Towards a physical understanding of climate feedback spatial distributions and links to climate sensitivity

The sensitivity of the climate system to any external radiative forcing is determined by the strength of internal feedbacks that can amplify or dampen the initial system response. In the effort to understand differences between Global Climate Model (GCM) projections of future climate change under anthropogenic radiative forcing, emphasis has been placed on global mean feedback strength differences in accordance with the widely accepted linear forcing-response-feedback paradigm. However, global mean feedback strengths result from highly variable regional responses. This study investigates the variability of these spatial feedback patterns using the radiative kernel technique and IPCC AR4 CMIP3 GCMs. The results indicate significant model-to-model variability between all feedbacks with the largest differences found in cloud feedback. Using the spatial feedback patterns, we are able to investigate the model processes driving the spatial feedback patterns and links between various feedbacks, i.e. cloud and water vapor feedbacks and water vapor and lapse rate feedbacks. Significant correlations are found between the model tropical convective response and water vapor and cloud feedback. However, the tropical atmospheric lapse rate feedback exhibits no significant dependence on the spatial convective response pattern but rather on initial state convective distribution. The result is no significant spatial correlation between lapse rate and water vapor feedbacks. Lastly, results suggesting links between model climate sensitivity and the feedback spatial distributions will also be presented with a discussion of observable quantities linked to these spatial distributions.