Anna Shlyaeva, NOAA Earth System Research Laboratory and Cooperative Institute for Research in Environmental Sciences, Boulder, USA

Correlated observation errors in the perturbed observation ensemble data assimilation

Mark Buehner (Environment Canada)


Most of the current data assimilation systems assume observation errors to be uncorrelated. Recent studies, however, have shown that correctly accounting for the correlation of observation errors may improve the analysis, especially for the small scales. On the other hand, there has been a growing interest in a scale-dependent background error localization where more severe localization is applied to small scales and less localization to large scales. Using scale-dependent background error covariance localization and including the effects of the correlated observation errors both contribute to how different scales are treated in the analysis. In this talk the standard approach (single-scale background error localization and ignoring observation error correlations) will be compared to a more sophisticated approach (scale-dependent background error localization and correctly accounting for the observation error correlations) in the context of the perturbed observation ensemble assimilation system. The examples will be presented for a simple 1D problem with isotropic and homogeneous covariances, as well as for a more realistic 2D sea-ice problem.

File available here