Localizing the Error Covariance by Physical Distances within a Local Ensemble Transform Kalman Filter (LETKF)

Abstract

An efficient implementation of the local ensemble transform Kalman filter (LETKF) with the error covariance localization by physical distances is introduced and assessed in this study. Instead of using local patches uniform in the model grid space to localize the error covariance, accurate physical distances are computed and used for the localization, so that the problem of analysis discontinuities in the Polar Regions is solved. Data assimilation cycle experiments with real observations are performed, which indicate less discontinuity in the Polar Regions. Moreover, the computational time is shorter and more robust for various localization scales. Thus, the implementation introduced in this study is a promising choice of future LETKF systems.