Variational Data Assimilation for the Sea Thermodynamics Model and Sensitivity of Marine Characteristics to Observation Errors

Abstract

The methodology of variational assimilation of observational data for the reconstruction of the initial state and heat fluxes for the mathematical model of sea thermodynamics is presented. An algorithm is developed for estimating the sensitivity of a model solution to errors in observational data. The calculation of the gradient of the response function of the model solution is based on the use of the Hessian of the cost functional. The results of numerical experiments for the Black Sea dynamics model developed at the Institute of Numerical Mathematics, Russian Academy of Sciences, are presented.