Ensemble Kalman-Filtering of Earth rotation observations with a global ocean model

We study the changes in ocean angular momentum which lead to changes in the Earth's rotation. The focus lies on the consistency of model, model-forcing and observations. To check this consistency an ensemble-based Kalman-Filter approach was applied to ocean angular momentum time series which we derived from Earth rotation observations. The filter propagates a reduced rank error-covariance matrix with an optimal ensemble. This way, a complex system like the utilized global ocean circulation model can be assimilated with 32 ensemble members only. The Kalman-Filter improved the ocean-model's trajectory with respect to the observations, i.e., the observed ocean angular momentum was better reproduced by our model. A subsequent analysis of the changes which were induced by the filter revealed that the utilized atmospheric forcing is insufficient. When the filter is not entitled to change the forcing fields no improvement in the model trajectory was possible.

► We run an ensemble of OGCMs.
► We used state-of-the-art Kalman-Filter techniques.
► We assimilated ocean angular momentum observations.
► The assimilation resulted in an improved model trajectory.
► The differences between non-assimilated model and assimilated model reveal a mass-budget problem in the atmospheric forcing.