Quantifying morphologic features of remotely-sensed data for its use in a tropical cyclogenesis predictor

Tropical cyclone genesis is a complex process that involves interactions on many scales. The classification and prediction of these events has been difficult, in part because of a lack of basic understanding of the physical mechanisms involved. Although some progress has been made in this area in recent years by several research groups, classification of an actual genesis event remains a subjective process until the tropical cyclone is well defined. The development of an objective model that can detect and predict the early stages of cyclogenesis would enable forecasters to warn on nascent systems with some confidence. Another potential benefit would be to enable better representation of the emerging tropical cyclone in numerical weather prediction models, perhaps improving forecasts from an earlier time.

In this paper we discuss a technique for obtaining features associated with the shape and the dynamics of structures embedded in cloud clusters that have potential for development into tropical cyclones. As the tropical cyclone develops and intensifies these structures become more axisymmetric. Using satellite-based remotely-sensed variables, the proposed technique uses gradient vectors and coordinate transformation techniques to measure the level of symmetry of each structure. We will show examples of our technique and discuss future development.