Assessing high-resolution global climate models in simulating subtropical cyclones over the southeast coast of Brazil

Subtropical cyclones when occurring close to the coast can be very dangerous for human activities bringing high amounts of precipitation, intense winds and gusts in the coastal cities. This can lead to natural hazards and risks, such as floods, inundations, and even deaths. Over the southeast coast of Brazil, seven subtropical cyclones occur on average each year, with higher frequency in austral summer and autumn.  However, there are still few studies focusing on its global models climatology and future projections. It is crucial to evaluate how accurate are the global climate models of the new HighResMIP-CMIP6 dataset, with fine horizontal high-resolution, in representing subtropical cyclones in the historical period. Thus, this study assesses the classification of the subtropical cyclones based on two reanalyses (ERA5 and ERA-Interim) to evaluate the fine-resolution HighResMIP-CMIP6 datasets.  First, we tracked all cyclones over the South Atlantic Ocean applying an automatic scheme using relative vorticity at 925 hPa. Then, the vertical structure of the cyclones are accessed by calculating three parameters (symmetry, thermal wind at low and upper levels) from the cyclone phase space approach. Finally, we classified subtropical features using an automatic scheme based on a pre-establish threshold. In general, the approach is able for classifying subtropical cyclones providing realistic climatology. Overall, for the total of cyclones, ERA5, ERAInterim and HighResMIP-CMIP6 reproduce similar areas of great cyclogenetic activity over the eastern coast of South America.  In terms of frequency, it is greater in ERA5 than ERAInterim, for both total and subtropical cyclones, while a similar behavior is noted in relation to the seasonal frequency. HighResMIP-CMIP6 tends to overestimate the total of cyclones in subtropical latitudes, impacting directly the frequency of the subtropical ones.