Assessment of the capability of CMIP6 global climate models to simulate Arctic cyclones

Arctic cyclones are one of the important synoptic-scale systems that affect weather variability over the Arctic and can cause intense weather phenomena and disasters. Thus, research on the assessment of the climate characteristics, activity laws, and variability trends of Arctic cyclones using climate models has practical significance. On the basis of the fifth-generation European Center for Medium-Range Weather Forecasts reanalysis (ERAS) data, the spatiotemporal variations of Arctic cyclones during 1981-2014 are analyzed by detecting and tracking cyclones using the Lagrangian method. Then, the simulation results of 14 global climate models from the Coupled Model Intercomparison Project Phase 6 (CMIP5) are compared with the results of ERAS. Both the individual models and their ensemble mean can simulate the spatial distribution of the density of cyclone tracks with reasonable capability, the correlation coefficients of track density are approximately 0.6. Furthermore, in boreal winter, the Atlantic zonal negative bias of track density is stronger than that in summer. By contrast, in boreal summer, the negative bias over the Arctic Ocean region is stronger than that in winter. Moreover, the simulations of density -field-related variables (i.e., cyclolysis, cyclogenesis, track, and lowest center pressure densities) of Arctic cyclones are generally better in winter than in summer, and the models can simulate well that the number of external Arctic cyclones entering the Arctic region from the midlatitudes is more than the internal Arctic cyclones generated inside the Arctic region (60 degrees-90 degrees N). Furthermore, we show that the capability of models to capture Arctic cyclones with a short lifespan (<3 d) is somewhat poor. Except for the simulation of the minimum pressure of Arctic cyclones, the performance of high-resolution models is better than that of low-resolution models. The simulation of Arctic cyclone radius is poor among all of the variables related to Arctic cyclones, and the observed trends of intensities are not well simulated. In general, the simulation of location-field-related variables (i.e., intensity, radius, deepening rate, and center pressure) of Arctic cyclones is better in winter than in summer.