1. Introduction
To understand current and future climate change, it is crucial to understand past changes. During the past century, short-lived climate forcers such as aerosols and ozone have impacted climate, however, their effects are challenging to simulate in climate models. These pollutants are more dynamic than long-lived climate forcers such as CO2, with emissions increasing during industrialization and decreasing from pollution emission controls and with these changes distributed heterogeneously both spatially and temporally. In this study, we describe a set of coupled atmosphere and deep-ocean climate experiments using online and interactive aerosol–ozone chemistry, including direct radiative effects and the effect of black carbon (BC) on snow albedo.
Aerosols and ozone affect climate “directly” by scattering and absorbing radiation. Aerosols primarily scatter shortwave (SW) radiation; however, dark particles such as BC are also absorbing. Dust absorbs both short- and longwave (LW) radiation but is mostly scattering. Ozone absorbs both long- and shortwave radiation. Deposition of dark particles such as BC on snow reduces the surface albedo and promotes snowmelt, called the BC-snow-albedo effect. This effect is highly uncertain and very difficult to constrain with measurements; recent model-based radiative forcing estimates range from 0.01 to 0.2 W m−2 (Koch et al. 2009a). Our experiments also include the “semidirect” effects of aerosols on clouds or the changes in clouds that result from aerosols altering the thermal structure of the atmosphere; however, the aerosol microphysical effects on clouds, or the “indirect effects” on cloud albedo and lifetime, are not included.
One challenge for modeling short-lived species during the twentieth century is obtaining good quality historical emissions estimates. Pollution levels during industrialization have generally increased. In regions such as North America and Europe, subsequent air quality legislation reduced pollution. Meanwhile, technology and pollution controls have evolved in poorly documented ways, making accurate estimation of emissions difficult. Particularly challenging is estimating emissions from open biomass burning during the century, as we will discuss further below. Although emissions estimates are uncertain, observational records from ice and lake cores are increasingly available to help constrain historical pollution changes.
Our experiments include coupling among aerosols, ozone chemistry, and climate and between atmosphere and ocean. These “real-time” interactions enable feedbacks within the climate system. The sulfur oxidation to form sulfate depends upon the oxidants provided by the chemistry. Nitrate and sulfate can form on dust and this coating renders the dust hygroscopic. This also reduces the number of radiatively active sulfate and nitrate particles. The aerosol and ozone are carried “online” in the model, so that transport, removal, and radiative effects respond instantaneously to the climate. Until recently, running a model with interactive aerosols was too computationally intensive. However, some recent studies have included full interaction in transient twentieth-century simulations (e.g., Stier et al. 2006; Roeckner et al. 2006; Nagashima et al. 2006; Rotstayn et al. 2007); the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4) simulations of the Hadley Centre Global Environmental Model version 1 (HadGEM1) and Model for Interdisciplinary Research on Climate (MIROC) models also included interactive aerosols.
Our climate experiments are run in transient mode, so that the impacts of the changing pollution levels can be studied during the century. Coupling the atmosphere to a dynamic deep ocean allows time-dependent ocean heat uptake. In a previous study (Koch et al. 2009a), we considered aerosol impacts on climate over the century by contrasting equilibrium Qflux simulations for the late 1890s with the 1990s. These experiments were run to equilibrium for each simulation, allowing the climate to reach a steady temperature state, which therefore had an exaggerated climate response to the aerosol effects. The Qflux mixed layer experiment permits ocean response to atmospheric temperature change by adjusting energetically rather than including the full dynamics of a deep ocean. By simulating transient climate with a deep and dynamic ocean, we have the greatest chance of capturing realistic interactions between short-lived species and climate throughout the century. Furthermore, the climate system is not in equilibrium during the late twentieth century, with ocean heat uptake moderating temperature change and feedbacks from the cryosphere occurring incrementally; this disequilibrium can only be captured in a transient experiment.
The paper begins with the model description (section 2), changes in emissions, concentrations, and loads (section 3), radiative flux changes at the top-of-atmosphere (TOA) and at the surface (section 4), and climate responses including temperature, snow and ice cover, clouds, and precipitation (section 5). We then present results from pollution reduction sensitivity studies for the end of the century (section 6), followed by the discussion (section 7).
2. Model description
We use the Goddard Institute for Space Studies (GISS) GCM Model E (Schmidt et al. 2006) with 20-level vertical resolution up to 0.1 hPa, and 4° × 5° horizontal resolution. With the exception of the aerosol and ozone components, the climate model is very similar to the version used for the twentieth-century climate simulations of Hansen et al. (2007) and submitted to AR4, with the atmosphere coupled to the deep-ocean model of Russell et al. (1995). We use the same forcings from long-lived greenhouse gases (GHGs), volcanoes, solar changes, and vegetation changes. The model resolution is coarse compared to many contemporary models; however, this model version was documented, was used in previous climate simulations, and was fast enough to permit carrying many aerosol and chemical species. However, the resolution limits the model’s ability to capture some smaller-scale features.
Our model version includes aerosols and ozone chemistry run online with climate feedbacks and with an interactive scheme for the effect of BC on snow albedo. The chemistry simulation includes both tropospheric and stratospheric ozone and oxidant photochemistry (Shindell et al. 2006). Aerosols include sulfate and sea salt (Koch et al. 2006), carbonaceous aerosols (Koch et al. 2007), dust (Miller et al. 2006), nitrate (Bauer et al. 2007), and heterogeneous interaction of nitrate and sulfate with dust (Bauer and Koch 2005). Chemistry and aerosols are coupled together, so that oxidant changes affect sulfate oxidation and the aerosols affect photochemistry (Bell et al. 2005). Our sulfur chemistry includes aqueous oxidation of SO2 by H2O2 but not ozone; ozone oxidation would be about 6% of the SO2 sink according to Berglen et al. (2004). The species are also coupled to the model climate, including, for example, uptake and rainout of soluble species and production of sulfate within clouds. Removal of species by dry deposition uses a resistance-in-series scheme (e.g., Koch et al. 1999; Wang et al. 1998) as well as aerosol gravitational settling. Aerosols and ozone affect long- and shortwave radiative fluxes and aerosol semidirect effects on clouds are included.
Our BC-snow-albedo effect scheme (Koch et al. 2009a) reduces snow albedo depending on the BC snow concentration. The albedo reduction also depends on snow grain size, which is parameterized to depend on snow age and surface temperature. Black carbon affects snow albedo only, so that it may influence sea ice and land ice (glaciers) only by changing the albedo of the snow deposited on the ice. The GISS model snow has up to three levels, and the snow fraction depends on snow amount and topography. Snow in forests is located between the canopy and the ground unless the snow depth surpasses the canopy height. Snow aging also affects its albedo and depends on its wetness. Schmidt et al. (2006) has a detailed description of the model cryosphere.
Anthropogenic historic 1890–1980 gaseous emissions (for SO2, NH3, NOx, NMVOCs, CO, and N2O) are from the Emission Database for Global Atmospheric Research (EDGAR)–History Database of the Global Environment (HYDE) emissions (Van Aardenne et al. 2001); for 1990 we used EDGAR32 and for 2000 we use EDGAR 32FT2000 emissions (http://www.mnp.nl/edgar/model). However, the EDGAR–HYDE and EDGAR32 emissions are not fully compatible, which causes some discontinuity between 1980 and 1990. Historic carbonaceous anthropogenic aerosol emissions for all years are from Bond et al. (2007). Compared with previous estimates (e.g., Novakov et al. 2003; Ito and Penner 2005), the Bond et al. (2007) carbonaceous aerosol emissions increase more gradually during the century, especially after 1950, which is due to the consideration of technology improvements. The BC emissions are about double that of Ito and Penner (2005) early in the century but are similar by the end of the century; they are similar to Novakov et al. (2003) for fossil fuel early in the century but smaller by a factor of two by the end of the century. Compared with previous inventories, Bond et al. (2007) also estimates substantial biofuel carbonaceous aerosol emissions at the beginning of the century. For all anthropogenic emissions, we used decadal average emissions and interpolated for intermediate years. Natural emissions are described in Shindell et al. (2006), Miller et al. (2006), Koch et al. (2006, 2007), and Bauer et al. (2007). Steady volcanic SO2 emissions are included in the sulfur sources, however, explosive volcanism is applied as an external forcing as in Hansen et al. (2007). Year 2000 biomass burning emissions are from the Global Fire Emissions Database (GFED) v3 averaged from 1997 to 2002 (van der Werf et al. 2003, 2004). For the historic biomass burning we assumed the year 1890 source is 50% of year 2000 in tropical and African burning regions and increases linearly to year 2000; other regions are assumed to have a constant (year 2000) source during the century. This assumption for biomass burning is similar to that used in Hansen et al. (2007) and is probably more appropriate than the assumption that preindustrial burning is only 10% of present day as used in some previous studies (e.g., Stier et al. 2006) since substantial burning and land clearing occurred during the nineteenth century (e.g., Mouillot and Field 2005). For most species our emissions are similar to those in the new Coupled Model Intercomparison Project phase 5 (CMIP5) transient experiments (Lamarque et al. 2010). One exception is the SO2 anthropogenic emissions. Global CMIP5 SO2 emissions peak at about 60 Tg S y−1 in 1980 and then decline 30%; ours peak at about 75 Tg S y−1 in 1990 and decline less than 10%. Also, although our global biomass burning emission change from the late 1800s to the late 1900s is similar to CMIP5, the trend is quite different. The CMIP5 biomass burning source decreases from 1900 to the 1950s followed by a steep rise. Also, the CMIP5 global BC biomass burning emissions at the end of the century are about 25% smaller than ours.
Our preindustrial model spinup was initiated from a well-equilibrated long (thousands of years) 1850 deep-ocean-coupled run from the Hansen et al. (2007) simulations that had used offline aerosols and ozone. After turning on our coupled species with 1890 emissions, we ran for about another 100 years to equilibrium. This spinup control simulation for the standard simulation (STD-CON; see Table 1 for list of simulations) was fairly stable, with an average imbalance of −0.001 W m−2 yr−1 and average temperature drift of 0.0009°C yr−1 for the 25 years following the transient ensemble branches. A doubled CO2 simulation was performed for the spinup (STD-2CO2). Ensembles of three or five experiments were performed for each transient case. The standard simulation (STD) included all species and effects. Simulation BCALB is like STD but did not include the BC-albedo effect and was initiated from the STD spinup. Most results presented below are the mean of the ensembles.
Climate experiment definitions.
3. Emissions and loads
Figure 1 shows global, decadal mean trends for the short-lived, radiatively important species and some of their precursors. Sulfur dioxide, NOx, and CO emissions rise steeply beginning midcentury. Black carbon emissions increase earlier in the century and BC and organic carbon (OC) enhancements during the century are generally smaller because of the substantial biofuel and biomass burning sources already at the beginning of the century. Decadal mean dust and sea salt change slightly, mostly from changes in model winds; however, their interannual variability can be large (section 5, Fig. 10). The species emissions, with natural fraction in parentheses, and burden values at the beginning and end of the century are in Table 2.
Decadal and ensemble mean for global emissions and STD burdens (normalized to 1890 levels); clear-sky aerosol optical depths (×100); radiative forcing at top-of-atmosphere and surface relative to 1890 (W m−2 × 100).
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
Standard simulations and sensitivity simulations: emissions, burdens, aeroseol optical depth, and radiative flux changes.*
The relative increase in burden is greatest for BC, owing to its increased lifetime (Table 2, burden/emissions or B/E is the lifetime). BC has 20% larger lifetime at the end of the century because of increased emissions in regions where rainout is less efficient. Although tropospheric ozone increases throughout the century, the ozone burden in Fig. 1 is dominated by stratospheric ozone. Stratospheric ozone increases early in the century due to cooling stratospheric temperatures, however, later in the century the ozone declines in polar regions where ozone holes develop. The sulfate lifetime (based on SO2 emission) decreases by 40% as emissions increase because its oxidants also increase. For example, the burden of in-cloud aqueous oxidant H2O2 increases by a factor of 1.8 (Fig. 1).
The dust lifetime decreases 6% because of the increased coating by sulfate and nitrate and therefore increased hygroscopicity. The change in dust burden is in Fig. 2, along with coating of sulfate and nitrate on dust. During the century, the pure dust decreases slightly, especially as the sulfate coating increases.
Global burdens of dust, sulfate, nitrate, sulfate coating dust, and nitrate coating dust during the century (Tg).
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
Over the century, the aerosol optical depth (AOD) increases only about 30% since there is a substantial background natural aerosol component, especially sea salt and dust (Fig. 1). The AOD increase is dominated by sulfate beginning in the 1960s. Changes in AOD over the century and for segments of the century are shown in Fig. 3. Over the century, the global AOD increases by 0.03, and 60% of this occurs during the 1940s–80s. In the Arctic, the AOD increases by 0.02, with 70% of this occurring during the 1940s–80s. From the 1980s to the 1990s the AOD declines in some regions, such as in parts of Europe. The model simulates increased Southern Ocean sea salt during the second-half of the century, discussed further in section 5. Tomasi et al. (2007) described observations of declines in background (summertime) Arctic AOD during the final three decades of the century. The model does not have decreased Arctic AOD, although the change is small at the end of the century.
Changes in clear-sky aerosol optical depth (×100) from the 1890s to the 1990s, 1890s to 1940s, 1940s to 1980s, and 1980s to 1990s, where decadal means from the STD simulation are used to calculate the changes. Global (Arctic) means are given above each panel.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
Observational constraints on historical trends of species
Ice and lake core records of species provide some regional constraint on the model species during the century. These comparisons have several difficulties. First, the model precipitation, and therefore the species accumulation rate, does not always match the observed. For the ice cores where we have accumulation rates, we adjust the model snow concentration by the ratio of model to observed accumulation averaged over the period (since the variation over the record is smaller than the disparity between model and observed accumulation). Second, if the core is at high altitude and the model topography for the grid box is lower than the core altitude (e.g., in the Alps), and if there are large vertical gradients in species concentrations, model errors result. Finally, particularly for the remote core sites, model errors can result from errors in transport as well as removal and emissions. The observational records are therefore more useful to test the model species trends rather than magnitudes.
Figure 4 compares model BC and sulfur snow concentrations with Greenland ice core records at D4 (McConnell et al. 2007), ACT2 (McConnell and Edwards 2008), Summit, and Humboldt (McConnell 2010). The Greenland ice cores have peak BC concentrations at around 1910 when U.S. BC pollution was maximum (McConnell et al. 2007). This peak is strongest in the south of Greenland and is less pronounced in the north. The model captures the trend, although the peak is not as high as observed. The model overestimates BC at the end of the century at ACT2 and Summit, which is possibly due to the excessive contribution from Eurasian sources that became more important toward the end of the century as discussed below. The observed sulfur has a peak around 1910–20 and again around 1970. These also match likely maxima in North American pollution, from coal early in the century and from gasoline and residual oil burning prior to sulfur controls later in the century. The model lacks the earlier peak, and the later peak occurs too late with insufficient decline at the end of the century. Generally the model underestimates Greenland sulfur levels, especially at D4; since the model simulates present-day sulfate reasonably well in the Arctic (both in these cores and as shown in Koch et al. 2006), the underestimation earlier in the century may be due to emission deficiencies. Using the emissions of Smith et al. (2004) might improve model performance in terms of both trend and magnitude since globally these peak in the early 1970s.
Greenland ice core (left) BC and (right) sulfur snow concentrations observed (black) and modeled (red). Model snow concentrations are adjusted by the ratio of model to observed accumulation rate (using the mean accumulation value during the century in each case). (bottom) Locations from north to south are Humboldt, Summit, D4, and ACT2 cores.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
Figure 5 compares sulfur deposition observations from the National Acid Deposition Program (NADP; National Acid Deposition Program 2008) for 1980–2000 within four regions of North America and one site in Alaska and sulfate air concentrations from 1980 to 2000 from the European Monitoring and Evaluation Programme (EMEP). In the U.S. observations (solid curves), the sulfur deposition was much larger in eastern North America and declined from 1980 to 2000 by nearly a factor of 2. In the west and in Alaska, the sulfur deposition was much smaller and had smaller decline. The model captures these magnitudes and trends extremely well. However, the data begin around 1980 so we cannot test the model prior to that time to see whether the observed deposition peaked earlier, as indicated by the Greenland cores. In Europe, the observed sulfate concentrations are generally larger in the south and east. The model simulates magnitudes and trends reasonably well from the late 1980s to the end of century; however, it underestimates sulfate in the early 1980s.
Model (dashed) and observed (solid) sulfur deposition (kg S ha−1) in North America from the NADP and sulfate air concentration (μg m−3) from EMEP. Each dataset is divided into 4 regions and Alaska. 144 NADP and 47 EMEP sites were used, first by averaging the data over model gridboxes, and then averaging within regions as shown on the map. The corresponding model gridbox results for the STD simulation were also averaged by region.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
Figure 6 compares BC, OC, and sulfur with midlatitude ice and lake core records. In the United States, BC air concentrations were derived from an Adirondack lake core (Husain et al. 2008). As in the Greenland ice cores, the BC is observed to peak in the 1910s. The model matches this peak; however, it underestimates BC for the end of the century. Husain et al. (2008) show a similar discrepancy when comparing the BC trends with fossil fuel emissions estimates of Novakov et al. (2003). The remaining panels in Fig. 6 are from Alpine ice core records. Model BC and OC are compared with decadal averages from Lavanchy et al. (1999). The model agrees reasonably well with the BC record, although it does not increase as much as observed for the first-half of the century. The model OC trend is almost flat while the observed shows a doubling from 1900 to 1970. On the right side of Fig. 6, model Alpine sulfur, reduced by a factor of 10, is compared with observed trends at Fiescherhorn (Schwikowski et al. 1999) and Col du Dome (Preunkert et al. 2001). The model sulfur in the Alpine gridboxes is much larger than observed because of the lower mean topography in the model gridbox. The ratio of the air concentrations between the model surface altitude and the Feischerhorn glacier altitude for SO2 and sulfate are 19 and 7, so a factor of 10 overestimate in deposition is consistent with these gradients. Overall the observed trend, with peak at 1980, is matched quite well by the model, especially for the Col Du Dome record.
(top left) Comparison of BC air concentrations derived from lake core sediments (black, with standard deviations) and from the model (two ensemble members). Other panels are model (red) and observed (black) snow concentrations for (left) BC and OC and (right) sulfur in Alpine glaciers. Model snow concentrations are adjusted by the ratio of model to observed accumulation rate (using the mean accumulation value during the century in each case) for all but Colle Gnifetti. The sulfur model results were reduced by a factor of 10.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
Recently, there has been concern about impacts of BC on snow albedo in the Arctic and whether that has contributed to melting of Arctic sea ice and snow. Some studies have focused on changes in Arctic BC since the 1980s when measurements were first made. Sharma et al. (2004) found a 60% decrease in atmospheric BC at Alert between 1989 and 2002. Recent Arctic snow measurements (e.g., Grenfell et al. 2009; Hegg et al. 2009) found BC concentrations to be about 5–15 ng g−1 in Canada, Alaska, and the Arctic Ocean, about a factor of 2 lower than measured in the 1980s (e.g., Clarke and Noone 1985). Contemporary Russian measurements are larger than the western Arctic, ranging from about 15–80 ng g−1, while BC concentrations in the Barants and Kara seas were measured at about 15–25 ng g−1 (Grenfell et al. 2009; Hegg et al. 2009). The Greenland ice sheet has relatively very low BC levels, about 2–3 ng g−1, similar to the measurements in the 1980s (Grenfell et al. 2009).
Figure 7 shows model BC deposition in the Arctic during the 1890s, 1920s, 1950s, and 1990s. Arctic model BC deposition peaked in the 1950s and has declined since. The decline is especially evident in northern Europe, where model concentrations have dropped by a factor of 2, consistent with the observed Svalbard record (Hicks and Isaksson 2006). The declines in Greenland and Alert are not as large as observed. The BC deposition distribution for the 1990s agrees well with the contemporary observations described above. Model BC snow concentrations at the end of the century in northern Asia are typically larger than those in Greenland and Alaska by a factor of 2–3, also consistent with observations (Hegg et al. 2009). The model Arctic mean value (shown in brackets above each map) is almost as low in the 1990s (13 ng g−1) as in the 1890s (12 ng g−1). In parts of northern Canada and western Europe, the concentrations are less in the 1990s than the 1890s, while in northern Asia the concentrations are larger.
Model BC deposition normalized by precipitation (ng g−1) in the 1890s, 1920s, 1950s, and 1990s, with global (Arctic) mean values above each map. (bottom right) Deposition during the century for five locations.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
4. Direct radiative forcing changes
a. Top of atmosphere
The net TOA radiative flux change over the century is positive from BC (+0.23 W m−2) and ozone (+0.21 W m−2) and negative from sulfate (−0.39 W m−2), nitrate (−0.12 W m−2), and organic carbon (OC) (−0.10 W m−2) (Table 2, Fig. 1). The net ozone forcing is at the low end of the AR4 Third Assesment Report (TAR) estimate of +0.35 ± 0.15 W m−2 (Forster et al. 2007), mostly owing to the strong negative forcing from excessive stratospheric ozone loss. Tropospheric ozone increases late in the century due to increasing NOx and CO pollution. However, the polar TOA radiative flux change due to ozone changes is dominated by stratospheric ozone loss beginning in the 1970s, with excessive Northern Hemisphere (NH) ozone loss. The sea salt negative forcing (−0.07 W m−2) is due to increased winds beginning in the 1970s in the southern oceans, as discussed below. Modest positive global mean forcing occurs for dust due to a slight decrease in dust load (+0.04 W m−2). The BC-albedo effect also has a small positive forcing over the century (+0.02 W m−2). The net TOA forcing from aerosols is −0.40 W m−2 while the surface forcing is −1.20 W m−2 because both scattering and absorbing aerosols have negative surface forcing. The net TOA forcing over the century, including all direct and BC-albedo effects, is −0.17 W m−2 and −0.40 W m−2 from aerosol direct effects, +0.21 W m−2 from ozone, and +0.02 W m−2 from BC-albedo effects.
Figure 8 shows maps of the radiative flux changes at the TOA for all aerosols, sulfate, BC, the BC-albedo effect, and net ozone (SW and LW) over the century and for beginning, middle, and end of the century. From the 1890s to the 1940s, both BC and sulfate effects increase, with BC atmospheric and snow forcing dominating and causing positive forcing change at high latitudes but sulfate dominating with negative forcing change over the United States and Europe. Net ozone flux increases due to tropospheric ozone production. From the 1940s to the 1970s, the Arctic aerosol flux change is very negative (−0.19 W m−2) because of both increased sulfate and decreased BC from the eastern United States and Europe. The BC-albedo effect change is also negative in the Arctic over the middle of the century. From the 1970s to the 1990s, the forcing strengths of sulfate and BC decrease over the eastern United States and parts of Europe; the sulfate decrease in Europe is greater during the final two decades (1980s–90s) since it reaches peak levels in the 1980s (not shown). The simulated net Arctic aerosol TOA flux change from the 1940s to the end of the century is negative, so that aerosol changes have overall contributed to Arctic cooling during that time. During this time, tropospheric ozone increases and stratospheric ozone decreases in polar regions. The increase in negative aerosol forcing at high latitudes of the Southern Hemisphere at the end of the century (Fig. 8, top) is due to the increased sea salt load as discussed above. The wind speed enhancements may result from an enhanced southern annular mode due to the formation of the ozone hole and/or increased GHG concentrations (e.g., Lenton et al. 2009).
Change in TOA radiative flux for (top) net aerosol direct effect, (second row) sulfate, (third row) atmospheric BC, (fourth row) the BC-snow-albedo effect, (fifth row) and ozone for decadal means of the (first column) 1890s–1990s, (second column) 1890s–1940s, (third column) 1940s–70s, and (last column) 1970s–90s. Units are 100 × W m−2.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
b. Surface radiation
There has been recent interest in whether aerosols are responsible for changes in surface radiation during the last half of the twentieth century, as measured by networks of surface radiation instruments such as the Baseline Surface Radiation Network. The observations indicate a general reduction in surface radiation from the 1950s to the 1980s (dimming), followed by an increase in surface radiation in many places after the 1980s (brightening) (e.g., Wild 2009a). The trends typically occurred for both all-sky and clear-sky conditions, suggesting that the radiation changes may be caused by aerosol load changes. The IPCC AR4 models underestimated these trends and possibly correspondingly tend to overestimate warming over land from the 1950s to the 1980s (Wild 2009a). Similarly, the diurnal temperature range was observed to decrease owing to the decreased maximum temperature during the 1950s–80s dimming period, a trend also missing in the models.
Figure 9 shows modeled changes in surface radiation for all-sky and clear-sky conditions. The strongest dimming occurs during the 1950s–80s, consistent with observations. The dimming is also generally greatest over land and close to pollution sources. Over global mean land areas, the model has a decrease in surface radiation of −1.0 W m−2 for all sky and −2.0 W m−2 for clear sky over the century, mostly since the 1950s. As in previous studies, the modeled dimming is much weaker than observed, which ranged from −1.5 to −5.1 W m−2 decade−1 since the 1950s, based on 140–400 global land sites (as summarized in Wild 2009b).
Surface radiation changes for (top) all sky and (bottom) clear sky. Changes for decadal means are given, from (first column) 1890s to 1990s, (second column) 1890s to 1950s, (third column) 1950s to 1980s, and (last column) 1980s to 1990s.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
Dimming and brightening trends differ by region (Wild 2009b). In the model, Europe and North America dimmed up to the 1980s then brightened in some parts of northern Eurasia and North America. Strongest dimming, both in the model and observations, occurred in Asia. The modeled Southeast Asian dimming leveled off near the end of century but did not brighten (reverse), generally consistent with observations. The model Arctic brightened over the century for all-sky conditions but varied with region and time for clear-sky conditions. Observed Arctic trends indicate dimming of −3.6 W m−2 decade−1 from the 1950s to the 1990s. Such dimming occurred in the model but only in some parts of the Arctic.
Overall, the pattern and decadal variability of global dimming/brightening is consistent with observations but is much weaker than observed. The all-sky dimming is stronger than clear sky, consistent with an important role for aerosols. One way to get stronger dimming without excessive cooling would be to have more black carbon or more absorbing black carbon. Our model’s present-day simulation underestimates aerosol absorption optical depth by about a factor of 2 (Koch et al. 2009b). Doubled BC absorption would significantly reduce the surface radiation; however, such doubling might not improve model agreement with the observed dimming/brightening patterns during the last half of the century. Including aerosol indirect effects would amplify the dimming/brightening changes for all-sky conditions.
5. Climate responses during the century
a. Temperature changes
Figure 10 shows the annual average TOA radiative forcing changes for long-lived greenhouse gases, stratospheric water vapor, solar, stratospheric volcanic aerosols, and the short-lived species. This figure shows interannual variability of short-lived species resulting from their interaction with climate, especially for sea salt and dust that vary due to wind speed–driven source changes and for stratospheric ozone that is affected by stratospheric temperature changes after volcanic eruptions. The lower panel of Fig. 10 shows global mean surface air temperature (SAT) changes for the STD simulation, the BCALB simulation, and observations. In general, the simulations begin and end the century at approximately the correct global mean SAT and are too warm midcentury. Note that the SAT observations have some missing data regions early in the century (shown on the maps in Fig. 11) and these are replaced with zonal mean values where available, while the model is a complete global mean. Because missing data values are mostly polar and near the beginning of the century, the data may be biased high early in the century.
(top) Top-of-atmosphere (annual mean) model radiative flux change relative to 1880 due to various short-lived species forcings); (middle) net aerosol direct, BC-albedo, ozone, volcanic aerosol, long-lived greenhouse gases, and net of all forcings (W m−2 × 100). GHG = greenhouse gases, volcanos = stratospheric aerosol forcing from volcanic eruptions, strat H2O = stratospheric water vapor, land = land-use changes. (bottom) Global mean surface air temperature trends for observed and model simulations. The average of the STD ensemble simulations is shown in bold red. BCALB is like STD but without BC-albedo. Observations are from the GISS temperature analysis available online at http://data.giss.nasa.gov/gistemp/.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
Surface air temperature changes, (top) observed from GISS surface air temperature analysis (white is missing data), (middle) in STD model, and (bottom) due to the BC-albedo effect (STD minus BCALB). SAT changes for decadal means are given from (first column) 1890s to 1990s, (second column) 1890s to 1940s, (third column) 1940s to 1970s, and (last column) 1970s to 1990s. Global (Arctic) mean values are above each panel. For the model changes, heavy and light contours show 90% and 95% for STD and 75% and 90% significance for the BC-albedo effect, according to a Student’s t test.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
The climate sensitivity, or temperature change of the control spinup (1890) simulations from doubled CO2, was calculated using the STD-CON and STD-2CO2 runs. The latter was run in the same configuration as the spinup experiments, with deep ocean, for over 100 years. The sensitivity was 3.4°C to doubled CO2, higher than for the model of Hansen et al. (2007), which was 2.7°C. The experiments are not directly comparable because our short-lived species emissions are different than those used in Hansen et al. (2007). Nevertheless, higher sensitivity may result from short time scale interactions between climate and species for our online simulations. In a separate model experiment, we found that forcing for online aerosols is smaller than if the same fields are saved and used in offline mode, as was done for Hansen et al. (2007). Both the negative forcing from scattering aerosols and the positive forcing from absorbing aerosols were smaller for the online case. Since the net short-lived effect is negative, the aerosol effects would be less potent in online mode than offline mode so that the doubled CO2 experiment would warm more and indicate higher sensitivity. Such online versus offline differences may come from changes in interaction between the aerosols and clouds and relative humidity. For example, since online sulfate in the model tends to be anticorrelated with clouds (Koch et al. 2003), it would undergo less hydration (enhancing of scattering by increased relative humidity) than the offline sulfate would. There may also be nonlinearities in the impacts of pollution, such that episodic pollution events during a month would have smaller radiative forcing than the same pollution applied evenly in a monthly mean field. The episodic pollution may saturate in its impacts on radiative scattering and absorption. This is similar to the argument of Jones et al. (2001), that the aerosol-cloud indirect effect is stronger for monthly mean aerosols because of the concavity of the functional relationship between the cloud droplet number and aerosol number concentrations. Further resolution of these online versus offline differences is beyond the scope of this study.
Figure 11 shows maps of observed SAT changes, over the century and for three segments of the century: 1890s–1940s (warming period), 1940s–70s (stable period), and 1970s–90s (warming). All three periods have increases in TOA radiative flux, as shown in Fig. 10. The STD run warms about the correct amount over the century, however, it warms too much over subtropical oceans; from the 1890s to the 1940s it warms too much over the continents, from the 1940s to the 1970s, it warms too much especially in the tropics and the Arctic, and from the 1970s to the 1990s it does not warm enough over the continents and in the Arctic.
From differences between the STD and BCALB simulations, we estimate the BC-albedo effect on SAT (Fig. 11, bottom row). The statistical significance is weak because the forcing changes are small and because we performed an ensemble of only three simulations. The BC-albedo effect has strongest impact on SAT in the Arctic. It caused warming during the early and middle parts of the century when North American and European emissions peaked. This effect may be too strong during the 1940s–70s when the Arctic is observed to cool but the BC-albedo effect caused strong warming. Since the 1970s, the BC emissions near the Arctic declined and the reversal in the BC-albedo effect contributed to cooling, especially in northern Europe. Although the BC-albedo SAT impacts may be overestimated, the features in northern Europe do appear in the observed patterns: warming from the 1940s–1970s and reduced warming during 1970s–1990s compared with other regions of the Arctic.
b. Snow and ice cover changes
Figure 12 shows snow and ice cover changes for the full model and from the BC-albedo effect. Overall the steady decline in Arctic sea ice is consistent with observations (Lemke et al. 2007) although the GISS model sea ice is too stable (Hansen et al. 2007). The modeled ocean-ice decline over the century is −0.24%, similar to that of Hansen et al. (2007) and about half the observed (Rayner et al. 2003). However here we focus on the relative change in snow/ice due to the BC-albedo effect. The model indicates that the BC-albedo effect caused 19% snow/ice loss globally and 16% in the Arctic over the century, although the statistical significance of the results are weak.
Model snow and ice cover changes (%) (top) for STD and (bottom) from the BC-albedo effect (STD minus BCALB). Snow/ice cover changes for decadal means are given from (first column) 1890s to 1990s, (second column) 1890s to 1940s, (third column) 1940s to 1970s, and (last column) 1970s to 1990s. Global (Arctic; Land) mean values are above each panel. Heavy and light white contours show 90% and 95% significance for the STD run and 75% and 90% significance for the BC-albedo effect.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
Consistent with the SAT changes, we simulated the largest snow/ice loss early in the century. During the 1890s–1940s, 29% of the Arctic and 44% of land snow and ice losses (given in parentheses above Fig. 12 panels) may be attributed to the BC-albedo effect. After the 1940s, modeled Arctic snow/ice continued to decline but land snow did not. In the Arctic, the BC-albedo effect continued to cause large losses during 1940s–70s; however for the 1970s–90s reduction of high-latitude BC caused increased snow-ice.
Flanner et al. (2009) argued that the BC-albedo effect should contribute to springtime snow cover reduction and hence warming over Europe during the last half of the century, a feature that was lacking in the IPCC AR4 climate model simulations. In our model simulations, this effect is important during the first half of the century. However during the final three decades, BC reductions from Europe, and consequent reduction in BC deposition on snow, caused cooling in that region rather than warming. The land-snow losses during the end of the century are distributed similarly to observations, with losses in the northern U.S., Canada, northern Europe and Asia. However the increased snow-cover is overestimated along with insufficient warming.
Generally the BC-albedo effect impact on snow/ice cover is anticorrelated with the BC-albedo radiative forcing (Fig. 8). However snow-cover changes over southeast Asia and eastern Europe correlate positively with forcing changes during the 1940s–70s. In southeast Asia there is positive forcing change and increased snow cover; there is also cooling (Fig. 11) in this region, and enhanced cloud cover off the coast of Asia. In eastern Europe there is negative BC-albedo forcing change, loss of snow/ice cover, regional warming (Fig. 11), decreased clouds and precipitation. Thus in some cases shifts in local climate and hydrology cause responses not expected based on the BC-albedo radiative forcing changes. However again, the statistical significance of many of the snow/ice changes is small.
c. Cloud and precipitation changes
Figure 13 shows changes in-cloud cover and precipitation for the STD simulation. The GISS model typically has a net decrease in cloud cover under global warming conditions (Koch et al. 2009a). This dominates the simulation, with a global cloud cover decrease of −0.35% over the century. The decrease is distributed relatively evenly over the century, with −0.12, −0.08, and −0.16% occurring from 1890–1940s, 1940s–70s, and 1970s–90s, respectively. The losses are larger over the continents especially during the beginning of the century. Particularly large decreases occur over central Africa, southern Europe, southwestern North America, and western South America, with increased cloud cover over northern Europe.
(top) Model cloud cover changes (%); (bottom) model precipitation changes (mm day−1). Changes for decadal means are given from (first column) 1890s to 1990s, (second column) 1890s to 1940s, (third column) 1940s to 1970s, and (last column) 1970s to 1990s. Global (Arctic) mean values are above each panel. Heavy and light contours show 75% and 90% significance, according to a Student’s t test.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
Figure 13 also shows model precipitation changes. The global mean changes are small; however, there are some large regional changes. The simulation generally has moistening in the tropics and drying in the subtropics, a feature common with AR4 simulations and evident in observations (Seager et al. 2007). The precipitation changes are distributed similarly to those in our Qflux equilibrium model simulations (Koch et al. 2009a), although those simulations had net decreased precipitation, −0.06 mm day−1, compared with +0.01 mm day−1 in these experiments. The tropical drying/moistening to the north–south corresponds to a southward shift of the intertropical convergence zone (ITCZ). This shift has been simulated in previous models and attributed to the cooling of aerosols, especially for simulations including indirect effect, which is larger in the Northern Hemisphere than the Southern Hemisphere (e.g., Kristjansson et al. 2005; Takemura et al. 2005; Rotstayn et al. 2000; Zhang et al. 2007).
Many regional patterns may be linked with the ITCZ shift. Most notable is the drying of the Sahel, simulated in our model and in previous studies. Rotstayn and Lohmann (2002) simulated this shift and compared the consequent drying of the Sahel and southeastern Asia with observations. The Sahel was observed to dry most strongly up to the 1980s. Folland et al. (2001) argued that the strongest shifting/drying also corresponds to the timing of the greatest increase in aerosols. They also attributed the increased precipitation at higher latitudes, also occurring in our simulations, to increasing greenhouse gases. Biasutti and Giannini (2006) found similar behavior in many of the AR4 IPCC simulations, showing that the drying in the northern part of the ITCZ over the Sahel occurred for models with aerosol effects but not for models with only greenhouse gas forcing. Our model simulates strongest Sahel drying midcentury, as observed.
6. Pollution reduction experiments, 1970–2000
Now we present two pollution reduction sensitivity studies. From these we may learn how aerosols have affected recent climate change. Also, we can consider whether changing aerosol loads might improve the model performance near the end of the century. Finally, we gain some insight into how future pollution reductions may affect climate.
Beginning from the STD simulation, we branched two experiments in 1970 and continued to the year 2000 (BC70 and S70 in Table 1). We eliminated BC and SO2 pollution, respectively, and show how these reductions affect model climate by comparing results with those from the STD run during 1970s–90s. BC70 had all nonbiomass BC set to zero and biomass burning BC and OC, which are always coemitted, set to 1890 levels from 1970 to 2000. S70 had all anthropogenic SO2 set to zero from 1970 to 2000. Note that nonbiomass BC and pollution SO2 also have coemitted species, so our experiments are highly idealized.
a. Impacts on burden, aerosol optical depth, forcing
Table 2 gives burden, AOD, and forcing changes for the final decade (1990s) of the sensitivity studies. Note that for the S70 case nitrate formation increases 75% in year 2000 since nitrate and sulfate formation compete for ammonia (e.g., Bauer et al. 2007).
BC and BC-albedo forcings for BC70 are much smaller (half and 30% less) than for the 1890s, because of the substantial 1890s anthropogenic emissions, especially in the residential sector.
In the 1990s, the TOA and surface forcings for S70 relative to STD are +0.20 and +0.37 W m−2; the TOA and surface forcings for the BC reduction are −0.32 and +0.74 W m−2. Therefore, sulfur reduction has a “warming” effect at both top and bottom of the atmosphere; however, BC reduction is “cooling” at the TOA but exerts a strong positive forcing at the surface.
The radiative forcing differences between the STD and the sensitivity studies, averaged over the three decades, are −0.3 and +0.3 W m−2 for BC70 and S70.
b. Impacts of pollution reduction on climate
Figure 14 shows the global mean SAT trends for the experiments and the differences between the sensitivity and STD simulations. Sulfur reduction has a much larger effect, with warming of about 0.2°C. The warming does not occur immediately but increases gradually during the three decades. The BC reduction experiments are not much cooler on average than the STD simulation.
Global mean surface air temperature trends for observed and model simulations for STD and for 1970–2000 pollution reduction experiments. The average of each ensemble is shown in bold. (bottom) The difference between reduction experiment means and STD experiment is shown. BC70 and S70 are the BC and sulfur reduction experiments.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
Figure 15 shows map views of the change between STD and each case, averaged over the three decades, for surface radiation, SAT, snow–ice cover, and low-cloud cover. The first panel shows the particularly strong reduction in surface radiation from BC reduction.
Differences between pollution reduction simulations and STD run, averaged over the three decades 1970s–1990s. Pollution reduction cases are (top) BC (BC70) and (bottom) sulfur (S70). Changes for surface radiation (W m−2), surface air temperature (°C), snow/ice cover (%), and low cloud cover (%) are shown. Global (Arctic) means are above each panel. Heavy and light contours show 75% and 90% significance, according to a Student’s t test.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
The SAT change (averaged over the three decades) per forcing change is 0.1° and 0.5°C W−1 m2 for BC70 and S70. Therefore, sulfur “unmasking” has a stronger surface temperature response than BC reduction. The stronger impact of sulfur reduction may be due to two probably related factors. The reduction in aerosol causes decreased cloud cover, particularly at low levels in the model, −0.14% and −0.05% for BC and S (Fig. 15, last column). Figure 16 (second column) shows the zonal mean cloud cover differences for the 1990s. BC enhances low-level stability and low-level clouds so that reduction of BC decreases stability and causes cloud loss especially at mid–high latitudes of the NH. To test this, we consider the change in atmospheric column temperature per unit radiative forcing, which is 0.4°C W−1 m2 for BC and 0.6°C W−1 m2 for S. Thus, sulfur reduction affects surface and column temperatures in a similar way, suggesting that atmospheric stability has not changed very much. However, BC reduction cools the column much more than the surface. Figure 16 (first column) shows the differences in zonal mean temperature. Because of its impact on stability and clouds, BC caused strong surface cooling, so that when BC is removed the surface temperature does not cool very much more, while the column temperature does cool. Note also that in Fig. 15, the regions of cooling for BC reduction are typically regions where low cloud cover has increased. For the sulfur reduction experiments warming is strongest in regions where cloud cover has decreased, particularly over North America and Europe.
Zonal mean differences between pollution reduction simulations and STD run, averaged over the 1990s. Pollution reduction cases are (top) BC (BC70) and (bottom) sulfur (S70). Changes for (left) temperature (°C) and (right) cloud cover (%) are shown.
Citation: Journal of Climate 24, 11; 10.1175/2011JCLI3582.1
The S70 warming also enhances snow–ice cover loss, and this is exacerbated especially where there are continental reductions in cloud cover; note that these experiments do not include the indirect effect, if they did the unmasking–warming effect would have been stronger. The strong warming over continents from reduced sulfate would improve our STD simulation compared to observations (Fig. 11). However, even if all sulfur pollution were removed, the STD simulation warming would still not be as much as observed.
7. Discussion and conclusions
We have presented results for coupled aerosols and ozone run online in a transient twentieth-century climate simulation including a dynamic ocean, with particular focus on aerosol effects on climate. We considered aerosol and ozone trends and compared them with available observations. We then considered the climate response and isolated the BC-snow-albedo effect on climate. Finally, we conducted sensitivity studies to see how aerosol reduction near the end of the century would have affected climate. In general, the model simulated the net temperature change over the century fairly well; however, it failed to capture the observed midcentury cooling and end-of-century warming (Fig. 11).
To test the historical emissions and model simulations, we compared modeled trends in aerosol species with available historical ice core, lake core, sulfur deposition, air concentration, and surface radiation observations. In general, the model trends agreed reasonably well with BC observations from ice and lake sediment cores in Greenland, North America, and Europe. Historical sulfur in ice cores presented a greater challenge. Sulfur in the Greenland ice core peaked at around 1970, while the model peaked around 1990 (Fig. 4). According to McConnell et al. (2007), the Greenland (BC) cores correlate best with emission trends from North America. Interestingly, the model sulfur deposition for 1980–2000 in North America compared very well with data from NADP, both in magnitude and trend (Fig. 5). However, we cannot check this prior to 1980 to determine whether the source region sulfur should have peaked in 1970 as indicated by the Greenland ice core records. The model also does not produce the smaller sulfur maximum around 1910 in the Greenland cores, presumably resulting from coal combustion. In Europe, the model sulfur trends compared reasonably well with Alpine ice core records, with both indicating peak European sulfur around 1980 (Fig. 6). However, compared with surface concentration sulfur measurements, the model underestimates the magnitude in 1980 while simulating the late century concentrations quite well. Therefore, the decline in sulfate is probably underestimated for Europe.
Deficiencies in model species are to some extent due to problems in the historical emissions. A shortcoming of our setup was to use the EDGAR–HYDE historical sulfur pollution emissions up to the year 1980 and then switch to the EDGAR32 emissions for the years 1990 and 2000. The EDGAR–HYDE and EDGAR32 are constructed differently and include different sources and are therefore not strictly consistent. We note that the emissions estimates of Smith et al. (2004) peak higher in the 1970s and decline more afterward compared with ours. This is because Smith et al. (2004) included sulfur emission controls applied to fossil fuel and smelting sources that were implemented later in the century; these were not included in the EDGAR inventories. The 1970s peak of Smith et al. (2004) is indeed due mostly to North American emissions peaking at around that time; European sulfur emissions peaked in the late 1970s.
Historical biomass burning is a particularly uncertain source of the short-lived species. Biomass burning TOA aerosol radiative forcing is estimated to be slightly negative while ozone contributes positive forcing (Fig. 8); the aerosols exert a strong negative surface forcing (Fig. 9). The net climate impact of these effects cannot be isolated in our experiments but could be significant. Because the history of biomass burning is poorly known, we chose a simple emission history, with no change in boreal/temperate burning over the century, and with half of present-day tropical forest and African grasslands burning in the late 1800s with a linear increase to the year 2000. According to Mouillot and Field (2005), Mouillot et al. (2006), and Marlon et al. (2008), temperate and boreal burning in some regions such as North America may have been greater in the late 1800s than at present owing to the extensive land-clearing activities at that time. These studies also suggest that tropical burning in the late 1800s would have been quite small and then increased more steeply toward the end of the century, while African grassland burning would have been relatively steady during the century. Future climate experiments (e.g., AR5) will include updated biomass burning histories. However, further constraint on these historical sources is needed.
Although our model included full coupling between aerosols, ozone, and climate, our treatment was simplified in two important respects. First, this model version does not include aerosol microphysics, or mixed particles, other than sulfur and nitrate mixing with dust. If sulfate and nitrate were also allowed to mix with BC and other particles, the amount of sulfate scattering would probably decrease and the amount of BC absorption would increase (e.g., Ackerman and Toon 1981). Future simulations of the twentieth century will include our aerosol microphysical scheme (Bauer et al. 2008, 2010). This was not available at the start of this study and would also have been too computationally slow to include together with ozone chemistry.
A second model component that should be included is the aerosol microphysical effects on clouds, or indirect effects. Excessive midcentury warming (e.g., Figs. 10 and 11) would be reduced by indirect effects because the resulting cloud enhancement would coincide with the large aerosol load increase at that time. Although we did include indirect effects in our initial experiments, we found it quite difficult to avoid excessive cooling toward the end of the century.
An important point of this study is that the changes in short-lived species that occur during segments of the twentieth century are more dynamic than what we see from an “IPCC bar chart” view of radiative forcing changes since the preindustrial. While the centennial view is useful for long-lived greenhouse gases, it underestimates the impacts of short-lived species since they have had large regionally varying changes throughout the century. For example, we showed that BC impacts are greatest early in the twentieth century, while sulfate impacts peak later in the century. And since both of these pollutants were later reduced in developed regions such as North America and Europe, the reductions sometimes caused reversals in radiative forcing.
For example, the BC-albedo effect underwent local forcing sign reversal late in the century. Over the century, about 20% of Arctic and global snow/ice loss was attributed to the BC-albedo effect and about 14% of Arctic warming (Figs. 11 and 12). However, the strongest impact was early in the century, when 30%–50% of warming and ice melting was estimated to be from the BC-albedo effect. Late in the century, reductions in Arctic BC contributed to Arctic cooling and increased snow/ice cover. Note that even though the BC-albedo effect has apparently not caused recent snow/ice cover Arctic losses, the model results imply that there is remaining potential for further reductions of Arctic BC to contribute to Arctic cooling. As shown in Fig. 7, the large BC deposition in the Eurasian region at the end of the century suggests that there are still opportunities for further BC reduction.
Our results support the role of aerosols in contributing to the interdecadal variability of surface radiation or dimming and brightening. Reduced warming and global dimming, modeled midcentury (1940–70s), coincide with strongly increased sulfate (Figs. 9 and 11). The dimming is generally stronger for clear sky than all sky, suggesting an important role for aerosols. The model simulated local brightening at the end of the century, especially in Europe and North America. Although our model simulates the correct decadal variability, the dimming/brightening magnitudes are weaker than in surface radiation measurements, similar to previous model studies (Wild 2009b). This deficiency may correspond to the lack of decadal variability in SAT during the last half of the century, with insufficient cooling midcentury and insufficient warming at the end of the century. Including aerosol indirect effects and stronger BC absorption, as would occur for mixed particles, should increase dimming and brightening magnitudes; the indirect effects would also amplify the aerosol cooling effects. On the other hand, the midcentury cooling may have resulted from natural variability that our model was not able to simulate.
Our coupled aerosol-chemistry–climate model simulated ozone changes, including reduced polar stratospheric ozone in both hemispheres (Fig. 8) from the 1970s to the 1990s. Concurrently, sea salt aerosol optical depths and radiative forcing also increased in the Southern Hemisphere around Antarctica (Fig. 8). Previous modeling studies have shown that the ozone hole and increased greenhouse gases are probably responsible for the intensification of the Southern Hemisphere annular mode and therefore intensification of the polar vortex (e.g., Arblaster and Meehl 2006). It is plausible that the sea salt enhancement also contributed to south polar cooling toward the end of the century (Fig. 11). In future studies, the climate effect of the sea salt enhancement could be isolated by comparing climate simulations with and without sea salt climate coupling.
Our study suggests that using online aerosols may reduce the potency of the aerosol forcing. The model sensitivity, or warming of the control run to doubled CO2, was larger (3.4°C) than found in a similar model version that used offline species (2.7°C; Hansen et al. 2007). Because our aerosol/ozone sources were different, it is difficult to quantify the reasons for the changed sensitivity. However, in separate experiments, we found weaker aerosol forcings using online aerosols compared with identical offline aerosols. This difference may result from saturation of forcing during pollution events that would occur less if using monthly mean fields. Thus, using online aerosols may further reduce the ability of the aerosols to contribute to midcentury cooling.
To isolate the effects of aerosol pollution on climate change during the last part of the century, we performed two pollution-reduction experiments beginning in the 1970s, removing sulfate and BC pollution, respectively. Although the magnitudes of the forcing and the column temperature responses were similar for the two experiments, there was large disparity in the surface air temperature response. Reduction in scattering aerosols caused a much stronger surface warming compared to the cooling from reduction in absorbing (BC) aerosols. We argued that since BC stabilizes the atmospheric column, it increases low-level cloud cover which contributes to surface cooling. Therefore, when BC was reduced, stability and cloud cover decreased, so that surface temperatures decreased only modestly. The cloud responses are probably model dependent and similar experiments should be performed in other models to confirm the result.
Our transient simulation of coupled short-lived species and climate was successful in some respects and indicates the need for improvement in others. The model agreed with most observed trends in species from lake/ice cores and with surface radiation trends, although it underestimated the magnitude of the surface radiation and SAT changes during the second have of the century. It is encouraging that there are ongoing efforts to derive additional historical trends in species, in ice cores and lake sediments. These are crucial for constraining emission histories and model performance. Our future experiments will use the newer (AR5) version of the GISS model, including higher horizontal and vertical resolution and updated climate model physics. We will also include our aerosol microphysical scheme (Bauer et al. 2008, 2010) to provide more physically realistic aerosol radiative forcing and aerosol indirect effects.
Acknowledgments
Support for this research is from the NASA Modeling, Analysis, and Prediction Program and the Clean Air Task Force. The work at Lawrence Berkeley National Laboratory was supported by the U.S. Department of Energy under Contract No. DE- AC02-05CH11231. S.M. acknowledges funding from the NASA MAP and the DOE ASR program. Collection and initial analyses of the Greenland ice cores was supported by a number of NASA and NSF grants. Reanalysis of the Humboldt and Summit ice cores for BC and sulfur was supported by NSF Arctic Section Grants 0909541, 0336450, and 0856845. We thank L. Husain for providing his lake core BC data.
REFERENCES
Ackerman, T. P., and O. B. Toon, 1981: Absorption of visible radiation in atmosphere containing mixtures of absorbing and nonabsorbing particles. Appl. Opt., 20, 3661–3667.
Arblaster, J. M., and G. A. Meehl, 2006: Contributions of external forcings to southern annular mode trends. J. Climate, 19, 2896–2905.
Bauer, S. E., and D. Koch, 2005: Impact of heterogeneous sulfate formation at mineral dust surfaces on aerosol loads and radiative forcing in the Goddard Institute for Space Studies general circulation model. J. Geophys. Res., 110, D17202, doi:10.1029/2005JD005870.
Bauer, S. E., D. Koch, N. Unger, S. M. Metzger, D. T. Shindell, and D. Streets, 2007: Nitrate aerosols today and in 2030: Importance relative to other aerosol species and tropospheric ozone. Atmos. Chem. Phys., 7, 5043–5059.
Bauer, S. E., D. Wright, D. Koch, E. R. Lewis, R. McGraw, L.-S. Chang, S. E. Schwartz, and R. Ruedy, 2008: MATRIX (Multiconfiguration Aerosol Tracker of mixing state): An aerosol microphysical module for global atmospheric models. Atmos. Chem. Phys., 8, 6003–6035.
Bauer, S. E., S. Menon, D. Koch, T. C. Bond, and K. Tsigaridis, 2010: A global modeling study on carbonaceous aerosol microphysical characteristics and radiative effects. Atmos. Chem. Phys., 10, 7439–7456.
Bell, N., D. Koch, and D. T. Shindell, 2005: Impacts of chemistry-aerosol coupling on tropospheric ozone and sulfate simulations in a general circulation model. J. Geophys. Res., 110, D14305, doi:10.1029/2004JD005538.
Berglen, T. F., T. K. Berntsen, I. S. A. Isaksen, and J. K. Sundet, 2004: A global model of the coupled sulfur/oxidant chemistry in the troposphere: The sulfur cycle. J. Geophys. Res., 109, D19310, doi:10.1029/2003JD003948.
Biasutti, M., and A. Giannini, 2006: Robust Sahel drying in response to late 20th century forcings. Geophys. Res. Lett., 33, L11706, doi:10.1029/2006GL026067.
Bond, T. C., E. Bhardwaj, R. Dong, R. Jogani, S. Jung, C. Roden, D. G. Streets, and N. M. Trautmann, 2007: Historical emissions of black and organic carbon aerosol from energy-related combustion, 1850–2000. Global Biogeochem. Cycles, 21, GB2018, doi:10.1029/2006GB002840.
Clarke, A. D., and K. J. Noone, 1985: Soot in the Arctic snowpack. Atmos. Environ., 19, 2045–2053.
Flanner, M. G., C. S. Zender, P. G. Hess, N. M. Mahowald, T. H. Painter, V. Ramanathan, and P. J. Rasch, 2009: Springtime warming and reduced snow cover from carbonaceous particles. Atmos. Chem. Phys., 9, 2481–2497.
Folland, C. K., and Coauthors, 2001: Observed climate variability and change. Climate Change 2001: The Scientific Basis, J. T. Houghton et al., Eds., Cambridge University Press, 99–181.
Forster, P., and Coauthors, 2007: Changes in atmospheric constituents and in radiative forcing. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 129–234.
Grenfell, T. C., S. G. Warren, V. F. Radionov, V. N. Makarov, and S. A. Zimov, 2009: Expeditions to the Russian Arctic to Survey Black Carbon in Snow. Eos, Trans. Amer. Geophys. Union, 90, 43, doi:10.1029/2009EO430002.
Hansen, J., and Coauthors, 2007: Climate simulations for 1880-2003 with GISS modelE. Climate Dyn., 29, 661–696, doi:10.1007/s00382-007-0255-8.
Hegg, D. A., S. G. Warren, T. C. Grenfell, S. J. Doherty, T. V. Larson, and A. D. Clarke, 2009: Source attribution of black carbon in Arctic snow. Environ. Sci. Technol., 43, 4016–4021.
Hicks, S., and E. Isaksson, 2006: Assessing source areas of pollutants from studies of fly ash, charcoal, and pollen from Svalbard snow and ice. J. Geophys. Res., 111, D02113, doi:10.1029/2005JD006167.
Husain, L., A. J. Khan, T. Ahmed, K. Swami, A. Bari, J. S. Webber, and J. Li, 2008: Trends in atmospheric elemental carbon concentrations from 1835 to 2005. J. Geophys. Res., 113, D13102, doi:10.1029/2007JD009398.
Ito, A., and J. E. Penner, 2005: Historical emissions of carbonaceous aerosols from biomass and fossil fuel burning for the period 1870 2000. Global Biogeochem. Cycles, 19, GB2028, doi:10.1029/2004GB002374.
Jones, A., D. L. Roberts, M. J. Woodage, and C. E. Johnson, 2001: Indirect sulphate aerosol forcing in a climate model with an interactive sulphur cycle. J. Geophys. Res., 106, 20 293–20 310.
Koch, D., D. Jacob, I. Tegen, D. Rind, and M. Chin, 1999: Tropospheric sulfur simulation and sulfate direct radiative forcing in the Goddard Institute for Space Studies general circulation model. J. Geophys. Res., 104, 23 799–23 822.
Koch, D., J. Park, and A. Del Genio, 2003: Clouds and sulfate are anticorrelated: A new diagnostic for global sulfur models. J. Geophys. Res., 108, 4781, doi:10.1029/2003JD003621.
Koch, D., G. A. Schmidt, and C. Field, 2006: Sulfur, sea salt and radionuclide aerosols in GISS, ModelE. J. Geophys. Res., 111, D06206, doi:10.1029/2004JD005550.
Koch, D., T. Bond, D. Streets, N. Unger, and G. R. van der Werf, 2007: Global impacts of aerosols from particular source regions and sectors. J. Geophys. Res., 112, D02205, doi:10.1029/2005JD007024.
Koch, D., S. Menon, A. Del Genio, R. Ruedy, I. Alienov, and G. Schmidt, 2009a: Distinguishing aerosol impacts on climate during the past century. J. Climate, 22, 2659–2677.
Koch, D., and Coauthors, 2009b: Evaluation of black carbon estimations in global aerosol models. Atmos. Chem. Phys., 9, 9001–9026.
Kristjansson, J. E., T. Iversen, A. Kirkevag, Ø. Seland, and J. Debernard, 2005: Response of the climate system to aerosol direct and indirect forcing: Role of cloud feedbacks. J. Geophys. Res., 110, D24206, doi:10.1029/2005JD006299.
Lamarque, J.-F., and Coauthors, 2010: Historical (1850–2000) gridded anthropogenic and biomass burning emissions of reactive gases and aerosols: Methodology and application. Atmos. Chem. Phys. Discuss., 10, 4963–5019.
Lavanchy, V. M. H., H. W. Gaggeler, U. Schotterer, M. Schwikowski, and U. Baltensperger, 1999: Historical record of carbonaceous particle concentrations from a European high-alpine glacier (Colle Gnifetti, Switzerland). J. Geophys. Res., 104, 21 227–21 236.
Lemke, P., and Coauthors, 2007: Observations: Changes in snow, ice and frozen ground. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 337–383.
Lenton, A., F. Codron, L. Bopp, N. Metzl, P. Cadule, A. Tagliabue, and J. Le Sommer, 2009: Stratospheric ozone depletion reduces ocean carbon uptake and enhances ocean acidification. Geophys. Res. Lett., 36, L12606, doi:10.1029/2009GL038227.
Marlon, J., and Coauthors, 2008: Climate and human influences on global biomass burning during the past two millennia. Nat. Geosci., 1, 697–702.
McConnell, J. R., 2010: New Directions: Historical black carbon and other ice core aerosol records in the Arctic for GCM evaluation. Atmos. Environ., 44, 2665–2666, doi:10.1016/j.atmosenv.2010.04.004.
McConnell, J. R., and R. Edwards, 2008: Coal burning leaves toxic heavy metal legacy in the Arctic. Proc. Natl. Acad. Sci. USA, 105, 12140–12144, doi:10.1073/pnas.0803564105.
McConnell, J. R., and Coauthors, 2007: 20th-century industrial black carbon emissions altered Arctic climate forcing. Science, 7, 1381–1384, doi:10.1126/science.1144856.
Miller, R. L., and Coauthors, 2006: Mineral dust aerosols in the NASA Goddard Institute for Space Sciences ModelE atmospheric general circulation model. J. Geophys. Res., 111, D06206, doi:10.1029/2005JD005796.
Mouillot, F., and C. B. Field, 2005: Fire history and the global carbon budget: A 1° × 1° fire history reconstruction for the 20th century. Global Change Biol., 11, 398–420.
Mouillot, F., A. Narasimha, Y. Balkanski, J.-F. Lamarque, and C. B. Field, 2006: Global carbon emissions from biomass burning in the 20th century. Geophys. Res. Lett., 33, L01801, doi:10.1029/2005GL024707.
Nagashima, T., H. Shiogama, T. Yokohata, T. Takemura, S. A. Crooks, and T. Nozawa, 2006: Effect of carbonaceous aerosols on surface temperature in the mid twentieth century. Geophys. Res. Lett., 33, L04702, doi:10.1029/2005GL024887.
National Atmospheric Deposition Program, 2008: National Atmospheric Deposition Program 2007 Annual Summary. NADP Data Rep. 2008-01. Illinois State Water Survey, University of Illinois at Urbana–Champaign.
Novakov, T., V. Ramanathan, J. E. Hansen, T. W. Kirchstetter, M. Sato, J. E. Sinton, and J. A. Sathaye, 2003: Large historical changes of fossil-fuel black carbon aerosols. Geophys. Res. Lett., 30, 1324, doi:10.1029/2002GL016345.
Preunkert, S., M. Legrand, and D. Wagenbach, 2001: Sulfate trends in a Col du Dome (French Alps) ice core: A record of anthropogenic sulfate levels in the European midtroposphere over the twentieth century. J. Geophys. Res., 106, 31 991–32 004.
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of SST, sea ice and night marine air temperatures since the late nineteenth century. J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670.
Roeckner, E., P. Stier, J. Feichter, S. Kloster, M. Esch, and I. Fischer-Bruns, 2006: Impact of carbonaceous aerosol emissions on regional climate change. Climate Dyn., 27, 553–571, doi:10.1007/s00382-006-0147-3.
Rotstayn, L. D., and U. Lohmann, 2002: Tropical rainfall trends and the indirect aerosol effect. J. Climate, 15, 2103–2116.
Rotstayn, L. D., B. F. Ryan, and J. E. Penner, 2000: Precipitation changes in a GCM resulting from the indirect effects of anthropogenic aerosols. Geophys. Res. Lett., 27, 3045–3048.
Rotstayn, L. D., and Coauthors, 2007: Have Australian rainfall and cloudiness increased due to the remote effects of Asian anthropogenic aerosols? J. Geophys. Res., 112, D09202, doi:10.1029/2006JD007712.
Russell, G. L., J. R. Miller, and D. H. Rind, 1995: A coupled atmosphere–ocean model for transient climate change. Atmos.–Ocean, 33, 683–730.
Schmidt, G. A., and Coauthors, 2006: Present day atmospheric simulations using GISS ModelE: Comparison to in situ, satellite, and reanalysis data. J. Climate, 19, 153–192.
Schwikowski, M., S. Brutsch, H. W. Gaggeler, and U. Schotterer, 1999: A high-resolution air chemistry record from an Alpine ice core: Fiescherhorn glacier, Swiss Alps. J. Geophys. Res., 104, 13 709–13 719.
Seager, R., and Coauthors, 2007: Model projections of an imminent transition to a more arid climate in southwestern North America. Science, 316, 1181–1184.
Sharma, S., D. Lavoue, H. Cachier, L. A. Barrie, and S. L. Gong, 2004: Long-term trends of the black carbon concentrations in the Canadian Arctic. J. Geophys. Res., 109, D15203, doi:10.1029/2003JD004331.
Shindell, D. T., G. Faluvegi, N. Unger, E. Aguilar, G. A. Schmidt, D. Koch, S. E. Bauer, and R. L. Miller, 2006: Simulations of preindustrial, present-day, and 2100 conditions in the NASA GISS composition and climate model G-PUCCINI. Atmos. Chem. Phys., 6, 4427–4459.
Smith, S. J., E. Conception, R. Andres, and J. Lurz, 2004: Historical sulfur dioxide emissions 1850–2000: Methods and results. Pacific Northwest National Lab Rep. PNNL-14537, 16 pp.
Stier, P., J. Feichter, E. Roeckner, S. Kloster, and M. Esch, 2006: The evolution of the global aerosol system in a transient climate simulation from 1860 to 2100. Atmos. Chem. Phys., 6, 3059–3076.
Takemura, T., T. Nozawa, S. Emori, T. Y. Nakajima, and T. Nakajima, 2005: Simulation of climate response to aerosol. J. Geophys. Res., 110, D02202, doi:10.1029/2004JD005029.
Tomasi, C., and Coauthors, 2007: Aerosols in polar regions: A historical overview based on optical depth and in situ observations. J. Geophys. Res., 112, D16205, doi:10.1029/2007JD008432.
Van Aardenne, J. A., F. J. Dentener, J. G. J. Olivier, C. G. M. Klein Goldewijk, and J. Lelieveld, 2001: A high resolution dataset of historical anthropogenic trace gas emissions for the period 1890-1990. Global Biogeochem. Cycles, 15, 909–928.
van der Werf, G. R., J. T. Randerson, G. J. Collatz, and L. Giglio, 2003: Carbon emissions from fires in tropical and subtropical ecosystems. Global Change Biol., 9, 547–562.
van der Werf, G. R., J. T. Randerson, G. J. Collatz, L. Giglio, P. S. Kasibhatla, A. F. Arellano, S. C. Olsen, and E. S. Kasischke, 2004: Continental-scale partitioning of fire emissions during the 1997 to 2001 El Nino/La Nina period. Science, 303, 73–76.
Wang, Y., D. J. Jacob, and J. A. Logan, 1998: Global simulation of tropospheric O3-NOx- hydrocarbon chemistry. J. Geophys. Res., 103, 10 713–10 725.
Wild, M., 2009a: How well do IPCC-AR4/CMIP3 climate models simulate global dimming/brightening and twentieth- century daytime and nighttime warming? J. Geophys. Res., 114, D00D11, doi:10.1029/2008JD011372.
Wild, M., 2009b: Global dimming and brightening: A review. J. Geophys. Res., 114, D00D16, doi:10.1029/2008JD011470.
Zhang, X., F. W. Zwiers, G. C. Hegerl, F. H. Lambert, N. P. Gillett, S. Solomon, P. A. Stott, and T. Nozawa, 2007: Detection of human influences on twentieth-century precipitation trends. Nature, 448, 461–466.