Gasparini et al., 2018), occurring mainly between 400 and 100 hPa altitude. They play an important role in the Earth's energy budget (Gasparini and Lohmann, 2016; Hartmann et al., 2018), in the transport of water vapor into the stratosphere (Randel and Jensen, 2013), and in the atmospheric heat and energy cycle (Crueger and Stevens, 2015). Cirrus clouds can form either by homogeneous nucleation by freezing of dilute sulfate aerosols or by heterogeneous ice nucleation in the presence of ice nuclei, most commonly dust (Ickes et al., 2015; Cziczo et al., 2017). Moreover, a large fraction of cirrus clouds have a liquid origin as the ice crystals were either nucleated at mixed-phase conditions and transported to lower temperatures or detrained from convective cloud tops (Krämer et al., 2016; Wernli et al., 2016; Gasparini et al., 2018). All mentioned formation processes except for the heterogeneous nucleation of ice crystals below the homogeneous freezing temperature (i.e., at cirrus conditions) are represented in by our model simulations. However, heterogeneous freezing on dust and black carbon aerosols is included in mixed-phase clouds (Lohmann and Hoose, 2009), for temperatures between freezing and −35^∘ C. Figure 10a–h show the impact of SO₂ emission changes on cirrus clouds. These panels show a decrease (5 %–30 %) in cirrus clouds over north India (20–35^∘ N) in the UTLS. The decrease in cirrus clouds coincides with a significant decrease in ice crystal number concentration by −0.15 to −0.5 cm⁻³ between 250 and 50 hPa (except in winter in Ind48Chin70 since the plume of sulfate aerosols does not reach the upper troposphere) (Figs. 10i–p).

Figure 10Seasonal distribution of anomalies in cirrus cloud (%) from Ind48-CRTL simulations for the (a) pre-monsoon, (b) summer-monsoon, (c) post-monsoon and (d) winter seasons; (e)–h) same as (a)–(d) but for Ind48Chin70-CTRL simulations. Seasonal distribution of anomalies in ice crystal number concentration (ICNC; cm⁻³) from Ind48-CTRL for the (i) pre-monsoon, (j) summer-monsoon, (k) post-monsoon and (l) winter seasons; (m)–(p) same as (i)–(l) but from Ind48Chin70-CTRL simulations. For the vertical cross section, averages obtained over latitudes or longitudes are indicated in each panel. The black hatched lines indicate the 99 % significance level. The black vertical bars indicate topography.

Our analysis indicates that an increase in the upper-tropospheric sulfate aerosol concentration leads to a temperature increase in the upper troposphere and lower stratosphere of about $\sim \mathrm{0.2}\pm \mathrm{0.15}$ to 0.8±0.72 K over north India and south China and to a cooling below (Fig. 8). This temperature changes causes a decrease in the upper-tropospheric temperature gradient and vertical velocity and concurrently an increase in the upper-tropospheric (200–100 hPa) static stability (Brunt–Väisälä frequency) (over 80–120^∘ E) (Fig. 9i–p) (Figs. 9a–h). A combination of decreased upper-tropospheric updraft motion and increased temperature decreases the likelihood of cirrus cloud formation in a similar way to the simulated responses to volcanic eruptions or stratospheric sulfur geoengineering (Kuebbeler et al., 2012; Pitari et al., 2016; Visioni et al., 2018a).

5.4 Discussion

Our model simulations presented here provide seasonal transport processes and estimates of radiative forcing for the year 2011. The interannual variability in the transport processes may impact the shallow or deep injection of sulfate aerosols into the lower stratosphere. The stratospheric warming produced in response to the transport of rising south Asian anthropogenic sulfate aerosol in the UTLS over Asia and further to the Arctic (Figs. 4 and 5) may modulate the QBO and thereby the transport of sulfate aerosol from the tropics to the extra-tropics. The QBO phases are modulated by the amount of sulfate and the height of the injection (Aquila et al., 2014; Niemeier and Schmidt, 2017; Visioni et al., 2018b). A previous study reports that the QBO slows down after an injection of 4 Tg (S) yr⁻¹ into the stratosphere and completely shuts down after the injection of 8 Tg (S) yr⁻¹ (Niemeier and Schmidt, 2017). However, another model study finds that the QBO, even for a larger amount of SO₂ injections, does not deviate much from present-day conditions (Richter et al., 2018). These studies indicate that there is a complicated interaction between UTLS aerosols, atmospheric dynamics and atmospheric chemistry (Richter et al., 2017; Niemeier and Schmidt, 2017; Visioni et al., 2018b). The QBO is known to modulate the tropical convection (Collimore et al., 2003; Fadnavis et al., 2013; Nie and Sobel, 2015). Thus, the transport of sulfate aerosols into the stratosphere would impact the tropospheric hydrological cycle in addition to the tropospheric aerosol loading. The increasing amounts of tropospheric sulfate aerosol loading are linked with droughts via changes in radiative forcing, stability and tropospheric temperature gradient (Yeh et al., 2015; Kim et al., 2016). Simulations for a longer time period and with the inclusion of QBO phases may reveal the influence of current SO₂ emission on tropospheric–stratospheric dynamics and the hydrological cycle. Nonetheless, the results of the current study show the impacts of sulfate aerosols on the UTLS for realistic emission perturbations over India and China.