Vadim Rezvov. Downscaling of Near-Surface Wind Fields in the Coastal Zones of the Barents and Kara Seas Using Neural Network Methods

The talk focuses on the development and application of statistical downscaling methods to enhance the spatial resolution of near-surface wind fields in the coastal areas of the Barents and Kara Seas. The relevance of this work stems from the need for more accurate wind forecasts in the Arctic, where climate change and the presence of mesoscale atmospheric phenomena, such as polar mesocyclones and the Novaya Zemlya bora, significantly affect navigation, sea ice forecasting, and wind resource assessment. The study explores modern downscaling approaches — from dynamical methods to neural network architectures specifically adapted for geophysical data. Special attention is given to the development of objective quality metrics for evaluating the reproduction of mesoscale structures. The results provide a foundation for more reliable high-resolution modeling of atmospheric processes in Arctic seas and their integration into applied tasks, including ensuring the safety of the Northern Sea Route and advancing renewable energy development.