Coupled Global and Regional Circulation Models for the Coastal Gulf of Alaska


A. J. Hermann

Joint Institute for the Study of Atmosphere and Ocean

University of Washington, Seattle


D. B. Haidvogel

Institute of Marine and Coastal Sciences

Rutgers University, New Brunswick


E. L. Dobbins and P. J. Stabeno

Pacific Marine Environmental Laboratory

National Oceanic and Atmospheric Administration, Seattle



As part of the US GLOBEC NE Pacific program, we are simulating currents in the Coastal Gulf of Alaska (CGOA) to explore sources of interannual and interdecadal variability. To do so, we are developing coupled modeling systems composed of linked regional and global circulation models. Our initial regional model (SCRUM), configured with 13-22 km resolution in the CGOA, is forced at the surface by observed heat fluxes and wind stresses, at the continental boundaries by observed runoff, and at the open ocean boundaries by a combination of tracer climatologies and subtidal velocity and tidal elevation provided by a global finite element model (SEOM). A improved regional model (ROMS), with finer resolution (10 km) and expanded spatial coverage (from California to the Bering Sea), is now running on a massively parallel, distributed memory computing architecture. Here we describe the present and anticipated coupled systems, including the present method of inter-model coupling, describe a series of multi-year model hindcasts, compare hindcast results with Eulerian and Lagrangian field data obtained in the CGOA in fall 1996, and assess the impact of global information (barotropic subtidal velocities and tidal elevations) on the regional model under the present coupling strategy. We find that the regional model produces appropriate current systems (Alaskan Stream, Alaska Coastal Current) and scalar fields, but with mesoscale variability (of SSH and velocities) at somewhat reduced strength relative to data, and with temperature gradients somewhat larger than those observed. Barotropic subtidal information from the global model penetrates the regional model interior, supplying additional mesoscale variability, and modifying regional velocity and scalar fields in both shallow and deep areas. Tidal information exerts a significant influence on subtidal scalar and velocity structure only in specific shallow areas, where the tides (and tidal mixing) are strongest.