Mathematical strategies for filtering complex systems: Regularly spaced sparse observations
- Department of Mathematics and Center for Atmosphere and Ocean Science, Courant Institute of Mathematical Sciences, New York University, 251 Mercer Street, New York, NY 10012 (United States)
Real time filtering of noisy turbulent signals through sparse observations on a regularly spaced mesh is a notoriously difficult and important prototype filtering problem. Simpler off-line test criteria are proposed here as guidelines for filter performance for these stiff multi-scale filtering problems in the context of linear stochastic partial differential equations with turbulent solutions. Filtering turbulent solutions of the stochastically forced dissipative advection equation through sparse observations is developed as a stringent test bed for filter performance with sparse regular observations. The standard ensemble transform Kalman filter (ETKF) has poor skill on the test bed and even suffers from filter divergence, surprisingly, at observable times with resonant mean forcing and a decaying energy spectrum in the partially observed signal. Systematic alternative filtering strategies are developed here including the Fourier Domain Kalman Filter (FDKF) and various reduced filters called Strongly Damped Approximate Filter (SDAF), Variance Strongly Damped Approximate Filter (VSDAF), and Reduced Fourier Domain Kalman Filter (RFDKF) which operate only on the primary Fourier modes associated with the sparse observation mesh while nevertheless, incorporating into the approximate filter various features of the interaction with the remaining modes. It is shown below that these much cheaper alternative filters have significant skill on the test bed of turbulent solutions which exceeds ETKF and in various regimes often exceeds FDKF, provided that the approximate filters are guided by the off-line test criteria. The skill of the various approximate filters depends on the energy spectrum of the turbulent signal and the observation time relative to the decorrelation time of the turbulence at a given spatial scale in a precise fashion elucidated here.
- OSTI ID:
- 21159383
- Journal Information:
- Journal of Computational Physics, Vol. 227, Issue 10; Other Information: DOI: 10.1016/j.jcp.2008.01.049; PII: S0021-9991(08)00080-6; Copyright (c) 2008 Elsevier Science B.V., Amsterdam, The Netherlands, All rights reserved; Country of input: International Atomic Energy Agency (IAEA); ISSN 0021-9991
- Country of Publication:
- United States
- Language:
- English
Similar Records
Mathematical test criteria for filtering complex systems: Plentiful observations
Improving filtering and prediction of spatially extended turbulent systems with model errors through stochastic parameter estimation