**INTRODUCTION:**

Observations in the region of the Pribilof Islands and Canyon (PIC) reveal a clockwise circulation around the group of islands and around each of the two largest islands, St. Paul and St. George. Six current meters deployed around St.Paul Island revealed a steady clockwise flow around the island, that was strongest south of the island, and weakest to the east. We use a high-resolution tidal model in the PIC region to show that this flow pattern results from tidal rectification over the shallow topography tides. Tidal residual currents of 10-15 cm/s were predicted by the model, compared to mean currents of 5-20 cm/s observed at the mooring sites. Both diurnal and semi-diurnal tidal oscillations contribute to the clockwise circulation around the islands. In the diurnal band the enhanced currents occur also at the shelf slope where a tidal wave generates a shelf wave due to resonance with topography. In the PIC region, the main shelf wave occurs in the Pribilof Canyon where both observations and measurements show enhancement of the diurnal tidal currents.

**SATELLITE-TRACKED DRIFTERS:**

Typically drifters deployed on the outer shelf (water depth 100m to 180m) and the slope, are advected northwestward to the vicinity of the Pribilof Islands by the mean currents depicted in Figure 1. The two drifters shown in Figure 2 were deployed in May 1995. They were transported northward along the 100 m isobath at 3-4 cm/s. After reaching the Pribilof Islands, the buoys were entrained into the flow around the islands. One drifter circled St.Paul and then continued northward. The second drifter circled both islands and then re-entered the Bering Slope Current. Tidal loops are clearly evident in the trajectories.

In 1993, a satellite-tracked buoy was deployed on the Bering Sea shelf south of the Pribilof Islands. It was transported northwestward to the vicinity of St. George Island where it circled the island seven times in 60 days before it lost its drogue. One period of consistent clockwise circulation is shown in Figure 3. Mean circular velocity was about 20 cm/s and mean speeds did not appear to differ markedly on the different sides of the island. Diurnal tidal ellipses are evident although very small.

**TIDAL MODEL:**

To obtain high resolution tides around St. Paul Island we start from the large-scale model of the entire Bering Sea with resolution of 5' along latitude and 10' along longitude (Figure 4, top). For the entire Bering Sea a spherical system of coordinates is used. The results of computations from this domain serve to construct the boundary conditions for the domain around Pribilof Islands (Figure 4, bottom). A rectangular system of coordinates is used for the Pribilof Islands with a resolution of 1.852 km. The results of computations from the latter domain are applied to establish boundary conditions for the computations around St. Paul with resolution of 617 m. For computations of tide amplitudes and phases we shall use the vertically integrated equations of motion and continuity.

**TIDES IN THE PRIBILOF ISLAND AND CANYON:**

To understand the origin of the trapped circulation around Pribilof Islands that was revealed by satellite-tracked drifters, we investigate tide dynamics around these islands. Four major tidal constituents are considered, i.e., M2, N2, K1,and O1. In this domain a resolution of 1852 m is used. A tide wave enters the Bering Sea as a progressive wave from the North Pacific Ocean through Aleutian Island passages. The PIC domain is located over both the Bering Sea shelf and slope. In this region the tides propagate from the basin into shallow domain and are enhanced in both amplitude and velocity. To identify differences in semidiurnal and diurnal tidal waves, results of calculations for the M2 and K1 waves are considered. In Figure 5, top the amplitudes of surface elevation for the M2 tide are given. Sea level amplitude slowly changes from 24 cm over the slope to 31 cm over the shelf. An interesting pattern in the sea level occurs around Pribilof Islands. The sea level depict a dipole structure with the minimum located at the southeastern shores and the maximum at northern shores. Especially conspicuous is 15 cm sea level change around St. Paul Island. The isolines of the current magnitudes tend to follow bathymetry contours (Figure 5, bottom). This is clear in Pribilof Canyon where complex bathymetry occurs. Around Pribilof Islands due to topographic amplification and nonlinear interactions in the shallow water, velocities as large as 50 cm/s are generated. The tidal dynamics in the diurnal band of oscillations is illustrated through the major K1 constituent. Figure 6, top depicts general pattern of slowly varying sea level from southwest to northeast, with the maximum of amplitude, (>39 cm) occurring along the shelf slope. A number of local sea level maxima can be discerned not only near the islands but also along the shelf slope close to the 200 m depth contour. Especially pronounced is the maximum at southeastern flank of Pribilof Canyon. The K1 maximum currents are given in Figure 6, bottom. Although in the large scale pattern one can see the enhancement of current from deep basin to the shelf the total picture is more complex than observed in M2 tide. The enhancement of currents around island is not the sole feature. Along the shelf slope, especially in the region of Pribilof Canyon, the local enhancement is also apparent. The local regions of enhanced diurnal currents are related to the sea level changes observed in Figure 6, top. The important conclusion is that the diurnal tide can generate enhanced currents not only in the shallow domains around islands but in the deeper domains in the shelf slope region as well.

**TIDES AROUND ST PAUL ISLAND:**

We extend the above approach into the local domain around St. Paul Island applying a high-resolution numerical grid. The model data will be compared to current meter data. Eight current meters were deployed in the vicinity of St. Paul Island (Figure 7) from September 13, 1995 till August 7, 1996. The compass in the instrument at mooring site 6 failed resulting in no current data. At mooring site 1 two current meters were deployed at the 25 m and 55 m in a water depth of 65 m. All other current meters were deployed close to 25 m isobath. Current records were separated into tidal (diurnal and semidiurnal periods) and subtidal oscillations. The fine-grid model (617 m resolution) has been run around St. Paul Island to provide model simulations for comparison with observations. Four major constituents are considered (M2, N2, K1 and O1). The general pattern of the tidal motion reveals that the largest tidal currents (above 70cm/s) occurred at the southern tip of St. Paul Island, close to the site 3 (Figure 7). The model shows that the currents of the same magnitude occur in the vicinity of the site 6 (where current meter failed). Thus, the major current amplification domains are off southern and northeastern peninsulas in the shallow water domains. The directions of the major axes of the tidal ellipses tend to follow the contours of constant depth. It is evident from observations and computations that M2 constituent dominates the tidal velocity in the vicinity of the island. An explanation of this phenomenon is related to the sea level distribution around St. Paul Island. Although, the sea level for both M2 and K1 is of the same order, the sea level gradients are much stronger for the M2 constituent. The dipole structure in the sea level distribution depicted in Figures 5 and 6 drives tidal currents in the island's proximity. This sea level pattern is much stronger for the M2 tide with approximately 15 cm sea level change around St. Paul Island as compared to about 5 cm difference for the K1 tide.

**RESIDUAL MOTION INDUCED BY TIDES:**

To define a slowly changing subtidal motion, a low-pass filter (half amplitude 35 hours) was applied to the current records obtained around St. Paul Island. Data were then resampled at six-hour intervals (Figure 8). This filter removes the diurnal and semidiurnal tidal signal from the record. The low frequency currents are directed along bathymetry and support clockwise circulation around the island as observed in drifter trajectories. The flow at site 2 is into English Bay. The flow around island for the whole measuring period depicts amazing stability in direction, only once (February 10), the flow reversal occurred. Currents around the island can be forced by density gradients or wind; but neither of these mechanisms would result in the described stable behavior through the various seasons (Figure 8). The only constant mechanism is rectifying tidal currents into low frequency currents. The high-resolution model constructed around St. Paul Island is used to study a permanent circulation induced through the nonlinear interactions. Residual velocities and sea level can be obtained by averaging an hourly time series over period of 29 days. The horizontal resolution was further increased around St. Paul Island to 617 m. Calculations with this fine grid in the St. Paul subdomain confirm general clockwise circulation (Figure 9). In the island's vicinity the residual circulation is small and tends to be organized in local coastal eddies. At a distance of about 1 to 3 km from St. Paul Island the residual velocity attains greatest values and circles the island in continuous fashion. The average residual currents in this region are of the order of 10 to 15 cm/s. Additional results obtained for the M2 constituent computed alone, without interaction with the remaining tidal constituents, show that this constituent is primarily responsible for the generation of residual currents around St. Paul Island.

A comparison of the measured residual currents with the currents obtained from the above computation is given in the Figure 10. Computed current magnitude and direction turned out to be in good agreement with the current obtained from the observations. The most interesting is current pattern at sites 2 and 5. There, due to the local interaction with the bottom slope and coastline, the measured currents have been deflected from the general clockwise circulation. The local coastal eddies, evident in the fine resolution model (Figure 9), are responsible for this deflected flow pattern.

**PARTICLE MOTION INDUCED BY TIDES:**

The trajectories of satellite-tracked drifters depict well developed clockwise flow around the Pribilof Islands. Although residual clockwise motion (Figures 8, 9 and 10) would result in the clockwise rotation the oscillating tidal motion is important as well, because it causes excursions of the water particle on the order of 3 to 12 km (the approximate diameter of St. Paul and St. George Is.). It is obvious from our observations that the general pattern of circulation plays an important role in generating trajectories trapped around islands, diverting the buoys into and out of close proximity of the islands. To depict pattern of the tidal motion in the island's proximity, a set of particles is released around St. Paul. From experiments carried out we can conclude that the primary factors are Eulerian oscillatory and residual velocity. Stokes velocity is of secondary importance in the particle motion around St. Paul Island.

To depict trajectories in the St. Paul Island region, a number of particles are released to the west from the island. The initial locations and trajectories of the particles are given in Figure 11. Out of five particles released, the one initially located 1 km from the shore describes the fastest trajectory. The particle released closest to the island is slowed down by the local coastal eddies. Only particles within 6 km of the island describe major features of the clockwise trapped motion. Particles initially located greater than 8 km west of the island were trapped into local offshore eddy generated around shallow bank, see Figure 9. We do not have a large enough set of measurements of buoy trajectories around the islands to perform statistical analysis and comparison against model computations. It is interesting to notice that trajectory of the fastest particle in Figure 11 (St. Paul Is. region), and satellite-tracked buoy trajectory in Figure 3 (St. George Is. region) show a markedly similar period of rotation (~6 days) and pattern of the very small tidal ellipses.