William Connolley British Antarctic Survey, High Cross, Madingley Road Cambridge CB3 OET, UK Email: wmc@bas.ac.uk
and
Siobhan P. O'Farrell CSIRO Division of Atmospheric Research Private Bag No 1, Aspendale Victoria 3195, Australia Email: spo@dar.csiro.au
One of the largest changes in climate response between equilibrium experiments where the atmosphere is coupled to a mixed layer ocean and transient simulations where a full dynamic ocean is used, occurs in the Southern Ocean around Antarctica, where heat transfer into the deep ocean significantly moderates warming in the coupled model studies. This subduction of heat in the Southern Ocean has consequences for climate change scenarios across the Southern Hemisphere.
An examination of the air temperature changes in three coupled transient climate model runs (UKMO, CSIRO, MPI) over the last century shows that the changes they simulate are not compatible with observed trends in the Antarctic region over the past 50 years, although the sparsity of data makes comparison difficult in all but a few places. One of the best places for such comparisons is the Antarctic peninsula where the density of stations is higher than elsewhere in Antarctica and station records are longer. Observations show that this area has seen temperature rises of up to 0.5 oC/decade over the last 40 years, which are larger than anywhere else in Antarctica. None of the three GCMs show trends of this magnitude and none show their greatest warming in the Peninsula region. A comparison with the standard deviation of temperature variability of the models shows that changes of the magnitude observed in the Peninsula region are unlikely to occur by chance over a 50 year period. Therefore, either all the models are making serious errors in their simulations or the Peninsula warming is a regional phenomenon unrelated to global climate change.
Also compared are the trends in atmospheric model experiments forced with observed SSTs and ice extents from the GISST dataset for the 1880-1992 period.
There are still a number of outstanding problems in coupled models which may be impacting on the success of these models in matching the observations. The UKMO seaice is too extensive although the control is commendably drift-free. The MPI has a substantial climate drift in the control run linked to excessive growth in seaice but the seasonal cycle is well simulated. The CSIRO model has a reasonable seasonal cycle of sea ice extent but its sensitivity to an increase in ocean temperature may be too low. A further issue under investigation is whether the high rates of heat subduction in the coupled models are being over estimated due to the models representation of water mass formation and mixing.