HR: 17:00h
AN: OS54B-05    [Abstracts]
TI: Progress in the Application of Tangent Linear and Adjoint Models for Predictability Studies in the Coastal Ocean.
AU: * Durski, S M
EM: sdurski@coas.oregonstate.edu
AF: Oregon State University, College of Oceanic and Atmospheric Sciences, Corvallis, OR 97331 United States
AU: Samelson, R M
AF: Oregon State University, College of Oceanic and Atmospheric Sciences, Corvallis, OR 97331 United States
AU: Allen, J S
AF: Oregon State University, College of Oceanic and Atmospheric Sciences, Corvallis, OR 97331 United States
AU: Egbert, G D
AF: Oregon State University, College of Oceanic and Atmospheric Sciences, Corvallis, OR 97331 United States
AB: The non-linear nature of coastal ocean dynamics raises questions about the predictability limitations that may be encountered when using coastal observing systems for forecast purposes. Developing an understanding of the transient growth phenomena that may be associated with a given environment can lead to system designs that maximize the information content of the observations and enhance the forecast skill. Regional circulation models, such as the Regional Ocean Modeling System(ROMS), based on the fully non-linear primitive equations, should exhibit transient growth phenomena representative of those which develop in the real ocean. Leading singular vectors (optimal perturbations) may be determined for a specific model configuration using the forward and adjoint tangent linear models. These modes describe the initial spatial structures which maximize growth in some norm (energy, enstrophy, etc.) and can be utilized in ensemble generation and assessments of predictability. Tangent linear and corresponding adjoint tangent linear models have recently become available for ROMS along with drivers for calculating the normal modes, adjoint normal modes and singular vectors. As preliminary steps in the assessment of these tools and the study of disturbance growth in coastal ocean flows, normal modes, adjoint normal modes and singular vectors are calculated for several idealized flows and compared to solutions of the corresponding discretized eigenvalue problems.
DE: 4219 Continental shelf processes
DE: 4263 Ocean prediction
DE: 3210 Modeling
SC: Ocean Sciences [OS]
MN: 2004 AGU Fall Meeting