Sharp Ridge Structures in Flow Visualization
Software Engineering and Programming Languages
Final Report Abstract
The project deals with problems in a subfield of Scientific Visualization: the visual analysis of (simulated or measured) flow data. In Flow Visualization, the visual analysis of the flow map (i.e. the consideration of all integrated field lines) recently moved into the focus of interest. Working with the flow map is challenging because of 3 reasons: flow maps are complex (for 3D time time-dependent flows resulting in a 5D field); flow maps are expensive (each sample requires a numerical integration of the flow field); flow maps tend to have strong gradients, in particular in the areas of maximal flow separation (making an extremely high sampling density necessary). Goal of this project was to come up with stable numerical approaches to visually analyze fields that are derived from the gradient of the flow map. In particular the ridges of such fields were in the focus of the project. They have something in common: with increasing integration time they tend to be “sharp” or “narrow”, making a numerical extraction challenging or even impossible. The main idea of the project was a simple insight: even if the ridges are sharp at a certain (longer) integration time, they used to be less sharp for shorter integration times. This motivated the general approach: instead extracting sharp ridges, we extracted smooth ridges with a subsequent ridge tracking until the final integration time is reached. This main idea was adapted in several ways during the project, resulting in several approaches to visualize fields that are derived from the gradients of the flow map. In particular, we made the following contributions: We have introduced approaches to formally define and visualize recirculation in 3D timedependent flows. - We have developed stable numerical methods for the extraction of ridge curves in Finite-Time Lyapunov Exponent (FTLE) fields for extremely long integration times. - We have developed a new approach for high-quality Monte-Carlo Rendering of 3D FTLE fields by adapting methods from global illumination and rendering. - We have developed approaches to extract and visualize FTLE (and similar) fields for inertial flows, i.e. particles that are equipped with mass and inertia. - We have described rotation invariant vortices, i.e., vortices that are invariant under the equal-speed rotation of the observer around a known axis, as relevant for flows induced by stirring devices, helicopters, hydrocyclones, centrifugal pumps, or ventilators.
Publications
-
Finite-Time Mass Separation for Comparative Visualizations of Inertial Particles. Computer Graphics Forum (Proc. EuroVis), 2015
T. Günther and H. Theisel
-
Backward Finite-Time Lyapunov Exponents in Inertial Flows. IEEE Transactions on Visualization and Computer Graphics (Proc. IEEE Scientific Visualization 2016)
T. Günther and H. Theisel
-
MCFTLE: Monte Carlo Rendering of Finite-Time Lyapunov Exponent Fields. Computer Graphics Forum (Proc. EuroVis), 2016
T. Günther, A. Kuhn and H. Theisel
-
Progressive Monte Carlo Rendering of Atmospheric Flow Features Across Scales. Gallery of Fluid Motion Posters (APS Division of Fluid Dynamics), 2016
T. Günther, A. Kuhn, H.-C. Hege, M. Gross and H. Theisel
-
Rotation Invariant Vortices for Flow Visualization. IEEE Transactions on Visualization and Computer Graphics (Proc. IEEE Scientific Visualization 2015), 2016
T. Günther, M. Schulze and H. Theisel
-
Progressive Monte Carlo Rendering of Atmospheric Flow Features Across Scales. American Physical Society, Phys. Rev. Fluids, 2017
T. Günther, A. Kuhn, H.-C. Hege, M. Gross and H. Theisel
-
FTLE Ridge Lines for Long Integration Times. Proceedings IEEE Scientific Visualization Short Papers, 2018
T. Wilde, C. Rössl and H. Theisel
-
Recirculation Surfaces for Flow Visualization. IEEE Transactions on Visualization and Computer Graphics (Proc. IEEE Scientific Visualization), 2018
T. Wilde, C. Rössl and H. Theisel
