tropical cyclones & machine learning

TCANE (Tropical Cyclone Artificial Neural-network Error) is a project aimed at improving tropical cyclone track and intensity error forecasting. This can be of use for forecasters and communities affected by these storms. In the paper below, I trained a machine learning model for predicting track forecast errors. The model is trained on historical data, including tropical cyclone forecast track errors, and the output is a storm specific error (bivariate normal) distribution.