83rd Annual

Tuesday, 11 February 2003
Advantages of data assimilation for understanding coastal ocean dynamics off Oregon
Alexander L. Kurapov, Oregon State University, Corvallis, OR
A data assimilation system (DAS), which combines a dynamical model and available data to provide an estimate of the time-varying three-dimensional circulation, will be an essential component of an operational coastal observatory. To understand issues of assimilating data into coastal ocean baroclinic models off Oregon, we consider a variety of options: models based on linearized dynamics are utilized with the most rigorous data assimilation methods (such as the generalized inverse method, GIM), and a model based on full non-linear dynamics is applied with a simple, suboptimal, sequential data assimilation algorithm (namely, "optimal interpolation", OI). In any case the models are constrained and validated with realistic data of the type that should be common for coastal observatories, e.g., HF radar surface currents and moored ADP velocity, temperature and salinity time series measurements. A linearized primitive equation model describing harmonic oscillations of the stratified ocean is utilized with the GIM to study internal tide propagation on the Oregon shelf. Assimilating HF radar surface currents improves the tidal solution at depth and as a result provides a uniquely detailed picture of the spatial and temporal variability of the M2 internal tide on the coastal shelf off Oregon. A DAS based on the full primitive equations combines Princeton Ocean Model (POM) with OI. It is implemented with observations from the COAST field experiment (spring-summer 2001) to learn about the potential of data assimilation for constraining modeled wind-driven circulation, using different sets of measurements. During this period, observational information assimilated in the model can be advected southward by shelf currents or propagated northward by coastal-trapped waves, providing corrections to the model solution away from the measurement site. We find that assimilating velocities from an across-shore line of ADPs helps to improve the quality of the prediction at alongshore distances of at least 90 km in both directions. Present efforts are directed at understanding multivariate capabilities of data assimilation and devising optimal data sampling strategies.

Supplementary URL: