Coastal Ocean Modeling: Processes and Real-time Systems

Introduction The coastal oceans are among the most challenging marine environments in the world. They are subject to the combined geometrical constraints of irregular coastlines and highly variable (steep and tall) bathymetry, and are forced both internally, laterally and surfacially by a complex array of tidal, wind and buoyancy forces on a broad range of space/time scales. The resulting coastal circulation patterns include both persistent and time-variable fronts, intense currents with strong spatial (offshore and/or vertical) dependence, coastal trapped waves, internally generated mesoscale variability, large horizontal water mass contrasts, strong vertical stratification, and regions of intense turbulent mixing in both surface and bottom boundary layers. An extended review of coastal physical processes, which complements the discussion below, is provided by Brink and Robinson (1998). Numerical modeling of these areas of the world's ocean clearly requires very flexible, highly optimized models of significant dynamical complexity. In the past, limited computer resources led to the development of physically and/or geometrically simplified models. In recent years, however, models developed for lake and shelf-sea dynamics have become increasingly complex, and are now typically based on the fully nonlinear stratified "primitive equations." Haidvogel and Beckmann (1998) summarize an international inventory of coastal ocean circulation models and describe their general attributes. This review has three objectives. First, we provide a brief description of the physical environment and processes which dominate regional circulation patterns on the U.S. continental shelves. Second, we describe three coastal circulation models in primary use in U.S. coastal waters, which taken together represent the present state-of-the-art in coastal circulation modeling in the U.S. Lastly, we discuss several ongoing real-time and near-real-time applications of these numerical circulation models.