Аннотация:The spatial-temporal variability of parameters of the model of ocean circulation Nucleus for European Modeling of theOcean (NEMO) with data assimilation by the generalized Kalman filter method (Generalized Kalman filter (GKF)) previously elaborated by the authors have been investigated. In this study, a principally new approach based on the theory of stochastic differential equations is proposed to determine the key parameters of the GKF method and it is shown how these parameters influence on the calculation characteristics of the model. Such an approach does not require preliminary constructing the ensemble of model states and does not use the assumption of the unbiasedness (the absence of a systematic error) of the model with respect to observations. The proposed approach allows obtaining the asymptotic behavior of model parameters, in particular, estimating the probability of their overrunning some fixed level on the considered time interval of modeling. Modeling of the spatial-temporal variability of ocean parameters is performed by using the NEMO model with application of the proposed data assimilation method and the use of observations from the Argo archive. Numerical calculations were carried out at the K-60 supercomputer in the Keldysh Institute of Applied Mathematics of the Russian Academy of Sciences.