Theses and Dissertations

Issuing Body

Mississippi State University


Mercer, Andrew E.

Committee Member

Wood, Kimberly M.

Committee Member

Dyer, Jamie L.

Committee Member

Wu, Tung-Lung

Committee Member

Clary, Renee M.

Other Advisors or Committee Members

Travis, Rick

Date of Degree


Original embargo terms

Visible to MSU only for 2 years

Document Type

Dissertation - Open Access


Earth and Atmospheric Sciences

Degree Name

Doctor of Philosophy


College of Arts and Sciences


Department of Geosciences


Forecasting rapid intensification (RI) of tropical cyclones (TCs) is considered one of the most challenging problems for the TC operational and research communities and remains a top priority for the National Hurricane Center. Upon landfall, these systems can have detrimental impacts to life and property. To support continued improvement of TC RI forecasts, this study investigated large-scale TC environments undergoing RI in the North Atlantic basin, specifically identifying important diagnostic variables in three-dimensional space. These results were subsequently used in the development of prognostic machine learning algorithms designed to predict RI 24 hours prior to occurrence. Using three RI definitions, this study evaluated base-state and derived meteorological parameters through S-mode and T-mode rotated principal component analysis, hierarchical compositing analysis, and hypothesis testing. Additionally, nine blended intelligence ensembles were developed using three RI definitions trained on data from the Statistical Hurricane Intensity Prediction Scheme- Rapid Intensification Index, Global Ensemble Forecast System Reforecast, and Final Operational Global Analysis. Performance metrics for the intelligence ensembles were compared against traditional linear methods. Additionally, a tenth ensemble was created using forecast data generated from Weather Research and Forecasting model simulations of TC RI events in the open North Atlantic and compared against linear methods. Results revealed modest classification ability of machine learning algorithms in predicting the onset of RI 24 hours in advance by including TC environmental spatial information of temperature and moisture variables, as well as variables indicative of ambient environmental interactions.



tropical cyclones||rapid intensification||machine learning||atmospheric science