Over the past 10 years, Pavel Sakov, colleagues and myself have developed iterative ensemble Kalman filters and smoothers. At the cost of propagating the ensemble a few more times (typically twice) per data assimilation cycle, they offer significanlty better accuracy than traditional ensemble Kalman filters and smoothers when the model dynamcs are nonlinear. We have recently incorporated additive model error into those schemes in [P. Sakov and M. Bocquet,Tellus A,70 (2018), 1414545] and [P. Sakov,J. Haussaire, and M. Bocquet, Quart. J. Roy. Meteorol. Soc., 144 (2018), pp. 1297-1309].

This new paper, led by my former PhD student, Anthony Fillion, generalizes these methods to the full (i.e. for any lag) iterative ensemble Kalman smoother with additive model error: the IEnKS-Q. Implementations and numerical simplifications of the IEnKS-Q are developped and tested on low-order chaotic models.

The paper, entitled An Iterative Ensemble Kalman Smoother in Presence of Additive Model Error, is published (open access) in SIAM/ASA Journal on Uncertainty Quantification.