On correction of model parameters with experimental data

A new method of the correction by experimental data of the computed model characteristics is considered and tested. The method is based on the known Kalman-filter theory but has a number of essential details. In particular, it doesn't require the linearity of the model operator and also has some advantages in the numerical realization. The method is verified by “PIRATA” experiment data, which is being carried out in the tropical Atlantic. The paper presents also the results of the comparison of numerical experiments by two different methods with the same data set.