Abstract This study investigates the use of topological data maps for extracting unique tropical cyclone (TC) wind field features. These maps are presented as graphs generated through a sequence of steps that filter, cluster, and identify data structure, and are used to characterize topological properties and shape in the data. The objective and scope of the method is explored through application to wind fields from the HURDAT2 data set, and its viability for detecting anomalous behavior in TCs is considered. We refer to the resulting graphs as wind field connectivity signatures (WFCS) or collective wind field connectivity map (CWFCM), depending on the data set. Our focus is Hurricane Sandy, where the method successfully identifies a complete 360-degree rotation of the high wind speed radii. This cyclical example of phase rotation of wind speed asymmetries corresponds to a distinct structural property of the graph. These methods have not been previously applied to wind field data and have only seen limited use in atmospheric sciences.
Justin Hoffmeier (Thu,) studied this question.