Impact of Mean-State Biases on Northern Hemisphere Winter Blocking in GFDL AM4 and CM4

Abstract Mean-state biases significantly affect the simulation of winter atmospheric blocking statistics in the Northern Hemisphere within global climate models (GCMs). While most models consistently underestimate blocking occurrences in the Atlantic–Europe region, the atmospheric model, version 4 (AM4), and the fully coupled model, version 4 (CM4), recently developed at the Geophysical Fluid Dynamics Laboratory (GFDL) display a pronounced overestimation in the Pacific–North America sector. This study partitions climatological biases in 500-hPa geopotential heights into zonal-mean and zonal eddy components to diagnose their respective impacts on winter blocking, which manifests as a high pressure system characterized by both a zonal wind bifurcation (i.e., reversal of meridional pressure gradients) and other regions within the same system without such reversal. By replacing the models’ mean state with reanalysis data, two historical simulations from AM4 and CM4 are analyzed to assess biases arising from atmospheric processes and coupled interactions. Results indicate that zonal-mean biases are substantially larger in CM4 than in AM4, exerting a stronger influence on reversal blocking. Zonal eddy biases, also more pronounced in CM4, reveal enhanced stationary wave patterns at zonal wavenumber 2, nearly out of phase with observations, causing a southward jet stream displacement. Replacing modeled zonal means with reanalysis corrects blocking frequency biases more effectively in the Atlantic–Europe region, while substituting zonal eddies mitigates biases in the Pacific–North America sector. These adjustments reduce positive shear vorticity in the Atlantic–Europe region and negative shear in the Pacific–North America sector. These findings underscore the importance of improving the simulation of climatological zonal eddy structures at wavenumber 2 and zonal-mean fields to enhance blocking and mean-state representations in GCMs.