From the meteorological perspective, the Gulf of Alaska is a region of extremes. The northern Gulf of Alaska experiences the potent consequences of vigorous marine extratropical cyclones making landfall in some of the most dramatic and extreme terrain in North America. The high frequency of storms- on average one every four to five days during the cold season- often create strong pressure gradients that interact with local topography to produce localized wind regimes in strong contrast to the larger scale circulation.While PWS and Cook Inlet are only a small part of the Gulf of Alaska coast, they experience most of the weather features and phenomena seen in the rest of the Gulf. PWS is a complex embayment composed of fjords, deeply incised river valleys, and steep mountain ridges. Much of PWS is surrounded by the Chugach Mountains, that here average about 2000 m in elevation, with peaks extending to near 3000 m. Three significant gaps in the terrain that can act to funnel and focus regional winds. Each of these gaps has the potential to permit exchange of air with a considerably different, and often continental, airmass. By contrast, Cook Inlet (CI), which borders PWS to the west, has a much simpler topographic structure: an elongated bay terminating in two perpendicular arms.

We use the Regional Atmospheric Modeling System (RAMS) to simulate several cases where terrain along PWS and CI plays a significant role in determining local coastal wind conditions. These cases were chosen because there is high resolution synthetic aperature radar (SAR) imagery of the surface wind field against which to verify the model. To capture the complex nature of terrain-weather interactions adequately, it is necessary to use a model with high spatial resolution. In this study, a double or triple nested-grid approach is used, with grid spacings of Dx,y=64 km, 16 km and 4 km and optionally 1 km for the parent grid and its 2 or 3 nested grids. This allows the model to simulate both the synoptic-scale weather at the scale of hundreds of kilometers and localized terrain influences that occur on a scale of just a few km.

Not surprisingly, the grid-spacing was found critically important to resolving ageostrophic, down-gradient flows in narrow channeled regions. This results in large part from the fact that at least 4 grid points in the horizontal are needed to adequately resolve a channel in the terrain. Runs without the adequate terrain resolution typically failed to produce mesoscale,terrain-influenced flow with a significant deviation from the large-scale circulation. This was more apparent in PWS, with it's complex system of fjords and bays, than in Cook Inlet where the channeling is simpler and of larger scale.

From a forecast perspective, these results have several implications. For short-term mesoscale marine weather forecasts, the better resolution of localized winds is clearly worth the extra computational expense required for the 4-km grid. While the configuration of the 1-km grid is not currently feasible to run as an operational forecast model, it clearly is required to adequately resolve some localized low-level jets, and the next generation of computers will make this kind of prediction possible on an operational basis.