Journal of the Meteorological Society of Japan, Vol. 77, No. 6, pp. 1299-1307, 1999 1299 NOTES AND CORRESPONDENCE Coupled Ocean-Atmosphere Model Experiments of Future Climate Change with an Explicit Representation of Sulfate Aerosol Scattering By Seita Emori, Toru Nozawa National Institute for Environmental Studies, Tsukuba, Japan Ayako Abe-Ouchi, Atusi Numaguti,l Masahide Kimoto and Teruyuki Nakajima Center for Climate System Research, University of Tokyo, Japan (Manuscript received 2. December 1998, in revised form 5 October 1999) Abstract The transient response of a coupled ocean-atmosphere model to increasing concentrations of greenhouse gases and sulfate aerosols, is investigated with an explicit representation of aerosol scattering. Experiments with an implicit representation of aerosol scattering, through the modification of surface albedo with the same parameter values as used in previous climate projections, are also conducted for comparison. The indirect effect of aerosol is yet to be included. As suggested by previous radiation computation studies, the estimated radiative forcing due to the direct effect of sulfate aerosol is significantly smaller with the explicit representation, than with the implicit one. A principal source of the overestimation by the implicit method is the neglect of the dependence of aerosol scattering on near-surface humidity. The projected surface air temperature change due to the addition of sulfate aerosols is considerably smaller in magnitude especially over dry regions with the explicit method, than with the implicit one. It is also suggested that the change in the Asian summer monsoon precipitation due to an increase in sulfate aerosols is particularly sensitive to the representation of sulfate aerosol scattering. 1. Introduction The increase in atmospheric concentrations of carbon dioxide (CO2) and other greenhouse gases due to anthropogenic emissions is expected to cause a substantially warmer climate in the next century (IPCC, 1996). Tropospheric sulfate aerosols are also increasing due to anthropogenic activity, which tend to cool the climate by scattering solar radiation (the direct effect of sulfate aerosol; IPCC, 1996). It is suggested that sulfate aerosols may increase the brightness and the extent of clouds, by acting as cloud condensation nuclei, resulting in a further cooling effect (the indirect effect; IPCC, 1996). Corresponding author: Seita Emori, National Institute for Environmental Studies, Tsukuba, 305-0053, Japan. E-mail: emori@nies.go.jp 1 Present affiliation: Graduate School of Environmetal Earth Science, Hokkaido University, Sapporo 060-0810, Japan. (c) 1999, Meteorological Society of Japan Several authors included the direct effect of sulfate aerosol in their coupled ocean-atmosphere model experiments for the past and the next centuries, with scenarios of atmospheric concentrations of greenhouse gases and sulfate aerosols (Hasselmann et al., 1995; Mitchell et al., 1995a; Haywood et al., 1997; Mitchell and Johns, 1997). According to their results, inclusion of the direct effect of sulfate aerosol significantly reduces the greenhouse gasinduced surface warming. The magnitude of the reduction is 0.7-1.1K in global average around 2050, and much more significant (up to 3K) over continents in the Northern Hemisphere. However, because their model could not represent the scattering by aerosols explicitly, they imposed the direct effect of sulfate aerosol on equivalent changes in surface albedo estimated by the method of Charlson et al. (1991). Kiehl and Briegleb (1993) estimated the present-day direct radiative forcing by a radiation model with an explicit repre-
1300 Journal of the Meteorological Society of Japan Vol. 77, No. 6 sentation of aerosol scattering, with monthly mean analyzed temperature and moisture fields from the European Centre for Medium Range Weather Forecasting. They gave a value considerably smaller than that estimated in the previous coupled model experiments (Table 1). They suggested that the method of Charlson et al. (1991) tends to overestimate the aerosol scattering primarily because their fixed value of specific extinction is valid only for visible wavelengths, and is too large for longer wavelengths. Mitchell et al. (1995b) pointed out that a large near-surface humidity (80%) assumed in this method causes a further overestimate over dry regions. Haywood and Shine (1995) showed that the same method of Charlson et al. (1991) with more appropriate parameter values could give a value of aerosol scattering as small as that estimated by Kiehl and Briegleb (1993). Because of the overestimation as suggested above, the impact of the direct effect of sulfate aerosol on the future climate projected with the method and the parameter values of Charlson et al. (1991) is also expected to be exaggerated. We perform coupled model experiments with an explicit and the conventional implicit representations of aerosol scattering, in order to reassess the impact of the direct effect of sulfate aerosol in future climate projections, and to discuss its uncertainty. Some authors started to include the indirect effect of sulfate aerosol in their coupled model experiments (Meehl et al., 1996; Le Treut et al., 1998). The present study concentrates on the direct effect, because the indirect effect is still uncertain. Though the other kinds of aerosols such as absorbing black carbon are also considered to be important, the present study considers only non-absorbing sulfate aerosol for an easier comparison with the previous studies. A further study including the direct effects of aerosols other than sulfate and the inderect effects is now in progress. 2. Model and experiments The model used in this study is a coupled oceanatmosphere model that consists of CCSR/NIES ALCM, CCSR OGCM, a thermodynamic sea ice model, and a river routing model. The spatial resolution is T21 spectral truncation (-5.6o), and 20 vertical levels for the atmospheric part and 2.8o horizontal grid and 17 vertical levels for the oceanic part. Flux adjustment for atmosphere-ocean heat and water exchange is applied to prevent a drift of the modeled climate. The atmospheric model adopts a radiation scheme based on the k-distribution, two-stream discrete ordinate method (DOM) (Nakajima and Tanaka, 1986). This radiation scheme can deal with absorption, emission and scattering by gases, clouds and aerosol particles in a consistent manner. The spectrum of solar and terrestrial thermal radiation Table 1. Global and annual mean radiative forcing due to the direct effect of anthropogenic sulfate aerosol (W m-2). t As the forcings at 1890 are taken to be zero in the present study, the values shown are corrected so that the pre-industrial levels are taken to be zero. is divided into 18 wavelength regions (channels). Each channel is divided into several k-distribution subchannels, and the total number of subchannels for the standard version is 37. The transmissivity, reflectivity, and source function of each atmospheric layers are calculated following Nakajima and Tanaka (1986) with delta-two stream approximation (Joseph et al., 1976), from the optical parameters of the layer, namely the optical thickness, the single scattering albedo, the asymmetry factor, and the cutoff factor. Finally, the radiative fluxes of each layer interface are calculated by use of the adding method. The optical parameters of gases are calculated from the HITRAN and LOWTRAN database. Two types of clouds, cumulus cloud and largescale cloud, are included in the radiation flux calculation. The fraction of cloudy area of a grid box is assumed to be proportional to cloud base mass flux for the cumulus cloud, while the mixing ratio of cloud water in the cumulus is estimated by the cloud model in the cumulus parameterization (Arakawa and Schubert, 1974). The cloud fraction and cloud water mixing ratio for the large-scale cloud is diagnosed by assuming a subgrid variation of total water mixing ratio, which is the prognostic variable of water in the atmospheric part of the model (Le Treut and Li, 1991). The random overlapping assumption is used for large-scale cloud, whereas the maximum overlapping is used for cumulus cloud. The cloud water is treated as cloud ice or cloud liquid water, depending on air temperature. The log-normal distribution of droplet radius with dispersion parameter of 1.5, and spherical shape is assumed for both types. The volumetric mode radii of cloud ice and cloud liquid water are assumed to be 40um and 8um, respectively. The volumetric mode radius of the sulphate particle in dry environment, rs0, is assumed to be 0.2am. Spherical shape and log-normal distribution with dispersion parameter of 4.12 are assumed. The hygroscopic growth of the sulphate is considered by an empirical fit of d'almeida et al. (1991):
December 1999 S. Emori, T. Nozawa, A. Abe-Ouch i et al. 1301 rs=rs0[1-f log {1-min(RH, 0.99)}], (1) where r3 is the volumetric mode radius, RH is the relative humidity of air given as a fraction, and a constant f is set to 0.5. The efficiency factor for scattering and absorption is calculated by Mie theory as a function of scattering angle, size parameter, and complex refractive index (van de Hulst, 1957). Once the mode radius of the hygroscopically grown particles is determined, the cross sections offfffff absorption and scattering and scattering phase function are calculated by use of this efficiency factor. These are calculated in advance and tabulated as a function of wavelength and hygroscopically grown mode radius. Because the complex refractive index value of sulfate particle is not available, that of 75% H2SO4 solution (World Climate Programme, 1986) is used for the dry sulfate particle. We assume that the aerosol particle is non-absorbing, and the imaginary part of the complex refractive index is taken to be zero. A weighted average of it and that of water is used for the complex refractive index of grown particles. In each step of radiation calculation, the optical properties of each layer such as optical thickness and single scattering albedo are calculated by use of tabulated cross sections and phase function from the specified aerosol mixing ratio and estimated mode radius rs for 18 wavelength regions. Note that, with this method, the effect of aerosol is properly suppressed over cloudy areas without any special manipulation, because the interaction between the radiative effects of cloud and aerosol is represented straightforwardly. The vertical distribution of the sulfate aerosol is assumed to be constant in the lowest 2 km of the atmosphere. The concentrations of greenhouse gases are represented by equivalent CO2. Four integrations are made for 200 model years (1890-2090). In the control experiment (CTL), the globally uniform concentration of greenhouse gases is kept constant at 345 ppmv C02 equivalent, and the concentration of sulfate is set to zero. In the experiment GHG, the concentration of greenhouse gases is gradually increased, while that of sulfate is set to zero. In the experiments SUL and ALB, the increase in anthropogenic sulfate as well as that in greenhouse gases is given. In SUL, aerosol scattering is explicitly represented in the way described above. In ALB, aerosol scattering is implicitly represented through the modification of surface albedo, with the same parameter values as in Mitchell and Johns (1997). The scenario of atmospheric concentrations of greenhouse gases and sulfate aerosols is given in accordance with Mitchell and Johns (1997). The increase in greenhouse gases is based on the historical record from 1890 to 1990, and is increased by 1% yr-1 (compound) after 1990. For sulfate aerosols, geographical distributions of sulfate loading for 1986 and 2050, which are estimated by a sulfur cycle model (Langner and Rodhe, 1991), are used as basic patterns. Based on global and annual mean sulfur emission rates, the 1986 pattern is scaled for years before 1990; the 2050 pattern is scaled for years after 2050; and the pattern is interpolated from the two basic ones for intermediate years, to give the time series of the distribution. The sulfur emission rate in the future is based on the IPCC IS92a scenario (IPCC, 1992). The sulfate concentration is offset in this study, so that it starts from zero at 1890. The seasonal variation of sulfate concentration is ignored as in the earlier studies. 3. Results Radiative forcing of sulfate aerosol is calculated on-line for selected years in SUL and ALB. The global and annual mean values are summarized in Table 1, with the results of previous studies that used essentially the same source of aerosol loading (Langner and Rodhe, 1991). As suggested by Kiehl and Briegleb (1993), the radiative forcing with explicit aerosol representation (SUL) is smaller than that with implicit representation (ALB), though the 1990 forcing in SUL is not as small as that of Kiehl and Briegleb (1993), mainly because of differences in aerosol optical parameters (the volumetric mean radius of a dry particle is assumed to be 0.27 um in Kiehl and Briegleb (1993) and 0.55um in this study, respectively). The forcing values in ALB are fairly close to those of Mitchell and Johns (1997). The geographical distributions of the annual mean radiative forcing due to sulfate aerosols at 2060 in SUL and ALB are shown in Fig. 1. The forcing is less pronounced in SUL (Fig. la) than in ALB (Fig. lb) over some dry regions such as northwest India, the Iranian Plateau, the Arabian Peninsula, the Kalahari Desert, the Mexico Plateau, and the Atacama Desert. This is mainly attributable to the dependence of aerosol scattering on near-surface humidity in SUL as shown by (1), which is neglected in ALB as well as in Mitchell and Johns (1997). We additionally conducted a sensitivity experiment, which is the same as SUL except that the relative humidity in (1) was set to a constant value of 0.8 instead of values simulated by the atmospheric model, and found a forcing value of -1.32W m-2 at 2060, which is considerably closer to ALB and Mitchell and Johns (1997). This indicates that the neglect of aerosol scattering on near-surface humidity is one of the principal sources of the overestimation of sulfate radiative forcing in ALB, and probably in Mitchell and Johns (1997). The projected time sequences of the global and annual mean surface air temperature change (in the lowest model level, r50m) in the experiments are shown in Fig. 2. Though CTL has a small trend due to an incompleteness of the flux adjustment
1302 Journal of the Meteorological Society of Japan Vol. 77, No. 6 Fig. 1. The geographical distributions of the annual mean radiative forcing clue to the direct effect of anthropogenic sulfate aerosol at 2060 in (a) SUL and (b) ALB. initialization, this does not seem to significantly affect the sensitivity of the modeled climate to aerosol radiative forcing. The difference between GHG and SUL, representing the effect of sulfate aerosol, seems to be within the range of their natural variability until around 2040. The difference around 2060 is approximately 0.35K, which is less than half of those in Mitchell and Johns (1997) and other studies using the method of Charlson et al. (1991). The reduction of warming by sulfate aerosols in ALB (-0.5K) is more significant than in SUL, although it is smaller than in Mitchell and Johns (1997), probably because of an overestimation of cloudy areas in ALB clue to the random overlapping assumption of large-scale clouds. Geographical distributions of the differences in the annual mean surface air temperature averaged over the last 50 years (2040-2090) between the experiments are shown in Fig. 3. The difference between SUL and CTL (SUL-CTL, Fig. 3a), representing the combined effect of increases in greenhouse gases and sulfate aerosols, has geographical features similar to the results of earlier transient experiments (IPCC, 1996). There is greater warming over land, especially in higher latitudes, than over the sea, and minimum warming in the high latitude southern ocean and the northern North Atlantic. As seen from the difference between SUL and GHG (SUL-GHG, Fig. 3b), the cooling due to increases in sulfate aerosols is significant over the continents in the Northern Hemisphere as expected from the aerosol forcing pattern (Fig. la), though the maximum cooling occurs further to the north of the maximum forcing, because of sea-ice and snow albedo feedback. The cooling spread over the North Pacific and the North Atlantic around 30N is due to the advection of cooled air mass from the continents, as discussed by Mitchell et al. (1995a). In the difference between ALB and GHG (ALB-GHG, Fig. 3c), cooling due to sulfate aerosols over dry re-
December 1999 S. Ernori, T. Nozawa, A. Abe-Ouchi el al. 1303 Fig. 2. The time sequences of global and annual mean surface air temperature change in the experiments. The average of CTL for the first 50 years (14.2C) is taken as zero. gions is more significant than in SUL-GHG as expected from the radiative forcing patterns (Fig. 1). In SUL-GHG (Fig. 3b), intensive cooling (larger than 2K) is found over the Canadian and West Siberian Arctic, while it is not found in ALB-GHG (Fig. 3c). This is because the aerosol radiative forcing over the Arctic regions is strongly suppressed by extensive cloud cover in ALB (Fig. Ib), while, in SUL, weak forcing is present over those regions (Fig. la), resulting in a substantial cooling through the amplification by sea-ice and snow feedback. The differences in annual mean precipitation averaged over the last 50 years between the experiments are shown in Fig. 4. In the total change, SUL-CTL (Fig. 4a), precipitation over the middle to high latitudes and the Asian summer monsoon regions is enhanced which is a common feature to many of the earlier transient experiments (IPCC, 1996). Intensification of tropical precipitation and strong suppression of subtropical precipitation are also evident in the present experiment, corresponding to the enhanced tropical warming. The impact of the addition of sulfate aerosol, SUL-GHG (Fig. 4b), is generally small in magnitude. The total change (Fig. 4a) and the change due to the addition of sulfate aerosol (Fig. 4b) are generally in the opposite sign over the middle to high latitudes and the subtropical continents as in Mitchell and Johns (1997), though it is not necessarily the case over the tropics and the subtropical oceans. In ALB-GHG (Fig. 4c), the reduction of precipitation over the Asian summer monsoon regions is much more significant than in SUL-GHG. Lal et al. (1995) and Mitchell and Johns (1997) found that the reduction of the Asian summer monsoon precipitation due to the addition of sulfate aerosol exceeds the greenhouse gas-induced enhancement, resulting in a net reduction of the precipitation. In the present study, however, the net change in the Asian summer monsoon precipitation is still an enhancement in SUL and ALB. As demonstrated in the above, the Asian summer monsoon precipitation appears to be especially sensitive to the representation of aerosol scattering. This is because the Asian summer monsoon precipitation is sensitive to the temperature over the South Asian countries, where the sulfate aerosol loading is expected to increase rapidly during the next century. Moreover, some part of the region has dry climate, where the estimated aerosol scattering is especially sensitive to the inclusion of humidity-dependence. 4. Conclusion The impact of the direct effect of sulfate aerosol in the future climate projection has been evaluated with an explicit representation of aerosol scattering, and the result was compared with experiments with the same implicit method and the parameter values as used in previous projections. As suggested by previous radiation computations (Kiehl and Briegleb, 1993), the estimated radiative forcing due to the direct effect of sulfate aerosol is significantly smaller with the explicit representation, than with the implicit one. A further sensitivity study suggested that a principal source of the overestimation, especially over dry regions, by the implicit method appears to be the neglect of the dependence of aerosol scattering on near-surface humidity. Accordingly, the
1304 Journal of the Meteorological Society of Japan Vol. 77, No. 6 Fig. 3. The geographical distributions of differences in annual mean surface air temperature averaged over the last 50 years (2040-2090) between the experiments, (a) SUL-CTL, (b) SUL-GHG, and (c) ALB-GHG.
December 1999 S. Ernori, T. Nozawa, A. Abe-Ouchi et al. 1305 Fig. 4. Same as Fig. 3 but for precipitation.
1306 Journal of the Meteorological Society of Japan Vol. 77, No. 6 impact of the addition of sulfate aerosols on the projected surface air temperature is considerably smaller in magnitude, especially over dry regions, with the explicit method than with the implicit one. The Asian summer monsoon precipitation is considered to be particularly sensitive to the representation of sulfate aerosol scattering, because of abundant sulfate aerosols expected over the South Asian regions in the next century, and a large uncertainty in aerosol scattering over dry parts of the regions. The present study implies that the impact of the direct effect of sulfate aerosol previously estimated with the method of Charlson et al. (1991) is considerably overestimated. Although it may seem contradictory to the significant improvement of the simulations of historical temperature with the inclusion of the sulfate direct effect demonstrated by Mitchell et al. (1995a), it is natural considering the indirect effect of sulfate and the effects of the other kinds of aerosols, which may play an essential role, are still ignored. It is also suggested that if the dependence on near-surface humidity is properly included, an implicit method may give a forcing value that is as small as that given by an explicit method. Even then, the use of an explicit method is preferable for reducing uncertainty because it is more physicallybased. It should be noted that an explicit representation of the direct effect has still some uncertainty in such aspects as particle size distribution, and its dependence on near-surface humidity. Moreover, the dependence of particle size on humidity is different for different chemical forms of sulfate, which is further unknown (Kiehl and Briegleb, 1993). The bias of the near-surface relative humidity simulated in the climate model is another major source of uncertainty. It is desired for future works to properly include the direct and indirect effects of sulfate aerosol and other kinds of aerosols, to reassess future climate change with reduced uncertainty. To validate the modeled present-day aerosol effects with satellite observations is also crucial (Nakajima and Higurashi, 1998). Acknowledgments We thank Y. Yamanaka, N. Suginohara, Y. Tsushima and H. Okamoto for the model development, and Y.N. Takayabu, A. Higurashi, I. Uno, H. Kanzawa and A. Sumi for valuable discussion and encouragement. Thanks are also due to T. Johns for advising on the scenario, to J. Langner and H. Rodhe for providing the sulfate data, and to the two anonymous reviewers for their valuable comments to the manuscript. Thanks are extended to D. Hall for proof-reading this paper. This work was partially supported by the Global Environment Research Program of the Environmental Agency of Japan. The calculations were made on a NEC SX-4 computer in the Center for Global Environment Research. The GFD-DENNOU Library was used for the drawings. References Arakawa, A. and W.H. Schubert, 1974: Interactions of cumulus cloud ensemble with the large-scale environment. Part I. J. Atmos. Sci., 31, 671-701. Charlson, R.J., J. Langner, H. Rodhe, GB. Leovy and S.G. Warren, 1991: Perturbation of the Northern Hemisphere radiative balance by backscattering from anthropogenic sulphate aerosols. Tellus, 43AB, 152-163. d'almeida, GA., P. Koepke and E.P. Shettle eds., 1991: Atmospheric Aerosols: Global Climatology and Radiative Characteristics. A. Deepak Publishing, 56lpp. Hasselmann, K., L. Bengtsson, U. Cubasch, G.C. Hegerl, H. Rodhe, E. Roeckner, H. von Stroch, R. Voss and J. Waszkewitz, 1995: Detection of anthropogenic climate change using a fingerprint method. Technical report, Max-Planck-Institute fur Meteorologie Report No. 168, Hamburg, 24pp. Haywood, J.M. and K.P. Shine, 1995: The effect of anthropogenic sulfate and soot aerosol on the clear sky planetary radiation budget. Geophys. Res. Lett., 22, 603-606. Haywood, J.M., R.J. Stouffer, R.T. Wetherald, S. Manabe and V. Ramaswamy, 1997: Transient response of a coupled model to estimated changes in greenhouse gas and sulfate concentrations. Geophys. Res. Lett., 24, 1335-1338. IPCC (Intergovernmental Panel on Climate Change), 1992: Climate Change 1992: The Supplementary Report to the IPCC Scientific Assessment. Houghton, J.H., BA. Callander and S.K. Varney (eds.), Cambridge University Press, Cambridge, UK, 205pp. IPCC, 1996: Climate Change 1995: The Science of Climate Change. Houghton J.H., L.G. Meira Filho, BA. Callander, N. Harris, A. Kattenberg and K. Maskell (eds.), Cambridge University Press, Cambridge, UK, 572pp. Joseph, J.H., W.J. Wiscombe and J.A. Weinman, 1976: The delta-eddington approximation for radiative transfer. J. Atmos. Sci., 33, 2452-2459. Kiehl, J.T. and B.P. Briegleb, 1993: The relative roles of sulfate aerosols and greenhouse gases in climate forcing. Science, 260, 311-314. Lal, M., U. Cubasch, R. Voss and J. Waszkewitz, 1995: Effect of transient increase in greenhouse gases and sulphate aerosols on monsoon climate. Current Science, 69, 752-763. Langner, J. and H. Rodhe, 1991: A global three dimensional model of the tropospheric sulfur cycle. J. Atmos. Chem., 13, 225-263. Le Treut, H. and Z.-X. Li, 1991: Sensitivity of an atmospheric general circulation model to prescribed SST changes: feedback effects associated with the simulation of cloud optical properties. Clim. Dyn., 5, 175-187. Le Treut, H., M. Forichon, O. Boucher and Z.-X. Li, 1998: Sulfate aerosol indirect effect and CO2 greenhouse forcing: equilibrium response of the LMD
December 1999 S. Emori, T. Nozawa, A. Abe-Ouchi et al. 1307 GCM and associated cloud feedbacks. J. Climate, 11, 1673-1684. Meehl, GA., W.M. Washington, D.J. Erickson III, B.P. Briegleb and P.J. Jaumann, 1996: Climate change from increased CO2 and direct and indirect effects of sulfate aerosols. Geophys. Res. Lett., 23, 3755-3758. Mitchell, J.F.B. and T.C. Johns, 1997: On modification of global warming by sulphate aerosols. J. Climate, 10, 245-267. Mitchell, J.F.B., T.C. Johns, J.M. Gregory and S.F.B. Tett, 1995a: Climate response to increasing levels of greenhouse gases and sulfate aerosols. Nature, 376, 501-504. Mitchell, J.F.B., R.A. Davis, W.J. Ingram and C.A. Senior, 1995b: On surface temperature, greenhouse gases and aerosols: models and observations. J. Climate, 8, 2364-2386. Nakajima, T. and M. Tanaka, 1986: Matrix formulation for the transfer of solar radiation in a plane-parallel scattering atmosphere. J. Quant. Spectrosc. Radiat. Transfer, 35, 13-21. Nakajima, T. and A. Higurashi, 1998: A use of twochannel radiances for an aerosol characterization from space. Geophys. Res. Lett., 25, 3815-3818. van de Hulst, H.C., 1957: Light Scattering by Small Particles, John Wiley and Sons, New York, 470pp. World Climate Programme, 1986: A preliminary cloudless standard atmosphere for radiation computation, WCP-112, WMO/TD-No. 24, World Meteorological Organization, Geneva.