Diagnosing Climate Feedbacks in Atmospheric General Circulation Models

Speaker(s): Karen Shell

Many different feedbacks influence how the earth’s temperature, precipitation, and winds respond to changes in the energy budget of the planet (caused, for example, by increases in carbon dioxide or other greenhouse gases). Because there is only one realization of the actual climate, computational climate models are useful tools for studying different scenarios and climate configurations in a very controlled setting. They also allow us to isolate particular climate feedbacks to determine how much they amplify or damp climate change. However, different climate models do not always produce the same magnitude and sign for a given climate feedback. In addition, different models generate different steady-state temperature changes for the standard doubled-\( \mathrm{CO_2} \) experiment, indicating that the net combination of the different feedbacks varies from model to model. An added complexity is that there are multiple techniques for calculating climate feedbacks. Some of the differences in reported feedback values may be due to the use of different techniques.I will present a new technique for calculating climate feedbacks in climate models. The feedback is separated into two parts: the change in climate components in response to an imposed radiative heating or cooling and the effects that those climate changes have on the radiative budget at the top-of-the-atmosphere (TOA). The usefulness of this technique depends on the linearity of the feedback processes. The sum of the effects of individual clear-sky components (water vapor, temperature, and surface albedo) on the TOA clear-sky change is similar to the clear-sky flux changes directly calculated by the climate model. However, due to nonlinear effects and interactions between components, this technique underestimates net shortwave flux changes and results in more negative outgoing longwave flux changes. Nevertheless, the technique is still useful for comparisons of feedbacks between models. Cloudy-sky flux comparisons are more difficult due to cloud overlap and strong nonlinear effects. Changes in “cloud radiative forcing” are often used as a measure of the cloud feedback. Using this new technique, I examine the contribution of clear-sky components to the change in cloud radiative forcing in order to demonstrate the limitations of using the cloud radiative forcing method to estimate cloud feedbacks.