Tropical cyclones position and intensity in the Southwest Indian Ocean as represented by CFS and ERA5 atmospheric reanalysis datasets

Abstract

Tropical cyclones are the most devastating meteorological systems that affect the Southwestern Indian Ocean (SWIO) bordering nations. The understanding of their track and intensity evolution is essential for the characterization of the systems. The National Centers for Environmental Prediction–Climate Forecast System (CFS) dataset and the ECWMF fifth‐generation reanalysis (ERA5) are evaluated against best‐track data from the International Best‐Track Archive for Climate Stewardship (IBTrACS) from 1980 to 2019 cyclone seasons. The cyclones were tracked within the reanalyses using a tracking algorithm based on relative vorticity and identified using a spatiotemporal matching method. Climatology from IBTrACS showed significant negative (positive) trend in the total number (number of intense) of systems; however, the evaluation of trends has to be cautious because it may be limited due to uncertainties in the first decades of data. Results for reanalyses showed that both performed similarly although some peculiarities were noted. CFS presented a slightly higher probability of detection, related to its special treatment of tropical cyclone‐like vortex, being able to reproduce all systems between 2000 and 2019. For positions, the majority of the systems have average errors around 50–60 km and the errors generally decrease with the intensity of the systems in both reanalyses, with ERA5 being slightly better during the early stages, while CFS outperforms ERA5 for intensity beyond the IBTrACS's moderate tropical storm category. Intensity follows a reverse pattern of position, with reanalyses largely degrading systems' intensity as they intensify. CFS (ERA5) appeared to be slightly better in terms of maximum wind (minimum sea‐level pressure), representing stronger systems although both datasets were unable to reproduce wind intensities beyond the tropical cyclone category. All evaluated parameters showed improvements when the most recent data period (2000–2019) is considered, especially in ERA5, tailored to an improved quality of the observations that are used to generate best‐track data and also to nudge the reanalyses through data assimilation. Based on the reanalyses characteristics revealed here, the choice of the most suitable dataset to address a specific study within the SWIO region may be dependent on the desired parameters and period.