Summary: | This dissertation presents, applies, and evaluates a statistical approach to an ocean circulation problem. The objective is to produce a map of ocean velocity in the North Atlantic based on sparse measurements along ship tracks, based on a Bayesian approach with a physical model. The Stommel Gulf Stream model which relates the wind stress curl to the transport stream function is the physical model. A Gibbs sampler is used to extract features from the posterior velocity field. To specify the prior, the equation of the Stommel Gulf Stream model on a two-dimensional grid is used.Comparisons with earlier approaches used by oceanographers are also presented. Submitted Note: A Dissertation submitted to the Department of Statistics in partial fulfillment of the requirements for the degree of Doctor of Philosophy. Degree Awarded: Summer Semester, 2007. Date of Defense: JULY 5, 2007. Keywords: MCMC, Gibbs Sampler, Gulf Stream Bibliography Note: Includes bibliographical references. Advisory Committee: Fred W. Huffer, Professor Co-Directing Dissertation; Kevin G. Speer, Professor Co-Directing Dissertation; Craig Nolder, Outside Committee Member; Xufeng Niu, Committee Member; Wei Wu, Committee Member.
|