Numerical modeling study of the circulation in the Gulf of Alaska

Abstract

A series of numerical experiments are performed to simulate the Gulf of Alaska circulation and to examine the dynamical ocean response to the annual mean and seasonal forcing using a primitive equation model (Semtner 1974). The model domain encompasses the North Pacific north of 45$\sp\circ$ N and east of 180$\sp\circ$ and is surrounded by artificial walls in the south and west. Biharmonic diffusion is used in the interior to excite mesoscale eddies. A sponge layer with high Laplacian diffusion is incorporated near the western boundary. Horizontal resolution of 30$\sp\prime$ x 20$\sp\prime$ and 20 vertical levels are used to resolve the mesoscale topography and eddies. Wind stress computed from sea level atmospheric pressure and temperature and salinity data of Levitus (1982) are used. A diagnostic model produces a circulation in the Gulf of Alaska which agrees with observed patterns. In a three-layer flat-bottom baroclinic model, baroclinic Rossby waves propagate at 0.8 cm/sec and it takes a decade for spin-up to be completed. Baroclinic models forced by the annual mean wind and thermohaline forcings show the generation of eddies by baroclinic instability. The eddies in the flat-bottom model have a period of 75 days and are interpreted as barotropic Rossby waves. In the model with topography, the period of dominant eddies is 3-4 years and they are interpreted as baroclinic Rossby waves. Anticyclonic eddies near Sitka show similar characteristics as the Sitka eddy. They propagate westward and cause meanders in the Alaska Stream near Kodiak Island. The abnormal shift of the Alaska gyre in 1981 is probably due to the presence of one of these anticyclonic eddies. A flat-bottom model with seasonal forcing shows a large seasonal variability. When bottom topography is present, however, seasonal response is greatly reduced due to the dissipation of barotropic response by bottom topography. The seasonal baroclinic model shows a similar seasonal variability to the seasonal barotropic model indicating that the seasonal response is mainly barotropic. Eddies are also excited in the seasonal case and are almost identical to those of the annual mean case.