B-preconditioned minimization algorithms for variational data assimilation with the dual formulation

S. Gürol, A. T. Weaver, A. M. Moore, A. Piacentini, Hernan Arango, S. Gratton

Research output: Contribution to journalArticle

30 Citations (Scopus)

Abstract

Variational data assimilation problems in meteorology and oceanography require the solution of a regularized nonlinear least-squares problem. Practical solution algorithms are based on the incremental (truncated Gauss-Newton) approach, which involves the iterative solution of a sequence of linear least-squares (quadratic minimization) sub-problems. Each sub-problem can be solved using a primal approach, where the minimization is performed in a space spanned by vectors of the size of the model control vector, or a dual approach, where the minimization is performed in a space spanned by vectors of the size of the observation vector. The dual formulation can be advantageous for two reasons. First, the dimension of the minimization problem with the dual formulation does not increase when additional control variables are considered, such as those accounting for model error in a weak-constraint formulation. Second, whenever the dimension of observation space is significantly smaller than that of the model control space, the dual formulation can reduce both memory usage and computational cost. In this article, a new dual-based algorithm called Restricted B-preconditioned Lanczos (RBLanczos) is introduced, where B denotes the background-error covariance matrix. RBLanczos is the Lanczos formulation of the Restricted B-preconditioned Conjugate Gradient (RBCG) method. RBLanczos generates mathematically equivalent iterates to those of RBCG and the corresponding B-preconditioned Conjugate Gradient and Lanczos algorithms used in the primal approach. All these algorithms can be implemented without the need for a square-root factorization of B. RBCG and RBLanczos, as well as the corresponding primal algorithms, are implemented in two operational ocean data assimilation systems and numerical results are presented. Practical diagnostic formulae for monitoring the convergence properties of the minimization are also presented.

Original languageEnglish (US)
Pages (from-to)539-556
Number of pages18
JournalQuarterly Journal of the Royal Meteorological Society
Volume140
Issue number679
DOIs
StatePublished - Jan 2014

Fingerprint

data assimilation
oceanography
meteorology
matrix
ocean
monitoring
cost

All Science Journal Classification (ASJC) codes

  • Atmospheric Science

Keywords

  • 3D-Var
  • 4D-Var
  • Conjugate gradient method
  • Dual approach
  • Lanczos method
  • Ocean data assimilation
  • PSAS

Cite this

Gürol, S. ; Weaver, A. T. ; Moore, A. M. ; Piacentini, A. ; Arango, Hernan ; Gratton, S. / B-preconditioned minimization algorithms for variational data assimilation with the dual formulation. In: Quarterly Journal of the Royal Meteorological Society. 2014 ; Vol. 140, No. 679. pp. 539-556.
@article{76927938840247cf8a9cf850c524b840,
title = "B-preconditioned minimization algorithms for variational data assimilation with the dual formulation",
abstract = "Variational data assimilation problems in meteorology and oceanography require the solution of a regularized nonlinear least-squares problem. Practical solution algorithms are based on the incremental (truncated Gauss-Newton) approach, which involves the iterative solution of a sequence of linear least-squares (quadratic minimization) sub-problems. Each sub-problem can be solved using a primal approach, where the minimization is performed in a space spanned by vectors of the size of the model control vector, or a dual approach, where the minimization is performed in a space spanned by vectors of the size of the observation vector. The dual formulation can be advantageous for two reasons. First, the dimension of the minimization problem with the dual formulation does not increase when additional control variables are considered, such as those accounting for model error in a weak-constraint formulation. Second, whenever the dimension of observation space is significantly smaller than that of the model control space, the dual formulation can reduce both memory usage and computational cost. In this article, a new dual-based algorithm called Restricted B-preconditioned Lanczos (RBLanczos) is introduced, where B denotes the background-error covariance matrix. RBLanczos is the Lanczos formulation of the Restricted B-preconditioned Conjugate Gradient (RBCG) method. RBLanczos generates mathematically equivalent iterates to those of RBCG and the corresponding B-preconditioned Conjugate Gradient and Lanczos algorithms used in the primal approach. All these algorithms can be implemented without the need for a square-root factorization of B. RBCG and RBLanczos, as well as the corresponding primal algorithms, are implemented in two operational ocean data assimilation systems and numerical results are presented. Practical diagnostic formulae for monitoring the convergence properties of the minimization are also presented.",
keywords = "3D-Var, 4D-Var, Conjugate gradient method, Dual approach, Lanczos method, Ocean data assimilation, PSAS",
author = "S. G{\"u}rol and Weaver, {A. T.} and Moore, {A. M.} and A. Piacentini and Hernan Arango and S. Gratton",
year = "2014",
month = "1",
doi = "10.1002/qj.2150",
language = "English (US)",
volume = "140",
pages = "539--556",
journal = "Quarterly Journal of the Royal Meteorological Society",
issn = "0035-9009",
publisher = "John Wiley and Sons Ltd",
number = "679",

}

B-preconditioned minimization algorithms for variational data assimilation with the dual formulation. / Gürol, S.; Weaver, A. T.; Moore, A. M.; Piacentini, A.; Arango, Hernan; Gratton, S.

In: Quarterly Journal of the Royal Meteorological Society, Vol. 140, No. 679, 01.2014, p. 539-556.

Research output: Contribution to journalArticle

TY - JOUR

T1 - B-preconditioned minimization algorithms for variational data assimilation with the dual formulation

AU - Gürol, S.

AU - Weaver, A. T.

AU - Moore, A. M.

AU - Piacentini, A.

AU - Arango, Hernan

AU - Gratton, S.

PY - 2014/1

Y1 - 2014/1

N2 - Variational data assimilation problems in meteorology and oceanography require the solution of a regularized nonlinear least-squares problem. Practical solution algorithms are based on the incremental (truncated Gauss-Newton) approach, which involves the iterative solution of a sequence of linear least-squares (quadratic minimization) sub-problems. Each sub-problem can be solved using a primal approach, where the minimization is performed in a space spanned by vectors of the size of the model control vector, or a dual approach, where the minimization is performed in a space spanned by vectors of the size of the observation vector. The dual formulation can be advantageous for two reasons. First, the dimension of the minimization problem with the dual formulation does not increase when additional control variables are considered, such as those accounting for model error in a weak-constraint formulation. Second, whenever the dimension of observation space is significantly smaller than that of the model control space, the dual formulation can reduce both memory usage and computational cost. In this article, a new dual-based algorithm called Restricted B-preconditioned Lanczos (RBLanczos) is introduced, where B denotes the background-error covariance matrix. RBLanczos is the Lanczos formulation of the Restricted B-preconditioned Conjugate Gradient (RBCG) method. RBLanczos generates mathematically equivalent iterates to those of RBCG and the corresponding B-preconditioned Conjugate Gradient and Lanczos algorithms used in the primal approach. All these algorithms can be implemented without the need for a square-root factorization of B. RBCG and RBLanczos, as well as the corresponding primal algorithms, are implemented in two operational ocean data assimilation systems and numerical results are presented. Practical diagnostic formulae for monitoring the convergence properties of the minimization are also presented.

AB - Variational data assimilation problems in meteorology and oceanography require the solution of a regularized nonlinear least-squares problem. Practical solution algorithms are based on the incremental (truncated Gauss-Newton) approach, which involves the iterative solution of a sequence of linear least-squares (quadratic minimization) sub-problems. Each sub-problem can be solved using a primal approach, where the minimization is performed in a space spanned by vectors of the size of the model control vector, or a dual approach, where the minimization is performed in a space spanned by vectors of the size of the observation vector. The dual formulation can be advantageous for two reasons. First, the dimension of the minimization problem with the dual formulation does not increase when additional control variables are considered, such as those accounting for model error in a weak-constraint formulation. Second, whenever the dimension of observation space is significantly smaller than that of the model control space, the dual formulation can reduce both memory usage and computational cost. In this article, a new dual-based algorithm called Restricted B-preconditioned Lanczos (RBLanczos) is introduced, where B denotes the background-error covariance matrix. RBLanczos is the Lanczos formulation of the Restricted B-preconditioned Conjugate Gradient (RBCG) method. RBLanczos generates mathematically equivalent iterates to those of RBCG and the corresponding B-preconditioned Conjugate Gradient and Lanczos algorithms used in the primal approach. All these algorithms can be implemented without the need for a square-root factorization of B. RBCG and RBLanczos, as well as the corresponding primal algorithms, are implemented in two operational ocean data assimilation systems and numerical results are presented. Practical diagnostic formulae for monitoring the convergence properties of the minimization are also presented.

KW - 3D-Var

KW - 4D-Var

KW - Conjugate gradient method

KW - Dual approach

KW - Lanczos method

KW - Ocean data assimilation

KW - PSAS

UR - http://www.scopus.com/inward/record.url?scp=84896049360&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84896049360&partnerID=8YFLogxK

U2 - 10.1002/qj.2150

DO - 10.1002/qj.2150

M3 - Article

AN - SCOPUS:84896049360

VL - 140

SP - 539

EP - 556

JO - Quarterly Journal of the Royal Meteorological Society

JF - Quarterly Journal of the Royal Meteorological Society

SN - 0035-9009

IS - 679

ER -