CMIP6 is the latest iteration of the long-standing CMIP project, with a considerable expansion of contributing institutions, models and experiments comprising the latest phase. A key question however, is how do these latest models compare to previous phases when assessing key aspects of the mean climate state and responses to natural and anthropogenic forcings.

This presentation will focus on comparisons across the last 3 phases of the CMIP archive (CMIP3, 5, and 6), assessing improvements in the representation of mean climate and variability across multiple realms in the suites. We will provide an overview of key aspects of model performance contrasted to the latest available observations, as well as model-to-model comparisons of simulation performance. For the Pacific Decadal Oscillation, Taylor Diagrams indicate that there is substantial overlap in skill from CMIP3 through CMIP6. The suggestion is that in CMIP6 the skill dispersion is not as large, including the tendency for a more realistic spatial amplitude.

Acknowledgements: This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344. We also acknowledge the support of the U.S. Department of Energy, Office of Science, Climate and Environmental Sciences Division, Regional and Global Modeling and Analysis Program.