Storm Forecast

Machine Learning algorithms to predict 24-hour tropical and extra-tropical storm intensity

Project Summary

Today, the forecasts (track and intensity) are provided by a numerous number of guidance models. Dynamical models solve the physical equations governing motions in the atmosphere. Statistical models, in contrast, are based on historical relationships between storm behavior and various other parameters. Machine learning (and deep learning) methods have been only scarcely tested, and there is hope in that it can improve storm forecasts. The database is composed of more than 3000 extra-tropical and tropical storm tracks (total number of instants = 90,000), and it also provides the intensity and some local physical information at each timestep. Moreover, we also provide some 700-hPa and 1000-hPa feature maps of the neighborhood of the storm (from ERA-interm reanalysis database), that can be viewed as images centered on the current storm location.



Project Website:

Deployment Architecture:


View Code on Github