A Three-Dimensional Variational Data Assimilation Scheme for the Regional Ocean Modeling System

Zhijin Li(1,2), Yi Chao (1), James C. McWilliams(3) & Kayo Ide(3)

(1) Jet Propulsion Labratory and California Institute of Technology
(2) Raytheon
(3) University of California, Los Angeles

A three dimensional variational data assimilation (3DVAR) scheme, which is associated with the Regional Ocean Modeling System (ROMS) and thus called ROMS-DAS, has been developed. ROMS-DAS emphasizes the capability for predicting meso- to small-scale variations with temporal scales from hours to days in coastal oceans. To cope with particular difficulties that arise from complex coastline and bottom topography, unbalanced dynamics and sparse observations, ROMS-DAS is incorporated with novel strategies. These strategies include the implementation of three-dimensional anisotropic and inhomogeneous error correlations, application of weak dynamic constraints to the increments, and implementation of efficient and reliable minimization methods. During real-time experiments, ROMS-DAS has demonstrated satisfactory predictive skill, as well as high efficiency and reliability.

Last Update: March 25, 2005