Deep Learning Based Cyclone Intensity Estimation using INSAT-3D IR Imagery

Tropical cyclones stand out as one of the most financially burdensome natural disasters globally, given the numerous potential risks they pose. Thus, the development of a meticulous assessment model for gauging the severity of tropical cyclones becomes crucial in averting property damage and preserving lives. Various methods and approaches are currently employed, each varying in effectiveness, to assess the atmospheric velocity of tropical cyclones utilizing satellite information at given moment. This research contributes a diagnostic assessment of tropical cyclone strength rooted in deep learning from infrared satellite data, demonstrating RMSE measuring 13.24 knots. Moreover, the study introduces part of the inaugural visualization portals of its kind in a production system. This portal showcases contextual information and the outcomes, providing end-users with valuable insights.