Drivers of biases in the CMIP6 extratropical storm tracks. Part 1: Northern Hemisphere

Abstract The ability of climate models to represent extratropical storm tracks is vital to provide useful projections. In previous work the representation of the extratropical storm tracks in the Northern Hemisphere was found to have improved from the 5th to 6th coupled model intercomparison project. Here we investigate the remaining and persistent biases in models from the 6th phase of the Coupled Model Intercomparison Project (CMIP), by contrasting the atmosphere-only simulations (AMIP6) with the historical coupled simulations (CMIP6). The comparison of AMIP6 and CMIP6 simulations reveal that biases in sea surface temperatures (SSTs) in the coupled simulations across the North Pacific in winter modify the atmospheric temperature gradient, which is associated with an equatorward bias of the storm track. In the North Atlantic, cyclones do not propagate poleward enough in coupled simulations, which is partly driven by cold SSTs to the south of Greenland, decreasing the latent heat fluxes. In summer, excessive heating across central Asia and the Tibetan Plateau reduces the local baroclinicity causing fewer cyclones to form and propagate from eastern China into the North Pacific in both the coupled and atmosphere-only simulations. Several of the biases described in the coupled models are reduced considerably in the atmosphere-only models when the SSTs are prescribed. For example the equatorward bias of the North Pacific storm track is reduced significantly. However, other biases are apparent in both CMIP6 and AMIP6 (e.g. persistent reduction in track density and cyclogenesis over eastern Asia in Summer), which are associated with other processes (e.g. land surface temperatures).