Return to search

Variational assimilation of acoustic tomography

For the first time an ocean model is used to assimilate oceanic tomography data in an upper ocean model of the northeast Pacific with the goal of estimating the time independent density field, and thus the slow manifold circulation structure. / The assimilation procedure works by minimizing the cost function, which generalizes the misfit between the observations and their model counterparts, in a least-squares sense, plus a penalty term. This minimization is done consistently with the constraint that the model dynamics must be exactly satisfied. The model consists of integrating the model equations forward in time over the period which data are going to be assimilated. Data misfits between the model and the observation are then calculated and the adjoint equations of the model are integrated backward using the data misfits as forcing. It is necessary to determine the gradient of the cost function with respect to the control variables (the density field). The gradient is found using the model and adjoint variables and it is used in a minimization algorithm to determine a new density field. The minimization procedure utilizes a limited memory quasi-Newton method. / The results indicate that the assimilation procedure works very well. For the twin experiments, the final estimated density recovers the Levitus density field as expected and as fast as in 10 iterations. For the experiments with the Navy layered ocean circulation model (NRLM) output, the density can be estimated through the assimilation procedures. The estimated density field improves the the Levitus climatological density data which are biased and makes the subtropical gyre stronger in the northeast Pacific region. / The proof of the identity between the discretization of the continuous adjoint equations and the adjoint equations which are from discretized model equations with the Arakawa C has been carried out. / Source: Dissertation Abstracts International, Volume: 56-07, Section: B, page: 3667. / Major Professor: James J O'Brien. / Thesis (Ph.D.)--The Florida State University, 1995.

Identiferoai:union.ndltd.org:fsu.edu/oai:fsu.digital.flvc.org:fsu_77507
ContributorsLiu, Ming., Florida State University
Source SetsFlorida State University
LanguageEnglish
Detected LanguageEnglish
TypeText
Format106 p.
RightsOn campus use only.
RelationDissertation Abstracts International

Page generated in 0.0017 seconds