~N_A_TU-'-R~E~V~O~L-'-. -'-33-'-6~ 1 =D~E=C=E_M_B_ER~ l-'-98"-'8------ARTICLES 441

Are global and controlled by marine ? Stephen E. Schwartz Environmental Chemistry Division, Brookhaven National Laboratory, Upton, New York 11973, USA

The recent suggestion that dimethylsulphide emissions by marine phytoplankton control global albedo and mean temperature would also imply a strong climatic influence of man-made S02• Anthropogenic S02 emissions exceed marine emissions of dimethylsulphide globally and are confined largely to the Northern Hemisphere. But no such influence of S02 emissions is found either in the present cloud component of planetary albedo or in JOO-year temperature records.

RECENTLY, Charlson, Lovelock, Andreae and Warren1 gaseous sulphur emissions between the Northern Hemisphere (hereinafter CLAW) proposed that production of dimethylsul­ (NH) and Southern Hemisphere (SH) constitute a global experi­ phide (DMS) by is the major source of ment that tests the hypothesis of ref. 1, that planetary albedo aerosol sulphate in the remote marine atmosphere and, in turn, and mean temperature are regulated by marine DMS emissions. of cloud condensation nuclei (CCN). They suggest that this It will be seen that a comparison of the present NH and SH process, by influencing planetary albedo, comprises a com­ cloud component of planetary albedo and of 100-year tem­ ponent of a biological mechanism for climate regulation. perature records yields no indication of any influence of

Specifically, it is hypothesized that an increase in marine DMS anthropogenic SO2 emissions that is either qualitatively or quan­ emissions would increase the number density of aerosol sulphate titatively consistent with the expectations in this respect based particles and, in turn, the number density of cloud droplets, on the CLAW hypothesis. thus enhancing . This enhancement would reduce global temperature generally, including temperatures, and Emission rates would consequently reduce marine productivity and DMS Emissions of reduced sulphur compounds and of sulphur 5 7 emission; that is to say, the process is a system dioxide have been critically reviewed recently - and are sum­ for control of the Earth's temperature. Such a feedback system marized in Table 1. Andreae6 estimates total global biogenic 2 was subsequently and independently suggested by Meszaros • emissions of sulphur gases to be 95 ± 30 Tg of sulphur per year,

Here I examine the hypothesized link between gaseous precur­ DMS and H2S each accounting for about half of this total 1 6 sors of aerosol sulphate, and global albedo and mean tem­ ( 43 ± 20 and 32 ± 27 Tg S yr- , respectively). The estimate for 1 perature. Because marine DMS emissions are substantially lower marine DMS emissions, 40 ± 20 Tg S yr- , is based on concentra­ than present emissions of other gaseous precursors of atmos­ tions of OMS in ocean surface at a variety of locations pheric sulphate (principally SO2 associated with com­ together with estimates of sea-air exchange rates. A lower OMS bustion) globally and especially in the Northern Hemisphere, emission rate of 16 Tg S yr- 1 (uncertain to a factor of two) is 8 then consequently, if, as CLAW suggest, marine DMS emissions given by Bates et a/. • Natural emission rates in Table 1 are exert a major influence on global mean albedo and temperature, consistent with budget estimates based on considerations of such an influence should also result from anthropogenic SO2 turnover times and of concentrations of sulphur compounds in 9 emissions. The potential for such anthropogenic influence on the atmosphere • cloud albedo and global mean temperature has been noted On a global basis, SO 2 emissions are roughly equal to biogenic 4 previously3· • But CLAW explicitly "ignore the perturbations of emissions and are at least twice as great as marine OMS the atmospheric sulphur cycle by man-made fluxes of em1ss1ons. A prima facie case exists, therefore, that SO 2 .•• and ... consider only the natural fluxes, which currently anthropogenic sulphur emissions cannot be neglected in examin­ represent about 50% of the total flux of gaseous sulphur to the ing the possible influence of biogenic gaseous sulphur com­ atmosphere, and which still dominate the atmospheric sulphur pounds on global cloud albedo and climate. Also, as cycle in the Southern Hemisphere". I argue here that the increase anthropogenic SO 2 emissions have increased to their present in SO 2 emissions over the past 100 years and the contrast in level almost entirely within the last 100 years (ref. 5 and H. Rodhe, private communication cited in ref. 7), any climatic influence of these emissions should be discernible in records Gaseous sulphur emission rates 10 Table 1 over this period . Furthermore, the asymmetric distribution of anthropogenic SO emissions over the globe suggests that the Northern Southern 2 of gaseous sulphur emissions on climate may also be Emission source Hemisphere Hemisphere Global influence tested by interhemispheric comparisons, significant interhemis­ Marine DMS 17 23 40 pheric transport (on a timescale of one to two years) being precluded by the short atmospheric residence times of these Total marine biogenic 22 28 50 9 11 substances • • Terrestrial biogenic 32 16 48 Distribution of aerosol sulphate Total biogenic 54 44 98 For sulphate derived from anthropogenic SO to serve as a Anthropogenic SO2 98 6 104 2 surrogate to test the CLAW hypothesis, this material must be Total 150 50 200 widespread throughout the NH. The geographical distribution Data are from Cullis and Hirschler5, except for marine DMS of sulphate attributable to anthropogenic emissions of SO2 is 6 (Andreae ), apportioned according to ocean surface area: 42% NH and examined in Table 2, which presents concentrations of aerosol 58 % SH. Emission rates are in units of Tg S yr- '. sulphate in boundary-layer air at locations remote from

© 1988 Nature Publishing Group _44_2 ______ARTICLES------N_A_T_U_R_E_V_O_L_. _33_6_1_D_E_C_E_M_B_E_R_I_9_88

Table 2 Measured concentration of sulphate (or sulphur) aerosol at remote locations in the Northern and Southern Hemisphere

Sulphate concentration* 3 Location (ng S m- ) Comments Ref.

Northern Hemisphere High-latitude sites Alert, Mould Bay, Igloolik; NSS so~- '; I-week samples; 3- to 4-year data record 15 Canadian Arctic (66-83° N) (200-1,000) Winter-Spring (20-70) Summer Faeroe Islands (62° N) NSS so;;- 13 1,100 (700-1,400) British trajectory (-1,000 km); 4 I-day samples 140 (30-230) Atlantic trajectory; 4 1-day samples Velen, Sweden (58° N) Sub-µ.m S; 1-day samples 18 960 [ 1.4] British trajectory (-1,000 km); 12 samples 60 [2.2] North Sea trajectory (Northwest air); 14 samples Atlantic and Caribbean Western North Atlantic 800±500 Sub-µ.m S; 26 8- to 68-hour samples 18 (33-38° N, 65-70° W) Bermuda (32° N) 530 [3.0] Sub-µ.m S; 10 2- to 4-day samples 18 Bermuda Sub-2.5-µ.m NSS so:;-; 39 I-day samples 14 1,300 Northeast US trajectory ( -1,200 km) 630 Southeast US trajectory (-1,100-1,500 km) 390 Caribbean trajectory 330 Southeast trajectory 650±230 All Barbados ( 13° N, 60° W) NSS SO;;- ; 1-day samples 17 300± 190 'High dust'-Europe or Africa trajectory 120±65 'Low dust' 250± 180 All Pacific Eastern Pacific Ocean off Aircraft sampling, boundary layer; 3 flights; Pacific trajectories; 20 Washington State (47-48° N) May 140±60 Total NSS so:;­ 80±30 Sub-1.5-µ.m NSS So;;- Midway Island (28° N) NSS so~- ; I-week samples; onshore flow only 16 260±90 'Dusty'; 29 samples 100±30 "Clean'; 27 samples 190±260 All; 56 samples Oahu (21°N) NSS sot; I-week samples; onshore flow only 16 160± 130 'Dusty'; 24 samples 90±80 'Clean'; 32 samples 120± llO All ; 56 samples Guam (l7°N) 150±240 NSS so~- ; I-week samples; onshore flow only; 49 samples 16 Belau (7° N) 210± 190 NSS so~- ; I-week samples; onshore flow only; 40 samples 16 Fanning Island ( 4° N) 210±50 NSS so~- ; I-week samples; onshore flow only; 48 samples 16

Southern Hemisphere

Samoa (14° S) 60 [2.4] Sub-µ.m S; 17 3- to 5-day samples 18 Samoa 130± 50 NSS SO;;-; 42 I-week samples 19 New Caledonia (22° S) 170± 130 NSS SO~- ; 46 I-week samples 19 Norfolk Island (29° S) 110±50 NSS SO~-; 41 I-week samples 19 Cape Grim, Tasmania (41 ° S) 90±20 NSS SO;;- ; multi-year data record under 'baseline' conditions 43 Tasmania, off West Coast (22-70) NSS so~-; aircraft sampling, boundary layer; 8 flights; ocean 21 trajectories; Austral summer Punta Arenas, Chile (54° S) 52 [0.3] Austral summer; sub-µ.m S; 9 3- to 5-day samples 18 83 Austral summer; NSS S; 140 4-hour samples 11 Austral winter; sub-µ.m S; 1 5-day sample South Pole 76±24 Austral summer; total S; sea salt small; 40 several-day samples 44 29± 10 Austral winter; total S; sea salt 4-10%; 49 several-day samples

* Mean ± standard deviation or geometric mean [geometric standard deviation] or (range). t NSS so;;- : Non-seasalt sulphate. industrial activity. Such locations are pertinent because aerosol of only sub-micrometre (sub-µ,m) sulphate (that fraction arising sulphate concentrations are obviously high in industrialized from gas-to-particle conversion, which is of concern here). The regions and at locations immediately influenced by transport data in Table 2 represent measurements by numerous inves­ from these regions. By comparison, median 24-hour boundary­ tigators using a variety of techniques and protocols for widely layer sulphate concentrations at non-urban locations of the differing sampling periods, so that any comparison of data from north-eastern United States are -2,000 ng S m-3 (ref. 12). Total different investigators must be made with caution. Nonetheless, aerosol samples are corrected for sea salt by reference to Na or Table 2 indicates a pattern in which aerosol sulphate concentra­ Mg; this correction is generally unimportant for measurements tions at remote NH locations substantially exceed those at

© 1988 Nature Publishing Group _N_A_TU_R_E_V_O_L._3_3_6_l_D_E_C_E_M_B_E_R_l_98_8______ARTICLES------44_3 remote SH locations. Concentrations in the NH are mostly trajectory was similar, because H+ and NSS Soi- were highly 3 greater than -150 ng S m- , often much greater, whereas con­ correlated ( correlation coefficient R = 0.94; slope = 0.94). centrations at remote locations in the SH are almost invariably Recently, Whelpdale et al. 25 concluded that even the most less. Concentrations in the NH also exhibit substantially greater remote parts of the North Atlantic are undoubtedly affected by variability than those in the SH, indicating the intermittent anthropogenic and that only a few data are available presence, in the former case, of aerosol sulphate transported in which precipitation is derived from air from 'clean' regions. from distant regions. Their estimate of sulphate concentrations in 'background' 1 For several NH sites, high concentrations have explicitly been marine precipitation samples ( 6-8 µeq 1- ) substantially exceeds ascribed to long-range transport from continental regions or, concentrations reported from remote SH sites (Table 3 ), in di cat- more specifically, from regions of industrial activity, by the trajectory of the air arriving at the site or by the presence of other air constituents. For example, high sulphate concentra­ Table 3 Volume-weighted mean concentrations of non-seasalt sulphate tions at the Faeroe Islands are attributed to transport from the in precipitation of remote sites 13 United Kingdom, 1,000 km distant , and high sulphate con­ centrations at Bermuda are attributed to transport from North 14 NSS America (1,100 km) • The marked seasonality of sulphate con­ sot µeql- 1 centrations in the Canadian Arctic15 is attributed to transport Location Dates Ref. Poker Flat, Alaska (o7° N) 10.2 5/80-5/81 22 from Eurasian sources during late winter and spring; this sea­ Bermuda (32° N) 18.2 5/80-5/81 45 sonal cycle is exhibited also by a number of other anthropogenic 13.9 8/82-5/84 46 constituents. High non-seasalt (NSS) sulphate at several North San Carlos, Venezuela (2° N) 3.0 9/80-3/81 22 Pacific islands commonly coincides with soil dust, transported Katherine, Australia (14° S) 3.2 1980-1984 47 16 mostly from Asia . Similarly, high NSS sulphate at Barbados, Amsterdam Island (38° S) 5.0 5/80-8/83 47 coincident with high incidence of soil dust, is attributed to Torres del Paine, Chile (47° S) 2.1 1984-1985 47 transport from North Africa or Europe (distances of 5,000 km 17 or more) • Several data sets in Table 2 afford direct comparison of ing that even this background contains a substantial component measurements at NH and SH locations. Lawson and of anthropogenically derived sulphate. Considerably higher con­ Winchester18 present measurements of sub-µm S for which NH centrations were reported for samples obtained with trajectories concentrations were an order of magnitude greater than SH from North America or Europe. concentrations. Prospero et al. 16 and Saltzman et al. 19 present an extensive set of measurements from several North Pacific16 Sulphate in polar ice and South Pacific19 islands; the mean NSS sulphate concentra­ Substantial increases in sulphate concentrations over the past tion for five NH islands was 30% greater than that for three century have been noted in polar ice at NH sites. Concentration SH islands. of NSS soi- in ice at Dye 3 Station, Greenland (65° N) has 20 21 26 27 Andreae et al. and Berresheim et al. report aircraft tripled since -1900 • • Similarly, there has been a substantial measurements of NSS sulphate by similar techniques in boun­ increase in conductivity of melt water from core samples dary-layer air off the coast of Washington state20 and off the obtained at Ellesmere Island (81° N) from 1910 to 1980, which 1 west coast of Tasmania2 , both sets of data relating to air behind is attributed to increased deposition of sulphate, nitrate and 28 cold fronts moving over open ocean without contact with land associated cations • In contrast, a corresponding increase in for several days. The NSS sulphate concentrations for Washing­ sulphate concentration is not observed in ice cores obtained at 3 29 30 ton state ranged from 90 to 190 ng S m- , whereas those for Antarctica • • One may conclude that deposition of NSS Soi­ 3 Tasmania ranged from 20 to 70 ng S m- • NSS sulphate con­ at these remote NH sites has increased as a consequence of centrations in the free troposphere at Tasmania (30-60 ng S m-3 anthropogenic emissions and that the relative increase in this at STP) were comparable to those in the boundary layer, but deposition is comparable to that in NH sulphur emissions. those at Washington state were substantially greater (150- Thus, comparisons of concentrations of NSS sulphate in 270 ng S m-3 at STP). This situation argues strongly against a aerosol, precipitation and ice cores at remote locations in the marine source. Indeed, Andreae et al. 20 suggest, in the case of NH and SH establish that there are substantially greater con­ the Washington state measurements, that the sulphur cycle in centrations throughout the NH than in remote SH locations, the free troposphere was dominated by transport from Asia. mainly as a result of anthropogenic emissions. These observa­ tions confirm my premise that if marine DMS emissions were Sulphate in precipitation to exert a control over global cloud albedo and mean tem­ Table 3 presents volume-weighted mean concentrations of NSS perature, then there should be an influence on these variables sulphate obtained in long-term records of wet-only event samp­ from anthropogenic SO2 emissions. 22 ling at several remote NH and SH sites • The measurements indicate markedly greater concentrations in the NH, supporting Interhemispheric comparison of cloud albedo the present assertion that sulphate derived from anthropogenic A key component of the CLAW hypothesis is that the cloud emissions exerts an influence on air composition throughout the contribution to planetary albedo should increase with increasing NH. emissions of sulphate precursors. In the absence of measure­ Long-range transport of anthropogenic sulphur is indicated ments that would permit comparison of present and pre-indus­ also by the dependence of the composition of precipitation on trial , I seek to identify the effect of anthropogenic back trajectory. For samples collected in the eastern North sulphur emissions on planetary albedo by a comparison of the Atlantic, Nyberg23 found that for trajectories from North present albedos of the two hemispheres. The expected magnitude America (4,000-5,000 km), NSS soi- concentrations were 12- of the interhemispheric albedo difference may be determined 1 37 µeq 1- , whereas for a trajectory emanating from the Azores, from the example given in ref. 1, which indicates that a 30% 1 24 concentrations were -6 µeq 1- • Similarly, Ji ckells et al. estab­ increase in CCN concentration would increase planetary albedo lished that, in precipitation at Bermuda, systematically greater above marine by 0.016, and average global albedo by acidity was associated with trajectories from North America 0.005, where the latter quantity takes into account the fraction 1 (mean H+ concentration= 41 µeq 1- ) than with trajectories of the Earth's surface covered by marine clouds. Following this 1 1 from the Caribbean (8 µeq 1- ) or the east (12 µeq 1- ); one can example, and using a NH: SH CCN ratio of 3: 1 ( compare Table assume that the dependence of NSS sot concentration on 1), I obtain a difference between average NH and SH albedos

© 1988 Nature Publishing Group -444------ARTICLES------N_A_T_U_R_E_ V_ O_ L_._3_3_6_1_D_E_C_E_M_B_E_R_19_88

Latttude ("N or 0 S)

0 15 7590 0.8

0.7 Fig. l Average annual zonal mean over 10° latitude bands for 0.6 Northern and Southern Hemispheres of total planetary albedo aT and of cloud and clear-sky components, Fcac and ( I - FcJas, respectively. The key is as follows: e, total albedo (NH); 0, total 0.5 albedo (SH); •, Cloud component (NH); 0, cloud component (SH); A, clear-sky component (NH); 'v, clear-sky component 0.4 (SH). Single-hatching indicates that the NH cloud component exceeds the SH component; cross-hatching indicates the reverse. 0.3 Equal distances on the lower abscissa correspond to equal areas on the Earth"s surface. 0.2

0.1

0 0 0.2 0.4 0.6 0.8 1.0

Sin (latitude)

of 0.023. Essentially the same result is obtained using the tant equilibrium decrease in mean global temperature is AT= logarithmic dependence of albedo on CCN concentration pro­ AQ/ A, where A, the parameter, is estimated34 4 2 1 posed by Twomey et a/. • Although such an albedo difference to be 1.8 W m- K- (uncertain to a factor of two); this yields is undoubtedly an overestimate, in view of the non-uniform AT= -1.0 K. (This value of AT, and others evaluated here, are distribution of sulphate concentrations in the NH, even a 50% all uncertain to a factor of two because of uncertainty in A. difference in CCN concentrations between the two hemispheres Evidently CLAW used a somewhat lower value of A, yielding (see Tables 2 and 3) would, in the absence of other factors, lead AT= -1.3 K.) Analogously, an increase of 0.023 in mean NH to an expected mean interhemispheric albedo difference of0.008 albedo ( which, according to the CLAW hypothesis, would result (NH greater than SH). from a tripling of NH sulphur emissions) should give rise to a In view of differences in land area and surface features of decrease in mean NH temperature of 4.4 K if, along with the the two hemispheres (and resultant differences in surface increase in emissions and concentrations, the cooling effect were 35 36 albedo) I have looked for indication of an influence of NH SO2 confined entirely to the NH · . Alternatively, if the temperature emissions by comparing not the total albedo of each hemisphere change were distributed over both hemispheres, an average but the component of this albedo that is due to clouds, under global temperature decrease of 2.2 K would be indicated. Again, the assumption that this quantity was more nearly comparable because of spatial non-uniformities in CCN concentration, such in the pre-industrial atmosphere. The cloud component of temperature changes may be overestimates. But even a NH albedo is a direct measure of the contribution of clouds to solar albedo increase of 0.008, which corresponds in the CLAW of the Earth's heat budget31 and is thus an hypothesis to a uniform 50% increase in CCN concentration appropriate indicator of a possible influence of gaseous sulphur over pre-industrial values, would result in a NH temperature emissions on this budget. decrease of 1.5 K in the absence of interhemispheric coupling, Figure 1 shows, for each hemisphere, the annual-average or a global temperature decrease of 0.8 K if the cooling was mean of total albedo and of the cloud and clear-sky components distributed globally. of this albedo, over 10° latitudinal zones. The cloud component These estimates represent equilibrium temperature changes, is evaluated31 as whereas any currently observable temperature decrease corre­ sponding to the indicated change in radiative forcing would be less, by a factor of about two, because of the lag time in reaching 10 39 thermal equilibrium • • It should be noted that such predicted 32 where Fe represents average fractional cloud coverage ,

© 1988 Nature Publishing Group _N_A_TU_R_E_' _V_0_L_. 3_3_6_1_D_E_C_E_M_B_E_R_1_9_88______ARTICL£S 445

Summary and discussion One such influence that must be addressed is that of warming caused by CO2 and other greenhouse gases. Current evalu­ The hypothesis that emissions of DMS by marine phytoplankton ations34·37 of expected global temperature changes since 1850 exert a regulatory influence on mean global cloud albedo and resulting from increased concentrations of CO2 and other temperature has been tested by searching for a possible influence infrared-active gases ( CH4, N 20, chlorofluoromethanes) indi­ of anthropogenic SO2 emissions on these properties. These cate an expected mean temperature increase of -1 K. (This emissions exceed estimated global marine DMS emission by a temperature change is evaluated from modelled radiative forc­ factor of two, are comparable to total biogenic gaseous sulphur ing34·37 with the same sensitivity parameter A employed above emissions globally, are confined largely to the NH, and have and exhibits the same factor of two uncertainty.) Such an expec­ reached their present levels within the past 100 years. Moreover, ted temperature increase is comparable to but somewhat larger aerosol sulphate is widespread throughout the NH at concentra­ than estimates of actual temperature changes38 -41 , the difference tions well in excess of those in remote SH locations. These being attributed10·37 both to the time lag in response of global features of anthropogenic SO 2 emissions and resultant sulphate temperature to the change in radiative forcing and to the uncer­ concentrations suggest that the influences of anthropogenic SO2 tainty in A, although in principle some of that difference might emissions on mean hemispheric cloud albedo and on mean be due to enhanced cloud albedo from anthropogenic sulphate. hemispheric or global temperature would be evident if these The several greenhouse gases are, however, all long-lived relative aspects of global climate were controlled by sulphate derived to interhemispheric mixing time, and are hence uniformly mixed from marine DMS emissions. The absence of any excess mid­ globally, leading to an essentially symmetric forcing function latitude cloud component of albedo in the NH relative to the on temperature, whereas anthropogenic sulphate, confined SH, and the lack of any global cooling or of any reduced warming largely to the NH, would exert its influence largely in that region. in the NH relative to the SH, indicate that these variables are The lack of hemispherical asymmetry in the rate of change in not controlled by anthropogenic sulphates, and by extension, global temperature38 -41 thus suggests that compensation of are not controlled by marine DMS emissions. cooling from enhanced cloud albedo by warming caused by The apparent lack of observable influence of emissions of greenhouse gases does not account for the observations. Also, gaseous sulphur compounds on climate is puzzling. There is greenhouse heating would not account for the lack of enhanced little doubt that SO 2 and reduced sulphur compounds are precur­ cloud contribution to NH albedo. sors to aerosol sulphate42 . It is also well established that sul­ In conclusion, it would seem that the lack of discernible phate-containing aerosol particles are effective cloud condensa­ response of mean global or hemispheric albedo or temperature 1 3 1 34 tion nuclei ' . There is also persuasive theoretical ' ' and to anthropogenic SO2 emissions indicates that control of these observational3 evidence that the albedo of liquid-water clouds properties is too complex to be governed by a single variable increases with cloud droplet number density. It thus seems of this type. Nonetheless, the potential for a substantial human reasonable that an influence on the cloud contribution to global influence on global climate by the mechanism examined here mean albedo and temperature might be expected, as Twomey makes it mandatory to gain a thorough understanding of the et al.4, CLAW 1 and Meszaros2 have suggested. The observational processes that control cloud albedo and its influence on global evidence, however, suggests otherwise. Thus, unless there is a climate. countervailing influence upon global or hemispheric cloud I thank Peter Daum, Paul Falkowski, Martin Hoffert, Paul albedo or temperature that has masked the influence of SO2 Michael, Leonard Newman and Douglas Wallace for valuable emissions, it seems unlikely that global cloud albedo and discussions. This work was supported by the National Acid temperature are controlled by emissions of gaseous sulphur Precipitation Assessment Program through the PRECP Program compounds. funded by the US Department of .

Received 16 May; accepted 18 October 1988. 22. Galloway, J. N., Likens, G. E., Keene, W. C. & Miller, J.M. J. geophys. Res. 87, 8771-8786 (1982). I. Charlson, R. J., Lovelock, J.E., Andreae, M. 0. & Warren, S. G. Nature 326,655-661 (1987). 23. Nyberg, A. Q. J. R. met. Soc. 103, 607-615 (1977). 2. MCsz

© 1988 Nature Publishing Group SCIENTIFIC CORRESPONDENCE greater than the spin frequency, stratifica­ Note added in proof: Purely Alfvenic et at. 6 for the total planetary albedos, tion can be important!'. In the case W,ib oscillations as a cause of the 8-hour an and the clear-sky planetary albedos, » Q » Vja, the low frequency hydro­ periodicity have also been noted by Katz!6. a,. Ellis's data are an inadequate source magnetic waves have a more complicated R. L. McNun, JR of statistics on Earth's albedo. The values character. Taking W,'b == IN x A1/K (N is Department of Physics and used by Schwartz are derived from eight the Brunt-Vaiasala frequenci'), frequen­ Center for Space Research, weeks of Nimbus 3 satellite data recorded cies of the short wavelength modes (wave Massachusetts Institute of Technology, during four two-week periods in 1969. For vector K) are w! = W;'b + (2Q . KlK)' + Cambridge, Massachusetts 02139, USA one of these periods, most of the Pacific (VA' K)' and W'- = (VA' K)' (For N = 0, Ocean north of 45° was not sampled. 1. Kristian, J. et al. Nature 338,234-236 (1989). W _ becomes the hydromagnetic inertial 2. Gaske", C. M. Nature 338, 121-122 (1989). There are now more complete and more mode). Purely Alfvenic oscillations are 3. Wang, Q. etal. Nature 338, 319--320 (1989). accurate data sets of satellite-measured decoupled from the rotation frequency, 4. Ray, A. & Datta, B. Astrophys. J. 282, 542-549 (1984). albedos. 5. Easson, I. Astrophys. J. 228, 257-267 (1979). and the inferred field strength is again 6. Friedman, J. L., Ipser, J. R. & Parker, L. Astrophys. J. 304, Similarily, the data of Beryland et at. 6 -5 x 109G. 115--139 (1986). are a sound, valid estimate of global cloud 7. Ruderman, M. A. Nature 225,619--620 (1970). Vibrational modes could, in principle, 8. Boriakoff, V. Astrophys. J. 208, L43--L46 (1976). cover. At the time of its production it was remove the phase accumulation problem 9. Van Horn, H. M. Astrophys. J. 236, 899--903 (1980). arguably the best global cloud climatology 10. Cowling, T. G. Proc. R. Soc. A233, 319--322 (1955). via the beating of two closely spaced 11. Kendall, P. C. Q. J. Mech. appl. Math. 13, 285--299 but it has naturally been superseded by modes (ilw = 2rr/8 hours). However, (1960). very many satellite-based nephanalyses hydromagnetic effects and rotation lead to 12. Alpar, M. A., Langer, S. A. & Sauls, J. A. Astrophys. J. and global archives of surface-based 282,533--541 (1984). splitting at the rotation frequency'4 and 13. Feibelman, P. G. Phys. Rev. 4, 1589--1597 (1971). observations of clouds (for example, ref. 7). this only when the Alfvenic oscillation 14. Papaioizou,J. &Pringle, J. E. Mon. Not. R astr. Soc. 182, We think that Schwartz should have 423--442 (1978). frequency dominates as in the surface 15. Acheson, DJ. & Hide, R. Rep. Prog. Phys. 36, 159 (1973). compared the curves of planetary albedo, layers of the star. 16. Katz, J.1. Nature 339,263--264 (1989). an in his Fig. 1. These curves show that NH = SH at the equator, NH > SH at 5°, SH> NHat 15°, NH > SH at 25°, equality Sulphate aerosols and climate at 35° and 45" and SH > NH in the remain­ SIR-One of Schwartz's! conclusions is CCN in the United States, Canada and the ing four latitude bands, 55° to 85°. The that the cloud component of the albedo of Arctic and thus compound the cooling for differences are less than 0.02 and seem to the Southern Hemisphere (SH) is substan­ which Schwartz is searching. On the other alternate between 0° and 60° (poleward of tially greater than that due to clouds in the hand, these clouds could be predominantly 60° both the cloud and albedo data are Northern Hemisphere (NH). He asserts cirrus type and hence augment the green­ highly suspect). Thus, in some latitudes that this difference, being of opposite sign house heating and perhaps thereby com­ the hypothesis of Charlson et at. 2 seems to to that which would result from the pres­ pensate for Schwartz's hemispherically be validated by Schwartz's Fig. 1. ence of anthropogenically elevated quan­ asymmetrical CCN. We do not know how Schwartz does not, however, examine tities of cloud-condensation nuclei (CCN) cloud amounts have varied this century, the total planetary albedo differences; in the NH, contradicts the thesis of much less whether cloud types have altered. rather he considers the "cloud component Charlson et al'. Schwartz's second argu­ It seems foolhardy, at present, to assume of the albedo". The curves plotted by ment depends on his assertion, taken from that all other features of the climate Schwartz are of the quantity a, F" where 3 elsewhere , that warming trends this system except CCN and greenhouse gases, F, is the fractional cloud cover, derived century show "no discernible interhemi­ have remained hemispherically symmetri­ directly by subtracting the product spheric difference". cal and unchanging this century. Without (1 - F,) aT' This term is itself a product of There are many factors known to affect this assumption, which is highly question­ cloud amount and cloudy sky albedo. temperatures. We have shown (for able, and without temporal analysis of Schwartz derives larger values for this example, ref. 4) that Northern American CCN variations, Schwartz cannot, in our product in the SH for most latitudes which cloud amounts have increased this century opinion, deduce anything from hemispheric indicates either that F, is greater in the SH, by as much as 10 per cent. Such an increase temperature trends. that a, is greater in the SH, or both. The in cloud amount could, of course, be asso­ Schwartz uses Ellis' as the source for cloud amounts, F" which Schwartz uses ciated with increased concentrations of latitudinal cloud amounts, F" and Berlyand are indeed generally larger in the SH and 0.45,------,

0.4 0.8 /- --- / \ / \ , \ 0.35 \ 0.3 C 0.6 :::l 0 0 1:J 0.25 E Q)

Latitude (deg) Latitude (deg) FIG, Latitudinally averaged values of total cloud amount derived from Berlyand and Strokina6 and the three-dimensional nephanalysisB derived predominantly from surface-based observations for many years and satellite-based observations for 1979, respectively. Note that the satellite• based cloud amounts are latitudinally more coherent and that both climatologies have SH cloud> NH cloud in most latitudes. 436 NATURE· VOL 340 . 10 AUGUST 1989 © 1989 Nature Publishing Group SCI ENTI FIC CORRESPON DENCE other global nephanalyses' also tend We do not know why different analyses at face value, we see that it contains data to have greater cloud cover in the SH of the historical record of surface temp­ from both 'dusty' and 'clean' episodes in (see figure). That the latitudinal cloud eratures lead to opposite conclusions the North Pacific and North Atlantic. amounts differ rather considerably from about the trends. But even if those studies Much of the sulphate in the dust episodes one another in the figure is not surprising were to be reconciled they might still not is actually on coarse particles and there­ when it is recognized that the USAF three­ offer a clear test of the effect of sulphate fore provides few cloud condensation dimensional nephanalysis is predomin­ aerosol on cloud albedo, because there nuclei (CCN). During the 'low-dust' antly a satellite archive for 1979 whereas are probably climate factors other than season, Barbados shows about the same the Berlyand et al. data are predominantly cloud albedo that could cause the temper­ values as Samoa, and the summer Arctic surface-based and accumulated over a ature trends to differ in the two hemis­ and North Sea trajectory data from Velen much longer period. In any case, it is well pheres. (See also Wigley8.) are about the same as Cape Grim and known that evaluations of latitudinally Using a published cloud climatology Punta Arenas. We conclude that many of averaged values of total cloud amount and published measurements of planetary the NH sites are about as clean as the differ by at least ± 10% (absolute cloud albedo, Schwartz also concludes that climatically and biogeographic ally com­ amount)9. Thus if Schwartz had chosen to average cloud albedo is no greater in the parable SH sites for much of the time. The use a different source for his F, values, the NH than in the SH. He takes this as only significant difference arises during a, F, curves in his Fig. 1 would have been evidence against an influence of industrial dust transport events (where much of the modified. Moreover, these curves are sulphur pollution on cloud albedo, despite SO;-is coarse and does not provide many highly suspect as they represent the plane­ the fact that the lifetime of sulphur in the CCN) or at some obviously and expec­ tary albedo as viewed from the top of the atmosphere is typically not more than a tedly polluted sites - for example, atmosphere in areas designated, from few days, so the effect of pollution­ Bermuda. Much of the dust transport is of another incompatible data source, to be derived aerosol may not extend much course natural, and would not contribute cloudy. beyond the NH continents. Therefore it is to a potential time trend. It seems that We conclude that the hypothesis of likely that the geographical footprint of anthropogenic inputs are significant for Charlson et al. has not been refuted. the pollution aerosol does not greatly boundary-layer n.s.s. SO;- levels only A. HENDERSON-SELLERS overlap that ofthe sensitive marine stratus over the Atlantic north of 30 oN, and School of Earth Sciences, clouds, particularly in the subtropics. possibly over the northern North Pacific, Macquarie University, However, sulphate aerosols, whether but sulphate distributions there are New South Wales 2109, Australia derived from industrial pollution or from presently not well known. Looking at our dimethylsulphide, can be expected own cruise data from the tropical Pacific, ll K. MCGUFFIE to influence only one of the determinants we find that the NH n.s.s. SO;- data are Department of Applied Physics, of cloud albedo, namely the effective actually lower than the SH and equatorial University of Technology, cloud-droplet radius, r,f[' As Stephens', ones. Consequently, 40 per cent at most of Sydney, PO Box 123, Broadway, for example, has shown, cloud albedo is a the area over the NH might New South Wales 2007, Australia function principally of three variables: experience significantly elevated CCN vertically integrated liquid water content concentrations. 1. Schwartz. S.E. Nature 336, 441-445 (1988). (liquid water path, LWP), solar zenith The aircraft data from the North Pacific 2. Charlson, R.J. Nature 326, 655--661 (1987). 3. Hansen, J. & Lebedefff, S. J. geophys. Res 92. 13345- angle, and r,", It is unlikely that the aver­ off Washington State and the south Indian 13372 (1987). age L WP of clouds in the SH is the same as Ocean off Tasmania do suggest significant 4. Henderson-Sellers, A. & McGuffie, K. Atmos. Ocean (in the 1 press). in the NH. As the average LWP for each differences in n.s.s. SO;-levels ,.13. But the 5. Ellis, J.E. thesis, Colorado State Univ. (1978). hemisphere is unknown, the effect of total aerosol sulphur in the atmosphere is 6. Berlyand. T.G., Strokina. L.A. & Greshnikova, L.E. Sov. sulphur on r,ffcannot be inferred just by Met. Hydrol. 3, 9--13 (1980). comprised of sulphate and methanesul­ 7. Hahn. C.J. et al. NCAR tech. Note TN-201 (National Center comparing hemispheric-average albedos, phonic acid. Thus, when methanesul­ for Atmospheric Research, Boulder, Colorado,1982). as SlingolO emphasized. Long-term trends phonic acid is added to n.s.s. SO;-, the 8. Hughes, N.A. & Henderson-Sellers. A. l. Clim. appl. Met. 24,669--686 (1985). in planetary albedo are also not available, total aerosol sulphur in the sub-cloud 9. Hughes, N.A. J. Clim. appl. Met. 23, 724-751 (1984). even though they have been measured for mixed layer is essentially the same for the three decades, because of changes in the Tasmania and Washington data sets. We SIR-Claiming that man-made emissions satellite instruments. know that the size distributions of of SO, have had no effect on surface­ Much of Schwartz's argument is based methanesulphonic acid and n.s.s. SO;- are temperature trends or planetary albedo, on a comparison of non-sea-salt sulphate similar, and we may assume that they are SchwartzI attempts to refute our hypo­ (n.s.s SO;-) concentration over the oceans internally mixed, and that both contribute thesis' of a biological mechanism for in the Northern and Southern hemispheres. to the mass of soluble particles, and thus climate regulation. His arguments are He compares measurements made by to CCN active at low supersaturation. flawed for several reasons. a variety of authors with a variety of But in the free troposphere there is a First, although Schwartz interprets the techniques. Unfortunately, there is con­ pronounced difference between the two ~ analyses of Jones et at. 3-5 to show that the siderable uncertainty about the accuracy changes in surface temperature in the past of such measurements. Very few sampler 1. Schwartz, S.E. Nature 336,441-445 (1988). 2. Charlson. R.J. et al. Nature 326,655--661 (1987). century for the Southern Hemisphere intercomparisons have been made, none 3. Jones, P.D., Wigley, T.M.L. & Wright, P.B. Nature 322, (SH) is not significantly greater than that of them conforming to the rigorous 430-434 (1986). 4. Jones, P.D. et al. J. Clim. appl. Met. 25, 161-179 (1986). for the Northern Hemisphere (NH), standards that are now expected to be part 5. Jones, P.D .• Raper, S.C.B. &Wigley. T.M.L.l. Clim. appl. 6 Jones et al. have themselves also said , of intercomparison protocols. Often the Met. 25, 1213-1230 (1986). large size fraction (> 1-2 !lm, depending 6. Jones, P.D. etal. Nature 332, 790 (1988). "the warmth of the 1980s is most evident 7. Oort, A.H .. Pan, Y.H., Reynolds, R.W. & Ropelewski, C.F. in the Southern Hemisphere". That state­ on the size cut of the sampler), which Clim. Dyn. 2, 29--38 (1987). ment refers to a single decade only, but an typically contains about 30-40 per cent of 8. Wigley, T.M.L. Nature 339, 365-367 (1989). 9. Stephens. G.L. l. atmos. Sci. 35, 2123-2132 (1978). independent analysis' has extended it back n.s.s. SO;-, is either not collected or its 10. Slingo. T. Nature 336.421 (1988). through two more decades and concluded n.s.s. SO;- content is ignored. Samples of 11. Raemdonck. H., Maenhaut, W. & Andreae. M.O. l. geophys. Res. 91, 8623-8636 (1986). that the 20-year trend of surface air temp­ fine n.s.s. SO;- and total n.s.s. SO;­ 12. Andreae, M.O. et al. l. atmos. Chem. 6,149--173 (1988). erature (1958-77) over the oceans has a should therefore not be compared, as has 13. Berresheim, H., Andreae, M.O., Ayers. G.P. & Gillet. been done by Schwartz'. RW. in Biogenic Sulfur in the Environment (Saltzman, latitudinal dependence, with the SH E.S. & Cooper, W.J. eds) (ACS Symposium Series 393, warming more rapidly. If we take the data in Schwartz's Table 1 352-366 1989). NATURE· VOL 340 . 10 AUGUST 1989 437 © 1989 Nature Publishing Group © 1989 Nature Publishing Group