Start Date/Time: Friday, September 06, 2013, 10:30 AM
Ending Date/Time: Friday, September 06, 2013, 11:30 AM
Location: ATG 310C
Atmospheric Sciences Special Seminar
Speaker: Dr. Peter Caldwell, Research Scientist, PCMDI, Lawrence Livermore National Labs, California
Title: Data Mining the CMIP5 Archive for Constraints on Climate Sensitivity
Quantifying the sensitivity of earth's climate to increased CO2 is the central goal of climate scientists, but progress on this front has been slow. Several recent efforts have focused on identifying observable quantities which are good predictors of future climate in global models, then using the resulting relationships to make observationally-based predictions. In this study we search through more than 42,000 current-climate fields from Phase 5 of the Coupled Model Intercomparison Project (CMIP5) for the best predictors of model equilibrium climate sensitivity (ECS). We show that more fields in the CMIP5 archive are highly correlated with ECS than can easily be explained by chance. Determining which of these strong correlations are physically meaningful is a challenge, however. To connect correlations to physical processes, we derive an approximate equation partitioning correlation with ECS into a weighted sum of correlations with constituent feedbacks. This partitioning reveals that shortwave cloud amount and scattering feedbacks are the main contributors to strong correlation with ECS. The best correlated fields are associated with shortwave cloud feedback over the southern ocean; investigating cloud/climate interaction in this region is identified as the most likely place to search for current-climate constaints on ECS.