Abstract/Summary

The Fifth Coupled Model Intercomparison Project (CMIP5) provides an unprecedented opportunity to comprehensively explore biases and changes in storm tracks across all seasons using storm-centered feature tracking methods. 29 CMIP5 models in fully-coupled and atmosphereonly configurations are compared to a reanalysis to assess biases, and two Radiative Concentration Pathways simulations are contrasted against the present-day climate to assess future projections. The models are found to have a pervasive equatorward bias in storm track position, and a zonal bias in their tilt. The primary cyclogenesis and storm growth regions have severe negative biases, contributing to an overall lack in the simulated total number of storms. The models generally under-represent the extratropical cyclone intensity, in part related to low spatial resolution. A particularly novel result is that the coupling with ocean and sea-ice components often contributes to the equatorward bias in the storm tracks, but is found to reduce the negative cyclogenesis biases and produce more storms. Many of the storm track biases are associated with equatorward and slow jet biases, and temperature biases. In general, the future projections show a small poleward shift in the storm tracks and a slight contraction in their width, which are strongly associated with changes in the mean flow. This work has given several new insights, for example in demonstrating the robustness of the equatorward decreases in the storm tracks across the seasons. Exceptions to this include the North Atlantic in the winter and spring; and the North Pacific in summer, with the latter example having large uncertainty associated with uniquely large biases. There is a global, year-round decrease in the total number of storms. Cyclone intensity distributions are generally projected to shift slightly towards stronger storms in the Southern Hemisphere while there are few changes in the Northern Hemisphere