Scotian Slope circulation and eddy variability from TOPEX/Poseidon and frontal analysis data

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 109,, doi: /2003jc002046, 2004 Scotian Slope circulation and eddy variability from TOPEX/Poseidon and frontal analysis data Guoqi Han Northwest Atlantic Fisheries Centre, Fisheries and Oceans Canada, St. John s, Newfoundland, Canada Received 15 July 2003; revised 16 December 2003; accepted 15 January 2004; published 17 March [1] Sea level observations from TOPEX/Poseidon (T/P) altimetry, in conjunction with concurrent frontal analysis data from infrared imagery, are used to study temporal and spatial variability of sea-surface currents over the Scotian Slope. Geostrophic surface current anomalies normal to ground tracks are derived from the sea level anomalies relative to local means for the study period. The altimetric results reveal prominent spatial variability of cross-track currents with an overall intensification toward the west and south, typically cm/s over the upper and lower slope, cm over the continental rise, and up to cm/s near the Gulf Stream northern wall. A rapid westward intensification on the western slope contrasts with nearly uniform distribution on the eastern slope. The intensification seems to be associated with the high occurrence of Gulf Stream warm core rings (WCR) and with the close proximity to the Gulf Stream. Crossover analyses of altimetric current anomalies indicate that the total current variability can be estimated by a factor of 1.5 from the cross-track current variability and the variance is isotropic to within 14%. The rotational speed of the Gulf Stream rings can reach 1 2 m/s over the Scotian Slope, with the root-mean-square variability of 60 cm/s. The relative vorticity of the interior core is estimated to be /s. When combined with CTD data, T/P observations can produce the vertical profile of the total geostrophic current that is in approximate agreement with ADCP measurements. INDEX TERMS: 4512 Oceanography: Physical: Currents; 4520 Oceanography: Physical: Eddies and mesoscale processes; 4528 Oceanography: Physical: Fronts and jets; 4536 Oceanography: Physical: Hydrography; KEYWORDS: circulation, eddy, satellite altimetry Citation: Han, G. (2004), Scotian Slope circulation and eddy variability from TOPEX/Poseidon and frontal analysis data, J. Geophys. Res., 109,, doi: /2003jc Introduction [2] The Scotian Slope (Figure 1) and Rise about 200 km south of Nova Scotia features moderate mean currents and strong temporal and spatial variability [Smith and Petrie, 1982]. Major circulation features include an equatorward flow over the shelf edge and upper continental slope mainly composed of the colder and fresher water of Labrador Current origin and the Gulf of St. Lawrence outflow water and a northeastward current over the lower continental slope and rise carrying warmer and more saline water. The two currents are part of a broad slope water cyclonic gyre [Csanady and Hamilton, 1988]. Meanders and anticyclonic WCRs pinched off from the Gulf Stream often modify the slope water circulation. These rings generate significant temporal and spatial variability in currents [Joyce, 1991] and provide an important mechanism for shelf/deep-ocean exchange processes [e.g., Garfield and Evans, 1987]. [3] Slope water circulation off Nova Scotia was extensively investigated through water mass analyses in the 1950s [McLellan et al., 1953; McLellan, 1956, 1957]. Published in 2004 by the American Geophysical Union. McLellan et al. [1953] indicated that the slope water was composed of surface coastal waters from the Scotian Shelf, surface Gulf Stream waters, Labrador Shelf waters, and deep Atlantic waters upwelled under the Gulf Stream, with specific composition depending on depth. They also showed pronounced north-south excursions (300 km) of the slope water northern edge and highly variable width ( km) of the slope water region. Subsequently, McLellan [1956] reported that slope water could have extremely sharp boundaries with temperature gradients of as large as 3 C/m in the vertical and 1.5 C/100 m in the cross-isobath direction. Furthermore, McLellan [1957] estimated the contributions from source waters and the production rate of slope water (20 Sv) and reported that the slope water left the region as an eastward flow and the production occurred from the Tail of the Grand Bank to Georges Bank. [4] Smith and Petrie [1982] examined low-frequency responses of Scotian Slope currents to surface wind and offshore oceanic forcings. Longshore seasonal mean currents over the Scotian Slope are strongly influenced by topographic Rossby waves [Louis et al., 1982]. Csanady and Hamilton [1988] summarized the key features of Scotian Slope circulation: an inflow of the coastal Labrador Seawater across the Grand Banks and an isopycnal advec- 1of16

2 Figure 1. Map showing the Scotian Slope and adjacent shelf and deep oceans with a schematic representation of the circulation. The numerical labeled lines (thin dashed lines) are the selected T/P ground tracks on which the analysis is performed. The thick segment indicates the location of a hydrographic survey section. Also depicted are positions of the shelf/slope front (thick dashed lines) and the Gulf Stream northern boundary (dash-dotted line) for the period from 1992 to SEC: shelf-edge current; SWC: Slope Water current. LC: Laurentian Channel. Gulf Stream rings are depicted as ellipse. tion from the Gulf Stream thermocline, the total draining eastward. Han et al. [1993] used Geosat altimetry to study an annual cycle of geostrophic sea surface currents over the Scotian Shelf and Slope and found the shelf-edge current to be stronger in winter/fall. Pickart et al. [1999] investigated the mean structure of an eastward slope current and its interannual variability over the eastern Scotian Slope and the southwestern Newfoundland Slope. The velocity structure of the slope current varies with longitudes. The latitudinal position and the strength of the current vary on interannual scales. State-of-the-art eddy-resolving basinscale models are still experiencing difficulties in reproducing the slope water circulation [e.g., Smith et al., 2000] without assimilating data. [5] So far, there have been limited efforts on current variability over the Scotian Slope including both observational programs and theoretical studies, compared with those of the Gulf Stream and the adjacent shelf areas. Therefore, quantitative knowledge on the spatial and temporal variability of circulation in the slope water region is limited and dynamics underlying the circulation variability is poorly understood. Recent moored observations (J. Loder, personal communication, 2003) together with drifter studies [Fratantoni, 2001] have started a new phase of observing the slope water currents variability. Fratantoni [2001], from satellite-tracked surface drifters, constructed 1 gridded fields of velocity and eddy kinetic energy in the North Atlantic for the 1990s. In particular, his results indicate that the jet-like structure of the Gulf Stream remains well organized offshore of the Scotian Slope. However, the horizontal resolution and data density are insufficient for a continuous flow along the upper Scotian Slope that was hypothesized by Chapman and Beardsley and found in Han et al. s [1997] model. To complement these studies and to provide a new perspective of understanding the slope water circulation, we have used T/P altimeter data to study current variability. T/P satellite altimetry has been providing synoptic high-accuracy measurements of instantaneous sea surface heights since Over the Atlantic Canadian Shelf, T/P altimeter data were used to help determine the open boundary condition for a regional ocean model [Han et al., 1999] and to study seasonal sea level and surface current variability [Han et al., 2002]. [6] The primary purpose of this study is to use T/P satellite altimeter data for quantifying and understanding mesoscale surface current variability over the Scotian Slope. In conjunction with frontal analysis data [Drinkwater et al., 1994] and hydrographic data, this manuscript describes surface current statistics and discusses major current features including WCRs in this region. Knowledge of currents and their variability is also useful to offshore oil and gas activities in the deep waters of the Scotian Slope. [7] In section 2, we briefly describe techniques of the T/P data processing, methods of deriving altimetric currents, and frontal analysis data. Section 3 presents sea level and current statistics and features. A detailed examination of 2of16

3 the Gulf Stream rings is given in section 4. Section 5 discusses limitations pertinent to altimetric current estimation and potential solutions, and section 6 provides a brief summary. 2. Methodology 2.1. T/P Altimetric Heights and Currents [8] We used corrected T/P sea-surface height data for the period from August 1992 to March 2000, obtained from NASA Pathfinder Project. Four T/P descending tracks and three ascending tracks were selected across the Scotian Slope and Rise off Nova Scotia (Figure 1). The satellite has a nominal repeat cycle of 10 days, and there are 276 observations at each location. The along-track resolution is about 6 km. The data were corrected based primarily on the principles of Benada [1997] for various atmospheric and oceanographic effects, as outlined by Han et al. [2002]. [9] A mean sea surface for the study period was constructed from the available T/P data. We then calculated the sea surface height anomalies relative to the mean sea surface. Both the marine geoid and mean oceanic topography are removed by this procedure. An along-track secondorder Butterworth filter with a width of 36 km (seven along-track points) was performed to reduce noise influences on the current estimates. The smoothing may also remove some geophysical variability. The smoothing parameters were chosen somewhat subjectively, in consideration of the characteristic cross-isobath scale of the shelfedge and slope currents. The T/P data have a nominal accuracy of 2 cm for the time-varying component. A crossover analysis at the six crossovers (see Figure 1 for locations) estimated a root-mean-square (RMS) error of 2 cm associated with the smoothed sea level anomalies. The results presented will be based on the smoothed height data unless indicated otherwise. The sea level anomalies for Track 071 are shown in Figure 2 for August November [10] From the T/P sea surface height anomalies, geostrophic surface current anomalies normal to the track were derived, as shown in Figure 2 (positive westward) for Track 071 for August November An analysis of the effects of the along-track smoothing for Track 071 indicated that it removed 13% of the overall variance of the unsmoothed data in the time domain (temporal variations at a fixed along-track location) or 5% in the space domain (alongtrack variations at a fixed repeat cycle). The RMS error of altimetric geostrophic current anomalies is estimated to be 5 cm/s (based on the RMS error of 2 cm for the sea level anomalies and the filter width of 36 km). Note that the above derived altimetric currents are estimates of surface current anomalies normal to the satellite ground tracks about the mean only, associated with the along-track pressure gradient derived from the slope of sea surface. Therefore neither local wind-driven Ekman flows, along-track current anomalies, or mean currents are included. [11] An approximate way of constructing absolute surface currents is to combine altimetric surface current anomalies with a mean circulation field from a numerical ocean model. In this study we have used climatological annual-mean currents from Han et al. s [1997] diagnostic finite element model solutions available inshore of the 4500-m isobath. The model surface currents are interpolated onto the satellite ground tracks. The components normal to the track were then derived. The major shelf-scale features include a southwestward flow along the shelf edge and the upper continental slope and a northeastward current along the lower continental slope. The model southwestward shelfedge current is broader with a maximum of cm/s in the east and narrower with a maximum of cm/s in the west. A comparison with moored measurements indicates approximate agreement. The model northeastward slope current is generally broad with a maximum of cm/s, consistent with Fratantoni s [2001] drifter estimates that are scarce over the Scotian Slope. At the 55 W the mean slope current has a maximum of 8 cm/s [Pickart et al., 1999]. Nevertheless, it is likely that the model underestimates the mean slope current since it used false bottom topography offshore of the 1000-m isobath Frontal Analysis Data [12] Quantitative information on the location of surface temperature fronts, such as shelf/slope front (separating cool shelf water from the warmer slope water immediately offshore), Gulf Stream WCRs, and Gulf Stream northern boundary (separating the stream from the slope water), is obtained for the Scotian Slope and adjacent deep oceans based on satellite infrared imagery [e.g., Drinkwater et al., 1994]. Frontal positions have been digitized from Oceanographic Features Analysis charts published by NOAA up to the end of September 1995 (K. Drinkwater and R. Pettipas, personal communication, 2001). Starting in April 1996, the charts have been referred to as Jennifer Clark s Gulf Stream ( No data are available from October 1995 to March After digitization, the positions of the Gulf Stream northern boundary and the shelf/slope front were averaged at each degree of longitude for each chart. [13] Satellite infrared imagery has been used to study the Gulf Stream rings and the shelf- and deep-water exchanges in the slope water region off northeast America [e.g., Halliwell and Mooers, 1979; Brown et al., 1986; Garfield and Evans, 1987]. In the present study we will focus on the contribution of WCRs on the Scotian Slope current variability and the ability of T/P satellite altimetry in detecting WCRs and providing information on their rotational speed and relative vorticity. 3. Current Statistics [14] The sea surface height anomalies on Track 071 (Figure 2) show significant along-track variations in the slope water region (39 N 42 N), where the magnitude of the positive sea level anomalies was as high as 0.7 m from August to November 1999, with an along-track scale of about 200 km. The positive sea level anomalies are apparently associated with fluctuations in the Gulf Stream position and occurrence of WCRs. The associated current anomalies (normal component) often have magnitudes of order 1 m/s, particularly south of 42 N. Anticyclonic ring circulation apparently occurred in September October 1999, indicated by the negative (eastward) current anomalies around 41 N between September 11 and October 20. A slow southwestward movement of the ring center can be 3of16

4 Figure 2. Along-track profiles of the smoothed T/P sea level anomalies (thin line) and associated crosstrack geostrophic surface current anomalies (thick line, positive westward) for track 071 from August to November The two bottom panels show the along-track profile of bottom topography. 4of16

5 Figure 3. T/P current anomalies at nominal 1000-m (thin line) and 4000-m (thick line) water depths on the four descending tracks. (a) Time series. (b) Power spectra. 5of16

6 Figure 3. (continued) 6of16

7 Figure 4. Sea level variability over the Scotian Slope calculated from the T/P altimetry data from 1992 to 2000 and plotted as twice the RMS values. Also depicted are positions of the shelf/slope front (thick dashed lines) and the Gulf Stream northern boundary (dash-dotted line) for the period. inferred. Substantial sea level and current variations over the shelf edge and upper continental slope (from the 200- to 1000-m isobaths) are also evident north of 42 N. [15] There are significant mesoscale (100 day) variations in the current anomalies, as seen in time series of the anomalies near the and 4000-m isobaths and their power spectra (Figure 3) on the descending tracks that are approximately normal to the shelf edge (Figure 1). The seasonal changes are also evident. On the seasonal scale and mesoscale the currents variability for Tracks 12 and 50 at the 4000-m water depth is much larger than that at 1000-m depth, while differences for Tracks 88 and 126 are smaller. This is consistent with the increasing distance of the Gulf Stream from the continental slope as one proceeds eastward from Georges Bank toward Grand Bank. The largest offshore-inshore contrast in the current anomalies is found on Track 050. Substantial interannual variations are also indicated at the 4000-m water depth for Track 050. Note that variability around the M2 aliasing period of 62 days near the 1000-m isobath for track 012, 088, and 126 may be associated with the error of the ocean tide correction. [16] The RMS values of the altimetric sea level anomalies (Figure 4) indicate generally increased variability as one proceeds offshore toward the Gulf Stream. Typical values are 5 10 cm over the upper slope, 10 cm over the lower slope (from the to 3000-m isobaths), and cm over the continental rise (from the to 4500-m isobaths). The magnitude increases westward, particularly over the continental rise. Near the center of the Gulf Stream the RMS sea level variability is about 50 cm. The maximum sea level variability is located about 100 km south of the Gulf Stream northern boundary, showing little along-stream difference. [17] The spatial distribution of RMS values of the crosstrack currents on the seven tracks (Figure 5) show substantial current variability over the Scotian Slope. Overall, the RMS values increase westward and offshore. However, on Tracks 126 and 088, the RMS values at the shelf break are large relative to the ones just offshore, which may be associated with the shelf-edge current whose strength and variability decreases west of Track 088. The westward intensification is more evident over the western Scotian Slope, and the offshore enhancement occurs mainly between the shelf/slope front and the Gulf Stream northern boundary. Typical RMS values of the cross-track current anomalies are cm/s over the upper and lower slope and cm/s over the continental rise. The maximum cross-track current variability, cm/s, is located close to the Gulf Stream northern boundary except for Track 050. [18] The westward and offshore intensification is apparently associated with the proximity of the Gulf Stream to the 7of16

8 Figure 5. Twice RMS values of the altimetric current anomalies (subsampled spatially) from 1992 to 2000 (thick shaded lines). The thin lines represent twice the RMS values when Han et al. s [1997] model means are included. Also depicted are positions of the shelf/slope front (thick dashed lines) and the Gulf Stream northern boundary (dash-dotted line) for the period. shelf edge (Figure 5). On the other hand, the Gulf Stream rings are known to be important to the slope water circulation variability. We have calculated the percentage occurrence of WCRs for a selected area as the ratio of the number of days when a ring s average radius is 90 km or larger and its average center is located inside the area to the number of total days. We also calculated the percentage occurrence with average radii of >60 and 120 km, respectively, and found that the relative occurrence among the four areas is insensitive to the choice of the radius. Therefore the westward intensification may also be attributed to increased WCR activity to the west (Figure 6). [19] To provide further statistical information on the spatial structure of the current anomalies over the Scotian Slope, four along-track bands were selected for the four descending tracks based on bathymetry: (1) m, (2) m, (3) m, and (4) m. Percentage occurrence and RMS values for the four bands are calculated. Overall, the variability is strongest with higher occurrence of extreme currents on Track 012 off Northeast Channel and Georges Bank, and weakest with lower occurrence of extreme currents on Track 126 off Sable Island Bank (Table 1). The RMS values for the four bands for each track also demonstrate the decrease in anomaly strength with distance from the Gulf Stream. [20] The inclusion of Han et al. s [1997] model mean currents (Figure 6) does not change the statistics significantly, especially over the lower slope, suggesting the eddy kinetic energy is overwhelmingly dominant over the mean kinetic energy for the Scotian Slope surface circulation. Nevertheless, it is likely that inclusion of a refined model mean with a more realistic representation of the Gulf Stream and associated currents would result in quantitative changes in the current statistics, particularly offshore of the 1000-m isobath. 4. Gulf Stream Rings [21] Earlier studies have revealed statistical properties of Gulf Stream WCRs, such as mean loci, lifetime, size, and movement in the Slope Water region off the northeast United States east coast [e.g., Brown et al., 1986]. The role of WCRs in entraining shelf water off Georges Bank was also examined [e.g., Garfield and Evans, 1987]. Satellite altimetry provides unique opportunity of inferring the surface rotational speed associated with WCRs. [22] Geostrophic current anomalies derived from T/P altimetry indicate anti-cyclonic eddies over the Scotian Slope from time to time. These mesoscale eddies are apparently associated with warm patches of surface water 8of16

9 Figure 6. Percentage occurrence of WCRs over the Scotian Slope based on the frontal analysis data (K. Drinkwater and R. Pettipas, personal communication, 2001) from 1992 to A total of 989 days was included in the analysis. Also depicted are positions of the shelf/slope front (thick dashed line) and the Gulf Stream northern boundary (dash-dotted line) for the period. seen from infrared images. When there are WCRs indicated in the frontal analysis data, the altimetric data also show anticyclonic rings (eddies) in the same location (Figure 7). It can be seen that the interior core of the ring rotates like a solid body. The geostrophic current anomalies associated with these rings can be over 1 m/s. Therefore, altimetric data not only confirm the occurrence of the Gulf Stream rings revealed from the infrared images, but also provide important quantitative information about associated currents. [23] On the basis of the frontal analysis data, we have traced two rings (see Figure 7 for their locations on respective dates) over the Scotian Slope for the period from November 1999 to February The center of a ring is calculated by averaging geographical coordinates (longitude and latitude) of the points defining the edge of ring in the frontal analysis data. The ring s radius is estimated by averaging distances between the points and the center. The results (Figure 8) indicate that the two rings, with a diameter of 200 km, slowly drifted westward (2 3 km a day) and lasted a few months. A detailed statistical analysis further indicates the radii for all WCRs vary from 20 to 150 km with a mean value of 80 km and a median value of 70 km over the Scotian Slope for the entire study period, which is consistent with Brown et al. s [1986] results for the short-lived ring formation between 66 W and 60 W. [24] Combining the T/P current and frontal analysis data, we have identified 24 snapshots of WCRs over the Scotian Slope for 1999, each of which has an average radius of greater than 75 km, with one of the seven tracks closely passing by its center. For each WCR, the relative vorticity associated with its interior core is derived and the RMS current variability within the ring perimeter is calculated from the T/P cross-track current anomalies on the passing track. The relative vorticity ranges from to /s with a mean value of /s, consistent with Joyce s [1984] estimate for a single ring. The RMS current variability varies from 34 to 80 cm/s with an average of 59 cm/s. There appears little correlation between the radius and the relative vorticity or between the radius and the RMS current variability. Table 1. Percentage Occurrence of T/P-Derived Cross-Track Current Anomalies Depth, m Range, m/s Track 012 Track 050 Track 088 Track > % 15.2% 15.7% 12.8% > % 0.7% 0.6% 0.9% > % 0.0% 0.0% 0.0% > % 0.0% 0.0% 0.0% > % 17.6% 17.5% 12.6% > % 2.3% 1.4% 0.7% > % 0.0% 0.0% 0.0% > % 0.0% 0.0% 0.0% > % 24.7% 16.1% 17.6% > % 4.8% 2.1% 0.9% > % 0.4% 0.4% 0.0% > % 0.1% 0.1% 0.0% > % 32.5% 24.0% 20.7% > % 10.6% 4.5% 2.4% > % 0.5% 0.3% 0.0% > % 0.0% 0.2% 0.0% 9of16

10 Figure 7. Frontal analysis data for December 1999, showing ring features (solid lines) and the positions of the shelf/slope front (thick dashed lines) and the Gulf Stream northern boundary (dash-dotted lines) and selected altimetric current anomalies (arrows). [25] On September 27 and 28, 1999, acoustic Doppler current profiler (ADCP) and conductivity-temperaturedepth (CTD) data were collected on a section across a Gulf Stream/WCR front on W off the southwest Nova Scotia [Smith et al., 1999]. The near-surface currents from ADCP data are presented in Figure 9. The ADCP section covered only half of the northern side of the WCR only, but recorded a maximum speed of about 2 m/s associated with the WCR. The change of direction of the ADCP current at N is consistent with the northern edge of the WCR as indicated in the frontal analysis data. The best T/P observations of the ring are from Track 071 on September 30, 1999, approximately passing along its major axis. The interior core of the ring rotated like a solid body with an estimated relative vorticity of /s. The maximum speed is about 1.5 m/s, located approximately 60 km to the northeast of the rotational center. To evaluate altimetric currents against ADCP observations, we have derived the ADCP current components normal to Track 071. The maximum cross-track speed, located at W and N, is about 1.8 m/s from the ADCP data. The maximum cross-track geostrophic current speed is 1.5 m/s from the T/P data, in fair agreement with the ADCP measurements. We can also see consistency of the location of the maximum speed between the altimetry and ADCP data. There are increased discrepancies toward the northern edge of the ring, probably attributed to the increased location mismatch and apparent ellipticity of the ring. It is worthwhile to point out that the present comparison is valid in spite of time differences of 2 days, since the center of the WCR moved very slowly. [26] Figures 10a and 10b are the salinity and temperature distribution from CTD data, which clearly shows that the temperature of WCR water is above 18 C and the salinity is 3 units greater than the ambient near-surface water. The CTD data indicate the shelf/slope front at 41.1 N, which is close to that suggested in the frontal analysis data. The steric height calculated from the temperature and salinity data relative to the 500 dbar is in good qualitative and approximate (within 0.1 m) quantitative agreement with the altimetric sea level anomalies on Track 12 (Figure 10c; note that they are translated to be equal at the intersection). Both show rapid increases of sea level toward the center of the WCR (associated with strong eastward currents) and slight rises toward the north (associated with substantial westward flows) from its northern edge. [27] From the steric height, we have derived geostrophic surface currents relative to the 500 dbar (Figure 11). The 10 of 16

11 Figure 8. Longitudinal locations and radius of two WCRs based on the frontal analysis data. Rings A and B were located on the western and eastern ends in Figure 7, respectively. The dashed lines are least squares fit to the data. 11 of 16

12 Figure 9. Comparison of the altimetric cross-track current anomalies (thin arrows) on Track 071 on September 30, 1999, and ADCP data (thick arrows) on September 27 28, Also depicted are positions of the Gulf Stream rings (solid lines), the shelf/slope front (thick dashed lines), the Gulf Stream northern boundary (dash-dotted lines), and satellite-measured sea surface temperature ( C) distribution (shaded image) on September 27, of 16

13 Figure 10. CTD temperature and salinity distribution across the ADCP section and comparison of the steric height relative to the 500 dbar and the T/P sea level anomalies on Track 12. See Figure 1 for location. 13 of 16

14 Figure 11. Comparison of the geostrophic surface currents relative to the 500 dbar (thick arrows) and ADCP cross-section current components (shaded arrows) on September 27 28, Also depicted are positions of the Gulf Stream rings (solid lines) and the shelf/slope front (thick dashed lines) on September 27, location of the maximum geostrophic speed is close to that of the maximum ADCP current, but the magnitude of about 0.9 m/s is much smaller than the ADCP and T/P maxima (1.8 and 1.5 m/s, respectively). The significant underestimation can in part be attributed to the uncertainty associated with the assumed level (500 dbar) of no motion. Across the WCR s northern edge and shelf/slope front the relative vorticity of the geostrophic current is comparable to that of the cross-sectional component of the ADCP current. However, north of the shelf-slope front the geostrophic current is in the opposite direction to that the ADCP current. 5. Discussion [28] The geostrophic current anomaly normal to the T/P ground track is expected to be an underestimate of the total geostrophic current anomaly due to the neglect of the alongtrack current component (dependent on sea surface anomalies normal to the track). Under the assumption of isotropy, the RMS values for the total geostrophic current anomaly can be estimated as those for the cross-track component multiplied by a factor of 1.4. [29] T/P data at crossovers of descending and ascending tracks allow us to estimate total RMS values of current anomalies at those locations. We have interpolated spatially and temporally geostrophic current anomalies normal to descending and ascending tracks to generate time series at crossovers (see Figure 1 for locations). The normal-to-track components are then transformed into the eastward and northward components. The total RMS current value is the square root of the sum of the mean square values. We have also calculated the ratio of the RMS variability for the cross- Table 2. RMS Values at Six Crossovers (See Figure 1 for Locations) for Cross-Track and Total Current Anomalies and the Ratios of the Former to the Latter a Location Descending, cm/s Ascending, cm/s Total, cm/s D/T A/T A B C D E F Mean ± SD 0.66 ± ± 0.03 a D/T and A/T are ratios for descending and ascending tracks, respectively; SD, standard deviation. 14 of 16

15 Table 3. Major and Minor Axes, Orientation of the Major Axis (degree, Positive Anticlockwise From the East), and Ratio of the Minor to Major Axes of T/P Current Anomalies at Six Crossovers (See Figure 1 for Locations) a Location Major, cm/s Minor, cm/s Orientation Ratio A B C D E F Mean ± SD 0.37 ± 0.11 a SD, standard deviation. track components to that for the total current (Table 2). We can see that the total RMS variability ranges from 40 to 50 cm/s at the offshore crossovers (A, B, C), while that is at cm/s at the crossovers near the shelf edge. The ratios vary from 0.58 to 0.74, with an average of There appears to be no systematic difference in ratios between the descending and ascending tracks. Therefore the total current variability over the Scotian Slope may be estimated by a factor of 1.5 from the cross-track current variability of either the descending or ascending tracks presented in section 3. [30] From the eastward and northward current components, we have derived major and minor axes, orientation of the major axis, and ratio of the minor to major axes. The ratios at the six crossovers vary from 0.28 to 0.59 with an average of 0.37 (Table 3). Therefore, on average, the variance of T/P altimetric current anomalies is isotropic to within 14% over the Scotian Slope. The dominant direction of variance is in the east-west direction. The variance seems to be more isotropic over the Scotian Shelf. However, 1 by 1 drifter data indicate that the variance offshore of the shelf/slope front is more isotropic with less dominant direction of velocity variance [Fratantoni, 2001]. [31] As demonstrated in the preceding sections, the altimetry data can be used directly to quantify the near-surface geostrophic current variability associated with anomalies in sea surface slope. Moreover, estimates of geostrophic currents at depth could be obtained by combining altimeter data with appropriate hydrographic data when available [Han and Tang, 1999]. For example, a CTD section coincident with a satellite ground track would allow inference of deep ocean currents. [32] We have chosen a location (65.48 W, N) on Track 071 that is close to an ADCP location (65.50 W, N). Combining the density calculated from the CTD data and the T/P surface currents, we have estimated a vertical profile of the cross-track geostrophic currents using the thermal wind relationship in which the vertical shear of current is balanced by horizontal gradient of density. The sea surface is used as the level of known motion inferred from T/P altimetry data. A comparison with ADCP currents shows fair agreement (Figure 12). For the top 250-m water column, the T/P-CTD estimate can account for more than 80% of the ADCP current. The remaining discrepancy is not surprising due to not only the geostrophic approximation but also the location and time mismatch and other assumptions. Nevertheless, the significance of this present evalua- Figure 12. Vertical profiles of cross-track current components at a Track 71 location. 15 of 16

16 tion is to have demonstrated the feasibility and robustness of combining altimetry measurements with hydrographic observations to quantify currents associated with WCRs. It is expected that careful planning can realize a better temporal and spatial sampling scheme of density to match altimetric measurements and therefore lead to improved results. 6. Summary [33] We have used T/P altimeter data for the period from 1992 to 2000 to study sea level and current variability over the Scotian Slope, in conjunction with frontal analysis data and hydrodynamic model solutions. The present results reveal significant current variability associated with the slope water circulation. [34] In general, the altimetric current anomaly variability increases westward and offshore. The increased current variability on the western Scotian Slope seems associated with both high occurrence of the anticyclonic rings and the close proximity of the Gulf Stream. Typical RMS values of the cross-track current anomalies are cm/s over the upper and lower slope and up to 50 cm/s over the western continental rise. The maximum cross-track current variability, about 75 cm/s is located close to the Gulf Stream northern boundary. [35] T/P data show that the Scotian Slope features frequent occurrence of the Gulf Stream WCRs, which usually drift westward and last several months. The altimetry reveals that the rotational speed of anti-cyclonic WCRs can exceed 1 m/s and may reach 2 m/s. Comparisons have demonstrated the consistency of the radar altimetry and infrared imagery in detecting and positioning rings and altimetry s unique ability in inferring current speeds associated with the rings. The RMS current variability and vorticity are estimated to be 60 cm/s and /s, respectively. [36] Analyses of T/P along-track data provide geostrophic cross-track surface current anomalies only. Nevertheless, the present study demonstrates feasibility of constructing absolute current fields by adding a model mean circulation field, deriving full geostrophic surface currents at crossovers of ascending and descending satellite ground tracks, and inferring subsurface geostrophic currents by combining altimetry with hydrography. The inclusion of the model mean current does not increase the RMS current values significantly, except over the shelf edge, which suggests that the eddy kinetic energy is overwhelmingly dominant over the mean kinetic energy for the Scotian Slope surface circulation. The crossover analysis indicates the altimetric current variance is isotropic to within 14% over the Scotian Slope. The total current variability over the Scotian Slope can be estimated by a factor of 1.5 from the cross-track current variability of either the descending or ascending tracks. [37] Acknowledgments. I thank J. Li for assistance in data analyses, R. Pettipas for providing the frontal analysis data, and P. C. Smith for providing CTD and ADCP data. J. Loder and P. C. Smith provided useful internal reviews. Constructive comments and suggestions were received from the two anonymous reviewers. The project was funded through the Offshore Environmental Factor Program of the Federal Program for Energy, Research and Development (PERD), the Climate Change Impacts on the Energy Sector of PERD, the Canadian Space Agency, the Canadian Hydrographic Service, and Transport Canada. T/P data were obtained from NASA Jet Propulsion Lab and Pathfinder Project. References Benada, R. (1997), Merged GDR (TOPEX/Poseidon) User Handbook, JPL D-11007, Jet Propul. Lab., Pasadena, Calif. Brown, O. B., P. C. Cornillon, S. R. Emmerson, and H. M. Carle (1986), Gulf Stream warm rings: A statistical study of their behavior, Deep Sea Res., 33, Csanady, G. T., and P. Hamilton (1988), Circulation of slopewater, Cont. Shelf Res., 8, Drinkwater, K. F., R. A. Myers, R. G. Pettipas, and T. L. Wright (1994), Climatic data for the northwest Atlantic: The position of the shelf/slope front and the northern boundary of the Gulf Stream between 50 W and 75 W, , Can. Data. Rep. Fish. Ocean Sci. 12, 103 pp. Fratantoni, D. M. (2001), North Atlantic surface circulation during the 1990s observed with satellite-tracked drifters, J. Geophys. Res., 106, 22,067 22,094. Garfield, N., and D. L. Evans (1987), Shelf water entrainment by Gulf Stream warm-core rings, J. Geophys. Res., 92, 13,003 13,012. Halliwell, G. R., Jr., and C. N. K. Mooers (1979), The space-time structure and variability of the shelf water-slope water and Gulf Stream surface temperature fronts and associated warm-core eddies, J. Geophys. Res., 84, Han, G., and C. L. Tang (1999), Velocity and transport of the Labrador Current determined from altimetric, hydrographic, and wind data, J. Geophys. Res., 104, 18,047 18,057. Han, G., M. Ikeda, and P. C. Smith (1993), Annual variation of sea-surface slopes over the Scotian Shelf and Grand Banks from Geosat altimetry, Atmos. Ocean, 31, Han, G., C. G. Hannah, P. C. Smith, and J. W. Loder (1997), Seasonal variation of the three-dimensional circulation over the Scotian Shelf, J. Geophys. Res., 102, Han, G., J. W. Loder, and P. C. Smith (1999), Seasonal-mean hydrography and circulation in the Gulf of St. Lawrence and eastern Scotian and southern Newfoundland Shelves, J. Phys. Oceanogr., 29, Han, G., C. L. Tang, and P. C. Smith (2002), Annual variations of sea surface elevations and currents over the Scotian Shelf and Slope, J. Phys. Oceanogr., 32, Joyce, T. M. (1984), Velocity and hydrographic structure of a Gulf Stream warm-core ring, J. Phys. Oceanogr., 14, Joyce, T. M. (1991), Review of U.S. contributions to warm-core rings, Rev. Geophys., 29, Louis, J. P., B. D. Petrie, and P. C. Smith (1982), Observations of topographic Rossby waves on the continental margin off Nova Scotia, J. Phys. Oceanogr., 12, McLellan, H. J. (1956), On the sharpness of oceanographic boundaries south of Nova Scotia, J. Fish. Res. Board Can., 13, McLellan, H. J. (1957), On the distinctness and origin of the slope water off the Scotian Shelf and its easterly flow south of the Grand Banks, J. Fish. Res. Board Can., 14, McLellan, H. J., L. Lauzier, and W. B. Bailey (1953), The slope water off the Scotian Shelf, J. Fish. Res. Board Can., 10, Pickart, R. S., T. K. McKee, D. J. Torres, and S. A. Harrington (1999), Mean structure and interannual variability of the slopewater system south of Newfoundland, J. Phys. Oceanogr., 29, Smith, P. C., and B. Petrie (1982), Low frequency circulation at the edge of Scotian Shelf, J. Phys. Oceanogr., 12, Smith, P. C., R. Boyce, and L. Petrie (1999), Report on C.S.S. Parizeau Cruise , report, Bedford Inst. of Oceanogr., Dartmouth, N. S., Can. Smith, R. D., M. E. Maltrud, F. O. Bryan, and M. W. Hecht (2000), Numerical simulation of the North Atlantic Ocean at 1/10, J. Phys. Oceanogr., 30, G. Han, Northwest Atlantic Fisheries Centre, Fisheries and Oceans Canada, P.O. Box 5667, St. John s, NL A1C 5X1, Canada. (hang@ dfo-mpo.gc.ca) 16 of 16