Willie Soon* and Sallie Baliunas
Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge
MA 02138, USA
The role of General Circulation Models (GCMs) has become predominantly important
as the practical interest in regional impacts from anthropogenic greenhouse gases such
as carbon dioxide (CO2) grows. This first report documents the quality of GCMs as a
tool for describing and predicting ‘global warming’ and the related geographical
pattern of climate change from the CO2 added to Earth’s atmosphere.
I Validation problems in climate modelling and General Circulation Models (GCMs)
Quantification of the impact of anthropogenic CO2 forcing, as well as its connection to
global warming and other natural or man-made climatic forcings, requires validation of
GCMs and reduction of their common deficiencies in simulating important climatic
variables. Recent reviews (Palmer, 2001; Pielke, 2001; Soon et al., 2001; Munk, 2002;
Pittock, 2002; van der Veen, 2002) call for a more comprehensive consideration of
nonlinear processes to describe the interactions among the atmosphere, oceans, land
and ice-covered surfaces. Those media, the reviewers argue, must be treated as
interactive, rather than mere unchanging surfaces or reservoirs, in order to progress in
the study of year-to-year and century-long climate change on both regional and global
scales (i.e., as distinct from the problem of weather prediction). The enormity and
urgency of the scientific task, with modern societal needs for coping with climate
change regardless of the added concerns about anthropogenic CO2 forcing, has led
Pielke (2001: 313), among others, to remind us that ‘there have been no model
experiments to assess climate prediction in which all important atmosphere–ocean–
land surface processes were included’. Observational capability must also advance significantly
before climate model validation can reach the next level of maturity (Goody
et al., 2002). Robust long-term monitoring of key climate quantities such as thermal and
dynamical outputs of the Sun (Parker, 2000), spectrally resolved infrared radiance
(Keith and Anderson, 2001), cloud properties (Wang et al., 2000; Rossow et al., 2002),
Progress in Physical Geography 27,3 (2003) pp. 448–455
© Arnold 2003 10.1191/0309133303pp391pr
*Author for correspondence. E-mail: wsoon@cfa.harvard.edu
surface radiative energy fluxes (Wild et al., 2001), mass balance of glaciers (Dyurgerov,
2002; van der Veen, 2002), sea ice thickness (Holloway and Sou, 2002; Preller et al., 2002)
and even a global mass-distribution constraint such as Earth’s mean dynamic
oblateness parameter J2 (Cox and Chao, 2002) would be important improvements.
II Simulations of climate variables
The current generation of GCMs have shown serious gaps and systematic deficiencies
in calculating both regional and global changes for many variables such as temperature,
precipitation, cloud properties and important oceanic variables, including oceanic
circulation, pattern of sea surface temperature, as well as sea surface elevation (sealevel)
and bottom pressure (Palmer, 2001; Pielke, 2001; Soon et al., 2001; Munk, 2002;
Schneider, E.K., 2002; van der Veen, 2002; Huang and Jin, 2002; Davey et al., 2002).
1 Temperature
As noted by Johnson (1997), the appearance of the 1990 Intergovernmental Panel on
Climate Change (IPCC) report (Houghton et al., 1990) marked the recognition that all
GCMs suffer from ‘the general coldness problem’, which is particularly acute in the
lower tropical troposphere (below 500 mb) and upper polar troposphere (above 500
mb). The general coldness problem is seen in 104 out of the 105 outcomes, covering the
entire troposphere, from 35 different simulations by 14 climate models. The ubiquitous
cold-bias problem persists to date, as shown in the collection of GCM simulations
compared with observed stratospheric and tropospheric temperatures in Pawson et al.
(2000).
Johnson (1997) suggested that the origin of the cold bias arises from an extreme
sensitivity of GCMs to systematic, aphysical entropy sources introduced by spurious
numerical diffusion, Gibbs oscillations or inadequate sub-grid-scale parameterizations.
Johnson estimated that a temperature bias of 10°C may be expected from only a 4%
error in modelling net heat flux that is linked to any number of aphysical entropy
sources, including those owing to numerical problems with the transport and phase
changes of water in vapor, liquid or ice, and the spurious mixing of moist, static energy.
A follow-on study by Johnson et al. (2000) shed further light on how the cold bias
associated with spurious, positive-definite entropy contaminates the computation of
hydrologic and chemical processes by virtue of their strong inherent dependence on
temperature. Johnson et al., estimated that the error in saturation-specific humidity
doubles for every 10°C increase in temperature. Johnson (1997: 2842) further stressed
that ‘erroneous sources of entropy in atmosphere and ocean models differ in both origin
and intensity. Efforts in coupled climate modeling to simulate accurately energy
exchange across the mutual atmosphere–ocean interface will be extremely difficult . . .
The implication is that atmospheric and oceanic energy balances within coupled
climate models that do not require flux adjustment are suspect’.
W. Soon and S. Baliunas 449
450 Global warming
2 Precipitation
Soden (2000) documented the inability of some 30 different atmospheric GCMs in the
Atmospheric Modeling Intercomparison Project (AMIP) to reproduce faithfully
interannual changes in precipitation over the Tropics (30°N to 30°S). Good agreement
was found between observations and the GCMs’ simulations of atmospheric water
vapour content, tropospheric temperature at 200 mb and outgoing longwave radiation.
However, observations and model simulations of precipitation and net downward
longwave radiation at the surface disagreed. Considering the direct association of latent
heat release by precipitation from moist air to the warming and cooling of the
atmosphere, Soden (2000) noted that the good agreement between the observed and
modelled temperature at 200 mb is surprising, given simultaneously the large
difference between the observed and modelled precipitation fields. One explanation for
the good temperature agreement at 200 mb is that it could be fortuitous because the
atmospheric GCMs were forced with observed sea surface temperatures. Meanwhile,
the modeled interannual variabilities of the hydrologic cycle, seriously underestimated
by a factor of three to four (Soden, 2000), correctly diagnose miscalculation of the precipitation
fields. Based on the models’ relatively constant values of downward
longwave radiation reaching the surface (see Wild et al., 2001 for further quantitative
comparisons around the globe), Soden (2000) points to possible systematic errors in
current GCM representations of low-lying boundary-layer clouds. However, the study
cannot exclude the possibility of errors in algorithms that retrieve precipitation data
from satellite observations, which would emphasize the need for improved precipitation
data products.
3 Clouds
Wielicki et al. (2002) offered observational evidence for large decadal variability of the
tropical mean radiative energy budget of the past two decades, which may be explained
by independently observed changes in tropical mean cloudiness. More significantly,
those results highlighted ‘the critical need to improve cloud modeling’ because several
GCMs failed to simulate that large observed variability in the tropical energy budget.
Grabowski (2000) emphasized the importance of proper evaluation of the effects of
cloud microphysics on tropical climate by using models that directly resolve mesoscale
dynamics. Grabowski pointed out that the main effect of cloud microphysics is on the
ocean surface rather than directly on atmospheric processes. Because of the great
mismatch between the timescales of oceanic and atmospheric dynamics, Grabowski
was pessimistic about quantifying the relation between cloud microphysics and tropical
climate. The parameterizations of cloud microphysics and cloud formation processes,
as well as their interactions with other variables of the ocean and atmosphere, remain
major challenges.
4 Ocean thermodynamics and dynamics: tropical ocean climatology and the stability
of the North Atlantic Thermohaline Circulation (THC)
In a systematic comparison of the performance of 23 dynamical ocean–atmosphere
W. Soon and S. Baliunas 451
models, Davey et al. (2002: 418) found that ‘no single model is consistent with observed
behaviour in the tropical ocean regions . . . as the model biases are large and gross errors
are readily apparent’. Without flux adjustment, most models produced annual mean
equatorial sea surface temperature (SST) in the central Pacific that are too cold by 2–3°C.
All GCMs except one simulated the wrong sign of the east–west SST gradient in the
equatorial Atlantic. The GCMs also incorrectly simulate the seasonal climatology in all
ocean sectors and its interannual variability in the Pacific ocean; surface wind stress is
diagnosed as the key parameter leading to those poor outcomes. The shortfall in
interannual variability is more pronounced for zonal wind stress than for SST.
Schneider, E.K. (2002) made the first progress in isolating and understanding specific
intramodel and intermodel disagreements in the simulations of the equatorial Pacific
ocean climatology and variability by using various flux-corrected experiments.
Russell and Rind (1999) noted that despite a global warming of 1.4°C around the time
of CO2 doubling, large regional coolings of up to 4°C were forecasted in both the North
Atlantic Ocean (56–80°N, 35°W–45°E) and South Pacific (near the Ross Sea, 60–72°S,
165°E–115°W) because meridional poleward heat transfer was reduced over the North
Atlantic and local convection was suppressed over the South Pacific. However, Russell
et al. (2000) subsequently demonstrated that their GCM predicted unreliable regional
changes over the Southern Ocean because of the model’s excessive sea ice variability.
Another GCM’s high-latitude Southern Ocean suffers from a large, unphysical drift
(Cai and Gordon, 1999). For example, within 100 years of coupling the atmosphere to
the ocean, the modeled Antarctic Circumpolar Current artificially intensified by 30 Sv
(from 157 to 187 Sv) despite the use of flux adjustments. Cai and Gordon identified the
instability of convection patterns in the Southern Ocean of the GCM to be the primary
cause of that large drift.
The reasons for the projected weakening of the North Atlantic THC in various models
remain unclear because, as Mikolajewicz and Voss (2000) caution, the GCMs give
contrasting roles to individual atmospheric and oceanic fluxes of heat, moisture,
salinity and momentum. Thus, although a complete breakdown of the North Atlantic
THC is predicted for sufficiently strong CO2 forcing in some simulations (Schmittner
and Stocker, 1999, Rahmstorf, 2000), Wood et al. (1999) and Mikolajewicz and Voss
(2000) cautioned that the predicted changes of the THC are very sensitive to parameterizations
of various components of the hydrologic cycle, including precipitation,
evapouration and river runoff. For example, without a perpetually enhanced influx of
freshwater (from any source) or extreme CO2 forcing, the transient decrease in THC
overturning eventually recovers as time progresses in the model (Mikolajewicz and
Voss, 2000; Holland et al., 2000). In addition, when a dynamic sea ice module is included
in a coupled atmosphere–ocean model, Holland et al. (2000) report a reduction (rather
than an enlargement) in the variance of the THC overturning flowrate under the
doubled CO2 condition, down to 0.25 Sv2 (or only 7% of value simulated for presentday
forcing) from the high value of 3.6 Sv2 simulated under the present-day forcing
level.
Furthermore, Latif et al. (2000) reported a new stabilization mechanism that counters
previous expectations of a CO2-induced THC weakening. Latif et al.,’s state-of-the-art
coupled ocean–atmosphere GCM of the Max Planck Institut für Meteorologie at
Hamburg (MPI) resolves the tropical oceans at a meridional scale of 0.5º, rather than the
more typical scale of 2–6º, and produces no weakening of the THC when forced by
452 Global warming
increasing CO2. Latif et al. showed that anomalously high salinities in the tropical
Atlantic, produced by excess freshening through the remote forcing at the equatorial
Pacific, were advected poleward to the sinking region of the THC. The effect was
sufficient to compensate for the local increase in freshwater influx at North Atlantic.
Updated experiments for a similar CO2 forcing scenario by Gent (2001) and Sun and
Bleck (2001) also confirm the relative stability of the THC because of compensating
effects among thermal perturbation, changes in surface hydrology and salinity.
Therefore, sophisticated GCMs results are consistent with both a stable and unstable
North Atlantic THC under a future high CO2 radiative forcing scenario. Latif et al.
(2000) concluded that the response of THC to enhanced greenhouse warming is still an
open question, and Gent (2001) reiterated that estimating the response of the THC in the
twenty-first century, important as the question may be, is ‘a very demanding question
to ask of current state-of-the-art coupled climate models’.
III The incorrect fingerprint of CO2 forcing in GCMs
No GCM has successfully simulated the observed relative warming trend of the surface
layer compared with the low troposphere. All GCMs consistently predict that the
troposphere should warm faster than surface air when radiative forcing is enhanced by
CO2 (Bengtsson et al., 1999). Santer et al. (2001) confirmed that the wrong fingerprint is
observed compared with that expected from CO2 forcing on the atmosphere. GCMs
simulate trend differences of surface-minus-lower troposphere temperature significantly
smaller than the observed results for 51 out of 54 simulations examined, even for
best-effort attempts to account for trend biases introduced by effects of volcanoes and
El-Niño–Southern Oscillation (ENSO).
Some theoretical proposals expect a warming of the surface relative to the lower
troposphere because of nonlinear climate dynamics (Corti et al., 1999). That expectation
is because of the differential surface response with the pattern of Cold Ocean and Warm
Land (COWL) that becomes increasingly unimportant with distance from the surface
(see Soon et al., 2001, for additional explanation). Nevertheless, no GCM has incorporated
such an idea into an operationally robust simulation of the climate system
response to greenhouse effects from added CO2. This incorrect forecast of the
fingerprint of carbon dioxide forcing on the surface and atmosphere temperature needs
resolution.
IV Summary
Climate models are now being used extensively to diagnose the causative, especially
anthropogenic, factors of observed climatic changes of the past few decades (Palmer,
2001; Stott et al., 2001; Thorne et al., 2002). These models are also used to make long-term
climate projections and climate risk assessments based on future anthropogenic forcing
scenarios (Saunders, 1999; Palmer, 2001; Houghton et al., 2001; Pittock, 2002; Schneider,
S.H., 2002). Many such exercises help to shape public policy recommendations
concerning future energy use and various ‘climate protection’ measures in order to
prevent ‘dangerous climate impacts’ (e.g., Schneider, S.H., 2002; O’Neill and
W. Soon and S. Baliunas 453
References
Andronova, N.G. and Schlesinger, M.E. 2001:
Objective estimation of the probability density
function for climate sensitivity. Journal of
Geophysical Research 106, 22605–11.
Bengtsson, L., Roeckner, E. and Stendel, M.
1999: Why is the global warming proceeding
much slower than expected? Journal of
Geophysical Research 104, 3865–76.
Cai, W. and Gordon, H.B. 1999: Southern highlatitude
ocean climate drift in a coupled model.
Journal of Climate 12, 132–46.
Corti, S., Molteni, F. and Palmer, T.N. 1999:
Signature of recent climate change in
frequencies of natural atmospheric circulation
regimes. Nature 398, 799–802.
Cox, C.M. and Chao, B.F. 2002: Detection of a
large-scale mass redistribution in the terrestrial
system since 1998. Science 297, 831–33.
Davey, M.K. et al., 2002: STOIC: a study of
coupled model climatology and variability in
tropical ocean regions. Climate Dynamics 18,
403–20.
Dyurgerov, M.B. 2002: Glacier mass balance and
regime: data of measurements and analysis.
Institute of Arctic and Alpine Research
(INSTAAR) Occasional Paper 55, University of
Colorado, Boulder.
Forest, C.E., Stone, P.H., Sokolov, A.P., Allen,
M.R. and Webster, M.D. 2002: Quantifying
uncertainties in climate system properties with
Oppenheimer, 2002). But meaningful and credible scientific confidence, resting either
on the traditional deterministic method of quantification or the probabilistic mode of
measuring change (as favoured by, for example, Washington, 2000; Räisänen and
Palmer, 2001; Schneider, S.H., 2002) cannot yet be made to such computer experiments
because climate models do not yield sufficiently reliable, quantitative results in
reproducing well-documented climatic changes around the world.
Govindan et al. (2002) highlighted, by performing the detrended fluctuation analysis
on daily maximum temperature records for six sites spread across the globe, that seven
leading coupled GCMs systematically underestimated the observed long-range
persistence of the atmosphere (roughly after timescales longer than 2 years or so) and
overestimated the daily maximum temperature trend. From that failure of computer
models to emulate the observed behaviour in the real atmosphere, Govindan et al.,
deduced that ‘the anticipated global warming is also [likely] overestimated’ by those
leading GCMs.
Another key uncertainty in climate modelling concerns the true sensitivity of the
Earth’s climate system to a given radiative forcing. Andronova and Schlesinger (2001)
deduced a large range of values for the climate sensitivity parameter, DT2x (i.e., the
equlibrium global-mean surface temperature warming in response to a doubling of
atmospheric CO2 concentration) owing to limited understanding of natural variability
and uncertainty in climate radiative forcing. That sensitivity parameter, so necessary for
an accurate prediction of CO2 radiative forcing, has a 90% confidence interval of 1.0ºC
to 9.3ºC. Using a different model and tuning scenarios, Forest et al. (2002) similarly
obtained 1.4°C to 7.7ºC as the 5% to 95% confidence interval for DT2x.
Acknowledgements
This work was supported by the Air Force Office of Scientific Research grant AF 49620-
02-1-0194 and by the National Aeronautics and Space Administration grant
NAG5-7635.
454 Global warming
the use of recent climate observations. Science
295, 113–17.
Gent, P.R. 2001: Will the North Atlantic ocean
thermohaline circulation weaken during the
21st century? Geophysical Research Letters 28,
1023–26.
Goody, R., Anderson, J., Karl, T., Miller, R.B.,
North, G., Simpson, J., Stephens, G. and
Washington, W. 2002: Why monitor the
climate? Bulletin of the American Meteorological
Society 83, 873–78.
Govindan, R.B., Vyushin, D., Bunde, A.,
Brenner, S., Havlin, S. and Schellnhuber, H.-J.
2002: Global climate models violate scaling of
the observed atmospheric variability. Physical
Review Letters 89, 28501(4).
Grabowski, W.W. 2000: Cloud microphysics and
the tropical climate: Cloud-resolving model
perspective. Journal of Climate 13: 2306–22.
Holland, M.M., Brasket, A.J. and Weaver, A.J.
2000: The impact of rising atmospheric CO2 on
simulated sea ice induced thermohaline
circulation variability. Geophysical Research
Letters 27, 1519–22.
Holloway, G. and Sou, T. 2002: Has Arctic sea ice
rapidly thinned? Journal of Climate 15,
1691–701.
Houghton, J.T., Jenkins, G.J. and Ephraums, J.J.,
editors, 1990: Climate change: the IPCC scientific
assessment. Cambridge: Cambridge University
Press.
Houghton, J.T., Ding, Y., Griggs, D.J., Noguer,
M., van der Linden, P.J., Dai, X., Maskell, K.
and Johnson, C.A., editors, 2001: Climate change
2001: the scientific basis. Cambridge: Cambridge
University Press.
Huang, R.X. and Jin, X.-Z. 2002: Sea surface
elevation and bottom pressure anomalies due
to thermohaline forcing. Part I: Isolated perturbations.
Journal of Physical Oceanography 32,
2131–50.
Johnson, D.R. 1997: ‘General coldness of climate
models’ and Second Law: implications for
modeling the Earth system. Journal of Climate
10, 2826–46.
Johnson, D.R., Lenzen, A.J., Zapotocny, T.H. and
Schaack, T.K. 2000: Numerical uncertainties in
the simulation of reversible isentropic
processes and entropy conservation. Journal of
Climate 13, 3860–84.
Keith, D.W. and Anderson, J.G. 2001: Accurate
spectrally resolved infrared radiance
observation from space: Implications for the
detection of decade-to-century-scale climatic
change. Journal of Climate 14, 979–90.
Latif, M., Roeckner, E., Mikolajewicz, U. and
Voss, R. 2000: Tropical stabilization of the thermohaline
circulation in a greenhouse warming
simulation. Journal of Climate 13, 1809–13.
Mikolajewicz, U. and Voss, R. 2000: The role of
the individual air–sea flux components in CO2-
induced changes of the ocean’s circulation and
climate. Climate Dynamics 16, 627–42.
Munk, W. 2002: Twentieth century sea level: an
enigma. Proceedings of the National Academy of
Sciences 99, 6550–55.
O’Neill, A. and Steenman-Clark, L. 2002: The
computational challenges of Earth-system
science. Philosophical Transactions of the Royal
Society of London, Series A 360, 1267–75.
O’Neill, B.C. and Oppenheimer, M. 2002:
Dangerous climate impacts and the Kyoto
Protocol. Science 296, 1971–72.
Palmer, T.N. 2001: A nonlinear dynamical
perspective on model error: A proposal for
non-local stochastic-dynamic parameterization
in weather and climate prediction models.
Quarterly Journal of the Royal Meteorological
Society 127, 279–304.
Parker, E.N. 2000: The physics of the Sun and the
gateway to the stars. Physics Today 53 (6), 26–31.
Pawson, S. et al., 2000: The GCM-Reality Intercomparison
Project for SPARC (GRIPS):
scientific issues and initial results. Bulletin of the
American Meteorological Society 81,
781–96.
Pielke, R.A. Sr., 2001: Earth system modeling –
an integrated assessment tool for environmental
studies. In Matsuno, T. and Kida, H.,
editors, Present and future of modeling global environmental
change: Toward integrated modeling.
Tokyo: Terra Scientific Publishing Company
(TERRAPUB), 311–37.
Pittock, A.B. 2002: What we know and don’t
know about climate change: Reflection on the
IPCC TAR. Climatic Change 53, 393–411.
Preller, R.H., Posey, P.G., Maslowski, W., Stark,
D. and Pham, T.T.C. 2002: Navy sea ice
prediction systems. Oceanography 15, 44–56.
Rahmstorf, S. 2000: The thermohaline ocean circulation:
a system with dangerous thresholds?
Climatic Change 46, 247–56.
Räisänen, J. and Palmer, T.N. 2001: A probability
and decision-model analysis of a multimodel
W. Soon and S. Baliunas 455
ensemble of climate change simulations.
Journal of Climate 14, 3212–26.
Rossow, W.B., Delo, C. and Cairns, B. 2002:
Implications of the observed mesoscale
variations of clouds for the Earth’s radiation
budget. Journal of Climate 15, 557–85.
Russell, G.L. and Rind, D. 1999: Response to
CO2 transient increases in the GISS coupled
model: regional coolings in a warming climate.
Journal of Climate 12, 531–37.
Russell, G.L., Miller, J.R., Rind, D., Ruedy, R.,
Schmidt, G. and Sheth, S. 2000: Comparison of
model and observed regional temperature
changes during the past 40 years. Journal of
Geophysical Research 105, 14891–98.
Santer, B.D., Wigley, T.M.L., Doutriaux, C.,
Boyle, J.S., Hansen, J.E., Jones, P.D., Meehl, G.
A., Roeckner, E., Sengupta, S. and Taylor, K.E.
2001: Accounting for the effects of volcanoes
and ENSO in comparisons of modeled and
observed temperature trends. Journal of
Geophysical Research 106, 28033–59.
Saunders, M.A. 1999: Earth’s future climate.
Philosophical Transaction of the Royal Society of
London Series A 357, 3459–80.
Schmittner, A. and Stocker, T.F. 1999: The
stability of the thermohaline circulation in
global warming experiments. Journal of Climate
12, 1117–33.
Schneider, E.K. 2002: Understanding differences
between the equatorial Pacific as simulated by
two coupled GCMs. Journal of Climate 15,
449–69.
Schneider, S.H. 2002: Can we estimate the
likelihood of climatic changes at 2100? Climatic
Change 52, 441–51.
Soden, B.J. 2000: The sensitivity of the tropical
hydrological cycle to ENSO. Journal of Climate
13, 538–49.
Soon, W., Baliunas, S. Idso, S.B., Kondratyev,
K.Y. and Posmentier, E.S. 2001: Modeling
climatic effects of anthropogenic carbon
dioxide emissions: unknowns and uncertainties.
Climate Research 18, 259–75.
Stott, P.A., Tett, S.F.B., Jones, G.S., Allen, M.R.,
Ingram, W.J. and Mitchell, J.F.B. 2001:
Attribution of twentieth century temperature
change to natural and anthropogenic causes.
Climate Dynamics 17, 1–21.
Sun, S. and Bleck, R. 2001: Atlantic thermohaline
circulation and its response to increasing
CO2 in a coupled atmosphere–ocean model.
Geophysical Research Letters 28, 4223–26.
Thorne, P.W., Jones, P.D., Tett, S.F.B., Parker,
D.E., Osborn, T.J. and Davies, T.D. 2002:
Ascribing potential causes of recent trends in
free atmosphere temperatures. Atmospheric
Science Letters, doi:10.1006/asle.2001.0046.
van der Veen, C.J. 2002: Polar ice sheets and
global sea level rise: how well can we predict
the future? Global and Planetary Change 32,
165–94.
Wang, J.-H., Rossow, W.B. and Zhang, Y.-C.
2000: Cloud vertical structure and its variations
from a 20-yr global rawinsonde dataset. Journal
of Climate 13, 3041–56.
Washington, R. 2000: Quantifying chaos in the
atmosphere. Progress in Physical Geography 24,
499–514.
Wielicki, B.A. et al. 2002: Evidence for large
decadal variability in the tropical mean
radiative energy budget. Science 295, 841–44
(see additional exchanges in Science 296,
2095a).
Wild, M., Ohmura, A., Gilgen, H., Morcrette, J.-
J. and Slingo, A. 2001: Evaluation of
downward longwave radiation in General
Circulation Models. Journal of Climate 14,
3227–39.
Wood, R.A., Keen, A.B., Mitchell, J.F.B. and
Gregory, J.M. 1999: Changing spatial structure
of the thermohaline circulation in response to
atmospheric CO2 forcing in a climate model.
Nature 399, 572–75.
