Compute and visualize discontinuity among neighboring integral curves of 2D vector fields

Abstract

This paper studies the discontinuity in the behavior of neighboring integral curves. The discontinuity is measured by a number of selected attributes of integral curves. A variety of attribute fields are defined. The attribute value at any given spatio-temporal point in these fields is assigned by the attribute of the integral curve that passes through this point. This encodes the global behavior of integral curves into a number of scalar fields in an Eulerian fashion, which differs from the previous pathline attribute approach that focuses on the discrete representation of individual pathlines. With this representation, the discontinuity of the integral curve behavior now corresponds to locations in the derived fields where the attribute values have sharp gradients. We show that based on the selected attributes, the extracted discontinuity from the corresponding attribute fields may relate to a number of flow features, such as LCS, vortices, and cusp-like seeding curves. In addition, we study the correlations among different attributes via their pairwise scatter plots. We also study the behavior of the combined attribute fields to understand the spatial correlation that cannot be revealed by the scatter plots. Finally, we integrate our attribute field computation and their discontinuity detection into an interactive system to guide the exploration of various 2D flows.