Performance Changes in Extratropical Modes of Variability Across CMIP Generations
Extratropical modes of variability represent substantial portions of climatic interannual variance in different geographical domains. Our objective is to gauge the evolution of performance across multiple generations of CMIP, focusing on extratropical modes of variability. We highlight performance changes in both patterns and amplitude.
We evaluate extratropical modes of variability in the ~800 simulations from the three most recent phases of the Coupled Model Intercomparison Project (CMIP3, 5, and 6) to gauge the improvement of climate models over time. Simulated spatial patterns of modes have been significantly improved in CMIP6, while little improvement and systematic overestimation in the mode amplitude found in the newer models, indicating bias in amplitude is still a challenge. We have demonstrated that many modes of variability can be diagnosed in atmospheric-only mode, making future improvements of models more trackable.
We evaluate extratropical modes of variability in the three most recent phases of the Coupled Model Intercomparison Project (CMIP3, 5, and 6) to gauge the improvement of climate models over time. A suite of high-level metrics is employed to objectively evaluate how well climate models simulate the observed Northern Annular Mode (NAM), North Atlantic Oscillation (NAO), Pacific North America pattern (PNA), Southern Annular Mode (SAM), Pacific Decadal Oscillation (PDO), North Pacific Oscillation (NPO), and North Pacific Gyre Oscillation (NPGO). We find simulated spatial patterns of those modes have been significantly improved in the newer models, although the skill improvement is sensitive to the mode and season considered. We identify some potential contributions to the pattern improvement of certain modes (e.g., the Southern Hemisphere jet and High-top vertical coordinate), however, the performance changes are likely attributed to gradual improvement of the base climate and multiple relevant processes. Less performance improvement is evident in the mode amplitude of these modes and systematic overestimation of the mode amplitude in spring remains in the newer climate models. We find that the post-dominant season amplitude errors in atmospheric modes are not limited to coupled runs but are often already evident in AMIP simulations. This suggests that rectifying the egregious post-dominant season amplitude errors found in many models can be addressed in an atmospheric-only framework making it more tractable to address in the model development process.