Guaranteeing the convergence of the saddle formulation for weakly constrained 4D‐Var data assimilation

This paper discusses the practical use of the saddle variational formulation for the weakly-constrained 4D-VAR method in data assimilation. It is shown that the method, in its original form, may produce erratic results or diverge because of the inherent lack of monotonicity of the produced objective function values. Convergent, variationaly coherent variants of the algorithm are then proposed whose practical performance is compared to that of other formulations. This comparison is conducted on two data assimilation instances (Burgers equation and the Quasi-Geostrophic model), using two different assumptions on parallel computing environment. Because these variants essentially retain the parallelization advantages of the original proposal, they often --- but not always --- perform best, even for moderate numbers of computing processes.

[1]  Stephen J. Wright,et al.  Numerical Optimization , 2018, Fundamental Statistical Inference.

[2]  Selime Gürol,et al.  Parallelization in the time dimension of four‐dimensional variational data assimilation , 2017 .

[3]  Luca Bergamaschi,et al.  Erratum to: Inexact constraint preconditioners for linear systems arising in interior point methods , 2011, Comput. Optim. Appl..

[4]  Craig H. Bishop,et al.  A Comparison of Hybrid Ensemble Transform Kalman Filter–Optimum Interpolation and Ensemble Square Root Filter Analysis Schemes , 2007 .

[5]  Philippe Courtier,et al.  Dual formulation of four‐dimensional variational assimilation , 1997 .

[6]  Nicholas I. M. Gould,et al.  Trust Region Methods , 2000, MOS-SIAM Series on Optimization.

[7]  Serge Gratton,et al.  Approximate Gauss-Newton Methods for Nonlinear Least Squares Problems , 2007, SIAM J. Optim..

[8]  D K Smith,et al.  Numerical Optimization , 2001, J. Oper. Res. Soc..

[9]  Serge Gratton,et al.  An observation‐space formulation of variational assimilation using a restricted preconditioned conjugate gradient algorithm , 2009 .

[10]  A. Piacentini,et al.  Variational data analysis with control of the forecast bias , 2004 .

[11]  M. Hestenes,et al.  Methods of conjugate gradients for solving linear systems , 1952 .

[12]  Andrew C. Lorenc,et al.  Improving ensemble covariances in hybrid variational data assimilation without increasing ensemble size , 2017 .

[13]  D. Zupanski.  A General Weak Constraint Applicable to Operational 4DVAR Data Assimilation Systems , 1997 .

[14]  A. Weaver,et al.  Correlation operators based on an implicitly formulated diffusion equation solved with the Chebyshev iteration , 2016 .

[15]  Serge Gratton,et al.  Range-Space Variants and Inexact Matrix-Vector Products in Krylov Solvers for Linear Systems Arising from Inverse Problems , 2011, SIAM J. Matrix Anal. Appl..

[16]  P. Courtier,et al.  A strategy for operational implementation of 4D‐Var, using an incremental approach , 1994 .

[17]  Y. Trémolet.  Model‐error estimation in 4D‐Var , 2007 .

[18]  Yousef Saad,et al.  Iterative methods for sparse linear systems , 2003 .

[19]  P. Courtier,et al.  Correlation modelling on the sphere using a generalized diffusion equation , 2001 .

[20]  F. L. Dimet,et al.  Variational algorithms for analysis and assimilation of meteorological observations: theoretical aspects , 1986 .

[21]  Melina A. Freitag,et al.  A low-rank approach to the solution of weak constraint variational data assimilation problems , 2017, J. Comput. Phys..

[22]  A. Weaver,et al.  Representation of correlation functions in variational assimilation using an implicit diffusion operator , 2010 .

[23]  Luca Bergamaschi,et al.  Erratum to: Inexact constraint preconditioners for linear systems arising in interior point methods , 2011, Comput. Optim. Appl..

[24]  A. Lorenc,et al.  Operational implementation of a hybrid ensemble/4D‐Var global data assimilation system at the Met Office , 2013 .

[25]  Y. Trémolet.  Accounting for an imperfect model in 4D‐Var , 2006 .

[26]  Y. Saad,et al.  GMRES: a generalized minimal residual algorithm for solving nonsymmetric linear systems , 1986 .

[27]  A. J. Wathen,et al.  Preconditioning , 2015, Acta Numerica.

[28]  Serge Gratton,et al.  Low rank updates in preconditioning the saddle point systems arising from data assimilation problems , 2018, Optim. Methods Softw..

[29]  Serge Gratton,et al.  Conjugate gradients versus multigrid solvers for diffusion‐based correlation models in data assimilation , 2013 .

[30]  M. Benzi,et al.  Some Preconditioning Techniques for Saddle Point Problems , 2008 .