1. Introduction
Recent studies suggest that a possible increase of global black carbon aerosol (BC) emissions could contribute to a significant warming in the future (Levy et al. 2008; Shindell et al. 2008), with BC evidently being the second strongest present-day source of anthropogenic climate forcing after carbon dioxide (Bond et al. 2013). The direct radiative effect of tropospheric BC is the warming of the troposphere due to absorption of solar radiation. The associated changes in the thermal structure of the atmosphere affect the relative humidity and tropospheric stability, which can impact the formation of clouds and their lifetime. This effect is known as the semidirect effect on clouds. The semidirect effect on clouds typically results in a positive (warming) top-of-atmosphere (TOA) radiative forcing caused by the evaporation of cloud droplets (e.g., Hansen et al. 1997; Ackerman et al. 2000). The semidirect effect on clouds can also result in a net negative TOA radiative forcing locally, depending on the vertical distribution of BC. For example, over cool subtropical waters off the subtropical eastern coasts, the heating of the free troposphere can increase the lower tropospheric stability (LTS), increasing the shallow marine stratocumulus cloud cover capping the boundary layer and resulting in an increase in the reflection of solar radiation (e.g., Johnson et al. 2004; Wilcox 2010; Sakaeda et al. 2011; Mahajan et al. 2012).
It is evident from several studies that TOA radiative forcing itself, without the knowledge of the forcing agents, appears to be a poor predictor of the climate response (e.g., Hansen et al. 1997; Allen and Sherwood 2011; Yoshimori and Broccoli 2008). Yoshimori and Broccoli (2008) find that the equilibrium global-average surface temperature and hydrological responses to the same strength of radiative forcing at the TOA depend on the characteristics of the forcing agents in idealized experiments with the Geophysical Fluid Dynamics Laboratory Atmospheric Model, version 2.1 (GFDL AM2.1), coupled to a slab ocean model (SOM). The distinct response to BC when compared to that of carbon dioxide was found to be due to cloud feedbacks and aerosol semidirect effects. Further, Ming and Ramaswamy (2009) find that the climate response of aerosols and greenhouse gases is not linearly additive in a version of the same model that also includes aerosol indirect effects. They suggest that the nonlinearity arises from changes in cloud distribution and aerosol indirect effects. Further, hydrological sensitivity to global-average surface temperature changes is also found to vary widely with combination of emissions in Model for Interdisciplinary Research on Climate, version 3.2 (MIROC3.2), future scenarios integrations that include the aerosol indirect effects (Shiogama et al. 2010). It is thus important to study the climate response to individual forcing agents independently.
The net global-average precipitation response to simulated BC-induced radiative forcing is found to be negative, opposite in sign to that of greenhouse gas warming (e.g., Yoshimori and Broccoli 2008; Ming et al. 2010; Frieler et al. 2011). Modeling results show that shortwave heating of the atmosphere by BC is largely balanced by a reduction in the surface latent heat flux into the atmosphere, which manifests as a reduction of global-average precipitation (e.g., Ming et al. 2010). This reduction in precipitation is a fast response to the quick atmospheric heating and is similar to the initial model response to increase in carbon dioxide in coupled models (e.g., Bala et al. 2010; Andrews et al. 2010). The precipitation response associated with the increase in surface temperatures to a positive radiative forcing on longer time scales, as the system equilibrates, is positive (e.g., Held and Soden 2006; Andrews et al. 2010). However, in the case of BC, studies show that the fast response is stronger than the response to surface temperature changes and hence the net equilibrium precipitation response is negative (Andrews et al. 2010; Ming et al. 2010; Frieler et al. 2011).
The distribution of BC is spatially nonhomogeneous, with large concentrations near the source regions, which are mostly located in the Northern Hemisphere. Their radiative forcing is thus different from the homogeneously distributed greenhouse gases. Modeling studies show that the interhemispheric asymmetry of radiative forcing results in heat transport from the warmer to the cooler hemisphere (e.g., Yoshimori and Broccoli 2008; Frierson and Hwang 2012). In the deep tropics, the cross-equatorial heat transport is manifested as a change in the Hadley circulation and a shift in the intertropical convergence zone (ITCZ) toward the warmer hemisphere (e.g., Frierson and Hwang 2012). The shift in the ITCZ in response to interhemispheric asymmetry in forcing is found to be a robust response in several modeling studies such as those investigating the large-scale impacts of simulated Atlantic meridional overturning circulation (AMOC) changes (e.g., Chiang and Bitz 2005; Mahajan et al. 2011), sulfate aerosols (e.g., Chen and Ramaswamy 1996; Ramaswamy and Chen 1997; Rotstayn and Lohmann 2002; Bollasina et al. 2011) as well as BC (e.g., Roberts and Jones 2004; Chung and Seinfeld 2005; Wang 2007; Jones et al. 2007; Yoshimori and Broccoli 2008; Ming and Ramaswamy 2009).
There is a large uncertainty in the present-day global BC aerosol distribution estimate both from natural as well as anthropogenic sources (e.g., Ramanathan and Carmichael 2008; Schultz et al. 2008). Further, the mean simulated TOA instantaneous radiative forcing of BC from fossil fuel and biofuel burning of 15 global aerosol models participating in the Aerosol Comparisons between Models and Observations project (AeroCom), Phase II, is 0.19 W m−2 (Myhre et al. 2012), whereas the latest calculations constrained by observations estimate it as 0.88 W m−2 from all BC sources (Bond et al. 2013), both with large uncertainties. The lower estimate in GCMs results from a number of reasons including a poor representation of a mixed state of BC with other aerosols, as well as the vertical distribution of BC (e.g., Ramanathan and Carmichael 2008).
Here, we are motivated to isolate the climate response to spatially varying BC with only direct and semidirect effects of aerosols included. We use a new estimate of spatial distribution of BC concentration (Mahajan et al. 2012) derived from phase 5 of the Coupled Model Intercomparison Project (CMIP5) emissions inventory (Lamarque et al. 2010) for our simulations. As discussed later, the simulated direct radiative forcing from this BC dataset is also much weaker than the Bond et al. (2013) estimate. We conduct a suite of idealized GCM experiments by progressively increasing the atmospheric BC concentration from our new estimate of present-day levels to investigate the sensitivity of climate response to BC forcing. Stronger forcing also increases the signal-to-noise ratio of the response (e.g., Roberts and Jones 2004). As discussed later in section 3, the simulated range of radiative forcing of BC in these experiments remains close to the estimated range of realistic total radiative forcing from BC.
The next section describes our model simulations and the BC data in more detail. In section 3, we discuss the radiative forcing of BC in our simulations. Section 4 presents the global surface temperature, precipitation, and cloud response to BC in those experiments, which we find to be linearly related to the imposed BC burden. Our results are summarized in section 5, where we also list the important caveats of our study.
2. GCM simulations
We use the spectral version of the Community Atmosphere Model, version 4 (CAM4), at a horizontal resolution of 1° coupled thermodynamically to an SOM for conducting experiments. The SOM contains an ocean prescribed with a finite mixed layer depth varying geographically. The lack of ocean dynamics in the model is compensated by prescribing a climatology of implied heat flux Q. The Q fluxes are computed from a fully coupled atmosphere–ocean model preindustrial control simulation climatology. More details on the SOM component used here can be found in Bitz et al. (2012). Additionally, we configure the model with noninteractive sea ice by prescribing it to the preindustrial control simulation climatology. While prescribing sea ice might result in a model climate that does not truly replicate the fully coupled model, it does allow for clearly isolating the impact of tropospheric black carbon aerosols directly without contamination of the climate response by changes in sea ice.
The present-day climatology of BC is computed from the period 1981–2000 of a late-twentieth-century atmosphere-only integration of CAM4 coupled to a bulk aerosol model (BAM) (Tie et al. 2005). CAM4–BAM is forced with a new monthly surface emissions dataset (Mahajan et al. 2012), which was derived from CMIP5 and Reanalysis of the Tropospheric chemical composition over the past 40 years (RETRO) wildfire emissions datasets (Lamarque et al. 2010; Schultz et al. 2008). The simulated aerosol optical depth compares well with satellite-derived estimates with exceptions. The simulated aerosol optical depth (AOD) is underestimated over carbonaceous aerosol dominant regions, which is a common bias in all participating AeroCom global models (Koch et al. 2009). More details on the new emissions and aerosol dataset can be found in Mahajan et al. (2012).
We conduct five experiments with CAM4–SOM prescribed with preindustrial levels of greenhouse gases and ozone. We first integrate CAM4–SOM with no aerosol forcing (0 × BC). We then conduct a BC aerosol forcing–only run, where we prescribe a realistic estimate of the present-day black carbon aerosol distribution to CAM4–SOM (1 × BC). We also conduct experiments forced with 2×, 5×, and 10× the present-day black carbon aerosol distribution (named 2 × BC, 5 × BC, and 10 × BC, respectively). The distribution of BC in these experiments is derived by scaling the present-day climatology of black carbon aerosols globally, thus maintaining their present-day horizontal and vertical distribution. All the models use the same Q fluxes for the SOM and are forced with preindustrial levels of greenhouse gases. The surface deposition of black carbon is prescribed to be at the preindustrial rates in all experiments and, thus, there are no changes in albedo due to black carbon deposition in these experiments. Each run is integrated for 40 years. All the experiments equilibrate by year 15, and we use the last 25 years for analysis.
Figure 1a shows the annual-mean distribution of present-day black carbon aerosol optical depth climatology simulated using CAM4–BAM. The distribution of BC is skewed toward the Northern Hemisphere, with large concentrations over Southeast Asia, China, the United States, and Europe. Large amounts of black carbon aerosols are also emitted from the southern tropical regions of African savannas in the boreal summer and South America in the boreal fall (not shown). Figure 1b shows the annual-mean zonal-average vertical distribution of BC. The largest amounts of BC are located in the lower troposphere in the Northern Hemisphere. Over the southern equatorial region, prevalent winds carry the aerosols into the midtroposphere particularly over the southeastern tropical Atlantic and Pacific Oceans. The zonal-average distribution of total atmospheric BC aerosol burden is shown in Fig. 1c for all the BC experiments.
BC distribution. (a) Annual-mean aerosol optical depth simulated by CAM4 forced with a climatology of present-day BC aerosol distribution. (b) Zonal averages of present-day annual-mean BC aerosol mixing ratio. (c) Zonally averaged vertically integrated annual-mean total atmospheric BC burden used to force the 1 × BC, 2 × BC, 5 × BC, and 10 × BC experiments.
Citation: Journal of Climate 26, 20; 10.1175/JCLI-D-12-00715.1
3. Radiative forcing of black carbon aerosols
We compute the radiative flux perturbation (RFP) following Haywood et al. (2009) as the change in the radiative flux from atmospheric GCM-only runs when aerosol forcing is included. The 5-yr CAM4 runs (termed as RFP integrations) forced with no aerosols and each of the BC forcings separately, with prescribed preindustrial SST and sea ice climatology from the coupled simulation, are integrated after a 1-yr spinup to compute the RFP. RFP includes changes in radiative flux from fast atmospheric feedbacks and processes (with time scales much shorter than the equilibrium time scale of CAM4–SOM) caused by the BC perturbations as forcing. Thus, RFP includes the semidirect effect of BC aerosols in addition to the direct effects. The TOA and surface all-sky and clear-sky RFP in the BC perturbation experiments, with respect to the 0 × BC experiment, are listed in Table 1 (positive values indicate warming). For the present-day BC burden, CAM4 simulates a TOA RFP of 0.2 W m−2. Bond et al. (2013) review numerous studies of BC and estimate that the total climate forcing of BC in the industrial era, including indirect effects and cryosphere forcing, is about 1.1 W m−2 with 90% uncertainty bounds from 0.17 to 2.1 W m−2. Their estimate is computed by appropriately scaling several climate model results after evaluating them with field observations and microphysical measurements. Although we do not include the effect of BC deposition and aerosol indirect effects, the CAM4 estimate of RFP for present-day BC is low like other AeroCom Phase II models, with values from 5 × BC and 10 × BC experiments being closer to the present-day estimate of Bond et al. (2013).
BC-induced radiative forcing (W m−2) at the TOA and the surface in idealized BC forcing experiments computed as RFP, F, and SDE.
Table 1 also lists the more commonly used instantaneous radiative forcing (F; Forster et al. 2007) due to BC perturbations with regard to the 0 × BC distribution. The value of F is computed by calling only the radiation sequence of CAM4 with prescribed atmospheric conditions for different atmospheric BC compositions. This is realized by integrating the Parallel Offline Radiative Transfer (PORT), which is distributed with the Community Earth System Model, version 1 (CESM1; Conley et al. 2012), for one year. The value of F is then computed as the difference in the radiative fluxes between a perturbed atmospheric composition run and a control run of PORT. Here, we use PORT with prescribed atmospheric conditions from a present-day simulation of the finite-volume configuration of CAM4 at a horizontal resolution of about 2°. The BC distributions were also remapped to match this coarser resolution. Stratospheric adjustment is not considered here, as it is expected to be small for tropospheric aerosols (Hansen et al. 1997). We assume the errors associated with calculating F from a coarser model to be small for the globally averaged values. Ramanathan and Carmichael (2008) estimate that F caused by present-day BC from all sources ranges from 0.4 to 1.2 W m−2. More recent estimates of F from fossil fuel and biofuel burning alone stand at 0.19 W m−2 with a standard deviation of 0.08 W m−2 from a range of AeroCom Phase II models (Myhre et al. 2012). Bond et al. (2013) estimate F caused by a BC of 0.88 W m−2 with 90% uncertainty bounds ranging from 0.17 to 1.48 W m−2. Our estimate of all-sky F of present-day BC from the 1 × BC experiment, which includes biomass burning in addition to fossil fuel and biofuel burning, is lower than the estimate of Ramanathan and Carmichael (2008) and Bond et al. (2013).
While F estimates the radiative forcing caused by the direct effect, it excludes the semidirect effect of BC. The radiative flux due to the total semidirect effect (SDE) of BC can, however, be estimated as SDE = (RFPall sky − Fall sky) − (RFPclear sky − Fclear sky) following Lohmann et al. (2010). Estimates of SDE for all BC perturbation experiments are also listed in Table 1. Koch and Del Genio (2010) and Bond et al. (2013) infer SDE to range from −0.4 to −0.08 W m−2 from several modeling studies that report the efficacy of BC with regard to carbon dioxide. This estimate of SDE is the total semidirect effect of BC and also includes other climate responses in addition to the cloud responses, similar to our estimate. Our estimate for the SDE from the 1 × BC of 0.05 W m−2 from CAM4 thus lies outside the range of estimates from these modeling studies.
Both the RFP and F at the TOA increase linearly with the increase in BC (Figs. 2a,b). The error bars in Fig. 2 represent the standard error computed from the interannual standard deviation of annual-mean values from 5-yr RFP integrations. We do not estimate the error in F, which is computed from 1-yr integration. The RFP and F at the surface also increase linearly with the BC burden (Table 1). The SDE for the 1 × BC experiment is positive. It becomes strongly negative and increases linearly as the BC burden increases (Fig. 2c). The cloud response of RFP integrations indicates that while there is a linear decrease in the medium- and high-level clouds, there is a linear increase in the low-level clouds as BC burden increases (Figs. 2d–f). The radiative forcing due to these changes in clouds in the short 5-yr RFP runs is included as forcing in RFP values. Although the cloud forcing (RFPall sky − RFPclear sky) in all RFP integrations is positive, the negative SDE suggests that semidirect effects other than those on clouds, such as changes in the water vapor and lapse rate (e.g., Yoshimori and Broccoli 2008), are playing an important role as BC burden increases. An increase in shallow clouds largely results in a negative forcing caused by an increase in reflection of solar radiation and has little impact on longwave radiation. As the BC burden increases globally, these changes in clouds (Figs. 2d–f) suggest that the negative semidirect radiative effects of BC including those due to increases in low-level clouds dominate despite the decrease in mid- and high-level clouds. We plan to study the partitioning of the SDE of BC more closely in a future study.
Radiative forcing. Globally averaged annual-mean TOA radiative forcing due to BC burden in the 1 × BC, 2 × BC, 5 × BC, and 10 × BC experiments with reference to the 0 × BC experiment computed as (a) RFP and (b) F. (c) An estimate of SDE from all-sky and clear-sky RFP and F (see text for details). Fractional change in (d) mid-, (e) high-, and (f) low-level clouds in the RFP experiments. The x axes represent the amount of BC aerosol burden compared to the present day. The squares represent the simulated values, and the line represents a linear fit to those values. The slope m of the linear fits is also indicated. Error bars represent the standard error computed from the interannual std dev of annual-mean values from the last 5 years of the RFP integrations.
Citation: Journal of Climate 26, 20; 10.1175/JCLI-D-12-00715.1
4. Equilibrium response to increased BC
a. Surface temperature
Figure 3 shows the change in annual-mean surface temperatures in the 10 × BC experiment as compared to the 0 × BC experiment, the zonal-average surface temperature response for all BC perturbation experiments, and a scatterplot of global-average surface temperature change versus RFP. Most of the globe warms significantly, with warming in the NH being stronger than the SH (Fig. 3a). Figure 4 shows a partitioning of the net surface heat flux at equilibrium into net shortwave, longwave, sensible, and latent heat fluxes in the 10 × BC experiment as compared to the 0 × BC experiment. The net shortwave flux reaching the surface is reduced in BC experiments because of additional scattering and absorption in the troposphere. The heating of the land surface and the ocean mixed layer is caused largely by the reduced sensible and latent heat flux release from the surface (Figs. 4c,d). Statistically significant cooling occurs over strong source regions of India, tropical African savannas, and East Asia because of the large reduction in shortwave radiation reaching the surface (Fig. 4a). Statistically significant cooling also occurs in the downwind regions of the southeastern Atlantic Ocean caused by the increase in low-level clouds through the semidirect effect, as is discussed later. The increase in latent heat flux in the southern equatorial regions is largely associated with increased trade winds there accompanying the shift of the ITCZ into the warmer Northern Hemisphere (Chiang and Bitz 2005; Mahajan et al. 2010) as discussed in section 4b.
Surface temperature response to BC forcing. (a) Annual-mean surface temperature response of the 10 × BC experiment as compared to the 0 × BC experiment. (b) Zonally averaged surface temperature response for the 1 × BC, 2 × BC, 5 × BC, and 10 × BC experiments as compared to the 0 × BC experiment. (c) Scatterplot of change in surface temperature vs the RFP for the 1 × BC, 2 × BC, 5 × BC, and 10 × BC experiments. The m indicates the climate sensitivity to BC. Colors of the squares indicate the experiment as indexed in (b). Error bars represent the standard error based on the interannual std dev of annual-mean values. Standard errors of RFP are not plotted. Hatching in (a) indicates regions where the difference is statistically significant at the 95% confidence level based on a two-tailed Student's t test.
Citation: Journal of Climate 26, 20; 10.1175/JCLI-D-12-00715.1
Heat flux partitioning at the surface. Difference in the annual-mean surface (a) shortwave, (b) longwave, (c) sensible, and (d) latent heat fluxes between the 10 × BC and 0 × BC experiments. Positive values indicate upward heat fluxes from the surface to the atmosphere. Hatching indicates regions where the difference is statistically significant at the 95% confidence level based on a two-tailed Student's t test.
Citation: Journal of Climate 26, 20; 10.1175/JCLI-D-12-00715.1
This surface temperature response to the increase in tropospheric BC is seen in all the BC experiments, although with weaker amplitudes. The warming becomes stronger with BC burden as is seen in the zonal averages in Fig. 3b, and the global-average temperature change ΔT is essentially a linear function of the RFP (Fig. 3c). The error bars in Fig. 3c represent the standard errors computed from the interannual standard deviation of the annual-mean values of the 25-yr of CAM4–SOM integrations. The slope of a least squares–fitted line to the scatterplot of the change in equilibrium global-average temperature against the RFP indicates a climate sensitivity to BC λRFP of 0.42 K W−1 m2. Using F instead of RFP yields a climate sensitivity λF of 0.22 K W−1 m2. These values in the absence of sea ice feedbacks and effects of BC deposition are lower than other model estimates. Intermodel differences in climate sensitivity could arise from well-known differences in the simulation of feedback processes that amplify or dampen the initial radiative perturbations (e.g., Cubasch and Cess 1990; Soden and Held 2006). Roberts and Jones (2004) estimate λF to be 0.56 K W−1 m2 from the Met Office Hadley Centre Atmosphere Model, version 4 (HadAM4), coupled to an SOM simulating only the direct and semidirect effects of BC and forced with 4 times the then-current fossil fuel BC estimated emissions. The value of F in their experiment of 0.78 W m−2 lies between our estimates for the 2 × BC and 5 × BC experiments. Chung and Seinfeld (2005) found λF of 0.6 K W−1 m2 in the Goddard Institute for Space Studies (GISS) General Circulation Model II coupled to an SOM, which excluded the indirect effects. A version of the GFDL Climate Model, version 2.1 (CM2.1), coupled to an SOM including the aerosol indirect effects exhibits λRFP of 1.1 K W−1 m2 (Ming et al. 2010). Jones et al. (2007) found that λRFP and λF of BC were not statistically different from each other in the Hadley Centre Global Environment Model, version 1 (HadGEM1), coupled to an SOM, with a λRFP of 0.62 K W−1 m2. Using the standard errors of RFP, we find that for 1 × BC and 2 × BC, F is statistically indistinguishable from RFP based on a one-tailed Student's t test at the 95% confidence level. However, for the 5 × BC and the 10 × BC experiments the difference is statistically significant at the 95% confidence level.
b. ITCZ and precipitation
Figure 5a shows the change in annual-mean convective precipitation in the 10 × BC experiment as compared to the 0 × BC experiment. Also, Figs. 5b and 5c show the change in zonal-average convective precipitation and the simulated northward heat transport computed from the TOA radiative fluxes for all BC perturbation experiments. The radiative forcing is greater in the Northern Hemisphere in the BC perturbation experiments. This interhemispheric imbalance induces a cross-equatorial southward atmospheric heat transport, which results in the northward migration of the ITCZ. A northward shift in the ITCZ is noted in all seasons, with increased (decreased) rainfall in the Northern (Southern) Hemisphere. This precipitation response is in contrast to the response to anthropogenic sulfate aerosols, where cooling of the Northern Hemisphere results in a northern cross-equatorial heat transport and a southward shift in the ITCZ (Chen and Ramaswamy 1996; Ramaswamy and Chen 1997; Rotstayn and Lohmann 2002; Bollasina et al. 2011). Figure 5d illustrates the linearity of the cross-equatorial northward heat transport response to the BC burden, with increased southward heat transport as the BC burden increases. Modeling studies (e.g., Yoshimori and Broccoli 2008; Frierson and Hwang 2012) show that the cross-equatorial transport is linearly correlated with the shift in ITCZ. We also find the same in our BC perturbation experiments. Figure 5e shows a scatterplot of the shift in the ITCZ center versus the change in cross-equatorial northward heat transport. The ITCZ center is computed as the centroid of zonally averaged convective precipitation in the deep tropical region (15°S–15°N) as in Frierson and Hwang (2012). The ITCZ center shifts to the north as the BC-induced cross-equatorial southward transport increases at the rate of about 4° PW−1.
Linearity of precipitation response. (a) Annual-mean convective precipitation response of the 10 × BC experiment compared to the 0 × BC experiment. Hatching indicates regions where the difference is statistically significant at the 95% confidence level based on a two-tailed Student's t test. Zonally averaged response for the 1 × BC, 2 × BC, 5 × BC, and 10 × BC experiments as compared to the 0 × BC experiment for (b) convective precipitation and (c) northward heat transport (positive values indicate northward). (d) Change in annual-mean cross-equatorial northward heat transport as compared to the 0 × BC experiment. The x axis represents the amount of BC aerosol burden with reference to the present day. Also shown are scatterplots of (e) ITCZ shift vs change in cross-equatorial northward heat transport, (f) fractional change in global-average total precipitation in CAM4–SOM BC experiments, (g) RFP experiments vs change in global-average atmospheric absorption, and (h) slow precipitation response vs change in global-average surface temperature. The squares in (d)–(h) represent the simulated values, and the line represents a linear fit to those values. The m of the linear fits is also indicated. Colors of the squares indicate the experiment as indexed in (b). Error bars represent the standard error of ordinate values based on the interannual std dev of annual-mean values. Standard errors of abscissa values are not plotted.
Citation: Journal of Climate 26, 20; 10.1175/JCLI-D-12-00715.1
Global-average precipitation is found to decrease with increase in BC despite the increase in equilibrium surface temperatures (Table 2, Fig. 5f). Recent studies (e.g., Yoshimori and Broccoli 2008; Andrews et al. 2010; Ming et al. 2010) have identified dual roles of BC in the precipitation response. While surface warming increases precipitation, absorption of solar radiation by BC in the troposphere suppresses it. Idealized experiments reveal that the role that dominates depends on the vertical distribution of aerosols (Ming et al. 2010). The BC perturbation experiments here reveal that for a relatively realistic vertical distribution of spatially varying BC, the suppression effect dominates more as the burden increases. We find that the change in global-average precipitation ΔP in CAM4–SOM is linearly related to the global-average atmospheric absorption ΔAA by BC with precipitation decreasing at a rate of 0.4% W−1 m2 (Fig. 5e), where ΔAA is computed as the difference in the TOA and surface RFP (Table 1).
Global-average precipitation response ΔP (%) at equilibrium in CAM4–SOM simulations, in the RFP integrations ΔPfast, and their difference ΔPslow.
The global-average precipitation response in the RFP integrations, termed the fast precipitation response ΔPfast [following the nomenclature used by Andrews et al. (2010)] is also linearly related to ΔAA, but at a stronger rate of 0.8% W−1 m2 (Fig. 5g). The linear relationship between ΔPfast and ΔAA supports the findings of Andrews et al. (2010) that atmospheric absorption is a good predictor of ΔPfast. Assuming that the total precipitation change at equilibrium ΔP can be represented as the sum of the slow adjustment and a fast component (ΔP = ΔPslow + ΔPfast) following Bala et al. (2010), we find that ΔPslow is positive for all BC experiments (Table 2). Further, a least squares linear fit to the scatterplot of ΔPslow versus ΔT indicates that the sensitivity ΔPslow/ΔT = 3.4% K−1 (Fig. 5). Andrews et al. (2010) showed that while ΔP/ΔT varies strongly with the forcing mechanism, ΔPslow/ΔT is independent of the radiative forcing agent in HadGEM1 coupled to an SOM and ranges from 2% to 3% K−1. Our estimate of 3.4% K−1 from CAM4–SOM is close to that range providing support to their result. Recent studies have also shown that the simulated equilibrium precipitation response in different idealized and future climate scenarios can be better explained by linear formulations that also include the effect of atmospheric absorption by radiative agents (Ming et al. 2010; Frieler et al. 2011). The inclusion of a ΔAA term mitigates the increase in precipitation expected only from surface temperature increases in different models. Our simulations support that result by clearly identifying the linear decrease in ΔPfast with a BC-induced increase in ΔAA and a linear increase in ΔPslow with an increase in ΔT.
c. Clouds
Similar to RFP integrations, CAM4–SOM also exhibits a linear decrease in mid- and high-level clouds and an increase in the low-level clouds as the BC burden increases. Figures 6a–c show the scatterplot of global-average annual-mean mid-, high-, and low-level clouds versus the RFP. Linear fits to a scatterplot of cloud changes in CAM4–SOM versus RFP integrations (not shown) reveal that while the ratio of change in high-level clouds is close to unity in the two experiment sets, changes in mid- and low-level clouds are amplified in CAM4–SOM by a factor of 1.67 and 2.11, respectively. These changes in clouds at equilibrium are associated with a linear decrease in global-average longwave cloud forcing (cooling) and a nearly equal increase in shortwave cloud forcing (warming, despite the increase in low-level clouds) shown in Figs. 6d and 6e. The net cloud forcing is thus statistically indistinguishable from zero in all experiments (Fig. 6f). Figure 7 shows the spatial distribution of the change in the annual-mean clouds in the 10 × BC experiments as compared to the 0 × BC experiment. The change in mid- and high-level clouds essentially follows the shift in ITCZ in the tropics with an increase in the northern tropical region and a decrease in the southern tropical regions. Significant decreases also occur over the northern midlatitude regions. Allen and Sherwood (2010) find that these changes in mid- and high-level clouds are caused by changes in the relative humidity due to atmospheric heating in CAM, version 3. We do not investigate these changes further here, but focus on the low-level stratocumulus clouds. The low-level clouds show an increase in the eastern tropical oceans off the coasts, where the marine stratocumulus clouds are the dominant cloud forms. The strong amplification of the low-level cloud response at equilibrium in CAM4–SOM when compared to RFP integrations suggests a strong role for air–sea thermodynamic feedbacks in the development of low-level clouds.
Cloud response. Scatterplot of change in annual-mean global-average (a) mid- (vertically integrated from 700- to 400-mb model level), (b) high- (vertically integrated from 400-mb model level to the model top), and (c) low-level cloud fraction (vertically integrated from surface to the 700-mb model level), and (d) longwave, (e) shortwave, and (f) net cloud forcing in the 1 × BC, 2 × BC, 5 × BC, and 10 × BC experiments with reference to the 0 × BC experiment vs RFP. The m of the linear fits is also indicated. Error bars represent the standard error of ordinate values based on the interannual standard deviation of annual-mean values. Standard errors of abscissa values are not plotted. Colors of the squares indicate the experiment as indexed in (a).
Citation: Journal of Climate 26, 20; 10.1175/JCLI-D-12-00715.1
Cloud response. Difference in (a) mid-, (b) high-, and (c) low-level cloud fraction between the 10 × BC and 0 × BC experiments.
Citation: Journal of Climate 26, 20; 10.1175/JCLI-D-12-00715.1
Previous studies have established that the subtropical marine stratocumulus decks strongly depend on the LTS (e.g., Klein and Hartmann 1993) and are also sustained by the SST–stratus feedback (e.g., Philander et al. 1996). In CAM4, the marine stratocumulus cloud fraction is parameterized to be linearly dependent on the LTS, which is defined as the local potential temperature difference between the lower troposphere at 700-mb pressure level and the surface. The cloud forms in the model vertical layer below the layer with the strongest stability jump between the 750-mb level and the surface (Neale et al. 2010).
Over the southeastern tropical Atlantic region, BC above the boundary layer heats up the free troposphere. Differential heating between the boundary layer and the free troposphere increases LTS, favoring the formation of marine stratocumulus clouds. Figure 8a shows the increase in low-level clouds in the 10 × BC experiment over the southeastern tropical Atlantic in the boreal summer, when the BC burden is the largest. Figure 8b shows the meridional average of the low-level clouds over the southern tropical Atlantic Ocean for all the experiments. Increasing BC results in a linear increase in the low-level cloud fraction (not shown) following the increase in stratification. Further, the westward extent of the low-level clouds also increases with the amount of BC. For example, while the low-level cloud fraction of 0.4 was found at 5°W in the 2 × BC experiment, it is found at about 15° and 20°W in the 5 × BC and 10 × BC experiments, respectively. Considering only the local change in LTS, a progressively increasing amount of free tropospheric BC over the westward stratocumulus regions in the experiments increases LTS and hence the stratocumulus cloud cover. An increase in marine stratocumulus cloud cover and its westward extent during high smoke days has also been observed over the southeastern tropical Atlantic using satellite-derived data (e.g., Kaufman et al. 2005).
Marine stratocumulus response over the southeastern tropical Atlantic in the boreal summer [June–August (JJA)]. (a) Low-level cloud fraction response in the 10 × BC experiment as compared to the 0 × BC experiment. (b) Meridional average of low-level cloud fraction for all experiments over the southern tropical Atlantic (25°–5°S, 35°W–10°E, ocean-only grid points). Vertical profile of the regional-average change in (c) stratocumulus cloud fraction, (d) diabatic shortwave heating (K day−1), and (e) temperature (K) over the boxed region in (a) (20°–10°S, 15°W–0°).
Citation: Journal of Climate 26, 20; 10.1175/JCLI-D-12-00715.1
Figures 8c–e show the vertical profile of change in marine stratocumulus clouds, the solar heating forcing, and the temperature response averaged over a region of the southeastern tropical Atlantic Ocean for all the BC experiments as compared to the 0 × BC experiment. An increase in air temperature due to heating at the 700-mb level increases the LTS, resulting in an increase in the marine stratocumulus in the BC simulations. The largest increase in temperature occurs just above the 800-mb level. The largest increase in marine stratocumulus hence occurs just below it. At the 800-mb level, the stratocumulus clouds decrease. It should be noted that a large part of the increase in marine stratocumulus clouds is already seen in the RFP integrations, which includes the warming of the free troposphere by BC as a fast atmospheric process, and considered as a radiative forcing in the RFP calculations. However, the complete response of the marine stratocumulus in equilibrium is captured in the CAM4–SOM runs, where the SST–stratus feedback operates. The increase in marine stratocumulus clouds reduces the SST beneath because of a reduction in the shortwave radiation reaching the surface. The resulting increase in LTS causes a further increase in the stratocumulus clouds. Over the southeastern tropical Atlantic, where the low-level cloud response is the largest, a local decrease in surface shortwave radiative flux of up to 40 W m−2 (Fig. 4a) is associated with a local cooling of the oceanic mixed layer beneath of up to about 0.4 K (Fig. 3a). At equilibrium, the strong decrease in surface shortwave radiation is largely balanced by a decrease in the latent heat fluxes there (Figs. 4a,d).
Similar, though smaller, increases in low-level stratocumulus cloud fraction are found over other subtropical regions including southeastern tropical Pacific, northeastern tropical Pacific, and northeastern tropical Atlantic. The increase in lower tropospheric stratification also increases the stratocumulus cloud fraction over the midlatitude Atlantic and Pacific. Changes in LTS can be induced by the vertical profile of radiative heating and cooling, latent heat release from condensation, diffusion, and advection. We do not explore all these mechanisms here but refer the interested reader to Persad et al. (2012), who partitioned heating rates over the southeastern tropical Pacific in response to increases in BC.
5. Summary and discussion
Progressively increasing BC aerosol concentration from 0 × BC to 10 × BC while maintaining a spatially varying estimated present-day distribution in CAM4–SOM results in a linear increase in global warming, decrease in global precipitation, northward shift of the ITCZ, decrease in mid- and high-level clouds, and increase in low-level clouds. CAM4–SOM only simulates the direct and semidirect effects of BC, and these simulations also exclude BC deposition changes and changes in sea ice. Our results broadly support the findings of Jones et al. (2007) that climate response is linearly additive for aerosols forcings, although specifically for BC. We also find support for several other studies (e.g., Andrews et al. 2010; Ming et al. 2010; Frieler et al. 2011) that the quick reduction in precipitation from atmospheric heating caused by BC is stronger than the increase in precipitation associated with surface warming on longer time scales. We also find that the increase in shallow marine stratocumulus cloud cover over the southeastern tropical Atlantic with increasing BC is largely caused by an increase in the lower tropospheric stratification.
Large model biases exist in the simulated BC radiative forcing in global aerosol models, along with a large uncertainty in the estimates of the total burden as well as spatial distribution of BC (e.g., Myhre et al. 2012; Bond et al. 2013). We have explored a range of BC burden scaled from an estimate of the present-day distribution. Although we scale the current BC burden up to a factor of 10, the radiative forcing from these idealized forcings in CAM4 is still close to the estimated range of realistic total radiative forcing from BC (Bond et al. 2013). The linearity of climate response to BC forcing implies that we can quantify the uncertainty of global-mean climate response due to direct and semidirect effects of BC given the uncertainty in BC burden, without conducting expensive GCM simulations. Our results confirm that BC can potentially contribute to substantial global warming and hydrological changes, with CAM4–SOM simulating a climate sensitivity of 0.4 K W−1 m2 using TOA RFP and 0.2 K W−1 m2 using TOA F and a global precipitation sensitivity to atmospheric absorption of −0.4% W−1 m2. It is important to improve our estimate of current BC burden and distribution in addition to improving our understanding of BC–climate interactions to quantify their effects more accurately in order to inform mitigation efforts. Further, the radiative forcing of BC is opposite to that of sulfate aerosols, which causes cooling. The interhemispheric contrast in sulfate aerosols causes a southward shift of the ITCZ, which is again the opposite effect of BC. These contrasting responses imply that aerosol mitigation efforts should account for climate impacts of each of the individual major species.
Our experiments with CAM4 exclude the indirect effects of aerosols by which aerosols impact cloud droplet/ice number concentration and cloud droplet sizes and also invigorate convection. Further, CAM4 also ignores the internal mixing of BC with other aerosols, which can substantially affect the direct radiative forcing. It is important to note that our experiments differ from other modeling studies of BC widely in terms of its distribution, mixing state, simulated radiative effects, and black carbon deposition. Our comparisons with these studies thus should be considered in context of these differences.
Some more caveats of our study need to be mentioned. The warming of the Northern Hemisphere along with deposition of more black carbon on Arctic sea ice would result in melting that will be amplified by sea ice–related feedbacks. Changes in the Arctic have been shown to impact the tropics, with the ITCZ shifting into the warmer hemisphere (Chiang and Bitz 2005; Mahajan et al. 2011). The simulated response in our BC experiments is, thus, conservative and it would probably amplify further in the presence of interactive sea ice and corresponding increases in aerosol deposition. The sea ice feedbacks might also result in a nonlinear climate response to black carbon forcing. Also, while previous studies suggest similarities in climate sensitivities between an AGCM coupled to an SOM and fully coupled models (e.g., Bitz et al. 2012), it will be interesting to test the robustness of our results in fully coupled models.
Acknowledgments
This work was funded by a grant from the Office of Science [Biological and Environmental Research (BER)] of the U.S. Department of Energy (DOE). This research used resources of the Oak Ridge Leadership Computing Facility at the Oak Ridge National Laboratory, which is supported by the Office of Science of the U.S. Department of Energy under Contract DE-AC05-00OR22725. The authors wish to thank three anonymous reviewers whose comments helped improve the manuscript significantly.
REFERENCES
Ackerman, A. S., O. B. Toon, D. E. Stevens, A. J. Heymsfield, V. Ramanathan, and E. J. Welton, 2000: Reduction of tropical cloudiness by soot. Science, 288, 1042–1047, doi:10.1126/science.288.5468.1042.
Allen, R. J., and S. C. Sherwood, 2010: Aerosol–cloud semi-direct effect and land–sea temperature contrast in a GCM. Geophys. Res. Lett.,37, L07702, doi:10.1029/2010GL042759.
Allen, R. J., and S. C. Sherwood, 2011: The impact of natural versus anthropogenic aerosols on atmospheric circulation in the Community Atmosphere Model. Climate Dyn.,36, 1959–1978, doi:10.1007/s00382-010-0898-8.
Andrews, T., P. M. Forster, O. Boucher, N. Bellouin, and A. Jones, 2010: Precipitation, radiative forcing, and global temperature change. Geophys. Res. Lett.,37, L14701, doi:10.1029/2010GL043991.
Bala, G., K. Caldeira, and R. Nemani, 2010: Fast versus slow response in climate change: Implications for the global hydrological cycle. Climate Dyn.,35, 423–434, doi:10.1007/s00382-009-0583-y.
Bitz, C. M., K. M. Shell, P. R. Gent, D. A. Bailey, G. Danabasoglu, K. C. Armour, M. M. Holland, and J. T. Kiehl, 2012: Climate sensitivity of the Community Climate System Model, version 4. J. Climate,25, 3053–3070.
Bollasina, M. A., Y. Ming, and V. Ramaswamy, 2011: Anthropogenic aerosols and the weakening of the South Asian summer monsoon. Science, 334, 502–505.
Bond, T. C., and Coauthors, 2013: Bounding the role of black carbon in the climate system: A scientific assessment. J. Geophys. Res.,118, 5380–5552, doi:10.1002/jgrd.50171.
Chen, C.-T., and V. Ramaswamy, 1996: Sensitivity of simulated global climate to perturbations in low cloud microphysical properties. Part II: Spatially localized perturbations. J. Climate,9, 2788–2801.
Chiang, J. C. H., and C. Bitz, 2005: Influence of high latitude ice cover on the marine intertropical convergence zone. Climate Dyn., 25, 477–496.
Chung, S. H., and J. H. Seinfeld, 2005: Climate response of direct radiative forcing of anthropogenic black carbon. J. Geophys. Res.,110, D11102, doi:10.1029/2004JD005441.
Conley, A. J., J.-F. Lamarque, F. Vitt, W. D. Collins, and J. Kiehl, 2012: PORT, a CESM tool for the diagnosis of radiative forcing. Geosci. Model Dev. Discuss.,5, 2687–2704, doi:10.5194/gmdd-5-2687-2012.
Cubasch, J., and R. D. Cess, 1990: Processes and modeling climate change: The IPCC scientific assessment. Climate Change: The IPCC Scientific Assessment, J. T. Houghton, G. J. Jenkins, and J. J. Ephraums, Eds., Cambridge University Press, 69–72.
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.
Frieler, K., M. Meinshausen, T. Schneider von Deimling, T. Andrews, and P. Forster, 2011: Changes in global-mean precipitation in response to warming, greenhouse gas forcing and black carbon. Geophys. Res. Lett.,38, L04702, doi:10.1029/2010GL045953.
Frierson, D. M. W., and Y.-T. Hwang, 2012: Extratropical influence on ITCZ shifts in slab ocean simulations of global warming. J. Climate,25, 720–733.
Hansen, J., M. Sato, and R. Ruedy, 1997: Radiative forcing and climate response. J. Geophys. Res.,102 (D6), 6831–6864.
Haywood, J. M., L. J. Donner, A. Jones, and J.-C. Golaz, 2009: Global indirect radiative forcing caused by aerosols: IPCC (2007) and beyond. Clouds in the Perturbed Climate System, J. Heintzenberg and R. J. Charlson, Eds., MIT Press, 451–467.
Held, I. M., and B. J. Soden, 2006: Robust responses of the hydrological cycle to global warming. J. Climate,19, 5686–5699.
Johnson, B. T., K. P. Shine, and P. M. Forster, 2004: The semi-direct aerosol effect: Impact of absorbing aerosols on marine stratocumulus. Quart. J. Roy. Meteor. Soc.,130, 1407–1422, doi:10.1256/qj.03.61.
Jones, A., J. M. Haywood, and O. Boucher, 2007: Aerosol forcing, climate response and climate sensitivity in the Hadley Centre climate model. J. Geophys. Res.,112, D20211, doi:10.1029/2007JD008688.
Kaufman, Y. J., I. Koren, L. A. Remer, D. Rosenfeld, and Y. Rudich, 2005: The effect of smoke, dust, and pollution aerosol on shallow cloud development over the Atlantic Ocean. Proc. Natl. Acad. Sci. USA, 102, 11 207–11 212.
Klein, S. A., and D. L. Hartmann, 1993: The seasonal cycle of low stratiform clouds. J. Climate,6, 1587–1606.
Koch, D., and A. D. Del Genio, 2010: Black carbon semi-direct effects on cloud cover: Review and synthesis. Atmos. Chem. Phys., 10, 7685–7696, doi:10.5194/acp-10-7685-2010.
Koch, D., and Coauthors, 2009: Evaluation of black carbon estimations in global aerosol models. Atmos. Chem. Phys.,9, 9001–9026, doi:10.5194/acp-9-9001-2009.
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.,10, 7017–7039, doi:10.5194/acp-10-7017-2010.
Levy, H., M. D. Schwarzkopf, L. Horowitz, V. Ramaswamy, and K. L. Findell, 2008: Strong sensitivity of late 21st century climate to projected changes in short-lived air pollutants. J. Geophys. Res.,113, D06102, doi:10.1029/2007JD009176.
Lohmann, U., and Coauthors, 2010: Total aerosol effect: Radiative forcing or radiative flux perturbation? Atmos. Chem. Phys., 10, 3235–3246, doi:10.5194/acp-10-3235-2010.
Mahajan, S., R. Saravanan, and P. Chang, 2010: Free and forced variability of the tropical Atlantic Ocean: Role of the wind–evaporation–sea surface temperature feedback. J. Climate,23, 5958–5977.
Mahajan, S., R. Saravanan, and P. Chang, 2011: The role of the wind–evaporation–sea surface temperature (WES) feedback as a thermodynamic pathway for the equatorward propagation of high-latitude sea ice–induced cold anomalies. J. Climate,24, 1350–1361.
Mahajan, S., K. J. Evans, J. E. Truesdale, J. J. Hack, and J.-F. Lamarque, 2012: Interannual tropospheric aerosol variability in late twentieth century and its impact on tropical Atlantic and West African climate by direct and semi-direct effects. J. Climate, 25, 8031–8056.
Ming, Y., and V. Ramaswamy, 2009: Nonlinear climate and hydrological responses to aerosol effects. J. Climate,22, 1329–1339.
Ming, Y., V. Ramaswamy, and G. Persad, 2010: Two opposing effects of absorbing aerosols on global-mean precipitation. Geophys. Res. Lett.,37, L13701, doi:10.1029/2010GL042895.
Myhre, G., and Coauthors, 2012: Radiative forcing of the direct aerosol effect from AeroCom Phase II simulations. Atmos. Chem. Phys. Discuss., 12, 22 355–22 413, doi:10.5194/acpd-12-22355-2012.
Neale, R. B., and Coauthors, 2010: Description of the NCAR Community Atmosphere Model (CAM4). NCAR Tech. Rep. NCAR/TN+STR-485, 120 pp.
Persad, G. G., Y. Ming, and V. Ramaswamy, 2012: Tropical tropospheric–only responses to absorbing aerosols. J. Climate,25, 2471–2480.
Philander, S. G. H., D. Gu, G. Lambert, T. Li, D. Halpern, N. C. Lau, and R. C. Pacanowski, 1996: Why the ITCZ is mostly north of the equator. J. Climate, 9, 2958–2972.
Ramanathan, V., and G. Carmichael, 2008: Global and regional climate changes due to black carbon. Nat. Geosci.,1, 221–227, doi:10.1038/ngeo156.
Ramaswamy, V., and C.-T. Chen, 1997: Climate forcing–response relationships for greenhouse and shortwave radiative perturbations. Geophys. Res. Lett.,24, 667–670.
Roberts, D. L., and A. Jones, 2004: Climate sensitivity to black carbon aerosol from fossil fuel combustion. J. Geophys. Res.,109, D16202, doi:10.1029/2004JD004676.
Rotstayn, L. D., and U. Lohmann, 2002: Tropical rainfall trends and the indirect aerosol effect. J. Climate,15, 2103–2116.
Sakaeda, N., R. Wood, and P. J. Rasch, 2011: Direct and semidirect aerosol effects of southern African biomass burning aerosol. J. Geophys. Res.,116, D12205, doi:10.1029/2010JD015540.
Schultz, M. G., and Coauthors, 2008: Global wildland fire emissions from 1960 to 2000. Global Biogeochem. Cycles,22, GB2002, doi:10.1029/2007GB003031.
Shindell, D. T., H. Levy, M. D. Schwarzkopf, L. W. Horowitz, J.-F. Lamarque, and G. Faluvegi, 2008: Multimodel projections of climate change from short-lived emissions due to human activities. J. Geophys. Res.,113, D11109, doi:10.1029/2007JD009152.
Shiogama, H., S. Emori, K. Takahashi, T. Nagashima, T. Ogura, T. Nozawa, and T. Takemura, 2010: Emission scenario dependency of precipitation on global warming in the MIROC3.2 model. J. Climate,23, 2404–2417.
Soden, B. J., and I. M. Held, 2006: An assessment of climate feedbacks in coupled ocean–atmosphere models. J. Climate,19, 3354–3360.
Tie, X., and Coauthors, 2005: Assessment of the global impact of aerosols on tropospheric oxidants. J. Geophys. Res.,110, D03204, doi:10.1029/2004JD005359.
Wang, C., 2007: Impact of direct radiative forcing of black carbon aerosols on tropical convective precipitation. Geophys. Res. Lett.,34, L05709, doi:10.1029/2006GL028416.
Wilcox, E. M., 2010: Stratocumulus cloud thickening beneath layers of absorbing smoke aerosol. Atmos. Chem. Phys., 10, 11 769–11 777, doi:10.5194/acp-10-11769-2010.
Yoshimori, M., and A. J. Broccoli, 2008: Equilibrium response of an atmosphere–mixed layer ocean model to different radiative forcing agents: Global and zonal mean response. J. Climate,21, 4399–4423.
