[home] [Personal Program] [Help]

---

Session: Experimental and numerical data assimilation I
Session starts: Thursday 06 November, 11:00
Presentation starts: 11:00
Room: Lecture room B

---

Marco Raiola (Universidad Carlos III de Madrid; Karlsruhe Institute of Technology)
Jochen Kriegseis (Karlsruhe Institute of Technology)

Abstract:
A novel framework for a data-driven decomposition technique to extract advecting flow structures from flow data is introduced. The technique is an extension to complex values of the Proper Orthogonal Decomposition (POD) dubbed Hilbert POD (HPOD). The Hilbert transform is computed either in time or in the advection direction of the traveling structure to obtain the analytic signal of the dataset before applying standard POD. This delivers two version of the Hilbert POD, defined, respectively, conventional and space-only. The method is applied both on a temporally-resolved set of Schlieren images of a flickering candle and on temporally undersampled velocity fields from a turbulent subsonic jet. The decomposition technique delivers physically-sound complex-valued modes, which represent wavepackets travelling in the main flow direction and undergoing spatial amplification and decay.