Predictable and Unpredictable Modes of Northern Hemisphere Atmospheric Circulation in CMIP6: Evaluation and Projections

Climate models are essential for understanding past, present, and future changes in atmospheric circulation, with circulation modes providing key sources of seasonal predictability and prediction uncertainties for both global and regional climates. This study assesses the performance of models participating in phase 6 of the Coupled Model Intercomparison Project in simulating interannual variability modes of Northern Hemisphere 500-hPa geopotential height during winter and summer, distinguishing predictable (potentially predictable on seasonal or longer timescales) and unpredictable (intraseasonal and essentially unpredictable at long range) components, using reanalysis data and a variance decomposition method. Although most models effectively capture unpredictable modes in reanalysis, their ability to reproduce dominant predictable modes—specifically the Pacific–North American pattern, Arctic Oscillation, and Western Pacific Oscillation in winter, and the East Atlantic and North Atlantic Oscillations in summer—varies notably. An optimal ensemble is identified to distinguish (a) predictable-external modes, dominated by external forcing, and (b) predictable-internal modes, associated with slow internal variability, during the historical period (1950–2014) and the SSP5-8.5 scenario (2036–2100). Under increased radiative forcing, the leading winter/summer predictable-external mode exhibits a more uniform spatial distribution, remarkably larger trend and annual variance, and enhanced height–sea surface temperature (SST) covariance under SSP5-8.5 compared to historical conditions. The dominant winter/summer predictable-internal modes also exhibit increased variance and height–SST covariance under SSP5-8.5, along with localized changes in spatial configuration. Minimal changes are observed in spatial distribution or variance for dominant winter/summer unpredictable modes under SSP5-8.5. This study, from a predictive perspective, deepens our understanding of model uncertainties and projected changes in circulations.