Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Adaptive Parameter Selection in Nudging Based Data Assimilation

Aytekin Çıbık, Rui Fang, William Layton, Farjana Siddiqua

TL;DR

This work tackles the practical challenge of selecting the nudging parameter $χ$ in nudging-based data assimilation by developing two self-adaptive strategies that respond to local flow behavior. It analyzes continuum nudging for the Navier–Stokes equations, derives $H$- and $χ$-conditions that guarantee uniform-in-time convergence, and interprets these bounds in 2D and 3D turbulence, illustrating their severity. The paper then introduces two adaptive algorithms—one heuristic (Algorithm 1) based on the projection error and one time-varying (Algorithm 2) grounded in a priori estimates—and demonstrates their effectiveness on manufactured solutions and complex flows, including flow between offset cylinders and flow over a flat obstacle. While adaptive nudging can yield smaller effective $χ$ values and improved short-to-moderate time accuracy, long-time convergence still depends on the $H$-condition, underscoring open problems such as time delays and model-error corrections in nudging data assimilation.

Abstract

Data assimilation combines (imperfect) knowledge of a flow's physical laws with (noisy, time-lagged, and otherwise imperfect) observations to produce a more accurate prediction of flow statistics. Assimilation by nudging (from 1964), while non-optimal, is easy to implement and its analysis is clear and well-established. Nudging's uniform in time accuracy has even been established under conditions on the nudging parameter $χ$ and the density of observational locations, $H$, Larios, Rebholz, and Zerfas [1]. One remaining issue is that nudging requires the user to select a key parameter. The conditions required for this parameter, derived through á priori (worst case) analysis are severe (Section 2.1 herein) and far beyond those found to be effective in computational experience. One resolution, developed herein, is self-adaptive parameter selection. This report develops, analyzes, tests, and compares two methods of self-adaptation of nudging parameters. One combines analysis and response to local flow behavior. The other is based only on response to flow behavior. The comparison finds both are easily implemented and yield effective values of the nudging parameter much smaller than those of á priori analysis.

Adaptive Parameter Selection in Nudging Based Data Assimilation

TL;DR

This work tackles the practical challenge of selecting the nudging parameter in nudging-based data assimilation by developing two self-adaptive strategies that respond to local flow behavior. It analyzes continuum nudging for the Navier–Stokes equations, derives - and -conditions that guarantee uniform-in-time convergence, and interprets these bounds in 2D and 3D turbulence, illustrating their severity. The paper then introduces two adaptive algorithms—one heuristic (Algorithm 1) based on the projection error and one time-varying (Algorithm 2) grounded in a priori estimates—and demonstrates their effectiveness on manufactured solutions and complex flows, including flow between offset cylinders and flow over a flat obstacle. While adaptive nudging can yield smaller effective values and improved short-to-moderate time accuracy, long-time convergence still depends on the -condition, underscoring open problems such as time delays and model-error corrections in nudging data assimilation.

Abstract

Data assimilation combines (imperfect) knowledge of a flow's physical laws with (noisy, time-lagged, and otherwise imperfect) observations to produce a more accurate prediction of flow statistics. Assimilation by nudging (from 1964), while non-optimal, is easy to implement and its analysis is clear and well-established. Nudging's uniform in time accuracy has even been established under conditions on the nudging parameter and the density of observational locations, , Larios, Rebholz, and Zerfas [1]. One remaining issue is that nudging requires the user to select a key parameter. The conditions required for this parameter, derived through á priori (worst case) analysis are severe (Section 2.1 herein) and far beyond those found to be effective in computational experience. One resolution, developed herein, is self-adaptive parameter selection. This report develops, analyzes, tests, and compares two methods of self-adaptation of nudging parameters. One combines analysis and response to local flow behavior. The other is based only on response to flow behavior. The comparison finds both are easily implemented and yield effective values of the nudging parameter much smaller than those of á priori analysis.
Paper Structure (13 sections, 2 theorems, 48 equations, 15 figures, 1 table, 2 algorithms)

This paper contains 13 sections, 2 theorems, 48 equations, 15 figures, 1 table, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related work
  3. Preliminaries
  4. Continuum Nudging
  5. Interpreting the Conditions
  6. Self-Adaptive Parameter Selection
  7. Numerical Tests
  8. Test of Accuracy
  9. Flow between offset cylinders
  10. DNS
  11. URANS
  12. Flow over a flat obstacle
  13. Conclusion

Key Result

Theorem 1.1

(The Ladyzhenskaya Inequalities, see Ladyzhenskaya ladyzhenskaya1969mathematical) For any vector function $u:\mathbb{R}^d\rightarrow \mathbb{R}^d$ with compact support and with the indicated $L^p$ norms finite,

Figures (15)

  • Figure 1: Relative errors. In a longer time, Algorithm 2 gives smaller errors.
  • Figure 2: $\chi$ in long time. $\chi$ values in Algorithm 2 reach the maximum value.
  • Figure 3: The relative error with the approximate true solution from DNS. The adaptive algorithms are effective for a shorter time. When the flow becomes more complex, the errors grow big and saturate at $\mathcal{O}(1)$.
  • Figure 4: The plot of the nudging parameter $\chi$ values in time. Both Algorithms 1 and 2 reach the maximum $\chi$ value in a longer time.
  • Figure 5: The relative error with the approximate true solution from a URANS model. When the flow becomes more complex, the errors grow big and saturate at $\mathcal{O}(1)$
  • ...and 10 more figures

Theorems & Definitions (3)

  • Theorem 1.1
  • Proposition 2.1
  • proof