Climate Change in the Tropical Pacific: Understanding and Quantifying

Uncertainties.

Matthew Collins,
Centre for Global Atmospheric Modelling,
Department of Meteorology,
University of Reading,
Early Gate,
Reading, RG6 6BB,
UK.

matcollins@met.rdg.ac.uk

The El Nino Southern Oscillation (ENSO) is a well know modulator of
global climate on seasonal to interannual time scales. The
quasi-period warming and cooling of Sea Surface Temperatures (SSTs) in
the eastern Tropical Pacific, and corresponding shifts in atmospheric
pressure and precipitation patterns cause climatic variations both
locally and via teleconnection patterns to remote areas. These short
term climate variations have significant impacts on ecology, society
and economics. What role then does the Tropical Pacific play the in
global change problem?

There are an increasing number of studies of climate and environmental
change which point to the Tropical Pacific as an important modulator
of the global climate system.  The studies of Latif et al. 2000 and
Thorpe et al. 2001 both examine the potential for a reduction in the
strength of the North Atlantic Ocean Thermohaline Circulation (THC) as
greenhouse gases increase. They find that the rate of the reduction in
strength (and hence the potential for a permanent shut-down) is
controlled by the advection of salty water from lower latitudes - the
saltier the Tropical Atlantic ocean, the smaller the reduction in THC
strength. During an El Nino event the Tropical Atlantic Ocean becomes
more salty through a combination of changes in evaporation,
precipitation and river run-off. In both the Latif et al. 2000 and
Thorpe et al. 2001 studies the climate models produce a climate change
towards more El Nino like mean conditions in the Tropical Pacific and
this results in changes to the freshwater cycle which cause a more
salty Tropical Atlantic ocean and thus limits THC slow down.

The work of Cox et al. 2000 also points to an important role for the
Tropical Pacific in global climate change. In their study they include
representations of the terrestrial and marine carbon cycles in a
global climate model and find a positive feedback in which the
terrestrial biosphere can flip from being a carbon sink to a carbon
source. The corresponding increase in atmospheric CO2 amplifies the
rate of global warming in the model considerably. One of the major
components of the flip is a die-back of the Amazonian rain forest
during the middle 21st century. A possible cause of this die-back is a
reduction in precipitation over the Amazon region caused by a shift in
atmospheric circulation and corresponding precipitation patterns
associated with, again, a shift towards more El Nino like mean
conditions in their model.

Understanding the climate of the Tropical Pacific is a complex problem
because of the tight coupling of the ocean and atmosphere in the
region. Coupled ocean-atmosphere global circulation models (AOGCMs),
which are now the principal tools for global change studies, show
large differences in their predictions for Tropical Pacific climate
change (Cubash et al., 2001). We can begin by examining two models
for changes in the mean climate and changes in the ENSO cycle.

Collins 2000a, using version 2 of the Hadley Centre Coupled Model
(HadCM2), found that at four times pre-industrial levels of CO2 ENSO
events became larger in amplitude and more frequent than present day
ENSO events. Thus in addition to the impacts of climate change, the
impacts of ENSO events would be felt more often and with greater
magnitude. Using version 3 of the Hadley Model (HadCM3) Collins 2000b
found that magnitude and frequency of ENSO events remained unchanged
as greenhouse gases increased, contradicting the results of the
earlier study and highlighting a level of uncertainty. This range
uncertainty in the future of ENSO is further widened when one examines
the responses of other AOGCMs (Cubash et al., 2001).

The difference in the ENSO response to global warming in HadCM2 and
HadCM3 was found by Collins 2000b to be due to differences in the
response of the mean climate of the two models in the Tropical Pacific
region. HadCM2 produced a pronounced broad maximum in SST warming on
the equator while HadCM3 had a more confined maximum and a change in
the south-north SST gradient (see figure). Forcing HadCM3 with the
pattern of mean SST warming from HadCM2 caused the HadCM3 ENSO cycle
to amplify and to become more frequent, much like the HadCM2 ENSO
response. Hence differences in the models mean pattern of SST change
caused differences in the response of ENSO to climate change.

But what caused the differences in the pattern of mean climate change
between HadCM2 and HadCM3? HadCM3 has a higher resolution oceanic
component than does HadCM2 and does not require a "flux-adjustment"
term to control climate drift (see Johns et al. 1997 and Gordon et
al. 2000 for details of the models). It would be tempting to attribute
the differences in mean climate response to these features. The reason
is much more subtle than this however. Williams et al. 2001 examined
changes in the physical parametrisations of cloud formation and the
representation of atmospheric boundary layer processes between HadCM2
and HadCM3 and found that rather small changes to these schemes could
combine in a non-linear way to produce the large differences in the
patterns of cloud, precipitation and SST change in the two models (see
figure). Cloud feedbacks are among the most important and most complex
of feedbacks in the climate system and it appears that even small
perturbations to the parameters of cloud models can cause non-linear
and far-reaching differences in global climate change.

Knowledge of the parameters of model physical schemes (particularly
those associated with clouds) may be ultimately limited by
observational errors or by uncertainties in parameters which have no
observable counterparts. How then can we resolve uncertainties in
climate change in the Tropical Pacific (and indeed globally) if models
are highly sensitive to small changes the the parameters of their
physical schemes? There are two methodologies we can adopt to
understand and quantify uncertainties in global climate change:
Palaeoclimate studies and mega-ensembles of AOGCMs.

Tudhope et al. 2001 have collected and analysed fossilised corals from
the Western Tropical Pacific. These corals provide windows of ENSO
variability in periods of history back 130,000 years. One fossilised
coral suggests that ENSO variability may have been very much weaker
(and perhaps even non-existent) around 6.5 thousand years ago, at a
time when global climate was not greatly different from that of
today. AOGCMs can be forced with boundary conditions from different
epochs and validated using the palaeo record. Given the sensitivities
of AOGCMs, the validation process needs to be quantitative and this
requires careful use of both the palaeo proxies which are indirect
measures of climate, and the model which simulates climate in
large-scale grid-boxes with dimensions of several hundreds of
kilometres (see e.g. Collins et al. 2001). One possible future
development would be to include models of coral growth within the
AOGCMs themselves.

A brute-force method of quantifying uncertainty in AOGCM estimates of
climate change is to systematically explore all the parameters of the
model physical schemes and map the potential climate change. This
would allow the construction of the probability density function (PDF)
of future climate scenarios for use by policy makers. Because of the
complexity of AOGCMs and because of potential non-linear interactions
between the parametrisation schemes (recall the Williams et al. 2001
study) this requires many thousands and perhaps millions of AOGCM
simulations. Such a study is beyond the capacity of the worlds current
supercomputer resource. A novel solution was suggested by Allen 1999
who proposed that if the AOGCM could be run on a Personal Computer
then members of the public could each produce a climate change
prediction with a combination of parameter perturbations, and the PDF
could be made. The climateprediction.com project is now underway -
more details are available from www.climateprediction.com.

It appears that, just as the Tropical Pacific is an important
modulator of global climate on seasonal and interannual time scales,
it also plays an important role in the global climate change
problem. The coupled AOGCMs currently show a wide range of possible
changes in both the mean and the ENSO variability in the region and
hence there is a large degree of uncertainty both in local and in
global human-induced climate change. It is hoped that future
palaeoclimate studies and mega-ensemble approaches will help in
understanding and quantifying these uncertainties.

References.

Allen, M. R. (1999): Do-it-yourself climate prediction, Nature, 401,
642.

Collins, M. (2000a): The El-Nino Southern Oscillation in the second
Hadley Centre coupled model and its response to greenhouse
warming. J. Climate, vol 13(7), 1299-1312.

Collins, M. (2000b): Understanding Uncertainties in the response of
ENSO to Greenhouse Warming. Geophys. Res. Letts., vol 27(21),
3509-3513.

Collins, M., Osborn, T. J., Tett, S. F. B., Briffa, K. R. and
Schweingruber, F. H. (2001): A comparison of the the variability of a
climate model with palaeo-temperature estimates from a network of
tree-ring densities. J. Climate, in press.

Cox, P. M., Betts, R. A., Jones, C. D., Spall, S. A. and Totterdell,
I. J. (2000): Acceleration of global warming due to carbon-cycle
feedbacks in a coupled climate model. Nature, 408, 184-187.

Cubash, U., Meehl G. A., et al. (2001): Projections of Future
Climate Change. Chapter 9, IPCC Third Assessment Report.

Gordon, C., Cooper, C., Senior, C. A., Banks, H., Gregory, J. M.,
Johns, T. C., Mitchell, J. F. B. and Wood, R. A. (2000): The
Simulation of SST, sea ice extents and ocean heat transport in a
version of the Hadley Centre coupled model without flux
adjustments. Climate Dynamics, 16, 147-168.

Johns, T. C., Carnell, R. E., Crossley, J. F., Gregory, J. M.,
Mitchell, J. F. B., Senior, C. A., Tett, S. F. B. and Wood,
R. A. (1997): The Second Hadley Centre Coupled Ocean-Atmosphere GCM:
Model Description, Spinup and Validation. Climate Dynamics, 13,
103-134.

Latif, M., Roeckner, E, Mikolajewicz, U. and Voss, R. (2000): Tropical
Stabilisation of the thermohaline circulation in a greenhouse warming
simulation, J. Climate, 13, 1809-1813.

Thorpe, R. B., Gregory, J. M., Johns, T. C., Wood, R. A. and Mitchell,
J. F. B. (2001): Mechanisms determining the Atlantic Thermohaline
Circulation reponse to greenhouse gas forcing in a non-fluxadjusted
coupled climate model. J. Climate, 14, 3102-3116.

Tudhope, A. W. et al. (2001): Variability in the El Nino-Southern
Oscillation through a glacial-interglacial cycle. Science, 291,
1511-1517.

Williams, K. D., Senior, C. A. and Mitchell, J. F. B. (2001):
Transient climate change in the Hadley Centre models: The role of
physical processes, J. Climate, 14, 2659-2674.

Figure Caption.

Sea Surface Temperature (SST) and precipitation changes at 4xCO2 from two
coupled ocean-atmosphere global circulation models. (a) SST change
from version 2 of the Hadley Centre Coupled Model (HadCM2), (b) SST
change from version 3 of the Hadley Centre Coupled Model (HadCM3), (c)
precipitation change from HadCM2 and (d) precipitation change from
HadCM3. The differences in the patterns of climate change between the
two models are caused by subtle changes in the physical representations
of clouds and these differences in mean climate also lead to very
different behaviour of the El Nino Southern Oscillation in a globally
warm world (see the text and Collins 2000b and Williams et al. 2001
for more details).