How does internal climate variability affect transient climate feedbacks?

Abstract:

A lot of attention has been focused on the inter-model spread in climate sensitivity, and the radiative feedbacks that contribute to it, as a measure of our uncertainty in the climate system's response to external forcing. Several of the studies investigating inter-model spread in climate feedbacks use transient climate simulations. The spread in feedback estimates obtained in this way, however, is due not only to structural differences among models, but also to internal climate variability. Here we explicitly assess the impact of internal variability on radiative feedback estimates in 40 transient SRES A1B scenario simulations with NCAR's Community Climate System Model Version 3 (CCSM3). We find that the spread in globally averaged feedbacks among members of this Large Ensemble corresponds to an average of 25% of the spread in feedbacks derived from the Coupled Model Intercomparison Project Version 3 (CMIP3) multi-model ensemble. The internal variability in the CCSM3 Large Ensemble is characterized using Empirical Orthogonal Functions (EOFs) and compared to the variability in feedback estimates from CMIP3. We identify EOFs that appear to represent similar processes in both ensembles using pattern correlation. Assuming that the multi-model ensemble includes variability due to both internal processes and model differences, we attempt to remove the impact of internal variability from CMIP3 feedback estimates by reconstructing feedbacks from their EOFs excluding those that exhibit a strong similarity to the leading EOFs in the Large Ensemble.