Interactive comparison of multifield scalar data based on largest contours

We present a feature-based interactive technique for comparing multiple scalar fields. It extracts largest contour features from the topologically simplified contour trees of each field, then measures similarities between features using information theory. These similarities are stored in a weighted graph on which we apply clustering. The clustering identifies coherent features across fields. Finally, the clusters are ranked and presented in a thumbnail gallery providing a quick overview of the whole dataset. We demonstrate the utility of our approach by applying it to a number of complex fluid flow datasets.

► We present a novel method for comparing multiple scalar fields.
► We use pairwise mutual information and graph clustering on field features.
► Our method performs well on fluid flow datasets to find physical phenomena.