CCSM3 simulated regional effects of anthropogenic aerosols for two contrasting scenarios: rising Asian emissions and global reduction of aerosols

Authors


Abstract

This paper examines the effects of two largely contrasting aerosol emissions scenarios on regional climate using National Center for Atmospheric Research Community Climate System Model version 3: (1) increasing the anthropogenic aerosols over China and India by a factor of three and (2) reducing the global anthropogenic aerosols by a factor of 10. Dynamic footprints of the increased Asian aerosols with monthly variations are obtained from Model for OZone And Related chemical Tracers simulations. Increasing Asian aerosol emissions would result in cooling and reduction of precipitation over China and India, with large warming over the USA and southern Canada in winter and cooling in summer. Additionally, large changes in rainfall rate are identified over the tropical regions. In contrast, reducing the global aerosol emissions by a factor of 10 would significantly warm the atmosphere especially over the polluted land areas of both hemispheres. Increases in rainfall over polluted land areas are also noted. Deepening of the Aleutian low and weakening of the Icelandic low in winter are noted in the 500-mb geopotential height under both scenarios suggesting a strengthening of the North Pacific storm track and weakening of the North Atlantic Oscillation. The polar regions of winter hemisphere are subject to large changes in the 500-mb geopotential height. Teleconnection patterns associated with ENSO play important roles in causing large changes in surface air temperature and rainfall far away from the source regions of the altered aerosol concentrations. Copyright © 2009 Royal Meteorological Society

1. Introduction

As the economies of China and India continue to grow they are emitting more and more anthropogenic aerosols (mainly black carbon, organic carbon, and sulphate aerosols) into the atmosphere. These aerosols constitute significant concerns to society due to their effects on human health, local and global ecosystems, visibility degradation as well as their potential effects on local, regional, and global climate (Menon et al., 2002; Ramanathan and Crutzen, 2003; Ramanathan et al., 2005). While greenhouse gases are warming the climate, aerosols are generally cooling the climate (IPCC, 2001, 2007). The major greenhouse gases tend to be well mixed in the atmosphere while aerosols display large variability both spatially and temporally. The impacts of aerosols on regional climate and environment are expected to be more significant than greenhouse gases in aerosol concentrated areas.

Scientific investigations of the effects of Asian anthropogenic aerosols are necessary for examining to what degree these aerosols might impact the regional and global climate while reconciling with the need of economic development in two of the most populous countries. On the other hand, aerosol concentrations in the atmosphere could change substantially in the future in response to emission control strategies and energy efficiency measures aimed at reducing anthropogenic emissions and improving air quality. Consequences of such future changes need to be carefully and systematically investigated in order to understand their potential effects and to prevent dangerous anthropogenic interference with the climate system and to maintain global warming below a specified threshold (United Nations, 1992). Recently, Brasseur and Roeckner (2005) examined the impact of removing all of the anthropogenic sulphate aerosols from the atmosphere on the future evolution of climate. They noted an increase in the globally averaged surface air temperature and precipitation by as much as 0.8 K and 3%, respectively, in less than a decade.

Menon et al. (2002) investigated the climate effects of black carbon aerosols in China and India using a global climate model. They noted that the absorbing black carbon aerosols heat the air, alter regional atmospheric stability and vertical motions, and affect the large-scale circulation and hydrological cycle with significant regional climatic effects. Ramanathan et al. (2005) conducted an ensemble of coupled ocean–atmosphere simulations for the period of 1930–2000 to study the impacts of atmospheric brown clouds (ABC, defined as layers of air pollution consisting of aerosols) in South Asia on regional climate and hydrological cycle. Their simulations indicated profound regional climate impacts that include: (1) altering the net solar flux both at the surface and at the top of the atmosphere (TOA), (2) reducing surface evaporation, (3) weakening the latitudinal SST gradients, (4) stabilizing the troposphere by differential warming of the atmosphere, and (5) decreasing monsoon rainfall and increasing numbers of droughts. Their simulations further suggested that absorbing aerosols in ABC may have masked as much as 50% of the surface warming due to the global increase in greenhouse gases.

Recently, Lau et al. (2006) examined the Asian summer monsoon anomalies that were induced by aerosol (mainly dust and black carbon) direct forcing through the Tibetan Plateau using a global circulation model (GCM). Their results suggested that increased dust loading coupled with black carbon emissions from local sources in northern India during late spring may lead to an advance of the rainy periods and subsequently an intensification of the Indian summer monsoon and suppressed rainfall over East Asia and the adjacent oceanic regions. Gu et al. (2006) studied the climatic effects of different aerosol types in China using the University of California at Los Angeles (UCLA) GCM. They noted that while sulphate aerosols mainly reflect solar radiation and induce negative forcing at the surface, black carbon and large dust particles absorb substantial solar radiation and have a positive solar forcing at the TOA but reduce the solar radiation reaching the surface. All these studies have provided insights into the impacts of primarily individual aerosol types (black carbon, dust or sulphate aerosols) on climate and hydrological cycles in and around the source regions. However, the collective impacts of aerosols emissions on regional and global climate are still yet to be examined. Most of the above-mentioned studies have emphasized the importance of including all types of aerosols in order to reproduce the observed climate trends and to generate more realistic climate response.

This work is an attempt in examining the collective effects of anthropogenic aerosols on regional and global climate using the National Center for Atmospheric Research (NCAR) Community Climate System Model version 3 (CCSM3). Two largely contrasting aerosol emissions scenarios that reflect the current and future emissions trends are examined: (1) increased anthropogenic aerosols over China and India and (2) decreased anthropogenic aerosols globally. This work is organized in the following way. Section 2 gives a brief description of CCSM3. Experimental design and aerosol forcing data are presented in sections 3 and 4, respectively. Model simulations under the two emissions scenarios are analyzed and discussed in sections 5 and 6. Teleconnection patterns associated with ENSO under the two scenarios are described in section 7. Major conclusions and discussions are presented in section 8.

2. CCSM3 climate model

CCSM3 (Collins et al., 2006) is a fully coupled climate model with components representing the atmosphere, ocean, sea ice, and land surface connected by a coupler that exchanges fluxes and state information among these components. CCSM3 supports several horizontal resolutions for the atmosphere and variable latitudinal and longitudinal resolutions for the ocean. In this work, the horizontal resolution for the atmospheric component is T31 with an equivalent grid spacing of 3.75° in latitude and longitude with 26 vertical levels in a hybrid terrain-following coordinate system are used. The land component (Oleson et al., 2004) operates on the same grids as the atmospheric component. The ocean model in CCSM3 is based on the Parallel Ocean Program (POP) (Smith and Gent, 2002). POP uses a displaced-pole grid (Smith and Kortas, 1995) with the logical North Pole displaced into the Greenland land mass. In this work, the longitudinal resolution in POP is 3.6°; the latitudinal resolution varies between 0.9° at the equator and 1.8° in the extratropics. The 25-level vertical grid in POP has variable spacing, starting with 12 m near the surface to better resolve surface and mixed-layer processes and expanding to 450 m in the deep ocean. The sea ice model in CCSM3 (Briegleb et al., 2004) employs the same displaced-pole grid and same resolution as the ocean component. Kiehl et al. (2006) showed that for CCSM3 the transient climate response, which is the surface temperature change at the point of doubling of CO2 in a 1% yr−1 CO2 increase scenario, is 1.48 °C. CCSM3 has been shown to produce realistic simulations over a wide range of spatial resolutions, enabling inexpensive simulations lasting several millennia as well as detailed studies of continental-scale dynamics, variability, and climate change (Collins et al., 2006).

The aerosol parameterization in CCSM3 incorporates five species of aerosols, including sea salt, soil dust, black and organic carbonaceous aerosols, sulphate, and volcanic sulfuric acid (Collins et al., 2006). Large soil dust and black carbonaceous aerosols are strongly absorbing in the visible wavelengths while other aerosols primarily scatter solar radiation. There are four size categories of soil dust with diameters spanning from 0.01 to 10 µm. Black carbon and organic carbon are represented by two tracers each for the hydrophobic (new) and hydrophilic (aged) components. CCSM3 includes the direct and semi-direct effects (Hansen et al., 1997) of tropospheric aerosols on shortwave fluxes and heating rates. The indirect effects of aerosols (Twomey et al., 1977; Albrecht, 1989) are not included in the model. The stratospheric volcanic aerosols are treated using a single species and the zonal variations in the stratospheric mass loading are not included.

3. Emissions scenarios and experimental design

Two aerosol emissions scenarios are investigated in this study. One is increasing the concentrations of anthropogenic black carbon, organic carbon, and sulphate aerosols over China and India by a factor of three (referred to as the dirty scenario). This scenario represents the current emissions trend in which China and India are releasing ever increasing amounts of pollutants into the atmosphere as their economies develop. The other one is reducing the concentrations of anthropogenic black carbon, organic carbon, and sulphate aerosols globally by a factor of 10 (referred to as the clean scenario). This scenario represents the future emissions trend in which society is striving to reduce anthropogenic emissions to combat global warming and improve the air quality. CCSM3 does not have a chemistry component to represent emissions and chemical reactions during model integration. Here for convenience we used emissions scenarios to refer to the two experiments, although what is really altered in the model is the aerosol loading.

CCSM3 was spun up with a 50-year simulation and was then run for another 50 years as a base experiment. The initial data came from one of the non-IPCC constant 1990 control runs (http://www.ccsm.ucar.edu/models/ccsm3.0/#input). Current climate setting with prescribed concentrations of greenhouse gases and aerosols at the 1990 level was adopted for the CCSM3 simulations. Runs for the dirty and clean scenarios were driven by the spin-up simulations and were conducted for a 50-year simulation for each scenario. Figure 1 is the schematic diagram showing the experimental design. In the runs for the dirty and clean scenarios, the only thing that has been changed from the base experiment is the aerosol mass concentrations that include dynamic footprints, which will be discussed shortly. In the clean scenario, the base concentrations of anthropogenic black carbon, organic carbon and sulphate aerosols are reduced by a factor of 10 at every grid point. The base concentrations were obtained using an aerosol assimilation system that includes transport and dispersion (Collins et al., 2004), and we have assumed that the anthropogenic aerosol footprint does not change from that of the base concentrations after the ten-fold reduction at every grid point.

Figure 1.

Schematic diagram showing the experimental design. Three ensemble CCSM3 runs were performed for each scenario

Three ensemble members with differing initial conditions were carried out for CCSM3 for each scenario. The differing initial conditions were simply the base run simulations that are 2-year apart (i.e. Year 01, Year 03, and Year 05). Analyses will be focused on the seasonal means of the last 30-year (Year 21 through Year 50) model simulations.

4. Aerosol forcing and distribution

CCSM3 uses three-dimensional time-dependent distributions of aerosol species that are generated by offline aerosol assimilation systems as base aerosols and are loaded into the atmospheric component during the initialization process (Collins et al., 2004). In this study, the dynamic footprints of increased anthropogenic aerosols over China and India are obtained by employing the Model for OZone And Related chemical Tracers (MOZART) simulations. MOZART (Horowitz et al., 2003) is a comprehensive global chemistry transport model of atmospheric composition that includes a detailed scheme for tropospheric ozone, nitrogen oxides and hydrocarbon chemistry. Its horizontal resolution is 2.8° longitude × 2.8° latitude with 34 hybrid vertical levels that extend from the surface to about 40 km. MOZART uses a 20-min time step for all chemistry and transport processes.

The MOZART runs were driven by meteorological data at 3-h time resolution from the middle atmosphere version of the NCAR Community Climate Model (MACCM3) (Kiehl et al., 1998). The dynamic footprints of the increased Asian emissions were obtained as two steps: first the differences between the MOZART simulated mass concentrations with and without aerosols over China and India were computed at each vertical levels, then the ratio of the differences to the MOZART simulated mass concentrations with base aerosols over China and India was computed. The footprint data are monthly output and include seasonal variations. Figure 2 shows the horizontal distributions of dynamic footprints of Asian anthropogenic aerosols at 700 mb for January and July. Clearly, aerosols emitted by China and India are transported well beyond their source regions and impact the North Pacific and North America as well as the Arctic region in the case of black carbon and sulphate aerosols. In January the pollutants are also transported over the North Indian Ocean as the prevailing northeast winds in this area bring in polluted continental air.

Figure 2.

Dynamic footprints of Asian anthropogenic aerosols (black carbon, organic carbon and sulphate aerosols) at 700 mb for January (left panels) and July (right panels). The footprint data range between 0 and 1 with 1 representing the source region

In our CCSM3 runs, aerosol mass for each species and aerosol properties do not change from year to year. Table I shows the annual, December–January–February (DJF) and June–July–August (JJA) mean aerosol optical depth (AOD), and single scattering albedo (SSA) at λ = 0.495 µm averaged over China and India as well as over the entire globe for the base experiment, dirty, and clean scenarios. The globally averaged AOD and SSA were computed as the Gaussian weighted average of the AOD and SSA at all sun-lit model points. The regionally and globally averaged SSA is presented here for reference purpose only. The AOD averaged over China and India for the base and dirty scenario is about 2–3 times higher when compared with the global average. The SSA averaged over China and India is slightly lower than the global average most likely due to the large amount of black carbon emissions over this region as black carbon absorbs solar radiation and reduces SSA.

Table I. Annual (ANN), DJF and JJA mean vertically integrated AOD, SSA in CCSM3 averaged over China and India (0°-50°N, 60°E-130°E) and the entire globe
 AODSSA
 China/IndiaGlobeChina/IndiaGlobe
BaseANN0.220.100.920.96
 DJF0.150.090.920.96
 JJA0.300.120.930.96
Dirty ScenarioANN0.370.110.930.96
 DJF0.260.100.920.96
 JJA0.500.150.940.96
Clean ScenarioANN0.090.060.920.97
 DJF0.050.060.930.98
 JJA0.130.070.910.97

Large differences in the AOD averaged over China and India between the clean scenario and the base experiment (Table I) indicate the dominant contribution of anthropogenic aerosol emissions in this region when compared with the other types of aerosols such as dust and volcanic aerosols. The footprint of the increased Asian aerosol emissions can also be identified in the differences of the globally averaged AOD between the dirty scenario and the base experiment. Table I also shows that over China and India the AOD in JJA is about twice as much as in DJF not only for the base experiment but also for the dirty and clean scenarios as well.

Figure 3 shows the horizontal distribution of the seasonal mean AOD at λ = 0.495 µm for the base experiment. High AODs are noted in all seasons over China and India, the eastern part of the USA, and the Western Europe, mainly due to emissions from fossil fuel combustion and biofuel burning. High AODs over Africa, South America, and the Middle East are primarily due to dust and biomass burning. For the polluted regions, the AOD is generally the highest in JJA and the lowest in DJF. Downwind transport and dispersion from the polluted regions are also evident. The elevated AODs over the Southern Hemisphere oceanic areas in DJF are mainly due to sea salt aerosols.

Figure 3.

Vertically integrated AOD at λ = 0.495 µm averaged for (a) DJF, (b) MAM, (c) JJA, and (d) SON based on CCSM3 base run simulations

Seasonal mean AOD differences at λ = 0.495 µm between the dirty scenario and the base run and between the clean scenario and the base run are presented in Figure 4. Increases in the range of 0.01–0.9 are the dominant feature for the dirty scenario (Figure 4(a)–(d)). These increases are exclusively over and downwind of China and India consistent with the transport of the elevated Asian aerosols emissions in the regional environment (Figure 2). The increases in AOD also extend more extensively in north–south direction and more eastward in March–April–May (MAM) and JJA than in DJF and September–October–November (SON), likely a result of transport by deep convections as the land heats up and monsoon systems develop in MAM and JJA.

Figure 4.

Mean AOD change at λ = 0.495 µm between the dirty scenario and the base run for (a) DJF, (b) MAM, (c) JJA, (d) SON and between the clean scenario and the base run for (e) DJF, (f) MAM, (g) JJA and (h) SON based on CCSM3 simulations

Decreases in the AOD in the range of − 0.9 to − 0.01 are associated with the clean scenario (Figure 4(e)–(h)). These decreases are located mainly over the now less-polluted regions with anthropogenic influence and their downwind areas. For both emissions scenarios, the largest changes in AOD are identified over the eastern part of China in all seasons.

Figure 5 shows the seasonal mean SSA at λ = 0.495 µm for the base experiment. The SSA over Africa, the South America, the Middle East, and Australia is relatively low (0.8–0.9) primarily due to the absorbing dust emissions and black carbon emissions from biomass burning. Over China, India, and the USA the SSA is also relatively low (0.85–0.95) most likely due to the black carbon emissions from residential activities and transport sectors in these populated areas. The SSA is close to 1 over the oceans in the high latitudes of both hemispheres.

Figure 5.

Vertically integrated SSA at λ = 0.495 µm averaged for (a) DJF, (b) MAM, (c) JJA, and (d) SON based on CCSM3 base run simulations

Figure 6 presents the seasonal mean SSA differences at λ = 0.495 µm between each scenario and the base run. Under the dirty scenario (Figure 6(a)–(d)), decreases in the range of − 0.02 to − 0.01 are identified over the northeastern part of China in DJF while increases in the range of 0.005–0.02 are noted for all seasons over India, Southeast Asia, the southern part of China and the Pacific Northwest of the USA. The largest increases (≥0.02) are located over India. The wintertime decreases over the northeastern part of China are mainly due to large black carbon emissions from industrial and residential heating activities when compared with the emissions of organic carbon and sulphate aerosols. The increased SSA over India and south China is related to the fact that the combined mass concentrations of organic carbon and sulphate aerosols are larger than those of black carbon under the dirty scenario (not shown). The increases over the Pacific Northwest of the USA are likely the result of long-range transport of the Asian pollutants.

Figure 6.

Mean SSA change at λ = 0.495 µm between the dirty scenario and the base run for (a) DJF, (b) MAM, (c) JJA, (d) SON and between the clean scenario and the base run for (e) DJF, (f) MAM, (g) JJA and (h) SON based on CCSM3 simulations. Shadings with solid lines and hatchings with dashed lines represent positive and negative changes, respectively

Under the clean scenario (Figure 6(e)–(h)), decreases in SSA in the range of − 0.04 to − 0.001 are identified over Asia and northern Africa in MAM, JJA and SON with the largest decreases located over the Gobi Desert in Central Asia. These decreases in SSA are likely related to the dominance of the absorbing dust aerosols after a ten-fold removal of black carbon, organic carbon and sulphate aerosols. In DJF (Figure 6(e)), decreases are located mainly over Central Asia while Southeast Asia, the eastern part of China and Europe exhibit increases (≥0.02) in SSA. Over Australia and the southern Africa, decreases in SSA are evident in DJF while increases are identified in the other seasons. Note that dust-storm frequency in Australia is generally the highest in January, February, and March. Over the North America, decreases in the range of − 0.01 to − 0.02 are identified in JJA and slight increases are identified in DJF with little changes in MAM and SON. The decreases in DJF over the North America are likely the result of local emissions changes as well as the trans-Atlantic transport of African dust aerosols. Relatively large increases are noted over South America in all seasons. Over the oceans downwind of the polluted land regions, the SSA is elevated when compared to the base experiment. This is due to the removal of the anthropogenic aerosols such that sea salt aerosols become dominant as sea salt aerosols have an SSA of 1. Polar regions witness slightly increased SSA in all seasons.

5. Dirty scenario

5.1. Clearsky shortwave radiation change and cloud forcing change

Figure 7(a)–(d) show the CCSM3 simulated seasonal mean differences in clearsky shortwave radiation at the TOA between the dirty scenario and the base experiment. These differences reflect the net direct and semi-direct effects of the increased anthropogenic aerosols through the entire atmospheric column, and include both radiative forcing of aerosols and the associated feedback. Large increases in TOA upward shortwave radiation fluxes (≥3 W m−2) that are statistically significant (p < 0.10) are noted over China and India and their downwind areas in all seasons with the maximum increases (>7 W m−2) occurring in JJA when solar radiation is the strongest in the Northern Hemisphere. In JJA, increases in upward shortwave radiation fluxes are also noted over the high latitudes of the Northern Hemisphere, mainly due to the Asian aerosols that are transported there (Figures 2 and 4(c)). Increases in downward shortwave radiation fluxes in the range of 1–5 W m−2 along the edge of the Antarctic Circle (∼60°S) are identified during DJF and SON but these increases are not statistically significant.

Figure 7.

CCSM3 simulated mean clearsky shortwave radiation change (W m−2, dirty scenario—base run) for (a) DJF, (b) MAM, (c) JJA, (d) SON and cloud forcing change (W m−2, dirty scenario—base run) for (e) DJF, (f) MAM, (g) JJA, and (h) SON at the TOA averaged over the last 30-model years (Year 21 through Year 50). Contours are drawn from − 9 to 9 W m−2 with an interval of 2 W m−2. Solid and dashed lines represent positive and negative changes, respectively. Shading indicates statistical significance (p < 0.10) is achieved with the Student's t-test

Appreciable increases in downward shortwave radiation fluxes (1–3 W m−2) that are not statistically significant are noted over the USA and the southern Canada in DJF and over the eastern Russia in MAM (Figure 7). These changes are likely related to (1) Asian black carbon aerosols that are transported over these regions (c.f., Figure 2(a)) and (2) changes in atmospheric waves and teleconnection patterns associated with the increased Asian aerosols that give rise to changes in clouds. Changes in the mass concentrations of aerosols would alter the modelled heating profiles in the atmosphere through the direct and semi-direct effects (as well as through the indirect effects which are not included in CCSM3) and the modelled heating profiles in the oceans through air–sea interaction, which in turn brings changes in the atmospheric circulations through induced atmospheric waves and teleconnection (Menon et al., 2002; Kim et al., 2006).

Figure 7(e)–(h) show the CCSM3 simulated seasonal mean differences in cloud forcing at the TOA between the dirty scenario and the base experiment. Cloud forcing is computed as the differences between the net solar radiation and net clearsky solar radiation, and represents the feedback of altered cloud fields as a result of aerosol forcing. Changes in cloud forcing correspond well to changes in vertically integrated total clouds that range between − 5% and 5% (not shown). Compared to changes in clearsky solar radiation that are concentrated primarily near the source regions of the altered aerosol concentrations, changes in cloud forcing are more spread. Large changes in cloud forcing that are statistically significant are noted over China and India and their downwind areas in all seasons (Figure 7(e)–(h)). These changes are always in opposite sign to those in clearsky solar radiation, thus giving rise to small changes in net solar radiation over these regions (not shown). Increases in upward cloud forcing in the range of 1–3 W m−2 along the edge of the Antarctic Circle are also identified during DJF and SON, which largely compensate the increases in downward shortwave radiation fluxes in the same area.

5.2. Surface air temperature change

Figure 8 presents the CCSM3 simulated surface air temperature differences between the dirty scenario and the base experiment for the four seasons. Salient features in Figure 8 include: (a) cooling over the Eastern Asia and India in the range of − 0.6 to − 0.2 °C in all seasons (statistically significant in DJF and JJA); (b) cooling in the range of − 1.0 to 0.6 °C over northern Russia in all seasons (statistically significant in DJF); (c) warming in the range of 0.6–1.0 °C over the USA and southern Canada in DJF that is statistically significant and cooling in the range of − 0.6 to − 0.2 °C in MAM and JJA; (d) cooling over the North Pacific in the range of − 0.6 to − 0.2 °C in all seasons except for the Aleutian Basin where slight warming is identified; and (e) warming over the high latitudes of the Southern Hemisphere in the range of 0.2–0.6 °C in DJF and SON and 0.6–1.0 °C in JJA and MAM that are statistically significant. Over Greenland, the model simulates statistically significant warming (0.2–0.6 °C) in all seasons except for JJA.

Figure 8.

CCSM3 simulated surface air temperature change ( °C, dirty scenario—base run) for (a) DJF, (b) MAM, (c) JJA, and (d) SON averaged for the last 30-model years (Year 21 through Year 50). Contours are drawn from − 1.8 to 1.8 °C with an interval of 0.4 °C. Solid and dashed lines represent positive and negative changes, respectively. Shading indicates statistical significance (p < 0.10) is achieved with the Student's t-test

Compared to clearsky shortwave radiation changes (Figure 7), large changes in surface air temperature are located far away from the source regions of the altered aerosol concentrations (Figure 8). This may imply the importance of teleconnection in association with air–sea interaction as will be discussed in section 7. Near the source regions, changes in cloud forcing largely compensate changes in clearsky shortwave radiation (Figure 7), thus resulting in small local changes in surface air temperature (Figure 8).

5.3. Rainfall change

Figure 9 shows the CCSM3 simulated rainfall differences between the dirty scenario and the base experiment for the four seasons. These changes are statistically significant most of the time. Reductions of rainfall in the range of − 0.3 to − 0.1 mm/day are noted over the southern and eastern parts of China in MAM and JJA and over India in JJA. There is cooling over the same region during the same time period (Figure 8). Further analyses (not shown) indicate that the cooling trend prevails in the lower troposphere. Such a cooling trend would result in the stabilization of the troposphere which would suppress the development of convection that normally brings rainfall there. Slight increases in rainfall over the Tibetan Plateau are indicated in JJA (Figure 9(c)), which may be due to the compensating effects of the inhibited convection over the southern and eastern parts of China and India.

Figure 9.

CCSM3 simulated rainfall rate change (mm day−1, dirty scenario—base run) for (a) DJF, (b) MAM, (c) JJA, and (d) SON averaged for the last 30-model years (Year 21 through Year 50). Contours are drawn from − 0.9 to 0.9 mm day−1 with an interval of 0.2 mm day−1. Solid and dashed lines represent positive and negative changes, respectively. Shading indicates statistical significance (p < 0.10) is achieved with the Student's t-test

Increases in rainfall in the range of 0.1–0.3 mm/day are evident along the coastal areas of the US Pacific Northwest in DJF and SON. As will be discussed shortly, the deepening of the Aleutian low in DJF would bring anomalous and moist southwesterly flow that is lifted up by the mountains over the Pacific Northwest and then precipitates out on the windward side of the mountains. A similar mechanism may also explain the increases in rainfall over the same region in SON.

Large changes in rainfall are noted over the tropical regions of both hemispheres in all seasons. These changes are likely related to teleconnection patterns associated with air–sea interaction. In general, decreases and increases in rainfall are noted north and south of the Equator, respectively.

5.4. 500-mb geopotential height change

CCSM3 simulated 500-mb geopotential height differences between the dirty scenario and the base experiment are shown in Figure 10. A broad band of decreases in the 500-mb geopotential height that is statistically significant is notable over Asia, the North Pacific and northern Canada in DJF. This band of decreases corresponds approximately to the path of the transported Asian pollutants (Figure 2) and where surface cooling is identified (Figure 8). The resultant deepening of the Aleutian low over the North Pacific would bring anomalous southwesterly flow to the western part of the USA as alluded to before. In MAM and SON, decreases in the 500-mb geopotential height are identified over the western part of Canada and northwestern part of the USA (statistically significant in SON) with slight increases noted over the eastern part of the USA and Greenland (Figure 10(b) and (d)).

Figure 10.

CCSM3 simulated 500-mb geopotential height change (m, dirty scenario—base run) for (a) DJF, (b) MAM, (c) JJA, and (d) SON averaged for the last 30-model years (Year 21 through Year 50). Contours are drawn from − 36 to 36 m with an interval of 8 m. Solid and dashed lines represent positive and negative changes, respectively. Shading indicates statistical significance (p < 0.10) is achieved with the Student's t-test

Decreases in the 500-mb geopotential height are dominant over the Northern Hemisphere in JJA (Figure 10(c)) with large decreases (<− 4 m) located over the Aleutian Basin, the western part of the USA, Greenland, and the Arctic. Over the Antarctic the 500-mb geopotential height increases especially in JJA and SON. These changes in 500-mb geopotential height are statistically significant. A statistically significant band with negligible changes in the 500-mb geopotential height is noted over the tropics in MAM and JJA. It is probably an artifact of the statistical computations.

The statistically significant decreases in the 500-mb geopotential height over the North Pacific in DJF suggest deepening of the Aleutian low under the dirty scenario with large implications on the mid-latitude jet stream and storm track (Zhang et al., 2007). This deepening of the Aleutian low appears to counteract the effects of greenhouse gases which tend to push the Aleutian low poleward (Salathé, 2006). Over the North Atlantic, statistically significant increases (decreases) in the 500-mb geopotential height north (south) of around 30°N are also noted in DJF (Figure 10). This pattern would help to reduce the North Atlantic Oscillation (NAO) positive index and would possibly cause mild winter in the Western Europe (Rodwell et al., 1999).

6. Clean scenario

6.1. Clearsky shortwave radiation change and cloud forcing change

Figure 11(a)–(d) show the CCSM3 simulated seasonal mean TOA clearsky shortwave radiation differences between the clean scenario and the base experiment. Statistically significant changes are located primarily near the source regions of polluted land areas. A prominent feature in the changes is the large increases in downward shortwave radiation fluxes (>3 W m−2) over the major polluted land areas (i.e. China, India, North America, and Western Europe) and their downwind regions throughout the year. Appreciable increases in downward shortwave radiation fluxes are also noted over the central Africa and the South America. These increased downward shortwave radiation fluxes are due to the removal of the anthropogenic aerosols, which would otherwise reflect a large portion of the incoming solar radiation back to space. Seasonal variations in the downward shortwave fluxes are evident in that the largest (smallest) changes in fluxes are noted over the polluted land areas in JJA (DJF) when the solar radiation is the strongest (weakest) over the Northern Hemisphere.

Figure 11.

CCSM3 simulated mean clearsky shortwave radiation change (W m−2, Clean Scenario—base run) for (a) DJF, (b) MAM, (c) JJA, (d) SON and cloud forcing change (W m−2, Clean Scenario—base run) for (e) DJF, (f) MAM, (g) JJA, and (h) SON at the TOA averaged over the last 30-model years (Year 21 through Year 50). Contours are drawn from − 9 to 9 W m−2 with an interval of 2 W m−2. Solid and dashed lines represent positive and negative changes, respectively. Shading indicates statistical significance (p < 0.10) is achieved with the Student's t-test

Figure 11(e)–(h) show the CCSM3 simulated seasonal mean TOA cloud forcing differences between the clean scenario and the base experiment. As in the case for the dirty scenario, large but opposite changes that are statistically significant are located where large changes in the clearsky shortwave radiation are identified, resulting in small changes in net solar radiation near the source regions of the altered aerosol concentrations (not shown). For clean scenario, changes in cloud forcing also correspond well to changes in vertically integrated total clouds that range between − 5% and 5% (not shown).

6.2. Surface air temperature change

Figure 12 shows the CCSM3 simulated surface air temperature differences between the clean scenario and the base experiment averaged for the four seasons. Warming is the dominant feature under this clean scenario. Warming in the range of 0.6–1.0 °C in DJF and SON and 0.2–0.6 °C in JJA and MAM is evident over China and India. Large warming in the range of 1.0–1.4 °C are noted over the eastern part of the USA and the Western Europe in DJF and over the northern part of Canada in SON as well as over the central South America in JJA. Warming in the range of 1.0–1.4 °C is also noted over Greenland in DJF and MAM. These warming are all statistically significant.

Figure 12.

CCSM3 simulated surface air temperature change ( °C, Clean Scenario—base run) for (a) DJF, (b) MAM, (c) JJA, and (d) SON averaged for the last 30-model years (Year 21 through Year 50). Contours are drawn from − 1.8 to 1.8 °C with an interval of 0.4 °C. Solid and dashed lines represent positive and negative changes, respectively. Shading indicates statistical significance (p < 0.10) is achieved with the Student's t-test

6.3. Rainfall change

Figure 13 shows the CCSM3 simulated rainfall rate differences between the clean scenario and the base experiment averaged for the four seasons. As in the case for dirty scenario, these changes are statistically significant most of the time. Increased rainfall in the range of 0.1–0.3 (0.3–0.7) mm/day are noted over the southern part of China, Southeast Asia and India in DJF and SON (JJA and MAM). Warming induced destabilization of the atmosphere over these areas (Figure 12) is likely responsible for the increased local rainfall. Slightly reduced rainfall (∼− 0.1 mm/day) is indicated over the northern part of China in JJA. Increased rainfall is also noted along the storm track over the North Pacific in DJF. As will be discussed shortly, intensification of the storm track over the North Pacific is suggested in the 500-mb geopotential height changes, which might be responsible for the increased rainfall there.

Figure 13.

CCSM3 simulated rainfall rate change (mm day−1, Clean Scenario—base run) for (a) DJF, (b) MAM, (c) JJA, and (d) SON averaged for the last 30-model years (Year 21 through Year 50). Contours are drawn from − 0.9 to 0.9 mm day−1 with an interval of 0.2 mm day−1. Solid and dashed lines represent positive and negative changes, respectively. Shading indicates statistical significance (p < 0.10) is achieved with the Student's t-test

Increased rainfall is identified over the eastern part of the USA and the South America in all seasons (Figure 13). These changes may be partly related to the warming-induced destabilization effects over these regions (Figure 12). Large changes in rainfall are indicated over the tropical regions throughout the year with increases towards the north of the Equator and decreases towards the south of the Equator.

6.4. 500-mb geopotential height change

CCSM3 simulated 500-mb geopotential height differences between the clean scenario and the base experiment are shown in Figure 14. Over the North Pacific, statistically significant decreases (increases) are noted north (south) of 35°N in DJF. Such changes would help to strengthen both the Aleutian low and the subtropical high-pressure system, which would in turn intensify the mid-latitude jet stream and storm track over the North Pacific. Over the North Atlantic in DJF, statistically significant increases in the 500-mb geopotential height north of 35°N appear to suggest the weakening of the NAO under the clean scenario.

Figure 14.

CCSM3 simulated 500-mb geopotential height change (m, Clean Scenario—base run) for (a) DJF, (b) MAM, (c) JJA, and (d) SON averaged for the last 30-model years (Year 21 through Year 50). Contours are drawn from − 36 to 36 m with an interval of 8 m. Solid and dashed lines represent positive and negative changes, respectively. Shading indicates statistical significance (p < 0.10) is achieved with the Student's t-test

Increases in the 500-mb geopotential height are noted over the polar regions of both hemispheres in all seasons except for the northern part of Canada in MAM where large decreases in the 500-mb geopotential height are evident (Figure 14). Statistical significance is achieved mainly for the polar regions of the winter hemisphere. As in the case for the dirty scenario, a spurious band of statistically significant changes is noted over the tropics in all seasons (Figure 14).

7. Teleconnection patterns associated with ENSO

In both scenarios, the largest clearsky solar radiation changes are located near the source regions of the altered aerosol concentrations. However, the largest changes in surface air temperature, rainfall rate and 500-mb geopotential height are located far away from the source regions. As mentioned above, this mis-match between the aerosol direct forcing and the examined fields may imply the importance of teleconnection patterns arising from air–sea interactions. As ENSO is the dominant forcing in the tropics that is responsible for much of the extratropical climate variability, it is logical to examine the teleconnection patterns in temperature and rainfall associated with the warm and cool phases of ENSO (i.e. El Niño and La Niña), and to see to what extent the temperature and rainfall anomalies for El Niño and La Niña years are reminiscent of the canonical ENSO teleconnection patterns. We will focus on winter (DJF) and summer (JJA) seasons.

Regional effects and teleconnection associated with absorbing aerosol forcings especially black carbon aerosols have been the focus of many recent works (e.g. Menon et al., 2002; Wang, 2004, 2007; Robert and Jones, 2004; Chung and Seinfeld, 2005; Ramanathan et al., 2005; Lau and Kim, 2006; Kim et al., 2006; Ramanathan et al., 2007; Meehl et al., 2008). These works have demonstrated that the solar absorption by black carbon is able to affect rainfall patterns over much of Asia including the monsoons, cause a significant change in tropical convective precipitation away from emission centres, and excite a planetary-scale teleconnection pattern in sea level pressure, temperature, and geopotential height spanning North Africa through Eurasia to the North Pacific. We will show in this section that changes in the concentrations of black carbon, organic carbon, and sulphate aerosols combined are also able to induce teleconnection patterns resembling those of ENSO.

The Oceanic Niño Index (ONI) has been used by National Oceanic and Atmospheric Administration (NOAA) for identifying El Niño and La Niña events in the tropical Pacific from the SST fields. ONI is defined as the running 3-month mean SST anomalies for the Nino 3.4 region (i.e. 5°N-5°S, 120°W-170°W). Events are defined as five consecutive months at or above the + 0.5° anomaly for warm phase and at or below the − 0.5° anomaly for cool phase. Figure 15 shows the ONI temporal distributions for the last 30-model years (Year 21 through Year 50) based on the CCSM3 simulations for each ensemble member. Table II presents the frequency (defined as the number of occurrence) of El Niño and La Niña events that last through DJF and JJA. For the base experiment, the numbers of El Niño and La Niña events range around 5 both for DJF and JJA. For the dirty and clean scenarios, the numbers of El Niño and La Niña events range from 3 to 8 with generally fewer events in JJA when compared to DJF. Differences in ENSO frequency among the ensemble members and among the experiments are noted (Figure 15 and Table II). CCSM3 simulates frequent ENSO events, with a periodicity of 2–3 years compared to the 2.5–8 years in observations. Deser et al. (2006) speculated that the overly narrow meridional scale of ENSO in the CCSM3 simulation, which is related to the coupling of the atmosphere and ocean components, may contribute to the excessive high frequency of ENSO.

Figure 15.

CCSM3 simulated running 3-month mean SST anomaly ( °C) for the Nino 3.4 region (5°N-5°S, 120°-170°W) between Year 21 and Year 50 for (a) base run member 1, (b) base run member 2, (c) base run member 3, (d) dirty scenario member 1, (e) dirty scenario member 2, (f) dirty scenario member 3, (g) Clean Scenario member 1, (h) Clean Scenario member 2, and (i) Clean Scenario member 3. Black and grey shadings represent positive and negative values, respectively. Dashed lines indicate the ± 0.5 °C reference lines

Table II. Frequency of CCSM3 simulated El Niño and La Niña events for the last 30-model year (Year 21 through Year 50) based on ONI index
 El Niño eventsLa Niña events
 DJFJJADJFJJA
BaseMember 15555
 Member 26555
 Member 36746
Dirty ScenarioMember 15675
 Member 27778
 Member 37484
Clean ScenarioMember 18475
 Member 25343
 Member 35355

Figure 16 shows the CCSM3 simulated DJF composite surface air temperature anomalies for El Niño and La Niña years. The anomalies are nearly opposite in sign between El Niño and La Niña years. Noticeable features for El Niño years that are consistent among the base experiment and the two scenarios include: (1) warming over the tropical Eastern Pacific Ocean and the northern South America (statistically significant); (2) warming over the eastern part of Russia, Alaska, and the Pacific Northwest; (3) warming over East Asia; (4) warming over Northern Australia; (5) cooling over New Zealand and the surrounding waters; (6) cooling over Southwest USA; and (7) cooling over the northeastern part of Canada. For La Niña years, the opposite of these features apply except for (1) Alaska where warming is identified in the base experiment and (2) the eastern part of Russia where warming is noted under the dirty scenario. These anomalous patterns in temperature are broadly consistent with the canonical teleconnection patterns associated with ENSO (Philander, 1989; Glantz, 1996). The anomalies in surface air temperature for El Niño and La Niña years in JJA (not shown) closely resemble those in DJF over the oceanic areas; however, over the land areas large differences are noted between DJF and JJA especially under the dirty and clean scenarios.

Figure 16.

CCSM3 simulated DJF mean surface air temperature anomaly ( °C) for (a) base run El Niño years, (b) base run La Niña years, (c) dirty scenario El Niño years, (d) dirty scenario La Niña years, (e) Clean Scenario El Niño years, and (f) Clean Scenario La Niña years. Contours are drawn from − 1.8 to 1.8 °C with an interval of 0.4 °C. Solid and dashed lines represent positive and negative changes, respectively. Shading indicates statistical significance (p < 0.10) is achieved with the Student's t-test

Rainfall anomalies associated with the El Niño and La Niña years are presented in Figure 17. Compared to the temperature anomalies, rainfall anomalies are primarily concentrated over the tropical regions. Noticeable features of rainfall anomalies for El Niño years that are statistically significant include: (1) wet anomalies over the equatorial Pacific Ocean with dry anomalies on both sides; (2) dry anomalies over the Maritime Continent and Northern Australia; (3) wet anomalies over the northeastern pacific; and (4) wet anomalies over New Zealand and the surrounding waters. The opposite of these features apply for La Niña years. The anomalies in rainfall for El Niño and La Niña years in JJA (not shown) also resemble those in DJF except for East Asia and Southern USA where large differences are noted between DJF and JJA.

Figure 17.

CCSM3 simulated DJF mean rainfall rate anomaly (mm day−1) for (a) base run El Niño years, (b) base run La Niña years, (c) dirty scenario El Niño years, (d) dirty scenario La Niña years, (e) Clean Scenario El Niño years, and (f) Clean Scenario La Niña years. Contours are drawn from − 2.0 to 2.0 mm day−1 with an interval of 0.5 mm day−1. Solid and dashed lines represent positive and negative changes, respectively. Shading indicates statistical significance (p < 0.10) is achieved with the Student's t-test

Figure 18 shows the DJF and JJA mean surface air temperature differences between the scenarios and the base experiment for El Niño and La Niña years combined. The differences are computed in three steps. First, the differences between the scenarios and the base experiment for El Niño years are obtained for each ensemble member and then averaged over the three ensemble members. Next, the above step is repeated for La Niña years. Thirdly, the differences from step one and step two are added together and these differences represent the changes in association with ENSO. Striking similarities are noted between these changes and the surface air temperature changes discussed under each scenario (sections 5 and 6), especially for regions that are far away from the aerosol concentrated areas. For the dirty scenario these include: (1) warming over USA and Greenland in DJF (Figures 8(a) and 18(a)); (2) cooling over the northern parts of Canada and Russia in DJF (Figures 8(a) and 18(a)); and (3) warming over the high latitudes of the Southern Hemisphere in JJA (Figures 8(c) and 18(c)). For the clean scenario these include: (1) warming over USA, Greenland and Western Europe in DJF (Figures 12(a) and 18(b)); and (2) warming over the eastern part of USA, the Mediterranean, the central South America, and the western part of Antarctic in JJA (Figures 12(c) and 18(d)).

Figure 18.

CCSM3 simulated surface air temperature differences ( °C) between dirty scenario and base run for El Niño and La Niña years combined for (a) DJF, (c) JJA, and between Clean Scenario and base run for El Niño and La Niña years combined for (b) DJF, (d) JJA. Contours are drawn from − 3.0 to 3.0 °C with an interval of 0.6 °C. Solid and dashed lines represent positive and negative changes, respectively

Figure 19 shows the DJF and JJA mean rainfall rate differences between the scenarios and the base experiment for El Niño and La Niña years combined. These differences match the rainfall changes discussed under each scenario rather well (i.e. Figure 9(a) and (c) for dirty scenario; Figure 13(a) and (c) for clean scenario). Similar arguments can also be applied to the 500-mb geopotential height differences (not shown). These analyses strongly suggest that teleconnection patterns associated with ENSO and possibly with land-surface processes that we have not examined here, play more important roles in causing the large changes in the examined fields far away from the source regions of the altered aerosol concentrations. Thus, local changes in aerosol emissions and concentrations can affect not only the source regions but also places that are far away.

Figure 19.

CCSM3 simulated rainfall rate differences (mm day−1) between dirty scenario and base run for El Niño and La Niña years combined for (a) DJF, (c) JJA, and between Clean Scenario and base run for El Niño and La Niña years combined for (b) DJF, (d) JJA. Contours are drawn from − 2.0 to 2.0 mm day−1 with an interval of 0.5 mm day−1. Solid and dashed lines represent positive and negative changes, respectively

8. Conclusions and discussion

We have examined the impacts of two largely contrasting aerosol emissions scenarios on regional climate using CCSM3: increasing the anthropogenic aerosols over China and India by a factor of three (the dirty scenario) and reducing the global anthropogenic aerosols by a factor of 10 (the clean scenario). Dynamic footprints of the increased Asian emissions are incorporated based on the MOZART simulations.

Increasing Asian aerosol emissions would result in cooling and reduction of precipitation over China and India. Large warming over the USA and southern Canada in DJF and cooling in JJA are identified under this dirty scenario. Warming is also noted over Greenland in DJF. Large changes in rainfall are indicated over the tropical regions.

Increasing Asian aerosols emissions would also bring large changes in the 500-mb geopotential height. Deepening of the Aleutian low in DJF would help to strengthen the North Pacific storm track with anomalous southwesterly flow and increased rainfall over the western part of the USA. Weakening of the Icelandic low and the subtropical high-pressure system over the North Atlantic in DJF would result in anomalous southeasterly flow over the eastern part of the USA and would possibly bring mild winter to the Western Europe. Large increases in the 500-mb geopotential height are noted over the Antarctic region in JJA.

Reducing the global anthropogenic aerosol emissions by a factor of 10 would significantly warm the atmosphere especially over the polluted land areas of both hemispheres. Increases in rainfall over the polluted land areas are also identified under this clean scenario. Changes in the 500-mb geopotential height under the clean scenario suggest strengthening of the North Pacific storm track and weakening of NAO in DJF under the clean scenario. The polar regions of both hemispheres are subject to large changes in the 500-mb geopotential height.

The largest changes in surface air temperature, rainfall rate and 500-mb geopotential height are located far away from the source regions of the altered aerosol concentrations where the largest clearsky solar radiation changes are identified. Our analyses show that teleconnection patterns associated with ENSO play important roles in causing the large changes in the examined fields far away from the source regions.

Our analyses show that simultaneously tripling the anthropogenic aerosols over China and India cause a small decrease (∼− 0.3 W m−2) in globally mean net shortwave forcing at TOA with negligible cooling effect (∼− 0.08 °C) on the globally mean surface air temperature and negligible reduction in globally mean precipitation (∼− 0.3%) (Table III). However, the far-reaching effects of the increased Asian aerosols on regional climate in and outside of the source regions are non-trivial.

Table III. CCSM3 simulated global mean changes in net shortwave forcing (W m−2) at the TOA, surface air temperature ( °C) and precipitation (%) between the scenarios and the base run averaged over the last 30-model years (Year 21 through Year 50)
 ΔForcing at TOAa (W m−2)ΔSurface air T ( °C)ΔPrecipitation (%)
  • a

    Computed as the changes in net shortwave radiation flux derived during the simulation

Dirty—Base− 0.3− 0.08− 0.3
Clean—Base0.50.31.7

Our analyses also show that a global reduction of anthropogenic aerosols by a factor of 10 would cause an increase in globally mean net shortwave forcing at TOA by 0.5 W m−2 with global warming of 0.3 °C and an increase in globally mean precipitation by 1.7% (Table III), with larger effects in regional climate and environment. Such strong influence has to be taken into account when designing strategies to reduce air pollutions while maintaining global warming below a specified threshold. Future emissions changes need to be carefully and systematically investigated in order to prevent dangerous anthropogenic interference with the climate system.

Acknowledgements

This work (LA-UR-07-3930) was supported by the Los Alamos National Laboratory through the Laboratory Directed Research Development (LDRD) Program (Project Number: LDRD200500014DR; PI: Dr. Manvendra K. Dubey). Support was also provided to Y. Zhang by the National Natural Science Foundation of China through the 973 Program (Grant Number: 2006CB403705; PI: Prof. Jinhai He). Three anonymous reviewers are acknowledged for their constructive comments and suggestions for improving the manuscript. The model simulations were performed at the NCAR Computational and Information System Laboratory (CISL). NCAR is sponsored by the National Science Foundation.

Ancillary