Wind forcing

The volume transport data were compared with NCEP (National Centre for Atmospheric Prediction) monthly surface wind data (sigma level = 0.995), which has a grid spacing of 2.5° latitude by 2.5° longitude (Peterson 2008). Multiple correlation coefficients with the observed transports were computed using monthly mean wind data at each grid point (i,j) north of 55° N, and a regression model of the form

where M is volume transport, Uj is the zonal wind component, Vj is the meridional wind component at grid point (ij), aj, bj, and c¡j are the regression coefficients, and e is the residual.

The maximum correlation coefficient (R = 0.68) is found in the Beaufort Sea west of M'Clure Strait at 75° N, 132.5° W, the western entrance of the NW Passage of the CAA. It is over 1,000 km from the mooring array in Lancaster

Fig. 11. Mean monthly volume flux through western Lancaster Sound derived from 8 years of mooring data (Aug 1998-Aug 2006).

Fig. 12. Map showing correlation coefficients between monthly anomalies of NCEP surface wind and volume transport (contour lines), and the optimum wind location and direction (thick black arrow). The position of the Lancaster Sound mooring line is marked by the asterisk. The thin black arrow shows the optimum wind location and direction using monthly total wind and volume transport.

Fig. 12. Map showing correlation coefficients between monthly anomalies of NCEP surface wind and volume transport (contour lines), and the optimum wind location and direction (thick black arrow). The position of the Lancaster Sound mooring line is marked by the asterisk. The thin black arrow shows the optimum wind location and direction using monthly total wind and volume transport.

Sound and its correlation coefficient is double that of the local wind forcing. The optimum wind direction at the western entrance to the NW Passage (thin arrow, Fig. 12) is to the northeast (43°T), which is parallel to the coastline. Similarly, an optimum wind direction parallel to the western Newfoundland coastline was reported for flow in the Strait of Belle Isle (Garrett and Toulany 1981), and was explained by the winds causing sea level setup or setdown at the ends of the strait, producing a sea level difference from one end to the other.

However, the location of the maximum correlation is not well defined. Since some portion of the correlation may be simply due to the wind and transport having similar annual cycles, the analysis was repeated using monthly anomalies of transport and wind components, computed by subtracting the 8-year mean values for each month. For this second case, the correlation coefficient reduced slightly from 0.68 to 0.62 and remained off the western entrance of the Northwest Passage (thick arrow in Fig. 12). The optimum direction is toward 44°T, and the time series for the wind anomaly component along this direction is plotted in Fig. 13. The gains, monthly wind anomalies, and the mean annual cycle of transport were used to produce estimates of transport (Fig. 13). The annual cycle of transport represents 34% of the variance of total transport, and the wind anomalies combined with the annual cycle account for 59% of the variance of monthly transport.

Fig. 13. Alongshore component of monthly wind anomalies at 75° N, 125.0° W (relative to 44°T) and annual cycle plus the wind anomaly effect (bold line), and the observed volume transport (thin line).

x -|q5 Observed Volume Transport vs. Wind Regression Model

1981 1982 1983 1984 1985 1986 1987 1988 1989 1990

Fig. 14. Monthly transport estimates for the 1980s including the Barrow Strait observations (bold line) derived from the NCEP winds.

The regression coefficients can be used to simulate the transport through Lancaster Sound for other years prior to 1998. Figure 14 shows the monthly simulated transports back to 1980 when the first transport observations were available from Barrow Strait just west of the present mooring array site (Prinsenberg and Bennett, 1987).

x -|q5 Observed Volume Transport vs. Wind Regression Model

" "" " "

— Wind Anomaly + Seasonal Cycle

■ ¥ -U \-

1981 1982 1983 1984 1985 1986 1987 1988 1989 1990

Fig. 14. Monthly transport estimates for the 1980s including the Barrow Strait observations (bold line) derived from the NCEP winds.

The observations compare well with the wind estimates; they are smaller at this site because some of the transport flux coming out of the Sverdrup Basin east of Cornwallis Island (Fig. 1) is not included, as it is in the present mooring observations. Assuming that Barrow Strait contributes three-quarters of the transport of Lancaster Sound (Prinsenberg and Bennett 1987), the 1981-82 observations should be increased by one third.

The annual mean transport (August to July) from 1999 to 2006 (Fig. 15) is also significantly correlated with the alongshore wind (r = 0.88) based on eight points (years). The gain is 0.24 x 106 m2, and using the relationship between annual mean transport and wind (1999-2006), transport was estimated for the full extent of the NCEP dataset (Fig. 15). The lowest estimated transport value over the entire 58-year record corresponded to the year 1999, while 2001 had the fourth highest value. The extreme low was captured in our mooring observation period. Estimated transport was particularly high from 1989 to 1997, and may have contributed to the freshwater gain observed in the Labrador Sea during this period (Yashayaev 2007).

Year

Fig. 15. Top panel shows annual mean transport (August-September) estimated from alongshore winds (thin solid line) and observed transport (bold line). Bottom panel shows the Arctic Oscillation (solid line). The 5-year running means are shown by the dashed lines in both panels.

Year

Fig. 15. Top panel shows annual mean transport (August-September) estimated from alongshore winds (thin solid line) and observed transport (bold line). Bottom panel shows the Arctic Oscillation (solid line). The 5-year running means are shown by the dashed lines in both panels.

The Arctic Oscillation index, defined as the leading principal component of Northern Hemisphere SLP, is plotted in Fig. 15, and is averaged between August and July as for the transport estimates based on the alongshore wind. At any given point, the correlation between the along-shore wind component and the AO index is expected to depend on the loading pattern of the AO index. The correlation of alongshore wind at 75° N, 125° W with the AO index is 0.60. Not surprisingly, the AO and the alongshore wind show similar trends and inter-decadal variability, with high values seen in the early 1990s. However the AO does not capture the interannual variability of the observed transport over the 8-year mooring deployment (r = 0.21).

0 0

Post a comment