Using a SEEK filter as data assimilation kernel in NEMOVAR inner loops

Data assimilation (DA) is divided in two main branches: (i) variational assimilation and (ii) stochastic assimilation. Unlike variational assimilation, stochastic assimilation does not rely on a functional minimization with iterative process and gradient estimations. The stochastic branch includes Kalman Filter-type approaches and low rank approximations such as SEEK Filter. NEMOVAR is designed for variational applications, however its two stages inner-outer loops allow for different assimilation kernel in the inner loop. The present report study the use of a SEEK filter within NEMOVAR.