Wednesday, 16 January 2002: 5:15 PM
A comparison of the behavior of different AOGCMs in transient climate change experiments
At the present time coupled atmosphere ocean general circulation models (AOGCMs) are widely used not only for studying present-day climate but also for making projections of possible climate change. However, results produced by different AOGCMs differ significantly even when models are forced by the same changes in greenhouse gases concentrations. In this study we use results of simulations performed in the framework of the Coupled Models Intercomparison Project (CMIP). In those simulations models were forced with a 1% per year increase in CO2 concentration for 80 years. The increase in surface air temperature at the time of CO2 doubling (average for years 61-80) simulated by the different models ranges from 1.32K to 2.15K. A transient response produced by a given model is, to a large part, determined by two integrated characteristics of the model: equilibrium sensitivity to an external forcing and the rate of heat uptake by the ocean. In the MIT 2D climate model, these characteristics can be varied by changing the strength of the cloud feedback and the value of the effective deep ocean diffusion coefficient, respectively. Thus, we can evaluate the above characteristics for different AOGCMs by obtaining the version of the MIT model that will reproduce their transient behavior. The equilibrium surface warming in simulations with the corresponding atmospheric model coupled to a mixed layer ocean model for a doubled CO2 concentration was used as a measure of the climate sensitivity for a given AOGCM. The value of the equivalent diffusion coefficient was chosen so that the transient change of surface air temperature produced by a given AOGCM was reproduced by the version of the MIT model with the same climate sensitivity. The quality of the fit was checked by comparing sea level rise due to thermal expansion of the ocean simulated by the AOGCM and the 2D model. Satisfactory fits have been obtained for a number of the AOGCMs in this manner. However, as has been noticed recently, the sensitivity of coupled AOGCMs changes with time. Alternative fits were obtained with use of the effective climate sensitivity at the time of CO2 doubling. Use of the effective instead of the equilibrium sensitivity leads to better simulations of the models' behavior, especially for the thermal expansion. We will present results for a number of models with sensitivities ranging from 1.7 to 4.2 K and effective diffusion coefficient ranging from 5 to 20 cm2/s.