Sensitivity of Environmental Predictors to Integrated Kinetic Energy Tendency in Atlantic Hurricanes

Integrated kinetic energy (IKE) is a useful metric that quantifies the energy and destructive potential of tropical cyclones (TCs) across a storm's entire wind field. By definition, IKE is tied to the square of wind speeds anywhere in the storm, meaning that the metric responds to fluctuations in size, strength, and/or intensity. As such, the signals between large scale fields and IKE are nonlinear. For instance, tropical environments with high SSTs and low shear promote intensification and development, but extratropical transition, trough interactions, and the associated wind field expansion with these baroclinic processes often occurs in an opposite spectrum of environments closer to the poles with lower SSTs and higher shear. To analyze these complex relationships, we use an artificial neural network system trained on historical IKE tendency in Atlantic TCs over the past 20 years and a series of corresponding environmental and storm-specific predictors from the developmental Statistical Hurricane Intensity Prediction Scheme (SHIPS) archive. Through a series of perturbation tests in the neural network, it becomes possible to statistically evaluate how IKE fluctuates in various environmental conditions. The results from this nonlinear sensitivity study are then presented alongside of known physical processes in an effort to determine the causes of IKE variability for the advancement of future IKE-specific prediction models.