Data Assimilation in a Quasi-geostrophic Open-Ocean Model of the Gulf Stream Region Using the Adjoint Method

Abstract

The method of adjoint data assimilation is applied in a quasi-geostrophic (QG) open-ocean model of the Gulf Stream region. The results of data assimilation experiments are presented in which simulated AXBT and satellite altimeter observations are assimilated into the QG model. The adjoint data assimilation scheme has the ability to correct for large errors in the speed and position of the Gulf Stream jet. These experiments provide valuable insight into the dynamics of the assimilation procedure, and highlight the limitations of the adjoint method in the Gulf Stream region. The adjoint variables of the linearized QG model are approximations of the Green's functions of the adjoint equations. By examining maps of the adjoint variables, one can determine the domain of influence of individual observations. The results of assimilation experiments in which GEOSAT sea surface height observations are assimilated into the QG model are also presented. The results of the GEOSAT assimilation experiments can be interpreted using the insight gained from the simulated data assimilation experiments. It is found that not only can the adjoint method correct the position of the Gulf Stream axis, but it can also reconstruct observed features present in the real ocean (such as Gulf Stream rings) which were originally absent in the model.