A dataset of Southwest China Vortex events based on deep learning and ERA5 reanalysis data

The Southwest China Vortex (SWCV) is a powerful mesoscale synoptic system that can lead to severe hazards to the east of the Tibetan Plateau. However, SWCV data have increasingly become a problem, impeding studies on SWCV evolution. In this study, we established a new SWCV event (SWCVE) dataset using a well-trained CNN model, a nearest-neighbor search method, and hourly ERA5 reanalysis data. The new dataset, containing genesis locations, moving paths, and lysis locations of 9,379 SWCVEs from 1940 to 2023, was validated by independent-sample test, case study, and cross-check analysis of the SWCVE genesis frequency. The results show that the new SWCVE dataset is characterized by efficiency, reliability, and robustness. Additionally, it can be updated following ERA5 reanalysis data in real time. Therefore, the new SWCVE dataset lays the foundation for a better understanding of SWCV evolution on time scales no less than one hour. The new SWCVE dataset is openly available at https://doi.org/10.5281/zenodo.15073815 or https://www.scidb.cn/doi/10.5281/zenodo.15073815.