Title: 4-Dvar spectral covariance with horizontal anisotropic transformation

Authors: Xudong Sun (Centre for Australian Weather and Climate Research Bureau of Meteorology)
Peter Steinle (Centre for Australian Weather and Climate Research Bureau of Meteorology)

The use of limited anisotropic horizontal covariances in spectral form in variational data assimilation will be presented. The importance of background error covariances to data assimilation are well established. Based on the previous studies of anisotropic covariances in physical space and its statistical evidence calculated in Australia region, the anisotropic horizontal covariance parameters are defined and the subsequent transform is performed from physical space to spectral space mathematically by using two-dimensional Fourier and Hankel transformations. We use SOAR correlation relationship as an example of these processes. This provides a way for correlations in 4Dvar to be calculated that not only it relies on distance but also on its horizontal directions. It is found that: with its Fourier transformation, the initial anisotropic angle can communicate in both the physical space and spectral space. Furthermore, such scheme is successfully applied in ACCESS regional 4D variational data assimilation and the model forecast verifications has shown some notable improvements in NWP model forecast accuracy.

GMAO Head: Michele Rienecker Global Modeling and Assimilation Office NASA Goddard Space Flight Center

Curator: Nikki Privé Last Updated: May 27 2011