The range of equilibrium temperature responses to a doubling of CO2, as estimated from climate models, has remained in the range 1.5°C to 4.5°C over the past thirty years. This is despite substantial progress in the parameterization of climate processes in the models. A key factor in the different model estimates of climate sensitivity to the doubling of CO2 is the parameterization of clouds and convection. A difficulty in estimating the role of convective parameterization in climate sensitivity has been that most climate models differ not only in their convection schemes, but also differ in many other aspects of their physical parameterizations.

This presentation addresses the role of moist convective parameterization scheme in determining a model’s climate sensitivity. A new global coupled atmosphere-ocean climate model has been developed and is known as HIRES. It has two choices of numerical schemes and a range of cumulus parameterizations available. The atmospheric model resolution is 1.5 degrees, and the ocean resolution is 1 degree, increasing to ½ degree in the tropics. A series of 110-year simulations (1990-2100) has been carried out with increasing CO2 concentrations based on the IS92a scenario, as well as a parallel control series with fixed CO2. Each series is comprised of six ensemble members generated from the two numerical schemes (high order Eulerian and Lagrangian schemes) and three moist convection parameterization schemes (moist convective adjustment, Betts-Miller; and simplified Arakawa-Schubert). All the control climate runs give reasonable simulations of the mean climate and its interannual variability. It was found that the climate change simulations are relatively insensitive to the choice of numerical scheme. However, there are different climate sensitivities in the simulations generated by the three convection schemes.

In this analysis, we concentrate on the Transient Climate Response (TCR), and not the equilibrium climate sensitivity, as the model simulations have not been run to equilibrium. The global mean temperature change at the time of CO2 doubling (around 2070) is the key measure of the magnitude of the TCR. There is a clear dependence of TCR on the convective parameterization scheme used. As expected, the global mean precipitation change is also dependent on the parameterization scheme.

The continental scale pattern of temperature change is not sensitive to the convective parameterization scheme, but the magnitude of the temperature change is. The patterns of precipitation change are more sensitive to the choice of convection scheme, particularly in the tropics.