Data-driven model reduction of low-frequency modes in turbulent flows
Abstract
Climate variability and geophysical turbulent flows in general exhibit recurrent large-scale patterns that evolve irregularly in time, but still exhibit characteristic dominant frequencies across a large range of spatio-temporal scales. Such large-scale patterns result typically from nonlinear interactions from which the identification of low-frequency modes (LFMs) becomes essential for the modeling and the prediction of irregularly occurring events. Our approach to LFMs identification and modeling makes use of two complimentary novel methodologies. In that respect, the Harmonic Koopman Analysis (HKA) and the Multilayered Stochastic Modeling (MSM) will be introduced. HKA is a data-adaptive spatio-temporal filtering method, and MSM is a data-driven model reduction strategy which leads to nonlinear reduced stochastic models conveying memory effects. These methods will be illustrated in the context of one-dimensional turbulence where the Kuramoto-Sivashinsky model will serve as a laboratory.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2013
- Bibcode:
- 2013AGUFMNG41A1663C
- Keywords:
-
- 4490 NONLINEAR GEOPHYSICS Turbulence;
- 4460 NONLINEAR GEOPHYSICS Pattern formation;
- 4420 NONLINEAR GEOPHYSICS Chaos;
- 3260 MATHEMATICAL GEOPHYSICS Inverse theory