A COMPARISON OF OPTIMAL LOW DIMENSIONAL PROJECTIONS OF A HURRICANE SIMULATION

We introduce the Signal Fraction Analysis (SFA) methodology for understanding large data sets associated with the simulation of a multi-scale physical system, i.e., hurricanes. We compare the results of this approach with the well-known Karhunen-Loève (KL) procedure which has been widely applied for characterizing coherent structures in a variety of physical systems. The SFA method separates the simulation data into subspaces with independent scales while the KL approach separates into components based on maximizing the capture of statistical variance. It appears that this study represents the first application of these methods to the hurricane problem.