AEFF HOME

Prince William Sound Domain

One of the first domains to be studied as part of the AEFF's mesoscale modeling program is that of Prince William Sound on the northern periphery of the Gulf of Alaska. Prince William Sound is known by many as the site of the Exxon Valdez oil spill incident. Huge oil tankers still ply the waters of the Sound as do many smaller craft, such as fishing boats, sight-seeing tours, recreational boaters and many others.

Some of the greatest hazards to marine interests in Prince William Sound are high winds and resulting rough seas. Strong winds can result from any of several meteorological forcing mechanisms. Furthermore, the rough terrain encircling the Sound on three sides often interacts with larger-scale winds and pressure gradients to produce highly variable wind regimes on scales of a few kilometers.

One of our goals in simulating the weather in Prince William Sound is to determine the predictibility of such localized winds and understand how they arise from the mesoscale atmospheric forcing. To that end, we are simulating events where high winds were known to have occurred. Here we show a "snapshot" of some preliminary modeling results from a nested-grid simulation of a storm that occurred on August 28-29, 2001. The following images also illustrate the several scales-spanning several orders of magnitude-that are involved in producing the complex weather observed in the North Gulf of Alaska. All the images are coincident in time.

**GRID 1**

Figure 1. Surface wind vectors, wind speed (m/s, shaded) and mean sea level pressure (mb, contours) on Grid 1. The horizontal grid spacing is 64 km with 45 points in both the zonal (E-W) and meridional (N-S). The wind vectors appear at approximately every other point in the zonal and meridional directions. .

Figure 1 shows the synoptic-scale weather at the planetary surface on Aug 28,18 UTC. As is often the case with high wind events, the Gulf of Alaska is dominated by a low pressure system, this one centered about 340 km south of Kodiak Island. A strong low-level jet (local wind maximum) has formed on the eastern and northern flanks of the low as the eastward propagation of this extratropical cyclone is hindered by the high terrain of the coastal ranges on the Alaska Panhandle.

The area shown in Figure 1 represents essentially the domain of Grid 1, the coarsest grid, within which the finer grids are nested. This grid needs to be large enough to capture the large scale dynamics associated with the propagating storm system.

**GRID 2**

Figure 2. Surface wind vectors, wind speed (m/s, shaded) and mean sea level pressure (mb, contours) on Grid 2. The horizontal grid spacing is 16 km with 70 points in the zonal (E-W) and 58 in the meridional (N-S). The wind vectors appear at approximately every 4 points in the zonal and meridional directions. .

The tongue of the jet seen in Figure 1 is seen in higher resolution in Figure 2, with a core maximum exceeding 18 m/s (~36 kt) impinging on the North Gulf Coast. A isallobaric jet is also apparent in Cook Inlet at this scale with NE winds exceeding 16 m/s crossing the isobars in the lower Inlet at nearly right angles.

Figure 2 spans the area covered by Grid 2, the intermediate grid in this simulation, with a grid spacing of 16 km. At this scale, the model is able to capture sub-synoptic flow features such as the Cook Inlet jet that are not resolved by larger-scale forecast models. Note however that the center of parent storm producing the pressure gradient responsible for the strong winds is well equatorward of the southern boundary of Grid 2. The integrity of the numerical solution on Grid 2 requires that Grid 1 accurately simulates the storm's large scale motion and evolution.

**GRID 3**

Figure 3. Surface wind vectors, wind speed (m/s, shaded) and mean sea level pressure (mb, contours) on Grid 3. The horizontal grid spacing is 4 km with 102 points in the zonal (E-W) and 78 in the meridional (N-S). The wind vectors appear at approximately every 4 points in the zonal and meridional directions.

The winds in Figure 3 show significantly more fine-scale variability than can be seen in Figure 2. The higher terrain on the Kenai Penninsula coincides with the local maxima (red variegations) in the wind speed. The down-gradient flow comprising the Cook Inlet jet (Figure 2) is seen to actually extend eastward of the Inlet over the lower terrain of the western Kenai Penninsula. At this scale it is also apparent that, for this case at least, Hinchinbrook and Montague Islands on the SW boundary of Prince William Sound act to shelter the Sound from the full brunt of the winds.

Figure 3 shows the full extent of Grid 3, the highest resolution grid in this simulation, with a grid spacing of 4 km. Note that there are many more points on this grid than would be suggested by the wind vectors in the figure. To appreciate the full resolution of local circulations possible with a 4 km grid spacing (and still have a graphic that is viewable with a web browser) it is necessary to look at a subdomain of Grid 3.

**GRID 3 Detail**

Figure 4. Same as Figure 3, except a subdomain of grid 3, where here the wind vectors appear at every point.

The domain in Figure 4 is a subset of Grid 3 centered on Prince William Sound. The gradient in wind speed from the center of the Sound to its periphery is pronounced, with a jet core now quite apparent in the west central part of the Sound. Other circulation features, such as the down-gradient flow out of Port Wells/Harvard Fjord (upper right of figure) and Valdez Arm (upper left of figure) are clearly tied to the complex terrain bounding Prince William Sound.

**GRID 3 Topography**

Figure 5. Six-hour precipitation totals ( mm water equivalent, shaded) and terrain height (m) representation on Grid 3.

There exists a strong correlation between simulated precipitation and terrain as well. Figure 5 shows simulated 6 hour precipitation totals and terrain height. Note the very high precipitation amounts on the windward side of the higher terrain along the western edge of the Sound-a consequence of the strong onshore flow seen in this event. The several tidewater glaciers found on the western edge of Prince William Sound attest to the large number of high-precipitation storms of this type that must occur each year. These results also suggest that such storms must contribute significantly to the fresh water input for the Sound on an annual basis.

In conclusion, the results presented here represent a snapshot of the ongoing mesoscale atmospheric modeling work underway at the Alaska Experimental Forecast Facility. We hope that this research will ultimately benefit a wide variety of Alaskans that use weather forecasts and other meteorological knowledge as part of their daily life and livelihood.