Hurricane Forcasting Using Deep Learning
This project aims to use machine learning and deep learning for hurricane forecasting. This mainly includes generating forecasting products that are useful to decision-makers, such as track (position + intensity) forecasts.
To evaluate the forecast, I compare the forecast errors from my models to the forecast errors of the official National Hurricane Center (NHC) forecasts. The goal is to achieve a relatively similar or smaller error than the official NHC forecast error.
Traditionally, the National Hurricane Center (NHC) models are simulation-based and compute-intensive. If I can achieve comparable performance but with much fewer computing resources, then there may be some interesting insights that can be discovered.
Below is some current results of all the models I have run. This is all evaluated on the test data split.