Machine learning for predicting hurricane and coastal flooding impacts

The goal of this project is to use data and machine learning to make better predictions of hurricane impacts on coastal areas. Hurricanes and extreme storms are expected to increase with climate change due to a number of factors including increasing ocean heat content, sea level rise, and changes in ocean-atmosphere physics. Currently, predictions of storm frequency, track, and coastal impacts rely on complex models developed over decades of research in meteorology, atmospheric and ocean science. However it may be possible to use modern machine learning methods with historical data sets to improve predictions. For example, remote sensing imagery, historical storm track databases, sea-surface temperature, coastal topography, streamflow data and even crowd-sourced data/photography on flooding could all be utilized to train prediction models. This project is very open-ended and can focus on any aspect within the range of hurricane prediction, coastal flooding, wind impacts, or combinations thereof. Student interest will determine the scope of the project.