Improving Prediction of Track and Intensity for TCs in Eastern NC Using a Machine Learning Based Model

Tropical cyclones (TCs) are among the most destructive environmental hazards on earth, causing hundreds of deaths and billions of dollars of damage every year in U.S.. Of the $2.6 trillion in damage associated with the U.S. billion-dollar weather events in 1980-2023, $1.4 trillion was due to TCs. As a poignant example, Hurricane Florence produced extensive flooding and damage across eastern North Carolina in September 2018. An accurate prediction for TC track and intensity with sufficient lead time is critical for protecting lives and property. This study focuses on the accurate prediction of track and intensity near landfalling. The considered storms are all TCs that made landfall at NC coast in the recent 30 years. The collected data include the historical storm data from HURDAT2 in NHC (National Hurricane Center), reanalysis data ERA5 from ECMWF (European Centre for Medium-Range Weather Forecasts), and operational forecast data from NHC. Statistical models are employed to approximate historical relationships between storm behavior and storm-specific features. A machine learning (ML) based model is then developed to combine the numerical weather prediction models (NWPs) with statistical models. The anticipated outcome of this research includes the pattern and categories of all the TCs (in time segments) with the aid of statistical models, and a system of combined models that produces prediction and forecasting on the track and intensity with comparable and/or improved accuracies when compared with the NHC official forecast errors. As an example, the developed model will be applied on Hurricane Florence 2018 to evaluate its performance and accuracy. Overall, these tools and methods can greatly improve the accuracy of the track and intensity prediction of future hurricanes like Florence and can help ensure better civilian preparedness for a hazardous storm.