The seasonal cycle in coupled ocean-atmosphere general circulation models

We examine the seasonal cycle of near-surface air temperature simulated by 17 coupled ocean-atmosphere general circulation models participating in the Coupled Model Intercomparison Project (CMIP). Nine of the models use ad ho c "flux adjustment" at the ocean surface to bring model simulations close t o observations of the present-day climate. We group flux-adjusted and non-f lux-adjusted models separately and examine the behaviol- of each class. Whe n averaged over all of the flux-adjusted model simulations, near-surface ai r temperature falls within 2 K of observed values over the oceans. The corr esponding average over non-flux-adjusted models shows errors up to similar to6 K in extensive ocean areas. Flux adjustments are not directly applied o ver land, and near-surface land temperature errors are substantial in the a verage over flux-adjusted models, which systematically underestimates (by s imilar to5 K) temperature in areas of elevated terrain. The corresponding a verage over non-flux-adjusted models forms a similar error pattern (with so mewhat increased amplitude) over land. We use the temperature difference be tween July and January to measure seasonal cycle amplitude. Zonal means of this quantity from the individual flux-adjusted models form a fairly tight cluster (all within similar to 30% of the mean) centered on the observed va lues. The nonflux-adjusted models perform nearly as well at most latitudes. In Southern Ocean mid-latitudes, however, the non-flux-adjusted models ove restimate the magnitude of January-minus-July temperature differences by si milar to5 K due to an overestimate of summer (January) near-surface tempera ture. This error is common to five of the eight non-flux-adjusted models. A lso, over Northern Hemisphere mid-latitude land areas, zonal mean differenc es between July and January temperatures simulated by the non-flux-adjusted models show a greater spread (positive and negative) about observed values than results from the flux-adjusted models. Elsewhere, differences between the two classes of models are less obvious. At no latitude is the zonal me an difference between averages over the two classes of models greater than the standard deviation over models. The ability of coupled GCMs to simulate a reasonable seasonal cycle is a necessary condition for confidence in the ir prediction of long-term climatic changes (such as global warming), but i t is not a sufficient condition unless the seasonal cycle and long-term cha nges involve similar climatic processes. To test this possible connection, we compare seasonal cycle amplitude with equilibrium warming under doubled atmospheric carbon dioxide for the models in our data base. A small but pos itive correlation exists between these two quantities. This result is predi cted by a simple conceptual model of the climate system, and it is consiste nt with other modeling experience, which indicates that the seasonal cycle depends only weakly on climate sensitivity.