Identification and Tracking of Rossby Wave Packets

Abstract

Rossby Wave Packets (RWPs) are regions of high amplitude meridional winds in the up- per troposphere that acts as precursors to extreme weather phenomena. Due to this, the identification and tracking of RWPs is an important field of research. We have implemented modifications to an identification algorithm that uses the underlying topology of the merid- ional wind field and techniques of computational geometry to identify RWPs from gridded data. These modifications have helped improve e ciency and reduce computational com- plexity. We have also implemented a feature-based tracking approach to track the evolution of RWPs across time. We recognize regions enclosing the identified RWPs as features. These features are then stored in the form of quadtrees for each time step which optimises stor- age and facilitates easy computation of unions and sizes of features. We then compute the overlap of features in consecutive time steps using their respective quadtree representations and formulate a weight that determines the extent of each overlap. Finally, we construct a tracking graph consisting of nodes representing features in each timestep and weighted edges representing the extent of overlap between these features. This representation allows for easy analysis of RWP tracks and contains information about splits and merges of RWP tracks.