SEARCH

SEARCH BY CITATION

Keywords:

  • black carbon;
  • aerosols;
  • climate sensitivity

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model Description
  5. 3. Experimental Design
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Appendix A:: Parameterization of Particle Scavenging by Clouds
  10. Acknowledgments
  11. References

[1] Black carbon aerosol particles from fossil fuel combustion are good absorbers of solar radiation and hence exert a positive radiative forcing, reinforcing the warming due to anthropogenic increases in CO2 and other greenhouse gases. However, it is unclear how the climate sensitivity to black carbon aerosol forcing compares with the sensitivity to greenhouse gas forcing. Here we investigate this question using the HadSM4 configuration of the Hadley Centre climate model, extended by the addition of interactive black carbon and sulphate aerosol schemes. The results confirm earlier suggestions that the climate sensitivities are not necessarily similar and indicate that the black carbon sensitivity may be weaker. Possible reasons for this are explored by studying several feedback mechanisms operating in the model.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model Description
  5. 3. Experimental Design
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Appendix A:: Parameterization of Particle Scavenging by Clouds
  10. Acknowledgments
  11. References

[2] The idea of quantifying the size of an externally imposed perturbation to the climate system in terms of its radiative forcing has been very useful in climate change studies [see, e.g., Ramaswamy et al., 2001]. However, from a practical viewpoint it is the response of the system to the perturbation that is of most interest. One may define the climate sensitivity λ as the change in annually averaged global mean surface temperature in °C per W m−2 of global annual mean radiative forcing. The frequent practice of comparing the radiative forcings due to different perturbations tacitly assumes that the associated climate sensitivities are not too different, for otherwise the relative magnitudes of the forcings may give a misleading impression of the relative significance of the perturbations. As reviewed by Ramaswamy et al. [2001], this assumption is often, but not always, justified.

[3] Black carbon (BC) aerosol from fossil fuel combustion exerts a positive radiative forcing and therefore reinforces the warming due to anthropogenic increases in greenhouse gas concentrations. In principle, a reduction in black carbon emissions is thus desirable and would make some contribution to slowing the rate of climate change (as well as having other benefits such as an improvement in air quality). This point has been well argued by Hansen and Sato [2001], who nevertheless stressed that a reduction in the rate of CO2 emissions remains important. Jacobson [2002] also emphasized the value of reducing black carbon emissions, but raised a further issue, namely that in some circumstances a reduction of CO2 emissions may be associated with an increase of black carbon emissions, leading to an apparent policy dilemma. In fact, Jacobson pointed out that although cars powered by diesel engines tend to emit, other things being equal, less CO2 per distance traveled than cars with petrol (gasoline) engines, diesel cars also emit much more BC.

[4] When comparing the climatic impact of any sources of BC and CO2 emissions, one must remember that there is a huge difference in their atmospheric lifetimes. Unless emitted into the stratosphere, e.g., by aircraft, BC particles may remain airborne for up to about 15 days, whereas anthropogenic CO2 emitted now will result in elevated atmospheric concentrations for centuries to come. As a consequence, there is some timescale τ such that for times less than τ, current BC emissions have a greater warming effect than current CO2 emissions; while at times greater than τ it is current CO2 emissions that are more significant. (Note that this statement contains no reference to the future course of emissions and is independent of it.) Of course, the size of τ depends on several imperfectly known factors, including the relative emission rates, the location of the BC emissions, the radiative forcings due to BC and CO2, and the associated climate sensitivities. In any specific case there will thus be some room for debate about the size of τ; a debate which is of more than academic interest as it may affect policy decisions. Under certain assumptions regarding engine emission factors, Jacobson [2002] estimated that in comparing petrol and diesel cars the time horizon beyond which diesel cars have less impact on the climate is further than a century.

[5] One of the interesting aspects of Jacobson's results is that they appear to indicate that the climate sensitivity to black carbon forcing, λBC, is significantly greater than the climate sensitivity to CO2 forcing, λC. (This remark is based on Jacobson's values for the climate response per unit direct forcing given in paragraph 63 of his paper.) Jacobson [2002] suggested that this is because BC forcing excites several feedbacks in the climate system that CO2 (and other greenhouse gases such as methane) forcing does not. This interpretation is very plausible, but since the result could be highly model dependent, it is important to investigate whether it is also obtained in other climate models. The object of the present paper is to study this question using the Hadley Centre climate model. Note that although the magnitude of the BC forcing is also a relevant issue, that will not be the main focus here. It should also be remarked that λBC and λC are not strictly independent, because (for example) λBC is affected by the distribution of sea ice and continental snow cover, which is likely to be much reduced in a climate warmed by increased CO2. Hence the responses to BC and CO2 are not simply additive; however, the investigation of this issue is outside the scope of the present study.

[6] An important paper in this context was written by Hansen et al. [1997], who used an idealized (“Wonderland”) climate model driven by a wide variety of prescribed perturbation forcings to explore the relationship between radiative forcing and climate response. They found that absorbing tropospheric aerosols (such as BC) “are a case in which the fixed proportionality between radiative forcing and climate response can break down,” a major underlying reason being the reduction of large-scale cloud due to extra atmospheric heating by the aerosol. Hansen et al. [1997] labeled this mechanism the “semidirect” aerosol effect. The relevance of this to the present work will be seen later. A more recent idealized study by Cook and Highwood [2004], using a somewhat more sophisticated model, also highlighted the role of cloud feedbacks in modulating the response to forcing by absorbing aerosols. Cook and Highwood investigated the effect of varying the aerosol single scatter albedo and found that for certain intermediate values of single scatter albedo, the radiative forcing and surface temperature change in their model were of opposite sign, implying a negative climate sensitivity.

[7] There have been a considerable number of previous studies in which the global distribution of fossil fuel black carbon aerosol has been modeled. These have often been combined with investigations of biomass burning aerosol, though for the sake of brevity this will not be mentioned further here. As far as can be ascertained, work in this field was initiated by Penner et al. [1993] using the Lawrence Livermore chemical transport model GRANTOUR. This model was also used, with an improved emissions inventory, in the study by Liousse et al. [1996]. Simultaneously, Cooke and Wilson [1996] published independent emissions data sets and used the MOGUNTIA chemical transport model to simulate the distribution of BC. Although there had been previous estimates of the radiative forcing due to prescribed distributions of BC [e.g., Haywood et al., 1997; Schult et al., 1997; Myhre et al., 1998], the first such estimate in which the BC distribution was interactively simulated using a three-dimensional global model was obtained by Penner et al. [1998], by coupling the GRANTOUR transport model to a version of the NCAR Community Climate Model. Their estimate of the global annual mean direct forcing due to fossil fuel BC was 0.20 W m−2, and it is interesting to note that this value remains the central estimate in the most recent report of Working Group I of the Intergovernmental Panel on Climate Change [see Ramaswamy et al., 2001]. Further studies have included those of Seland and Iversen [1999], Kirkevåg et al. [1999], Tegen et al. [2000], Koch [2001], Lohmann and Feichter [2001], Jacobson [2001], and Chung and Seinfeld [2002]. The paper by Tegen et al. [2000] investigated the evolution of the direct radiative forcing due to black carbon (and other aerosol species) from 1950 to 1990, and Koch [2001] investigated its possible evolution in the 21st century. The results of these two papers illustrate that significant temporal evolution of BC forcing is highly probable, but for simplicity this is not taken into account in the present study. Another interesting feature of the work by Koch [2001] is the comparison of three parameterizations of the formation of hydrophilic BC particles, including one which explicitly depends on the presence of sulphate. Wang [2004] has recently investigated the climate response to BC aerosol from both fossil fuel combustion and biomass burning, using an interactive aerosol-climate model broadly similar to the one to be presented here, though many of the details are of course different. In contrast to Jacobson [2002], Wang [2004] found that the results did not suggest a significant change in global mean surface temperature resulting from BC forcings, though regional changes were significant. Although Wang did not quote a value of λBC, his results appear to imply a relatively low central estimate of about 0.26 K W−1 m2, albeit with a fairly wide uncertainty range (0.06 to 0.47 K W−1 m2).

[8] A recent study by Sato et al. [2003] compared aerosol absorption in two models with Sun photometer data from the AERONET network, and concluded that current models typically underestimate the radiative forcing due to BC (including BC in biomass smoke) by a factor of at least 2. Sato et al. remarked that the results were reasonably consistent with an estimate of 0.5 W m−2 for radiative forcing from fossil fuel BC. If this is so, the need to evaluate the impact of BC forcing is reinforced.

[9] The rest of the paper is structured as follows. The next section describes the relevant aspects of the version of the Hadley Centre climate model we used. Section 3 describes the experimental design, and the results are presented in section 4, and discussed in section 5. Finally, the conclusions are summarized in section 6.

2. Model Description

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model Description
  5. 3. Experimental Design
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Appendix A:: Parameterization of Particle Scavenging by Clouds
  10. Acknowledgments
  11. References

[10] The model used in this study is based on the HadAM4 configuration of the Met Office's Unified Forecast-Climate model, described by Webb et al. [2001]. This is a finite difference grid point model with a meridional resolution of 2.5°, a zonal resolution of 3.75°, and 38 levels. The radiation scheme is the two-stream code by Edwards and Slingo [1996], with 6 spectral bands in the shortwave region of the spectrum and 9 in the longwave. For radiative purposes, stratiform cloud decks are assumed to overlap maximally when the clouds are in contiguous layers and randomly otherwise; convective clouds are treated as vertically coherent.

[11] As in the study by Williams et al. [2001], the atmospheric model is coupled to a 50m deep mixed layer ocean model, together with an interactive sea ice model, forming the “slab” model HadSM4. This slab model is a useful tool for exploring the equilibrium response to climate perturbations without the expense involved in running a full coupled atmosphere-ocean model, and similar models have been used in many previous investigations by a large number of modeling groups. Previous studies have shown that the atmospheric response of a slab model is a reasonable proxy for the equilibrium response of the corresponding coupled model, as discussed by Williams et al. [2001].

[12] For the purposes of this investigation, HadSM4 was extended in three ways. An interactive scheme for simulating the distribution of black carbon aerosol was essential, and details of this are given below. Second, an interactive sulphate aerosol scheme was used. The main reasons for this were to include the indirect aerosol effects on (liquid) clouds, and to enable the in-cloud scavenging of BC aerosol to be parameterized in terms of the interactively computed cloud droplet number concentration. Note that the latter effect is the only way in which sulphate and BC aerosols explicitly interact in this model, though since both aerosols alter the climate of the model, they consequently influence each other's distributions obliquely. Third, the parameterization of the albedo of snow lying on land was changed to a scheme based on the work of Marshall [1989]. This includes a dependence on the solar zenith angle, and allows the snow cover to “age” as the snow grain size (a prognostic variable) increases.

[13] At this stage it ought to be mentioned that the model does not yet incorporate an explicit representation of mixed particles containing both sulphate and BC (the parameterization of BC “ageing” described below contains no dependence on sulphate concentration). This is obviously not ideal, since mixed particles are indeed observed to occur [see, e.g., Hasegawa and Ohta, 2002], and such mixing enhances the absorption due to the BC, as has been pointed out by a number of authors, including Ackerman and Toon [1981], Fuller [1995], Chýlek et al. [1995], Haywood et al. [1997], Fuller et al. [1999], and Jacobson [2000, 2001]. However, while this issue undoubtedly influences the radiative forcing due to the BC, it is not obvious that it affects the associated climate sensitivity, which is the main focus of the present study. Nevertheless, it should be borne in mind when considering the results.

2.1. Black Carbon Aerosol Scheme

[14] The scheme involves three prognostic variables, representing the mass mixing ratios of “fresh” BC, “aged” BC, and BC within cloud droplets. The separation into fresh and aged modes follows the treatment of Cooke and Wilson [1996]: It is assumed that all BC is hydrophobic when emitted, but gradually “ages” because of condensation of hygroscopic material and/or oxidation by ozone (as measured by Decesari et al. [2002]) into a form that has at least some affinity for water. The ageing process is parameterized very simply as exponential decay, with an e-folding timescale of 1 day, slightly shorter than the value of 1.15 days used by Cooke et al. [1999, 2002]. (Note, however, that in those studies it was assumed that 20% of the BC emissions are hydrophilic. The sensitivity tests performed by Cooke et al. [2002] indicate that the combined changes of assuming 100% hydrophobic emissions and using a slightly shorter ageing timescale should mutually compensate.)

[15] The fresh and aged modes of BC are assigned the same lognormal size distribution, with median radius 40 nm and geometric standard deviation 2.0. All particles are assumed to be spherical. It has to be recognized that these assumptions are highly idealized. In reality, black carbon particles exhibit complex distributions of sizes and shapes, which differ significantly with location and evolve during a particle's atmospheric lifetime. See, for example, the papers by Ogren and Charlson [1983] and Liousse et al. [1993]. Substantial uncertainty in the estimated radiative forcing due to BC is therefore inescapable; however, as already noted, estimating the exact size of the forcing is not the main object of this study. The direct radiative effect of the BC aerosol is calculated in the Edwards-Slingo radiation scheme, using scattering and absorption coefficients computed using Mie theory and subsequently averaged over the spectral bands. The spectral refractive index data for black carbon were taken from the expert group report edited by Deepak and Gerber [1983]. The density of black carbon was taken to be 1900 kg m−3, around the median of the range of values tabulated by Fuller et al. [1999]. Hygroscopic growth of the BC particles is assumed to be negligible for radiative purposes. Since both fresh and aged BC particles are assumed to be interstitial within clouds (see below), their (direct) radiative effect is taken into account when radiative fluxes are computed in the cloudy portion of a grid box.

[16] Large-scale vertical and horizontal advection of each BC mode is handled by the model's tracer advection scheme, which uses the flux redistribution method of Roe [1985] to maintain positive definiteness. Vertical transport by convection and turbulent mixing in the boundary layer are also parameterized in ways similar to the model's treatment of the vertical transport of heat and moisture. Dry deposition is parameterized using an approach analogous to electrical resistances, similar to many previous studies [see, e.g., Seinfeld and Pandis, 1998]. This is consistent with the model's treatment of dry deposition for sulphate particles, as described by Jones et al. [2001], except that the particle diffusivities differ because of the different size distribution assumptions. Sedimentation is neglected.

[17] Since fresh BC is assumed to be hydrophobic, this is not subject to wet deposition. The particles in the aged mode are assumed to be insufficiently soluble to act as cloud condensation nuclei, but are subject to capture by cloud droplets due to diffusion and/or differential advection, producing BC in cloud droplets. The parameterization of this process is somewhat more sophisticated than that given by Jones et al. [2001]; details are given in Appendix A. Rainout of BC within cloud droplets is modeled in the same way as rainout of dissolved sulphate [see Jones et al., 2001], by assuming that if a certain proportion of the condensed water in a grid box is removed as precipitation, the same proportion of the BC within cloud droplets is removed. If cloud evaporates without precipitating, BC within the cloud water is instantaneously converted to aged BC. As in the model of Cooke et al. [2002], below-cloud scavenging of aerosol is not included.

2.2. Sulphur Cycle Scheme

[18] This is an improved version of the scheme described by Jones et al. [2001]. Five prognostic variables are used, representing the mass mixing ratios of sulphur dioxide, dimethyl sulphide, and three modes of sulphate aerosol. Of the latter, one represents the Aitken mode, another the accumulation mode, and the third represents sulphate dissolved in cloud water. Since the sulphur cycle plays a secondary role in the present context, the new features of the scheme will only be mentioned here.

[19] First, the parameterization of the collection of interstitial Aitken mode particles by cloud droplets has been reformulated, as described in Appendix A. Second, a parameterization of the Brownian coagulation of Aitken mode particles with accumulation mode sulphate particles has been introduced; this includes the effect of hygroscopic growth, which sharply reduces the coagulation rate. Note, however, that coagulation of sulphate and BC particles is not taken into account, as already implied.

[20] The direct radiative effect, and both the first (“albedo”) and second (“lifetime”) indirect effects of the sulphate aerosol are active in all the runs described below. The indirect effects are parameterized as described by Jones et al. [2001]; this also involves a simple diagnostic treatment of sea-salt aerosol. (Note, however, that the direct effect of the sea salt is not included in the experiments.) The direct effect is modeled taking into account the hygroscopic growth of sulphate particles, as described by Haywood et al. [1997], but the dry size distributions of the sulphate modes are as specified by Jones et al. [2001].

3. Experimental Design

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model Description
  5. 3. Experimental Design
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Appendix A:: Parameterization of Particle Scavenging by Clouds
  10. Acknowledgments
  11. References

[21] Since the slab model HadSM4 does not explicitly represent ocean heat transport, sea surface temperatures are maintained close to climatological values by adding a monthly varying heat flux field. This was computed in a preliminary calibration experiment, run for 20 years, in which the sea surface temperatures were reset to climatological values. In this calibration experiment there were no black carbon emissions, and the atmospheric concentration of CO2 was set to a typical modern-day value of 345 ppmv. Sulphur emissions were as described by Jones et al. [2001], including the GEIA 1B anthropogenic SO2 emissions for 1985, volcanic emissions from the data set of Andres and Kasgnoc [1998], and DMS emissions based on the work of Kettle et al. [1999], with the Liss and Merlivat [1986] parameterization of air-sea gas exchange.

[22] The climate sensitivities of HadSM4 to changes in BC and in CO2 were estimated by running a set of three parallel simulations: a control simulation, which had zero BC emissions and the same CO2 mass mixing ratio as in the calibration run; a BC experiment, which included anthropogenic fossil fuel BC emissions; and a CO2 experiment, which had double the CO2 concentration in the control, with no BC emissions. In each case the sulphur emissions were the same as specified for the calibration run. All simulations had the same initial conditions and were run for 20 years: Equilibrium was reached after 5 to 10 years, followed by a further 10 to 15 years of simulation. The results shown below are meaned over the final 10 years unless otherwise stated.

[23] The fossil fuel BC emissions field used here is based on the data set assembled by Cooke et al. [1999] using submicron particle emission factors. Although significant mass is emitted in particles larger than a micron, those particles have shorter atmospheric lifetimes and are also radiatively less powerful (per unit mass) than the submicron particles. Inclusion of the micron-sized emissions would be inconsistent with the modeling assumptions explained above and would be likely to exaggerate the impact on climate. To improve the signal-to-noise ratio in the experiment (by obtaining a larger forcing), the emissions actually used were four times those given by Cooke et al. [1999], making a total of 20.24 Tg of carbon per year. It is important to appreciate that the Cooke et al. data set does not include BC from biomass burning and is intended to represent fossil fuel emissions in the mid-1980s; significant changes in the distribution of emissions will have taken place since then, as illustrated by the recent study of Bond et al. [2004]. Furthermore, as discussed by Cooke et al. [1999], there is significant uncertainty in their emissions estimates.

[24] To determine the radiative forcing produced by the BC aerosol, an atmosphere-only experiment parallel to the BC slab model run was run for 5 years. In this simulation the radiation scheme is called twice: The first diagnostic call includes the radiative effects of the BC aerosols, whereas the second, which is used to advance the model's evolution, does not. The effects of the BC are therefore unable to influence the model's evolution, and so the difference between the results from the two radiation calls, in terms of the net top-of-atmosphere radiative flux, gives the radiative forcing by the BC aerosol. Stratospheric adjustment is not taken into account here as it is insignificant for tropospheric aerosols, as illustrated by results in the paper of Hansen et al. [1997]. For the CO2 experiment, the radiative forcing due to doubling CO2 in the HadAM3 model has previously been calculated (including stratospheric adjustment) as 3.74 W m−2, which is close to the best estimate of 3.7 W m−2 given by Ramaswamy et al. [2001]. Since the Edwards-Slingo radiation scheme is used in both HadAM3 and HadAM4, with the same radiative properties of CO2 in each case, the radiative forcing due to doubling CO2 is expected to be almost identical in the two model configurations (J. M. Edwards, personal communication, 2003), so we adopt the value of 3.74 W m−2 here.

4. Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model Description
  5. 3. Experimental Design
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Appendix A:: Parameterization of Particle Scavenging by Clouds
  10. Acknowledgments
  11. References

[25] Figure 1 shows the annual mean direct radiative forcing due to BC aerosol, averaged over the five years of the atmosphere-only experiment. Since there are far larger emissions in the Northern Hemisphere (NH) than in the Southern Hemisphere (SH), it is not surprising that the forcing is much greater in the NH. Also as expected, the areas of strongest forcing occur near and downwind of major industrial regions. It is well known that BC forcing is enhanced, other factors being equal, in areas of high surface albedo. This effect is manifested here in the relatively low forcing over the dark Mediterranean Sea compared with the stronger forcing over the neighboring bright deserts. For similar reasons the forcing is significant in the Arctic where the surface albedo is particularly high.

image

Figure 1. Annual mean direct radiative forcing by BC from fossil fuel combustion (quadrupled emissions) (W m−2).

Download figure to PowerPoint

[26] The global annual mean forcing is 0.779 W m−2, which at first sight may seem rather a high value. It must be remembered though that the emissions here are four times larger than the Cooke et al. [1999] estimates. In a similar experiment using the unmodified emissions, the global annual mean forcing was 0.195 W m−2, so the forcing has been scaled up in an approximately linear way. This estimate happens to be quite close to the central estimate of 0.2 W m−2 given by Ramaswamy et al. [2001]. This apparent agreement is fortuitous and of no particular significance; the forcing in reality might well be significantly different, as is clear from the uncertainty range indicated by Ramaswamy et al. [2001], and the subsequent study of Sato et al. [2003]. However, for the purposes of the present study, it is not crucial that the forcing be exactly correct, as already explained.

[27] The annual mean surface temperature change in the BC experiment (relative to the control) is shown in Figure 2. This field contains some interesting structure, which results both from changes in the large-scale circulation and responses to local forcing features. Given the preponderance of forcing in the NH, it is not surprising that the regions of greatest warming occur in middle and high northern latitudes. The warming in the Arctic is due to the significant forcing there, amplified by the well-known ice-albedo feedback. The region of cooling in China coincides with a local maximum in the top-of-atmosphere forcing shown in Figure 1. However, there is no paradox here, because the presence of large amounts of aerosol significantly reduces the solar flux reaching the surface (see Figure 5, later) and thus leads to surface cooling. The regions of cooling over India and Africa just north of the equator have a different cause, as will be explained later; note that there are also some areas of cooling in the SH.

image

Figure 2. Change in annual mean surface temperature due to BC from fossil fuel combustion (quadrupled emissions) (K).

Download figure to PowerPoint

[28] The global annual mean surface temperature change is 0.436 K, implying a climate sensitivity of approximately λBC = 0.56 K W−1 m2. The statistical uncertainty in this value (95% confidence limits, making the approximation that each annual mean is an independent sample, and combining the uncertainties in forcing and temperature change as shown in the appendix of Williams et al. [2001]) is estimated to be ±0.06 K W−1 m2. Of course, the uncertainty due to imperfections in model parameterizations is likely to exceed this.

[29] Figure 3 shows the annual mean surface temperature change in the doubled CO2 experiment (again relative to the control). This is qualitatively similar to results from many previous experiments of this type so does not need extensive discussion here. In contrast to Figure 2, both hemispheres show significant warming, though even here the NH warms relative to the SH as a result of the greater land/sea areal ratio in the NH. In the doubled CO2 experiment there are no regions of outright cooling, though some of the areas of relatively weak warming occur in similar regions to areas of cooling in the BC experiment, suggesting some common elements in the response. The global annual mean surface temperature change is 3.42 K, which is fairly close to the mean of 3.5 K (from a group of 15 models) cited in the most recent report of Working Group I of the Intergovernmental Panel on Climate Change [see Cubasch et al., 2001, Table 9.4].

image

Figure 3. As Figure 2 but for a doubling of CO2 concentration (K).

Download figure to PowerPoint

[30] The implied climate sensitivity to enhanced CO2 concentrations is approximately λC = 0.91 K W−1 m2. Thus in these experiments it turns out that λBC is significantly smaller than λC: a result not expected when work on this study commenced. Such a difference in sensitivity must be a consequence of differences in feedback mechanisms; these are investigated below.

[31] Figure 4 shows the mean precipitation change in boreal summer (June to August) in the BC experiment. The alteration in the thermal balance between the hemispheres leads to a northward shift in the average location of the ITCZ, a phenomenon also observed by Wang [2004]. Among other consequences, this involves a strengthening of the Indian summer monsoon and an increase in rainfall in the Sahel. This is the reason for the cooling in these regions seen in Figure 2. It is perhaps not surprising that these changes are mirror images of those seen in previous experiments on the climate response to indirect sulphate forcing, in which the NH cooled relative to the SH and the ITCZ migrated southward. These are discussed in more detail by Williams et al. [2001], but it is worth reiterating here that the change in the strength of the monsoon is a response to the large-scale circulation shift in the ITCZ, not to the local surface temperature change over India.

image

Figure 4. Change in mean June–August precipitation rate due to fossil fuel BC (mm d−1).

Download figure to PowerPoint

[32] The intensity of the global hydrological cycle does not change much in the BC experiment. In the control, the global annual mean precipitation rate is 2.92 mm d−1 and this only increases by 0.011 mm d−1 in the anomaly. In contrast, in the CO2 experiment the increase is 0.18 mm d−1, which is proportionately much greater in relation to the strength of the initial forcing. The review article by Allen and Ingram [2002] contains an illuminating discussion of such issues. The key point is that changes in the intensity of the global hydrological cycle are controlled by the ability of the troposphere to radiate away latent heat released by the formation of precipitation. Using the approximate argument given by Allen and Ingram [2002] following their equation (1), it is straightforward to show that the predicted increase in the global mean precipitation rate in the BC experiment is about 0.53%, i.e., about 0.015 mm d−1. For the CO2 experiment the predicted increase is on the order of 7.7%, i.e., about 0.225 mm d−1. Considering the approximations made, this theoretical argument seems to agree reasonably well with the experimental results.

[33] The annual mean change in clear-sky incident shortwave radiation flux at the surface in the BC experiment, relative to the control, is portrayed in Figure 5. The global mean reduction is 2.61 W m−2, and there is a significant flux reduction in populated regions of the NH, particularly Europe and China. Even taking into account the quadrupled emissions, these results suggest that BC aerosol from fossil fuel combustion may have played a role in the observed reductions of surface solar radiation, as discussed, for example, by Stanhill and Cohen [2001] and Liepert [2002]. Of course, increases in the concentrations of other types of aerosol and changes in cloud frequency and optical thickness are also likely to be important in this context. It is worth noting too that there is also a global mean reduction of 3.93 W m−2 in the clear-sky incident shortwave radiation flux at the surface in the doubled CO2 experiment, which we attribute to increased absorption by water vapor.

image

Figure 5. Change in annual mean clear-sky incident surface shortwave radiation due to fossil fuel BC (W m−2).

Download figure to PowerPoint

[34] We turn now to consider some feedback mechanisms, starting with the ice-albedo feedback. As in most experiments with enhanced CO2 concentrations, the present CO2 run produces significant reductions in sea-ice coverage in both polar regions. The annual mean sea-ice area in the NH is 10.79 million km2 in the control and 6.49 million km2 in the CO2 experiment, while the corresponding figures for the SH are 10.13 million km2 and 6.94 million km2. In the BC experiment the annual mean NH sea-ice area is again smaller than the control, at 10.00 million km2, but there is no significant change in ice extent in the SH (10.11 million km2). This is only to be expected, given the asymmetry in radiative forcing between the NH and SH already discussed. Taking the percentage reductions in hemispheric sea-ice areas in each experiment and dividing by the respective global annual mean radiative forcings (3.74 W m−2 and 0.779 W m−2) to get a rough comparison of the strengths of the feedback, one obtains figures of 10.7 (NH) and 8.42 (SH) for the CO2 experiment and 9.40 (NH) and 0.25 (SH) for the BC experiment. Given the nonlinearity of the system it would be premature to conclude that there is a significant difference in the strength of the feedback in the Arctic, but it is clear that this is the case in the Antarctic.

[35] The situation with regard to water vapor feedback has some similarities, but is more complex. Figure 6 displays the zonal annual mean changes in water vapor in each experiment, normalized by the respective global annual mean radiative forcings. Again the hemispheric asymmetry in the BC forcing produces a response concentrated north of the equator, and there are even some areas of drying in the SH. The response is much more symmetrical in the CO2 experiment though it is still slightly stronger in the NH than the SH. Note that the reductions in water vapor found in the midtroposphere south of the equator in the BC experiment are more significant than might at first appear, because the outgoing longwave radiation at the top of the atmosphere is more sensitive to changes in water vapor at high altitudes where the air is cool than to changes in the boundary layer where temperatures are relatively close to surface temperature, as discussed by Stocker et al. [2001]. Because of such effects it would probably be misleading to quantify the strength of the feedback in terms of the changes in global mean precipitable water content (though on that basis the feedback in the BC experiment is roughly 75% as strong as it is in the CO2 experiment). Nevertheless, Figure 6 suggests that the overall water vapor feedback is somewhat weaker in the BC experiment.

image

Figure 6. Change in annual average zonal mean water vapor mass mixing ratio normalized by global annual mean radiative forcing (g kg−1 W−1m2): (a) due to fossil fuel BC; (b) due to doubled CO2.

Download figure to PowerPoint

[36] Next we consider cloud feedbacks. Since clouds interact differently with shortwave and longwave radiation it is useful to consider the feedback effects in these wavebands separately. These can be measured by the respective changes in the global mean cloud forcings at the top of the atmosphere, ΔSWCF and ΔLWCF. (A full explanation of cloud forcings can be found in the paper by Senior and Mitchell [1993]; their method II has been followed here.) In the CO2 experiment, ΔSWCF = 1.363 W m−2 and ΔLWCF = −0.957 W m−2, so that the net cloud feedback is positive: ΔNCF = 0.406 W m−2. However, in the BC experiment ΔSWCF = 0.284 W m−2 and ΔLWCF = −0.271 W m−2, so that the net cloud feedback is virtually zero: ΔNCF = 0.013 W m−2. Here then is another reason why the climate sensitivity is lower in the BC experiment. Normalizing the changes in cloud radiative forcing by the relevant global annual mean radiative forcings produces normalized shortwave feedback strengths of 0.364 in the CO2 case and 0.365 in the BC case, and normalized longwave feedback strengths of −0.256 in the CO2 case and −0.348 in the BC case. Although the similarity in the shortwave values must to some extent be coincidental, it is clear that the main difference between the experiments lies in the longwave, where the BC run has a proportionately stronger negative cloud feedback. This indicates that high-altitude clouds may be responding differently in the two experiments.

[37] Figures 7 and 8 show the zonal annual mean changes in condensed water mass mixing ratios in the CO2 and BC experiments respectively. Once again, the pattern of response in the CO2 experiment is fairly symmetrical; it also shows the upward migration of ice cloud and its replacement at lower levels by water cloud that has been seen in previous doubled CO2 experiments, e.g., by Senior and Mitchell [1993]. In the BC case there is (as expected) a relatively weak response in the SH, apart from a reduction in tropical cirrus associated with the northward shift in the ITCZ noted earlier. The response in the NH shows some resemblance to that in the CO2 experiment, but there is a significant difference in the midlatitude upper troposphere, where the reduction of cirrus cloud extends much higher than in the CO2 case. This appears to be a response to increased heating of the air due to absorption of solar radiation by the BC aerosol; the zonal annual mean mass mixing ratio of BC is shown in Figure 9, where it is seen that appreciable BC amounts are present in the zone where cirrus cloud has been reduced. Thus this feedback is a variety of the “semidirect” effect.

image

Figure 7. Change in annual average zonal mean condensed cloud water mass mixing ratio due to doubled CO2 (g kg−1): (a) cloud liquid water; (b) cloud ice water. Negative values are indicated by dashed contours.

Download figure to PowerPoint

image

Figure 8. As Figure 7 but for fossil fuel BC.

Download figure to PowerPoint

image

Figure 9. Annual average zonal mean cross section of fossil fuel BC concentration from the experiment using 4 times the Cooke et al. [1999] emissions (ng kg−1).

Download figure to PowerPoint

[38] Some observations of BC aerosol concentrations at high altitudes are displayed in a paper by Pueschel et al. [1997]. In the midlatitudes of the NH, at a level of around 10 km (approximately 250 hPa), these suggest that typical values are of order 1 to 2 ng m−3. However, the simulated zonal mean BC concentrations in this region correspond approximately to 20 ng m−3 (after dividing by 4 to take account of the enhanced emissions in the experiment). Now, Fahey et al. [1999] considered the measurements to represent a lower limit for the actual values because of limitations of the sampling technique. Also comparing in situ data from a limited number of flights with zonal mean values may be misleading. Nevertheless, we believe that the simulated BC concentrations at high altitudes are probably several times greater than they should be, even taking the quadrupled emissions into account, and that this is mainly due to excessive vertical transport by the model's tracer advection scheme. This suggests that the simulated longwave cloud feedback is probably too strongly negative and the net positive cloud feedback may well be too weak. Thus the difference between the cloud feedbacks in the BC and CO2 experiments is to some extent an artefact of the model.

[39] Another process relevant to differences in climate sensitivity is the change in the fluxes of longwave radiation associated with alterations in the vertical temperature profile. Since the vertical distribution of the radiative forcing is different in the BC and CO2 cases, this would imply a different surface temperature sensitivity even in the absence of the feedbacks already discussed. Now the “ghost forcing” experiments of Hansen et al. [1997] showed that this factor favors a greater sensitivity in the BC case, because most of the energy absorption by the aerosol occurs in the lower layers of the troposphere whereas most of the CO2 forcing takes place higher up. This should still hold in the current experiment, since the BC concentrations are greater at low levels than at high altitude. As we have just seen, however, there is probably too much BC aerosol at high altitude, even allowing for the quadrupled emissions. Hence the temperature profile effect, though still operating to give a relatively higher sensitivity in the BC case, is probably somewhat weaker than it should be.

[40] Some interesting interactions with the sulphur cycle occur in these experiments. The annual mean global total loading of sulphate aerosol is 0.530 Tg(S) in the control run, 0.565 Tg(S) in the BC run, and 0.636 Tg(S) in the doubled CO2 run. The geographical distributions of the changes in sulphate aerosol loading (not shown) suggest strong links with changes in cloud amount in many regions, reductions in cloud being associated with increased sulphate, which is physically plausible. (Similarly, there is a reduction in sulphate loading over India in the BC experiment due to the strengthened summer monsoon already noted.) The increase in global sulphate burden is relatively stronger, in proportion to the radiative forcing, in the BC experiment than in the CO2 case. It is likely that this is primarily due to the intensification of the hydrological cycle in the doubled CO2 run, which would tend to accelerate wet deposition of sulphate relative to the BC case in which the hydrological cycle hardly strengthens at all. Of course, the increased abundance of sulphate aerosol implies a cooling in both experiments, reducing the sensitivity in each case. There is thus a possibility that this sulphate feedback makes a small contribution to the lower climate sensitivity in the BC experiment relative to the CO2 case. Unfortunately, it is not straightforward to isolate the impact of the increased sulphate in an unambiguous way, so at present this must be regarded as a speculative idea.

5. Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model Description
  5. 3. Experimental Design
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Appendix A:: Parameterization of Particle Scavenging by Clouds
  10. Acknowledgments
  11. References

[41] In the previous section it was seen that differences in the strengths of several feedback processes contribute to the lower climate sensitivity obtained in the BC experiment, relative to the doubled CO2 experiment. Some of these differences are physically reasonable, such as the weaker ice-albedo and water vapor feedbacks in response to the BC forcing, and the negative feedback via increased sulphate concentrations. However, it is appropriate to be more cautious about the differences in the temperature profile and cloud feedbacks, in view of the likelihood that the simulated BC concentrations at high altitude are too large. Nevertheless, it is interesting to find that the “semidirect” effect discussed by Hansen et al. [1997] does not necessarily increase the warming due to BC. The effect certainly operates in the present experiment, but because high-level cloud is reduced, the net result is to enhance the negative longwave cloud feedback more than the positive shortwave cloud feedback.

[42] Given the complexity of the system, many additional factors may contribute to the differences in sensitivity. For example, it is interesting to note the following side effect of a reduction in cloud due to the “semidirect” effect: The direct radiative effects of BC and sulphate aerosols would alter even if the aerosol concentrations themselves did not. This arises because of their radiative interactions with cloud: The positive forcing due to BC is strongly enhanced when the aerosol overlies low cloud decks (as discussed, e.g., by Haywood et al. [1997]), whereas the negative (direct) sulphate forcing is suppressed in the same situation (or when the aerosol is below the cloud). Hence a reduction in cloud amount would simultaneously weaken the heating due to the BC and strengthen the (direct) sulphate cooling (if the aerosol concentrations were fixed), and in each case these changes operate in a cooling direction. Obviously, a similar situation occurs with respect to reductions in the surface snow and sea-ice cover. These mechanisms may thus play a small role in reducing the climate sensitivity to BC forcing.

[43] Little has so far been said here about the potential indirect effects of BC aerosol, which are not represented in the present experiments. In principle there are at least four of these potential indirect effects. First, there is the enhanced absorption that will occur where cloud droplets contain BC inclusions. Liu et al. [2002] investigated this effect and concluded that it is probably insignificant except very close to major sources of BC. Some unpublished calculations we have made, using a different method, support Liu et al.'s findings. Next, if, contrary to the assumptions made here, some BC particles served as cloud condensation nuclei, there would be indirect effects analogous to the first (cloud albedo) and second (cloud lifetime) indirect effects due to sulphate aerosol. These effects would contribute a negative indirect forcing just as in the sulphate case, which would tend to offset the positive direct forcing from BC. Hence neglecting these effects cannot result in an underestimate of the warming due to BC aerosol. Finally, there is the reduction of the surface albedo of snow cover and sea ice due to contamination by BC particles. Hansen and Sato [2001] suggested this process may cause a significant forcing, and Hansen and Nazarenko [2004] subsequently presented results supporting the idea, and indicating that the associated climate sensitivity may be unusually large. This is certainly an intriguing possibility and cannot be excluded on the basis of the work we have presented here.

[44] While this paper was being written, we received the paper of Penner et al. [2003], which considers some of the issues and mechanisms discussed here from an alternative perspective. Penner et al. use the concept of “relaxed forcing”: the forcing computed from differencing two simulations using fixed sea surface temperatures. This approach, which has also been used in studies of the indirect sulphate effect [e.g., Jones et al., 2001], allows some feedbacks with relatively fast timescales to influence the diagnosed (relaxed) forcing. In contrast, the viewpoint adopted in the present study is that such feedbacks affect the climate sensitivity to the (pure) radiative forcing. We believe that both viewpoints are valid ways of approaching the same problem from different angles and that no confusion is caused, provided that the methodologies are explained clearly. Interestingly, Penner et al. [2003] found that feedbacks involving longwave radiation result in a significantly lower “relaxed forcing” due to fossil fuel BC than its instantaneous radiative forcing, which from the present perspective would correspond to a lower climate sensitivity than expected. There are thus some features in common with the results presented here. Penner et al. [2003] go further than the present study in that they take indirect effects into account, which apparently reduces the “relaxed forcing” to virtually zero: a quite surprising result. However, they do not explore the climate response to the forcing as done here.

[45] It should be acknowledged that there are several sources of uncertainty in this work. As with all complex models, there are various deficiencies in the one used here, many of which have been noted above. Perhaps the most concerning is the excessive vertical transport of tracers. A new model (HadGEM1) currently under development at the Hadley Centre uses a different advection scheme of the semi-Lagrangian type. Early results from a prototype version of HadGEM1 suggest better transport properties, with lower BC concentrations at high altitude than in this study. It is planned to repeat the present experiments with the new model in due course.

[46] Another source of uncertainty arises from the BC emissions data set. A new study by Bond et al. [2004] has produced an emissions inventory for the mid-1990s that differs significantly from the Cooke et al. [1999] data set used here, which represents the position in the mid-1980s. Bond et al. analyzed the differences and concluded that some are due to changes in economic activity, for example increased fuel consumption, while others are associated with differences in emission factors. For the purposes of the present paper the uncertainty in the magnitude of the global total emissions is relatively unimportant, as shown by the linear scaling of the radiative forcing when the emissions were multiplied by 4. However, changes in the geographical distribution of emissions may be significant for the sensitivity, since the results show that the strengths of some feedbacks are influenced by the spatial distribution of the BC aerosol. Hence a slightly different climate sensitivity would probably be obtained if the experiments were repeated using the Bond et al. [2004] data set, though since most of the emissions are still in the NH, the results might not differ greatly from those presented here.

6. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model Description
  5. 3. Experimental Design
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Appendix A:: Parameterization of Particle Scavenging by Clouds
  10. Acknowledgments
  11. References

[47] This study has investigated the climate sensitivity to anthropogenic emissions of fossil fuel black carbon aerosol (λBC), using the “slab” model HadSM4 extended to include interactive BC and sulphur cycle schemes. The climate sensitivity to doubling CO2, λC, was also explored in a parallel experiment. It transpired that in this model λBC is significantly less than λC, due to differences in the strengths of several feedback processes. The realism of these differences has been assessed: Some are physically reasonable, although others may partially arise from imperfections in the model. Of course, it is well known that different climate models often exhibit strikingly different cloud feedbacks and as a result differing climate sensitivities to greenhouse gases [see, e.g., Stocker et al., 2001, p. 430]. It appears that the same situation is likely to hold with respect to BC aerosol forcing, but the intermodel differences in cloud feedbacks will not automatically parallel those seen in response to greenhouse gas forcing, partly because of the “semidirect” effect.

[48] In view of the uncertainties, and the possibility that some aspects of the results may be specific to the model used, it is important to be careful when considering potential implications for policy. It may be premature to conclude on the basis of these experiments alone that λBC is much smaller than λC, though it is noteworthy that the results of Penner et al. [2003] and Wang [2004] seem to point in the same direction. What can be stated is that the present study offers no support for the idea that λBC is much larger than λC. This implies that the timescale τ (such that for times less than τ, current BC emissions have a greater warming effect than current CO2 emissions) may be shorter than suggested by Jacobson [2002]. (We refrain from estimating τ here since a slab model cannot be relied upon to capture the correct transient response to a climate perturbation.) Although a reduction in BC emissions would make some contribution to slowing the rate of climate change in the short term, reductions in the emissions of CO2 and other greenhouse gases would be more influential in the long run.

Appendix A:: Parameterization of Particle Scavenging by Clouds

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model Description
  5. 3. Experimental Design
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Appendix A:: Parameterization of Particle Scavenging by Clouds
  10. Acknowledgments
  11. References

[49] Here we outline the parameterization of the collection of interstitial aerosol particles, either aged BC or Aitken mode sulphate, by cloud droplets. Since this process strongly influences the simulated BC distribution and the associated cloud feedbacks, a detailed description will be given. Scavenging is assumed to be due to a combination of particle diffusion and relative advection (since cloud droplets fall faster than aerosol particles). The particles under consideration are sufficiently small that it is reasonable to neglect impaction. With somewhat less justification, thermophoresis and diffusiophoresis are also neglected; the results of Young [1974] suggest that the former can be significant under some conditions, though it should be noted that both these processes can operate to reduce collection as well as enhance it. It is also assumed here that particles are always captured once they reach the surface of a droplet.

[50] Let N denote cloud droplet number concentration, R cloud droplet radius, U(R) the fall speed of a droplet, W cloud liquid water content, ρ the density of water, μ the viscosity of air, g the acceleration due to gravity, r aerosol particle radius and D(r) particle diffusivity. We neglect the fall speed of the aerosol particles. As discussed by Pruppacher and Klett [1997], the relative importance of particle diffusion and relative advection is indicated by the Péclet number Pe = 2UR/D. For a typical cloud droplet, R is small enough that it is reasonable to use the Stokes's law approximation for U [see, e.g., Batchelor, 1967]: U = 2ρgR2/9μ, where the density of air has been neglected in comparison with ρ. Thus Pe = 4ρgR3/9μD. From equation (17–12) of Pruppacher and Klett [1997], since the Reynolds number for a cloud droplet is small, a fairly good approximation to the advective-diffusive flux F of particles to a single droplet is

  • equation image

where n is the ambient particle concentration. It follows that in a cloud of droplet number concentration N, the timescale for removal of the particles is τ, where

  • equation image

Λ being the scavenging coefficient.

[51] So far, it has tacitly been assumed that both cloud droplets and aerosol particles have monodisperse size distributions, which is obviously unrealistic. Let us generalize this first by allowing R to have a distribution f(R). The average scavenging coefficient equation image is then given by

  • equation image

Writing Y = equation image,

  • equation image

[52] Taking the Khrgian-Mazin droplet size distribution: f(R) = αR2 exp (−βR), and using the result

  • equation image

it follows that

  • equation image

Since N = 2α/β3 and W = 160πρα/β6,

  • equation image

[53] The next stage is to generalize this by taking into account the particle size distribution ν(r), which is assumed to be lognormal:

  • equation image

where equation image is the median radius and σ the geometric standard deviation.

[54] The aim now is to obtain a mass-weighted average diffusivity equation image which can be substituted for D in the expression for equation image.

  • equation image

From equations (11–15) of Pruppacher and Klett [1997], we have

  • equation image

where k is Boltzmann's constant, T is (absolute) temperature, Kn = λ/r is the Knudsen number, λ is here the mean free path of air molecules (not climate sensitivity), and γ is an empirical factor for which a recommended form is γ = A + B exp (−C/Kn). Here Pruppacher and Klett [1997] cite a number of authors as suggesting that suitable values are A = 1.257, B = 0.4, and C = 1.1. In order to evaluate the integral, we approximate γ by replacing r by equation image: γ ≈ equation image = A + B exp (−Cequation image/λ). With this approximation we have

  • equation image

Using the result

  • equation image

it follows that

  • equation image

Substituting this expression for D, and making the implied substitution for Y, in the expression for equation image produces the required effective scavenging coefficient equation image. Note that this is applied only in the cloudy portion of a grid box.

[55] It remains to specify μ and λ. These again follow the prescriptions of Pruppacher and Klett [1997, equations (10–140) and (10–141)], except that the latter is approximated by omitting the quadratic term in temperature.

[56] Finally, it should be pointed out that there is scope for further development of this scheme. Apart from the neglect of phoretic forces already mentioned, another aspect that is not ideal is that scavenging in mixed-phase and ice clouds is at present treated as though the cloud were liquid.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model Description
  5. 3. Experimental Design
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Appendix A:: Parameterization of Particle Scavenging by Clouds
  10. Acknowledgments
  11. References

[57] We are most grateful to William Cooke for making available the BC emissions inventory we used. It is also a pleasure to thank our colleagues John Edwards, for providing spectral files for the radiation scheme, and Luke Robinson, who helped to develop the previous version of the BC scheme. This work was supported by the UK Department for Environment, Food and Rural Affairs under contract PECD 7/12/37.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Model Description
  5. 3. Experimental Design
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Appendix A:: Parameterization of Particle Scavenging by Clouds
  10. Acknowledgments
  11. 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, 36613667.
  • Allen, M. R., and W. J. Ingram (2002), Constraints on future changes in climate and the hydrological cycle, Nature, 419, 224232.
  • Andres, R. J., and A. D. Kasgnoc (1998), A time-averaged inventory of subaerial volcanic sulfur emissions, J. Geophys. Res., 103(D19), 25,25125,261.
  • Batchelor, G. K. (1967), An Introduction to Fluid Dynamics, Cambridge Univ. Press, New York.
  • Bond, T. C., D. G. Streets, K. F. Yarber, S. M. Nelson, J.-H. Woo, and Z. Klimont (2004), A technology-based global inventory of black and organic carbon emissions from combustion, J. Geophys. Res., 109, D14203, doi:10.1029/2003JD003697.
  • Chung, S. H., and J. H. Seinfeld (2002), Global distribution and climate forcing of carbonaceous aerosols, J. Geophys. Res., 107(D19), 4407, doi:10.1029/2001JD001397.
  • Chýlek, P., G. Videen, D. Ngo, R. G. Pinnick, and J. D. Klett (1995), Effect of black carbon on the optical properties and climate forcing of sulfate aerosols, J. Geophys. Res., 100(D8), 16,32516,332.
  • Cook, J., and E. J. Highwood (2004), Climate response to tropospheric absorbing aerosols in an intermediate general-circulation model, Q. J. R. Meteorol. Soc., 130, 175191.
  • Cooke, W. F., and J. J. N. Wilson (1996), A global black carbon aerosol model, J. Geophys. Res., 101(D14), 19,39519,409.
  • Cooke, W. F., C. Liousse, H. Cachier, and J. Feichter (1999), Construction of a 1° × 1° fossil fuel emission data set for carbonaceous aerosol and implementation and radiative impact in the ECHAM4 model, J. Geophys. Res., 104(D18), 22,13722,162.
  • Cooke, W. F., V. Ramaswamy, and P. Kasibhatla (2002), A general circulation model study of the global carbonaceous aerosol distribution, J. Geophys. Res., 107(D16), 4279, doi:10.1029/2001JD001274.
  • Cubasch, U., G. A. Meehl, G. J. Boer, R. J. Stouffer, M. Dix, A. Noda, C. A. Senior, S. Raper, and K. S. Yap (2001), Projections of future climate change, in Climate Change 2001: The Scientific Basis, Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change, edited by J. T. Houghton et al., Cambridge Univ. Press, New York.
  • Decesari, S., M. C. Facchini, E. Matta, M. Mircea, S. Fuzzi, A. R. Chughtai, and D. M. Smith (2002), Water soluble organic compounds formed by oxidation of soot, Atmos. Environ., 36, 18271832.
  • Deepak, A., and H. E. Gerber (Eds.) (1983), Report of the experts meeting on aerosols and their climatic effects (Williamsburg, Virginia, March 1983), Rep. WCP-55, World Clim. Programme, World Meteorol. Organ., Geneva.
  • Edwards, J. M., and A. Slingo (1996), Studies with a flexible new radiation code. I: Choosing a configuration for a large-scale model, Q. J. R. Meteorol. Soc., 122, 689719.
  • Fahey, D. W., U. Schumann, S. Ackerman, P. Artaxo, O. Boucher, M. Y. Danilin, B. Karcher, P. Minnis, T. Nakajima, and O. B. Toon (1999), Aviation-produced aerosols and cloudiness, in Aviation and the Global Atmosphere: Special Report of the Intergovernmental Panel on Climate Change Working Groups I and III in collaboration With the Scientific Assessment Panel to the Montreal Protocol on Substances That Deplete the Ozone Layer, edited by J. E. Penner et al., Cambridge Univ. Press, New York.
  • Fuller, K. A. (1995), Scattering and absorption cross sections of compounded spheres: III. Spheres containing arbitrarily located spherical inhomogeneities, J. Opt. Soc. Am. A Opt. Image Sci., 12, 893904.
  • Fuller, K. A., W. C. Malm, and S. M. Kreidenweis (1999), Effects of mixing on extinction by carbonaceous particles, J. Geophys. Res., 104(D13), 15,94115,954.
  • Hansen, J. E., and L. Nazarenko (2004), Soot climate forcing via snow and ice albedos, Proc. Natl. Acad. Sci. U. S. A., 101, 423428.
  • Hansen, J. E., and M. Sato (2001), Trends of measured climate forcing agents, Proc. Natl. Acad. Sci. U. S. A., 98, 14,77814,783.
  • Hansen, J. E., M. Sato, and R. Ruedy (1997), Radiative forcing and climate response, J. Geophys. Res., 102(D6), 68316864.
  • Hasegawa, S., and S. Ohta (2002), Some measurements of the mixing state of soot-containing particles at urban and non-urban sites, Atmos. Environ., 36, 38993908.
  • Haywood, J. M., D. L. Roberts, A. Slingo, J. M. Edwards, and K. P. Shine (1997), General circulation model calculations of the direct radiative forcing by anthropogenic sulphate and fossil-fuel soot aerosol, J. Clim., 10, 15621577.
  • Jacobson, M. Z. (2000), A physically-based treatment of elemental carbon optics: Implications for global direct forcing of aerosols, Geophys. Res. Lett., 27, 217220.
  • Jacobson, M. Z. (2001), Strong radiative heating due to the mixing state of black carbon in atmospheric aerosols, Nature, 409, 695697.
  • Jacobson, M. Z. (2002), Control of fossil-fuel particulate black carbon and organic matter, possibly the most effective method of slowing global warming, J. Geophys. Res., 107(D19), 4410, doi:10.1029/2001JD001376.
  • 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(D17), 20,29320,310.
  • Kettle, A. J., et al. (1999), A global database of sea surface dimethylsulfide (DMS) measurements and a procedure to predict sea surface DMS as a function of latitude, longitude and month, Global Biogeochem. Cycles, 13, 399444.
  • Kirkevåg, A., T. Iversen, and A. Dahlback (1999), On radiative effects of black carbon and sulphate aerosols, Atmos. Environ., 33, 26212635.
  • Koch, D. (2001), Transport and direct radiative forcing of carbonaceous and sulfate aerosols in the GISS GCM, J. Geophys. Res., 106(D17), 20,31120,332.
  • Liepert, B. G. (2002), Observed reductions of surface solar radiation at sites in the United States and worldwide from 1961 to 1990, Geophys. Res. Lett., 29(10), 1421, doi:10.1029/2002GL014910.
  • Liousse, C., H. Cachier, and S. G. Jennings (1993), Optical and thermal measurements of black carbon aerosol content in different environments: Variation of the specific attenuation cross-section, sigma (σ), Atmos. Environ., Part A, 27, 12031211.
  • Liousse, C., J. E. Penner, C. Chuang, J. J. Walton, H. Eddleman, and H. Cachier (1996), A global three-dimensional model study of carbonaceous aerosols, J. Geophys. Res., 101(D14), 19,41119,432.
  • Liss, P. S., and L. Merlivat (1986), Air-sea gas exchange rates: Introduction and synthesis, in The Role of Air-Sea Exchange in Geochemical Cycling, edited by P. Buat-Menard, D. Reidel, Norwell, Mass.
  • Liu, L., M. I. Mishchenko, S. Menon, A. Macke, and A. A. Lacis (2002), The effect of black carbon on scattering and absorption of solar radiation by cloud droplets, J. Quant. Spectrosc. Radiat. Trans., 74, 195204.
  • Lohmann, U., and J. Feichter (2001), Can the direct and semi-direct aerosol effect compete with the indirect effect on a global scale? Geophys. Res. Lett., 28, 159161.
  • Marshall, S. E. (1989), A physical parameterization of snow albedo for use in climate models, NCAR Coop. Thesis 123, Natl. Cent. for Atmos. Res., Boulder, Colo.
  • Myhre, G., F. Stordal, K. Restad, and I. S. A. Isaksen (1998), Estimation of the direct radiative forcing due to sulfate and soot aerosols, Tellus, Ser. B, 50, 463477.
  • Ogren, J. A., and R. J. Charlson (1983), Elemental carbon in the atmosphere: Cycle and lifetime, Tellus, Ser. B, 35, 241254.
  • Penner, J. E., H. Eddleman, and T. Novakov (1993), Towards the development of a global inventory for black carbon emissions, Atmos. Environ., Part A, 27, 12771295.
  • Penner, J. E., C. C. Chuang, and K. Grant (1998), Climate forcing by carbonaceous and sulfate aerosols, Clim. Dyn., 14, 839851.
  • Penner, J. E., S. Y. Zhang, and C. C. Chuang (2003), Soot and smoke aerosol may not warm climate, J. Geophys. Res., 108(D21), 4657, doi:10.1029/2003JD003409.
  • Pruppacher, H. R., and J. D. Klett (1997), Microphysics of Clouds and Precipitation, 2nd ed., Kluwer Acad., Norwell, Mass.
  • Pueschel, R. F., K. A. Boering, S. Verma, S. D. Howard, G. V. Ferry, J. Goodman, D. A. Allen, and P. Hamill (1997), Soot aerosol in the lower stratosphere: Pole-to-pole variability and contributions by aircraft, J. Geophys. Res., 102(D11), 13,11313,118.
  • Ramaswamy, V., O. Boucher, J. Haigh, D. Hauglustaine, J. Haywood, G. Myhre, T. Nakajima, G. Y. Shi, and S. Solomon (2001), Radiative forcing of climate change, in Climate Change 2001: The Scientific Basis, Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change, edited by J. T. Houghton et al., Cambridge Univ. Press, New York.
  • Roe, P. L. (1985), Some contributions to the modeling of discontinuous flows, in Large-Scale Computations in Fluid Mechanics, Lect. Appl. Math., vol. 22, pp. 163193, Am. Math. Soc., Providence, R. I.
  • Sato, M., J. E. Hansen, D. Koch, A. Lacis, R. Ruedy, O. Dubovik, B. Holben, M. Chin, and T. Novakov (2003), Global atmospheric black carbon inferred from AERONET, Proc. Natl. Acad. Sci. U. S. A., 100, 63196324.
  • Schult, I., J. Feichter, and W. F. Cooke (1997), Effect of black carbon and sulfate aerosols on the global radiation budget, J. Geophys. Res., 102(D25), 30,10730,117.
  • Seinfeld, J. H., and S. N. Pandis (1998), Atmospheric Chemistry and Physics: From Air Pollution to Climate Change, John Wiley, Hoboken, N. J.
  • Seland, Ø., and T. Iversen (1999), A scheme for black carbon and sulphate aerosols tested in a hemispheric scale, Eulerian dispersion model, Atmos. Environ., 33, 28532879.
  • Senior, C. A., and J. F. B. Mitchell (1993), Carbon dioxide and climate: The impact of cloud parameterization, J. Clim., 6, 393418.
  • Stanhill, G., and S. Cohen (2001), Global dimming: A review of the evidence for a widespread and significant reduction in global radiation with discussion of its probable causes and possible agricultural consequences, Agric. For. Meteorol., 107, 255278.
  • Stocker, T. F.et al. (2001), Physical climate processes and feedbacks, in Climate Change 2001: The Scientific Basis, Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change, edited by J. T. Houghton et al., Cambridge Univ. Press, New York.
  • Tegen, I., D. Koch, A. A. Lacis, and M. Sato (2000), Trends in tropospheric aerosol loads and corresponding impact on direct radiative forcing between 1950 and 1990: A model study, J. Geophys. Res., 105(D22), 26,97126,989.
  • Wang, C. (2004), A modeling study on the climate impacts of black carbon aerosols, J. Geophys. Res., 109, D03106, doi:10.1029/2003JD004084.
  • Webb, M. J., C. A. Senior, S. Bony, and J.-J. Morcrette (2001), Combining ERBE and ISCCP data to assess clouds in the Hadley Centre, ECMWF and LMD atmospheric climate models, Clim. Dyn., 17, 905922.
  • Williams, K. D., A. Jones, D. L. Roberts, C. A. Senior, and M. J. Woodage (2001), The response of the climate system to the indirect effects of anthropogenic sulfate aerosol, Clim. Dyn., 17, 845856.
  • Young, K. C. (1974), The role of contact nucleation in ice phase initiation in clouds, J. Atmos. Sci., 31, 768776.