This paper, entitled Adaptive covariance inflation in the ensemble Kalman filter by Gaussian scale mixtures is a smart(?) continuation on the finite-size ensemble Kalman filter (EnKF-N). It describes four significant accomplishments:

  • A thorough review on multiplicative inflation as used in the ensemble Kalman filter (EnKF) literature.
  • Building on the paper Expanding the validity of the ensemble Kalman filter without the intrinsic need for inflation, a significant clarification on the role of nonlinearity in the need for multiplicative inflation and the appearance of sampling errors.
  • An extension of the finite-size EnKF (EnKF-N) to account for not only sampling errors but also model error.
  • A thorough numerical comparison on the two-scale spatially-extended Lorenz model that shows that the EnKF-N that also accounts for model error, systematically outperforms other adaptive schemes, although with a moderate margin.