Determining the Relative Importance of Factors Influencing Tropical Cyclone Rapid Intensity Changes: A Novel Approach and Its Outcome for TCs over the Bay of Bengal

Tropical cyclones (TCs) are the product of complex multi-scale processes and interactions. The role of the environment in favoring rapid intensity change has long been recognized without objective characterization. However, recent research has shown that convective-scale processes in the hurricane core might also play a crucial role in determining the rapid change in TC intensity and size. Several case studies have linked Rapid Intensification (RI) to the characteristics of the convective clouds (shallow versus deep), their organization (isolated versus wide-spread) and their location with respect to dynamical controls (the vertical shear, the radius of maximum wind).

Previously we developed a methodology to describe the 2D structure of precipitation and surface wind in a very concise way, using Wave Number Analysis (WNA). This approach can be applied to any 2D field, allowing us to condense the huge number of atmospheric state variables to a very manageable subset of integral quantities that faithfully capture the most important dynamical and thermodynamical characteristics of the environment and that of the vortex.

Here we use wavenumber representation of the state variables (using integral quantities) and present an empirical framework that quantifies the relative importance of the various factors that influence critical transitions in a TC's intensity. We use Linear Discriminant Analysis (LDA) to derive empirical predictors of rapid intensification and weakening based on the current state of variables in the environment and in the vortex. Our framework identifies a set of variables that are statistically significant in their difference during time periods that just precede a rapid intensification (RI) as opposed to a rapid weakening (RW). Our methodology also allows for the ranking of competing processes that take place near-simultaneously based on how their associated variability will be magnified over the course of model forecasts.

Our analysis of model simulations of recent TCs over the Bay of Bengal indicates that the environmental variables and the vortex variables have nearly equal contribution (45:55) to the predictive capabilities for discriminating RI cases from RW cases. A detailed analysis identifies the following variables within the vortex as the most important: i) the amplitude of wavenumber (WN) 1 of 700-850 mb horizontal moisture flux convergence; ii) the WN 1 amplitude of the precipitation in the rainband region; iii) the amplitude of WN 0 of precipitation within the radius of maximum winds. Likewise, the most important environmental variables identified are: i) the angle between the driest air and the shear vector; and ii) the magnitude of vertical wind shear.

These findings provide guidance-on-guidance for future observational efforts and data assimilation into TC forecasting models. Furthermore, the presented new framework that computes the relative importance of the competing processes driving tropical cyclone rapid intensity changes could be applied to model forecast over different ocean basins to provide an empirical forecast of RI and RW that would properly reflect the different environmental conditions (variability in SST, shear and water vapor distributions) characteristic for the different ocean basins.

The research described in this paper was performed at the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration (NASA).