Ensemble-based background error covariance implementations using spatial recursive filters in NCEP's grid-point statistical interpolation system

A methodology for specifying ensemble-based background error covariances within the recursivefilter (RF) formalism of the NCEP Grid-point Statistical Interpolation (GSI) is proposed and tested using a low-resolution version of the North-American Regional Data Assimilation System. Perturbation fields from six-hour forecasts from the 80-member NCEP Global Ensemble Forecast System are used to represent the background errors. The RF-generated covariances are found to agree fairly well with the exact covariances computed directly from the ensemble perturbation fields in the vicinity of the selected test points. In addition, they are necessarily free of the spurious correlations at long ranges that characterize the covariance-matrix of the low-dimension space of the sample ensemble perturbations. The anisotropies are found to be rather weak when the 80-member ensemble is used without any further treatment to prescribe the local aspect tensor required by the recursive filter. However, a remarkable enhancement of the anisotropy is obtained when the ensemble size is artificially increased through a special type of local averaging. Results from a case study reveal an improvement of the 500 hPa geopotential height skill scores when the convariances with enhanced anisotropy are applied.

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

[2]  W. Briggs.  Statistical Methods in the Atmospheric Sciences , 2007 .

[3]  Brian J. Hoskins,et al.  The Geostrophic Momentum Approximation and the Semi-Geostrophic Equations. , 1975 .

[4]  T. Hamill,et al.  A Hybrid Ensemble Kalman Filter-3D Variational Analysis Scheme , 2000 .

[5]  R. Purser,et al.  Three-Dimensional Variational Analysis with Spatially Inhomogeneous Covariances , 2002 .

[6]  Haixia Liu,et al.  Retrieval of Moisture from Simulated GPS Slant-Path Water Vapor Observations Using 3DVAR with Anisotropic Recursive Filters , 2007 .

[7]  Francis P. Bretherton,et al.  Atmospheric Frontogenesis Models: Mathematical Formulation and Solution , 1972 .

[8]  Comparison of strategies for the specification of anisotropies in the covariances of a three-dimensional atmospheric data assimilation system , 2006 .

[9]  N. Roberts,et al.  Numerical Aspects of the Application of Recursive Filters to Variational Statistical Analysis. Part II: Spatially Inhomogeneous and Anisotropic General Covariances , 2003 .

[10]  John Derber,et al.  A Global Oceanic Data Assimilation System , 1989 .

[11]  Lars Peter Riishojgaard,et al.  A direct way of specifying flow‐dependent background error correlations for meteorological analysis systems , 1998 .

[12]  Yuejian Zhu,et al.  Initial perturbations based on the ensemble transform (ET) technique in the NCEP global operational forecast system , 2008 .

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

[14]  A. O'Neill.  Atmospheric Data Assimilation , 2000 .

[15]  Gérald Desroziers,et al.  A Coordinate Change for Data Assimilation in Spherical Geometry of Frontal Structures , 1997 .

[16]  B. Hoskins,et al.  The Forcing of Ageostrophic Motion According to the Semi-Geostrophic Equations and in an Isentropic Coordinate Model , 1977 .