Data assimilation methods for the passive impurity transfer-diffusion problem
This work is devoted to the study of the issue of obtaining a joint assessment of the concentration of passive admixture and emission.
An extended problem is considered and variable formulas of stochastic methods are given, such as the Classical Kalman filter and the ensemble Kalman filter. Localization is applied in case of problems and it is demonstrated how this affects the result.
In the second part, we consider ways to correct the error covariance matrices using multiplicative inflation ― it is shown why and why it should be used. Model experiments show how this works.
The progress of work of Kalman Smoothter is implemented and shown, and the results are compared with the Kalman ensemble filter. Demonstrated various variations of input parameters and the dependence of the results on them.
All experiments were performed on model data.
Ms. Platonova Vladimirovna Marina Novosibirsk State University