A Study On Data Assimilation Of Sea Temperature With Kalman Filter

The Kalman Filter is introduced and explored by assimilating sea temperature observations into one-dimensional numerical temperature model, which is decomposed into two parts, one part is the certain, the other is the random. Then a data assimilation model is set up, it can assimilate sea temperature continuous observations data. A data assimilation experiment is carried out through combination the model with temperature measurements from the 165 E moorings on the Equator, which collected by TAO/TRITON (Tropical Atmosphere and Ocean Project) array of bugs. The result shows that Kalman Filter is efficient to improve the calculation of temperature with the above mentioned model.Through the study of this paper, it indicated that the Kalman Filter method provides not only the optimal sequential estimation of the system, but also system error information. It is suitable to assimilate the real-time observed data into forecasting model. Its drawback is high demand on the computational resources. So for the practical ocean numerical model (Three-dimension, nonlinear), some reduced-order scheme are employed, which give less loss in optimality, such as SEEK filter method (Singular Evolutive EKF).