Visualizing Statistical Complexity in 3D Turbulent Flows using a Robust Entropy Calculation Method
Arne Präger, Baldwin Nsonga, Gerik Scheuermann · Computer Science Research Notes · 2022
Highly resolved flow simulation data is becoming more common. These simulations frequently feature high tur- bulence with complex flow patterns. Finding these regions often requires expert knowledge, and in more complex cases, flow patterns of interest may remain hidden. The concept of statistical complexity was shown to be suitable to indicate regions of interest within flow data with limited prior knowledge. One way to determine the statistical complexity of flow fields is via the local vector field entropy and the correlation. In this work, we improve the method for calculating the Shannon entropy in vector fields. To this end, we introduce a robust entropy compu- tation that takes the scale of the corresponding regions into account. The improved method uses a novel way to determine the distributions required for the entropy calculation and is applicable to unstructured domains. We val- idate our method with analytic flow fields and apply it to fluid simulation data, visualizing the results via volume rendering. This work shows the applicability of our technique to highlight regions of interest in turbulent flows.