Comment on “Control of fossil‐fuel particulate black carbon and organic matter, possibly the most effective method of slowing global warming” by M. Z. Jacobson

Abstract

[1] In a recent paper, Jacobson [2002] indicated that abruptly eliminating the fossil-fuel (f.f.) black carbon (BC), and organic matter (OM) annual emissions (15.2 TgC) would reduce the global mean surface air temperature by 0.35°C within 5 years, whereas abruptly eliminating one third of anthropogenic CO2 emissions would have the same cooling effects but after 50–200 years. The author went on to argue that the existing U.S. and European vehicle particulate emissions standards and other fuel-related regulations favoring diesel vehicles actually promote global warming. The implication of any such dramatic conclusions is enormous, and independent verifications of the conclusions as well as a careful scrutiny of the assumptions leading to the conclusions are in order. Our comments address the issues of model assumptions, input information, and policy implications. [2] Jacobson [2002] states that the f.f.-BC inventory was obtained from Cooke et al. [1999], who estimated the global bulk emissions of f.f.-BC and f.f.-organic carbon (OC) to be 6.4 TgC/yr and 10.1 TgC/yr, respectively, and the submicron global emissions of f.f.-BC and f.f.-OC to be 5.1 TgC/yr and 7.0 TgC/yr, respectively. Jacobson [2002] states that the baseline OM:BC emission ratio from fossil fuels was set at 3.1:1; this was obtained from an average OC: elemental carbon (EC) fossil-fuel ratio of 2.4 and an OM:OC emission mass ratio of 1.3:1, which is probably too low according to Turpin and Lim [2001]. Use of the fossil-fuel OM:BC emission ratio of 3.1:1 would result in global bulk OM emissions of 19.9 Tg/yr and global submicron OM emissions of 15.9 Tg/yr, which is significantly higher than the values provided by Cooke et al [1999]. Adding to the confusion, Jacobson [2002] quotes a value of 5.1 Tg/yr for the f.f.-BC emissions and 10.1 Tg/yr for the f.f.-OM (note: not OC and not TgC/yr) emissions in the caption of Figure 1 of his paper. The first value is the same as the submicron f.f.-BC emissions given by Cooke et al. [1999], but the second value is in accordance with neither Jacobson's OM:BC ratio nor the value provided by Cooke et al. [1999]. Jacobson does not clarify which f.f.-BC and f.f.-OM emissions were actually used in the study. [3] Jacobson [2002] also indicates that the model underpredicted observed BC in urban and many rural areas, suggesting that the BC inventory did not lead to overestimates of BC climate effects. However, significant underpredictions of near-surface BC concentrations are expected over urban areas because the predicted concentrations are for a very large grid (typically 104–105 km2 in low latitudes to midlatitudes, which is much greater than typical urban areas) while the observations are for single points in space where high concentrations are anticipated within the urban areas. More important, Jacobson's model tends to overpredict the near-surface BC over marine regions, while the near-surface sulfates over marine regions tend to be underestimated. This is significant because we expect the BC concentrations to be more uniform over the marine regions. Accordingly, the predicted BC loading is likely larger than actual, which would exaggerate the actual climate impact of BC. The atmospheric loading for f.f. (BC + OM) is estimated by Jacobson [2002] to be 0.25–0.6 Tg, which yields a loading of 0.08–0.2 TgC for f.f. BC. Dividing these numbers by the emission rate of 5.1 TgC/yr yields a f.f. BC lifetime of 6 to 14.4 days. This lifetime is higher than those obtained by others (3.86, 4.29, 5.29, and 7.85 days, from different studies) [see Cooke et al., 2002]. [4] In Jacobson's [2002] paper, BC aerosols are assumed to be particles with a BC core coated with well-mixed shells. The description of this particular mixture configuration is an important contribution of Jacobson [2000, 2001]. The coated BC aerosols are thus hygroscopic and are good candidates for forming cloud condensation nuclei. Therefore the internally mixed BC plays a role in both the first (Twomey effect, Twomey [1977]) and the second (Albrecht effect, Albrecht [1989]) indirect forcing. Jacobson also asserts that his model includes these effects. However, we find that Jacobson [2002] has not included the second indirect effect. [5] According to Jacobson [2002], the cloud variables such as cloud, ice, and rainwater content are determined by the Arakawa-Schubert subgrid scheme and large-scale stratus schemes in the general circulation model (GCM, within the modeling system). The resulting bulk cloud parameters are redistributed into different size bins for further simulations on aqueous chemistry and cloud microphysics. Interactions between aerosols and cloud are described explicitly. The radiative properties of the subsequent size-resolved cloud constituents are determined from Mie calculations. Accordingly, the first indirect radiative forcing has been considered. [6] However, because (1) at the beginning of each time step, the bulk cloud liquid, ice, and precipitation from the cumulus and stratus schemes are evaporated or sublimated into the gas phase in each grid cell for regrowth onto size-resolved ice nuclei and cloud condensation nuclei, (2) at the end of each time step, water content from the nonprecipitated cloud liquid, ice, and graupel distributions are re-evaporated, and (3) newly updated bulk-water information from the cumulus and stratus schemes is used at the next time step, the change of the cloud lifetime and precipitation amount caused by aerosols actually cannot be described by the size-resolved scheme. Therefore there is no cloud lifetime modification due to a change in BC aerosols, and the second indirect effect and forcing has not been considered, counter to the assertion of Jacobson [2002]. Note that the forcing due to both indirect effects should be negative because of the increased optical depth, albedo, and lifetime of the clouds. [7] Jacobson [2002] uses a rather detailed gas-aerosol chemistry module coupled with a GCM with a rather coarse horizontal grid of 4° latitude × 5° longitude and a highly resolved vertical grid containing 39 sigma-pressure layers. One immediate concern is the potential mismatch in the resolution of the dynamics of the system. A highly resolved vertical structure coupled with a poorly resolved horizontal structure would make the model quasi-one dimensional. Another concern is the use of a detailed gas-aerosol chemistry module in a coarse (horizontal) grid. At 35° latitude, a grid with 4° latitude and 5° longitude has an area of 444(NS) × 455(EW) km2. One would not expect a rather uniform distribution of the different gaseous and aerosol species within such a domain, especially over land. Accordingly, the nonlinear chemistry would likely make the module predictions unrepresentative of the grid. A third concern is the limited ability of the coarse-grid GCM to describe regional climate expected to be important for black carbon. A case in point is that Krishnan and Ramanathan [2002] have demonstrated that absorbing aerosols lead to local cooling of the surface air temperature, but this observation is not evident in the paper by Jacobson [2002]. [8] In our view, the GCM used by Jacobson [2002] has not been adequately scrutinized or used by others, or compared with other GCMs. A rather smooth, sharp drop in temperature in 5 years after the removal of f.f.-BC and f.f.-OM in the 5-year simulation, as depicted in Figure 1 of Jacobson [2002], raises concerns about the ability of the model to simulate the annual and decadal variability of surface air temperature. Such variability is present in the predictions of more commonly used models and in observation [see, e.g., Stouffer et al., 2000]. This could reflect an important artificial curtailment in the short-end range of the timescales of the dynamical processes represented in the model. The shorter timescales are clearly relevant to regional-scale processes, which should greatly influence the climate impact of black carbon. There are also issues related to the long timescales due to the huge thermal inertia of the deep ocean. We are concerned as to whether the ocean model used in the GCM adequately accounts for this inertia and, if so, whether a 5-year interval is sufficient for the model to reach thermal equilibrium after a sudden removal of f.f.-BC and f.f.-OM. [9] In the radiative transfer, the role of halocarbons such as chlorofluorocarbons (CFCs) and hydrofluorocarbons (HFCs) seems to have been excluded by Jacobson [2002]. Their exclusion would further exaggerate the relative warming role of black carbon in the paper. [10] Jacobson [2002] concludes that eliminating all f.f. (BC+OM) with no other change might reduce the global warming by 0.35°C in 3 to 5 years, or more than 40% of net observed global warming between 1850 and 2000. However, this statement implicitly assumes that the observed temperature increase of 0.75°C between 1850 and 2000 is correctly predicted by his model. The paper did not provide any support for this assumption. Such support is necessary to assure self-consistency of the 40% claim. Jacobson [2002] treats the warming by BC and by CO2 linearly and establishes an “equivalent” ratio between them for a given time horizon, ignoring the difference in the spatial extents of the impacts. We believe that this extrapolation is premature given the problems associated with the model itself. [11] On a global scale, whether BC has a net and statistically significant warming impact at the surface remains to be resolved. More studies using established GCMs need to be conducted. Insofar as the diesel policy implication is concerned, the lifetimes of both BC and CO2 must be considered. Jacobson [2002] emphasizes the immediate benefit of controlling diesel particulate emissions compared to the slow benefit of controlling CO2. We believe just the opposite. Because the CO2 lifetime is some 5000 or more times longer than that of BC, it is more prudent not to aggressively reduce BC emissions to the extent of jeopardizing significant CO2 emission reduction. In fact, if we indeed find that BC has a significant detrimental climate impact, controlling it will quickly remove the problem in perhaps one to two decades. The next generation of diesel vehicles will use particulate filters to reduce the particulate emissions by a factor of about 10, to the emission levels of gasoline vehicles. Therefore a substantial reduction of BC from road traffic will soon be in place. On the other hand, the climate impact of CO2 reduction will only gradually show up much longer than a century later. Therefore every opportunity that significantly reduces overall CO2 emissions should be encouraged at an earliest possible time frame. The key benefit of diesel vehicles using affordable and proven technologies is their ability to reduce CO2 emissions by 20–25% relative to comparable gasoline vehicles. Therefore use of diesel vehicles in place of gasoline vehicles should serve as a practical and near-term step toward the reduction of CO2.