Spectral and spatial localization of background‐error correlations for data assimilation

In this study, the localization of background-error correlations in both the spectral and the spatial domains is examined. While spatial localization has become a standard approach for reducing the sampling error of background-error correlations, localization of spectral correlations has not yet been fully explored. It is shown that spectral localization results in a spatial smoothing of the correlation functions in grid-point space. The use of correlations that are diagonal in spectral space, resulting in globally homogeneous correlations, has been frequently employed with data assimilation applications for numerical weather prediction (NWP). More recently, correlations that are diagonal in the space defined by an expansion of wavelet functions have been used to implicitly localize the correlations in a particular way in both spectral and grid-point spaces simultaneously. In this study, the explicit localization of correlations by varying amounts in both the spatial and the spectral domains is applied, to evaluate their complementary ability to reduce sampling error. Spectral and spatial localization are first applied to an idealized one-dimensional problem where the true correlations are known. In this context it is found that there is an optimal combination of spectral and spatial localization that minimizes the sampling error for a given ensemble of error realizations. Then the implementation of spectral localization is demonstrated in both a realistic three-dimensional variational assimilation system for NWP and an ensemble-based data assimilation system applied to an idealized model of the atmosphere's mesoscale dynamics. Two very different practical approaches are used to implement spectral localization for these two types of data assimilation systems. For the application to the ensemble-based data assimilation system, spectral localization is shown to systematically reduce analysis error without requiring additional model forecasts to be performed. Copyright © 2007 Crown in the right of Canada. Published by John Wiley & Sons, Ltd

[1]  A. Staniforth,et al.  The Operational CMC–MRB Global Environmental Multiscale (GEM) Model. Part II: Results , 1998 .

[2]  G. Evensen.  Sequential data assimilation with a nonlinear quasi‐geostrophic model using Monte Carlo methods to forecast error statistics , 1994 .

[3]  D. B. Preston.  Spectral Analysis and Time Series , 1983 .

[4]  M. Buehner.  Ensemble‐derived stationary and flow‐dependent background‐error covariances: Evaluation in a quasi‐operational NWP setting , 2005 .

[5]  M. Buehner,et al.  Atmospheric Data Assimilation with an Ensemble Kalman Filter: Results with Real Observations , 2005 .

[6]  B. Sonnerup,et al.  Vortex laws and field line invariants in polytropic field-aligned MHD flow , 1994 .

[7]  M. Lesieur,et al.  Geostrophic versus Wave Eddy Viscosities in Atmospheric Models , 1996 .

[8]  P. Courtier,et al.  The ECMWF implementation of three‐dimensional variational assimilation (3D‐Var). I: Formulation , 1998 .

[9]  Mark Buehner,et al.  Evaluation of new estimates of background- and observation-error covariances for variational assimilation , 2005 .

[10]  John Derber,et al.  The National Meteorological Center's spectral-statistical interpolation analysis system , 1992 .

[11]  P. L. Houtekamer,et al.  Assimilation with an Ensemble Kalman Filter of Synthetic Radial Wind Data in Anisotropic Turbulence: Perfect Model Experiments , 2006 .

[12]  R. Daley.  Atmospheric Data Analysis , 1991 .

[13]  J. Whitaker,et al.  Distance-dependent filtering of background error covariance estimates in an ensemble Kalman filter , 2001 .

[14]  P. Houtekamer,et al.  Data Assimilation Using an Ensemble Kalman Filter Technique , 1998 .

[15]  A. Staniforth,et al.  The Operational CMC–MRB Global Environmental Multiscale (GEM) Model. Part I: Design Considerations and Formulation , 1998 .

[16]  Loïk Berre,et al.  A Wavelet Approach to Representing Background Error Covariances in a Limited-Area Model , 2005 .

[17]  Andrew C. Lorenc,et al.  The potential of the ensemble Kalman filter for NWP—a comparison with 4D‐Var , 2003 .

[18]  H. Mitchell,et al.  Revised Interpolation Statistics for the Canadian Data Assimilation Procedure: Their Derivation and Application , 1990 .

[19]  J. Derber,et al.  A reformulation of the background error covariance in the ECMWF global data assimilation system , 1999 .

[20]  P. Houtekamer,et al.  A Sequential Ensemble Kalman Filter for Atmospheric Data Assimilation , 2001 .

[21]  L. Berre,et al.  The Use of an Ensemble Approach to Study the Background Error Covariances in a Global NWP Model , 2006 .

[22]  S. Cohn,et al.  Ooce Note Series on Global Modeling and Data Assimilation Construction of Correlation Functions in Two and Three Dimensions and Convolution Covariance Functions , 2022 .

[23]  N. B. Ingleby,et al.  The statistical structure of forecast errors and its representation in The Met. Office Global 3‐D Variational Data Assimilation Scheme , 2001 .

[24]  Olivier Pannekoucke,et al.  Filtering properties of wavelets for local background‐error correlations , 2007 .

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