NOAA Tech Memo 170

Three cruises were completed off the northeastern United States during late 1981 and 1982 in cooperation with the National Science Foundation-sponsored Warm Core Rings Program in order to study entrainment of shelf water and associated biota by Gulf Stream warm-core rings (i.e., pinched-off meanders of the Gulf Stream which form clockwise/anticyclonic eddies). Hydrographic data were collected with a conductivity-temperature-depth recorder on board the NOAA R/V Albatross IV during 22 September - 6 October 1981, 19 April - 4 May 1982, and 17 June - 2 July 1982. These hydrographic data defined the "streamers" of entrained shelf water which were associated with Rings 81-E, 82-A, and 82-B twice, and formed the basis for the present results and discussion.

The amount of shelf water, delineated by salinity < 34 practical salinity units, varied from 58 km³ in a nearly isolated bowl associated with Ring 82-A, to 290 km³ associated with the second visit to Ring 82-B. Geostrophically calculated transports across sections of the streamers ranged from 488 x 10³ m³ sec^-1 in the same direction as the azimuthal velocity for Ring 82-B, to 345 x10³ m³ sec^-1 in the opposite direction as the azimuthal velocity for Ring 82-A.

Although each streamer was different in shape and size, a saddle point - a portion of the streamer with a pronounced cross-sectional narrowing - was present within its structure. This saddle point appears to be related to the inshore periphery of a counterclockwise/cyclonic eddy that develops near the head of the streamer.

Analysis of water properties leads to the conclusion that the source of shelf water for streamers is primarily seaward of the cold band, present during the stratified season (i.e., April-September), at water depths approximately 100 m. Therefore, passage of warm-core rings is not likely to have a major direct effect on the biota of the continental shelf, including larvae of commercially important fish species found between Cape Hatteras and Nova Scotia, since their distributions are generally concentrated at shallower depths.

Understanding variation of biological components is a focus of the ecological studies of the Northeast U.S. Continental Shelf Ecosystem (NCSE). For populations of commercially important species in the NCSE, a major source of that variation is recruitment - the number of individuals in the population that annually survive and grow to reach a catchable size. Recruitment in commercial fish populations is highly variable, and the causes of mortality that limit recruitment are not well understood (Hennemuth et al. 1980).

Advection of eggs and larvae from a preferred habitat has long been hypothesized as a large source of prerecruit mortality in commercial fish populations (Iles and Sinclair 1982), capable of causing occasional catastrophic losses (Walford 1938). One source for advective loss of shelf water from the NCSE is the interaction of warm-core rings (WCRs) with shelf water at the front separating shelf water and slope water (i.e., the shelf-slope front or shelfbreak front). Several authors describe WCRs and associated processes; see the review of Richardson (1983) and the special series of papers introduced by Joyce (1985). Garfield and Evans (1987) document statistics for the entrainment of shelf water, mainly derived from satellite thermal imagery over the 1979-85 period.

Morgan and Bishop (1977) found a feature with salinity <34‰ (i.e., a conservative, unambiguous indicator of shelf water) at least 90 km from the edge of the continental shelf and to the east of a WCR. However, the precise extent of the shelf water feature was not determined due to the sparse distribution of hydrographic stations. They concluded that a transport of 8.9 x 10³ m³ sec^-1 was likely for water with salinity <34‰.

Smith (1978) speculated on the interaction between a WCR well south of the Scotian Shelf and fluxes of heat and salt onto the shelf below 50 m. A volume of entrained shelf water on the eastern and western sides of the ring was calculated as 2500 km³, based on an analysis of an image derived from an advanced very-high-resolution radiometer (AVHRR) on board NOAA's polar-orbiting, environmental-sensing satellites, and on an assumed depth of 50 m for the entrainment. However, shelf water found west of the ring was probably caused by a second ring located farther toward the west.

Fornshell and Criess (1979) provided additional evidence for offshore flow from the continental shelf, finding water with proper temperature and salinity characteristics on the south side of a ring, and patches of similar water on the southeast side of it.

Bisagni (1983) reported on a direct measurement of flow from two satellite-tracked buoys, each equipped with a drogue at 10 m depth and placed at the 106-Mile Dumpsite (39°N, 72°W). The buoys first moved southwest at <25 cm sec^-1, then anticyclonically at up to 54 cm sec^-1 around WCR 80-A. (The nomenclature used in this paper for WCRs follows the scheme of the National Marine Fisheries Service's (NMFS's) Narragansett (Rhode Island) Laboratory: the two-digit year of formation is given first, and the alphabetical sequence of formation in that year is given second.) Interpretation of thermal images from satellites showed that these buoys were within shelf water, but always well seaward of the shelfbreak. Bisagni (1983) estimated the transport of the feature as 1.5 x 10⁵ m³ sec^-1, assuming a mean speed of 25 cm sec^-1, depth of 50 m, and width of 12 km.

During the U.S. Department of Energy's Shelf Edge Exchange Processes Experiment, WCR 83-D passed through a linear array of moored instruments across the continental shelf and slope south of Marthas Vineyard (Churchill et al. 1986). Based on temperature and salinity data from a mooring located at a 125-m depth, and on an expendable bathythermograph (XBT) section farther to the east that crossed the eastern half of the ring about 2 wk later, Churchill et al. (1986) hypothesized that subsurface water from the cold band encircled the WCR at approximately 50 m. The minimum temperatures of the water were <10°C over the shelf and <14°C at the southern end of the ring. Unfortunately, there were no salinity data to distinguish between shelf water and slope water at the southern end. There was slope water with similar temperatures nearby on the contoured section.

These various measurements of "streamers" of shelf water associated with WCRs (i.e., fingers of shelf water located at the trailing edge of WCRs as they move along the edge of the continental shelf towards Cape Hatteras) were mainly fortuitous. However, the National Science Foundation sponsored a specific program - the Warm Core Rings Program (WCRP) - to study WCRs (Warm-Core Rings Executive Committee 1982; Joyce and Wiebe 1983). The WCRP provided a unique opportunity to study cooperatively the interaction of WCRs with shelf water and associated biota. One purpose for NMFS involvement in the WCRP was to determine if the passage of a WCR could lead to loss of early life stages of major commercial fish populations. Such determination would require a description of the physical and biological characteristics, as well as the kinematics, of streamers. A series of interdisciplinary cruises, in conjunction with the WCRP, was completed between September 1981 and September 1982 in order to collect the necessary data.

This paper focuses on a description of the physical characteristics of streamers, as determined primarily from the oceanographic data. These data, even though collected during 1981 and 1982, represent the only extensive collection of in-situ information principally directed toward entrainment of shelf water by WCRs. First, the temperature and salinity characteristics, volume, and spatial structure of streamers of entrained shelf water are described. Then, the kinematics of the streamers are discussed, showing the variability of streamers in time and space. Finally, the primary source for shelf water found in streamers is directly substantiated.

Three cruises on board the NOAA R/V Albatross IV extensively sampled shelf water streamers associated with WCRs: 1) 22 September - 6 October 1981, associated with Ring 81-E (initially mislabeled as Ring 81-D (see Fitzgerald and Chamberlin (1984) for an explanation), and appearing as such in Joyce et al. (1984)); 2) 19 April - 4 May 1982, associated with Rings 82-A and 82-B; and 3) 17 June - 2 July 1982, associated again with Ring 82-B. The locations of all stations for the three cruises are shown in Figure 1. The station numbers for those stations are shown in Figures 2a-2c. Only a portion of the data collected during the series of cruises is used here.

Sea surface temperatures, derived from the NOAA-6 and NOAA-7 polar-orbiting, environmental-sensing satellites, were used extensively to guide the sampling at sea. Graphical interpretations of the processed, high-resolution images were transmitted from the University of Miami to the research vessel via the Application Technology Satellite III. A compilation of all images for Ring 82-B is found in Evans et al. (1984). Colder temperatures and sharp fronts delimited the shelf water. Later analysis of the images ashore aided interpretation of the in-situ measurements presented here.

Temperatures and salinities were sampled with a continuously recording conductivity-temperature-depth instrument (CTD), and with a surface thermosalinograph. Reporting of salinity follows the Practical Salinity Scale, and is expressed as practical salinity units (PSU). At most stations, a CTD cast was completed to 1000 m or near the ocean bottom when the depth was <1000 m. Sometimes, time constraints or special sampling led to shallower casts. On the return of each cast, water samples were collected for salinity analysis to compare with the continuous values. Also, XBTs were used to provide better resolution for the thermal field.

Eulerian currents and temperatures were measured from two arrays of moored current meters along the continental shelf. Positions and schematic views for the two arrays of current meters are presented in Ramp (1989) and Schlitz et al. (2001). The two arrays were identical except for their relative position along the continental shelf break. The first array, called WCR1, was deployed from 15 April to 1 July 1982, and was centered at roughly 40°N, 70°W (Ramp 1989). The second array, called WCR2, was deployed during August and September 1982 and was centered at roughly 40°20'N, 67°30'W (Schlitz et al. 2001).

Short-term Lagrangian currents were measured by drogues placed within streamers of shelf water. Details about the drifters are found in Schlitz et al. (2001).

For sampling plankton, especially ichthyoplankton, either a multiple-opening/closing net and environmental sensing system (MOCNESS) or bongo net haul generally followed the CTD cast at each station.

To estimate the volume of shelf water in streamers, a planimetric method was used, based on the isolines shown in Figures 3a-6a. Each isoline in Figures 3a-6a represents the depth of a layer measured from the surface to the deepest level in which the salinity is <34 PSU, the value normally at the inshore side of the shelf-slope front. The surface area at each 10-m isoline was determined. Two adjoining levels were averaged, then the value was multiplied by 10 m to obtain a volume. This computation was done for all levels containing shelf water.

To estimate the transport of shelf water along each section across the streamers in this study, geostrophic calculations -- relative to the deepest common depth between adjacent stations (generally about 1000 m) -- were used. Joyce et al. (1992) demonstrated that the azimuthal circulation in a streamer is generally geostrophic, based on a combined acoustic Doppler current profiler (ADCP)/"tow-yo" CTD section across the active streamer associated with Ring 82-B during June. Transport was estimated by using an average salinity at each depth level between pairs of hydrographic stations, for those stations which effectively described a section across the streamer. When the shelf water - slope water boundary crossed between paired stations for which a calculation was made, an underestimate was possible due to the increased gradient at the front. A total of 26 sections across the streamers were used for these calculations; three other sections were across the shelfbreak.

Inspection of satellite imagery by Chamberlin (1981), Evans et al. (1985), and Garfield and Evans (1987) shows that the surface expression of streamers is a common feature and can be viewed for extended periods. However, the balance of mass for continental shelf water and the direct measurement of shelf water features show that the flow of shelf water into streamers is not continuous. It is not clear how long shelf water might remain identifiable within slope water. Occasionally, shelf water can spiral into the center of a ring (Evans et al. 1985) and be rapidly mixed. Nevertheless, this is a rare event in the life of a ring. Without this spiraling into the center of the ring or without interaction with the Gulf Stream, the mixing of shelf water with surrounding slope water should be an important process to dissipate the shelf water.

To investigate a time scale for dissipation of a streamer by mixing of shelf water within slope water, a two-dimensional diffusion calculation was performed using an idealized model. The model is based on the diffusion of salt across the sides and bottom of a rectangular area. Salinity was chosen to model due to its relatively conservative nature throughout the year (i.e., seasonal variations are relatively small, and a clear contrast between shelf water and slope water can be defined). Correspondence was also good between the surface thermal gradients and salinity gradients sampled on many crossings of the streamer boundary. The diffusion model in two dimensions is

where S is salinity (PSU), K_h is the horizontal diffusivity (cm² sec^-1), K_z is the vertical diffusivity (cm² sec^-1), x is the distance across the region (cm), z is the depth (cm), and t is the time (sec). In this model, the third spatial dimension is assumed infinite (). With boundary conditions of S = 0 at each boundary for simplicity, no diffusion through the upper boundary, and a constant initial value (S = S_o) within the regime, the solution for salinity becomes

l is width of the streamer, and h is depth of the streamer. Five parameters are needed for the calculation: S_o, l, h, K_h, and K_z. The initial salinity deficit within the streamer, S_o, was estimated as 2 PSU based on differences in the average salinity within the streamer and surrounding slope water. The width and depth of the streamer, l and h, were set at 40 km and 60 m, respectively. Horizontal diffusivity, K_h, was estimated from the diffusion diagrams presented by Okubo (1971); for a length of 40 km, it was calculated as 4 x 10⁵ cm² sec^-1. Vertical diffusivity, K_z, was varied as 0.1, 1, 5, and 10 cm² sec^-1, following the vertical diffusivity parameterizations by Schmitt and Olson (1985). Calculations used a grid having horizontal resolution of 2 km and vertical resolution of 3 m, and covered a 20-day period.

Each ring was at a different stage of maturity and geographic location, and showed various types and intensities of interactions. The following descriptions of shelf water associated with each ring are named by the nearby ring.

Ring 81-E formed about 16 July 1981 (day 197) near 39N, 63W, and moved near 40N, 64W during mid-September, approximately 120 nm (220 km) southeast of the Northeast Channel (Fitzgerald and Chamberlin 1984). During 16-24 September, Ring 81-E vigorously interacted with a Gulf Stream meander, causing major changes in the ring's structure (Joyce et al. 1984). After the transformation of the ring, a survey of most of the streamer was completed between 23 September and 4 October (69-80 days after formation). The position of Ring 81-E during this study's series of stations is in the same area as the ring described by Smith (1978).

The general structure of the streamer is seen in Figure 3a. The thickness of this streamer ranged from >40 m at the northern end, to 12 m at a saddle point at the southern end. At the southern end, the shelf water bifurcated, with one portion with maximum depth of 29 m moving southeastward, and the other portion with maximum depth of 17 m moving westward along the periphery of the ring. Extension of the western portion was defined at the surface by data from the thermosalinograph, and below the surface by XBTs, as the ship moved toward the northwest after Station 68 (Figure 3a). Positions of the boundary were guided by surface temperatures derived from AVHRR images (Figure 3b).

The streamer associated with Ring 81-E was well away from the continental shelf. The surface expression of the shelfbreak front along the continental shelf north of Cape Hatteras generally separates from close association with the underlying topography near the southeast corner of Georges Bank.

Water with Gulf Stream characteristics (i.e., salinity >36 PSU) was measured underneath the shelf water at the southern end. Those measurements confirmed the satellite imagery that showed shelf water moving along the northern side of the Gulf Stream after interaction between the ring and Gulf Stream (Figure 3b).

Typical temperature (T) and salinity (S) values for Station 35 at the northern side and for Station 68 at the southern side, in proximity to the Gulf Stream, showed differences in water characteristics (Figure 3a). The values changed from S = 32.23 PSU and T = 16.8°C on the northern side, to S = 33.86 PSU and T = 18.4°C on the southern side. Also, the subsurface temperature minimum in the shelf water was no longer in the water column on the southern side. The reason for the change in characteristics is not clear, although different sources for the water in the streamer seem likely rather than mixing, if the events develop over approximately 3-5 days.

Ring 82-A formed around 15 January 1982 (day 15) about 110 nm (210 km) southeast of Corsair Canyon (Celone and Price 1985). During April, the ring interacted with a meander of the Gulf Stream, was transformed, and moved eastward (Celone and Price 1985). A new ring, Ring 82-D, formed to the south of Ring 82-A during the last half of April, and exchanged water with Ring 82-A, finally destroying Ring 82-A on 2 May 1982 (day 122). At the time of the survey of the streamer, during 22-27 April (97-102 days after formation), the ring was centered near 39°30'N, 66°30'W, off the southern side of Georges Bank.

The prominent feature of the streamer associated with Ring 82-A was a bowl of shelf water, >75 m thick at the deepest point, nearly separated from the shelf water source (Figure 4a). The series of satellite images during the survey period (Figure 4b) showed slope water moving counterclockwise/cyclonically around the northern side of the streamer. It finally surrounded shelf water at the surface, agreeing with the station and surface thermosalinograph data collected earlier during the process of separation.

The temperature and salinity values for a station within the center of the entrained shelf water, Station 8, and at the shelf end of the connecting neck, Station 27 (Figure 4a), both showed a subsurface temperature minimum in shelf water, which was cooled during winter and then became isolated by vernal warming. Temperature and salinity within the streamer, T = 5.51°C and S = 32.89 PSU, were in the range of nearby shelf water.

Ring 82-B was formed on 14 February 1982 (day 45) near 40N, 6825W, to the west of the New England Seamount Chain and farther west than most rings previously monitored (Celone and Price 1985), including Ring 82-A. Movement was westward until the end of March when the motion stalled between Block Canyon and Hudson Canyon where the orientation of the continental shelf changes from east-west to northeast-southwest. During this period no interaction occurred with other rings or the Gulf Stream.

The streamer was surveyed between 28 April and 3 May 1982 (73-78 days after formation). During this time, the shelf water within the streamer was generally in a layer <30 m thick over slope water, except for a layer >50 m thick in the northeast corner (Figure 5a). Although the entire streamer was not sampled due to time constraints, indications from satellite imagery were that the shallow layer extended further to the west along the southern side of the ring (Figure 5b).

Temperatures and salinities for one station in the northeast corner of the streamer, Station 39, and at the southern end of the area sampled, Station 59 (Figure 5a), both showed salinity near the transition between shelf water and slope water (S = 33.73 PSU for Station 39 and S = 33.99 PSU for Station 59). The surface temperature at the southern station was about 0.5°C higher.

Again, there was no way to resolve from these data the different sources for the streamer and the local diffusion of salt and heat into the streamer causing property changes. The lack of a subsurface temperature minimum and the salinities near 34 PSU indicate a relatively quiescent period without significant interaction with the nearby shelf. Also, current vectors were alongshelf on 29 April, potentially blocking the source of shelf water (see Ramp (1986), Figure 20).

The greater depth of water with salinity <34 PSU at Stations 34 and 39 (Figure 5a) can be interpreted as a new active outbreak of shelf water into slope water. This interpretation is supported by the thermal imagery on 30 April and 1 May (Figure 5b), and also by the daily current vectors with offshore directions (see Ramp (1986), Figure 21).

Between early May and the last half of June, Ring 82-B moved along the continental slope to a position southeast of Delaware Bay. An extensive survey of the streamer (including a closely sampled section in cooperation with other vessels) and adjoining waters over the continental shelf was carried out between 21 and 29 June (127-135 days after formation).

The greatest horizontal extent at the surface (approximately 90 nm, 145 km), depth of a layer with salinity <34 PSU (>80 m at Stations 65, 20, and 21), and volume of entrained shelf water were found in this streamer (Figures 2c and 6a). A prominent saddle point was observed where the layer of shelf water was only 20 m thick, as in Ring 81-E, and previously in Ring 81-B.

At Station 28 at the southern end of the streamer (Figure 6a), the surface temperature was slightly warmer (T = 17.04 vs. 16.64°C), and the surface salinity was slightly higher (S = 32.87 vs. 32.67 PSU) than at Station 65 at the 200-m isobath near the center of the streamer. An AVHRR image for a time during the CTD stations is shown in Figure 6b. At the subsurface temperature minimum, the water at Station 28 was both slightly cooler and slightly fresher than that at Station 65. Surface temperature and salinity values for Stations 65, 20, and 21 (Figures 2c and 6a) were T = 16.64, 16.27, and 15.87°C, and S = 32.67, 32.73, and 32.83 PSU, respectively. The temperatures at the transition from shelf water at those three stations were T = 9.10, 9.42, and 9.50°C, respectively. These data suggest rapid movement from the continental shelf without significant mixing.

The estimated volume for each streamer is presented in Table 1. The values were between 58 km³ for the nearly detached streamer associated with Ring 82-A, and 290 km³ for the streamer associated with Ring 82-B during June.

Each estimate, except for Ring 82-A, was a minimum since there was an area at the southern end containing shelf water that was not sampled. By comparing the shelf water determined from hydrography with the available satellite images for the same time, an underestimate of volume by 10-15% seems likely, but caution must be used since the subsurface structure of a streamer, which is highly variable in these four cases, must be extrapolated from the surveyed region.

Description of the hydrographic properties of shelf-water streamers provides one view of the variability in the structure and volume of entrained water in these streamers (Figures 3a-6a). Each streamer had a region where the depth of shelf water was at least 40 m, and for the streamer associated with Ring 82-B during June, the depth reached >80 m both near the shelfbreak and at the southern end.

In each streamer, there was a saddle point where the layer of shelf water reached a minimum thickness within the streamer. This saddle point can be striking as in Ring 82-B during June, or relatively subtle as in Ring 82-B during April, but each such saddle point indicates variations in the flow of shelf water associated with WCRs.

The encircling of shelf water by slope water (i.e., thinning of the depth at the saddle point to zero) as seen in the streamer associated with Ring 82-A is the first known report of the isolation of a relatively large mass (30 nm (55 km) in diameter, 60 km³ in volume) of largely unaltered shelf water within slope water, based on data collected in-situ. There have been other indications of streamers and eddies at the shelf-slope front causing losses of shelf water, but at smaller scales (Houghton et al. 1986; Garvine et al. 1988).

The overall mean transport of shelf water away from the continental shelf (i.e., for those 21 sections with transport in the same direction as the near-surface circulation of the associated WCR) was 165 x 10³ m³ sec^-1, with a standard deviation of 149 x 10³ m³ sec^-1. The overall mean transport for the 19 sections with transport of shelf water in the opposite direction as the near-surface circulation of an associated WCR was 62 x 10³ m³ sec^-1, with a standard deviation of 90 x 10³ m³ sec^-1. Table 2 shows offshore transport across sections ranging from 0 to 488 x 10³ m³ sec^-1. Onshore transports ranged from 0 to 344 x 10³ m³ sec^-1. The streamer associated with Ring 82-B during June was the most active in removing shelf water.

Based on the two-dimensional model for salt diffusion, the average salinity within the streamer for K_z = 1 cm² sec^-1 was halved, but did not quite reach the S_o/e value (0.74), at 14 days (Figure 7). The average salinity for K_z = 5 cm² sec^-1 was halved just before 8 days, and reached the S_o/e value at 12 days. Note that for a streamer with a given width and depth, these time scales are valid for any initial salinity value within the boundaries, since the decay depends only on the values of K_h and K_z (Equation 2).

Cross sections of the salinity differences are shown at 10 and 20 days in Figure 8. Several points become clear by using a salinity deficit of 1 PSU as a conservative indicator of the recognizable streamer. At 10 days, there was water with shelf characteristics clearly seen at the surface for all values of K_z from 0.1 to 10 cm² sec^-1. The width of shelf water at the surface was at least one-half of the original dimension for all values of K_z. The depth of shelf water was at least two-thirds of the original dimension for K_z = 5 cm² sec^-1, and was one-half of the original dimension for K_z = 10 cm² sec^-1. At 20 days, this streamer was still recognizable at the surface for values up to K_z = 5 cm² sec^-1. Calculations for both K_z = 0.1 and 1 cm² sec^-1 showed a substantial section of shelf water that was 24 km in width and at least 44 m in depth, little changed in 20 days from the initial condition.

The calculation for diffusion of salt indicates that a streamer of shelf water relatively isolated within slope water would remain cohesive for 2-3 wk in the absence of advective shearing processes or strong air-sea interaction with mixing. Moreover, this streamer could be viewed from satellites even when the original cross section of the streamer has been eroded to a large extent, thereby complicating interpretations based solely or primarily on thermal imagery.

It was assumed that the balance of geostrophic forces was the cause for movement, and thus the basis for calculating the velocity and transport, of streamers. Any high-frequency vertical variations of the density field could not be detected based on the sampling scheme, and are an unknown source of error.

One comparison between calculated geostrophic velocities and directly measured velocities was possible. The section sampled by Joyce et al. (1992) was concurrent and nearly parallel with Stations 17-25 sampled on 23 June 1982 in this study (Figure 2c (inset) and Figure 9). Calculated velocities for this study were plotted for the 20- and 100-m levels to the same scale as the measured velocities in the Joyce et al. (1992) study. (Shallow CTD casts were required in the Joyce et al. (1992) study in order to keep up with the second ship as it crossed the streamer.) Vertical shear compares well, changing signs at Station 21 from a subsurface maximum west of the station to a surface maximum east. The lower velocities and occasional reversals in the geostrophic calculation can be at least partially explained by the shallow level of no motion (300 m) for this section, by the inherent averaging in the geostrophic calculations, by the orientation of stations that are not perpendicular to the measured current, and by the spatial variations within the streamer.

Transport by salinity classes estimated by these geostrophic calculations (Table 3) can also be compared with those direct measurements obtained by Joyce et al. (1992). The transport of shelf water (S < 34 PSU) calculated geostrophically for Stations 17-25, 408 x 10³ m³ sec^-1, is 32 x 10³ m³ sec^-1 higher than that obtained by direct measurement. This difference could be partially caused by low salinity water not being sampled at the western end of the Joyce et al. (1992) section. The difference of 9% between these methods, in spite of the shallow reference level and slightly different sections, gives confidence that other transport calculations are reasonable for an overall characterization. The deficit in geostrophic transport for salinities between 34 and 35 PSU appears to be caused by the greater measured velocities at 100 m and the greater intrusion of water with salinity >34 PSU at the western end of the Joyce et al. (1992) section. In spite of the range of potential errors, the differences between geostrophic calculations and direct measurements are considered reasonable, and geostrophic calculations are considered acceptable for estimating velocity and transport of entrained shelf water.

Various modeling and observational studies using current meters (e.g., Beardsley et al. 1985; Ramp 1986; Wang 1992) indicate that the waters of the continental shelf are generally buffered from the passage of WCRs. Yet significant amounts of shelf water (Table 1) were found to be entrained within the streamers associated with this study's rings. The hydrographic data collected during this study's series of cruises provided sufficient spatial resolution to examine the proximate source of water for the streamers.

The most prominent characteristic of the Middle Atlantic Bight during the stratified season (i.e., April-September) is the continuous subsurface band of cold water found along the outer continental shelf (OCS), with its core found along the bottom between 50 and 80 m (Houghton et al. 1982). The low temperature and associated low salinity of cold band water are distinct, permitting the unambiguous identification of cold band water if present within a streamer. Alternatively, the absence of cold band characteristics in a streamer requires a source seaward of the cold band for active streamers.

The data collected for the streamer associated with Ring 82-B during June were examined for the presence or absence of cold band characteristics in order to determine the sources of shelf water within the streamer. The temperature and salinity structure on the shelf adjacent to the streamer associated with Ring 82-B during June is shown for three sections which were nearly perpendicular to the local bathymetry (Figure 2c, Figure 6a, and Figure 10-12). The cold band is prominent, with temperatures below 7°C and salinities generally between 32.75 and 33.00 PSU. Note that in the surface layer, salinities below 33.00 PSU extended further seaward only with temperatures above 8°C. A section across the streamer (Figure 13b) showed that only two stations off the continental shelf, Stations 48 and 50 (Figures 2c and 6a), had water with a subsurface salinity minimum below 33.00 PSU. At Station 48, there were two occurrences of water, each more than 3 m thick, separated from each other by about 5 m, and located about 23 and 35 m of depth. Temperatures of 9.0-9.5°C were associated with the upper occurrence, and 8.0-8.5°C with the lower occurrence. At Station 50, there was an occurrence of water about 2 m thick, located between about 28 and 30 m of depth, with a temperature of 7°C. The outcome is that the source of shelf water for the streamer is above the OCS, generally seaward of the cold band. The plume-like structure of the cold band detached from the bottom may occasionally contribute, but in small volume, according to these data. This inference contrasts with the inference of Churchill et al. (1986).

Other data also support this present inference. For the streamer associated with Ring 82-B during April, the salinity was entirely above 33.00 PSU except for a surface layer at Station 34 (Figure 2b and Figure 5a) that was 7 m thick. For the nearly-isolated bowl of shelf water associated with Ring 82-A, there was considerable water with temperature below 6°C and salinity below 33.00 PSU (Figure 2b, Figure 4a, and Figure 14). However the source of the water was still from the OCS, as shown in Figure 15 for Stations 24-28 in Figure 2b, since stratification from vernal warming formed a cold band having minimum temperatures below 4.5°C and salinities below 32.75 PSU. For the streamer associated with Ring 81-E, the data are consistent with a source above the OCS and generally seaward of the cold band, but the interpretation is equivocal because both the ring and interacting shelf water are well seaward of the continental shelf.

Each survey of streamers associated with WCRs seemed to give a unique result except for the observation of a saddle point at some distance along the streamer. The presence of a nearby continental shelfbreak was not required. For example, the streamer associated with Ring 81-E during September 1981 was well away from the continental slope, yet entrained at least 131 km³ of shelf water. The strength of interaction, defined by the amount of shelf water within slope water, did not depend on the age, and therefore energy, of the ring. For example, the most intense observed streamer was during late June 1982 associated with Ring 82-B after this ring had traversed along the continental slope from its formation south of Nantucket to southeast of Delaware Bay, about 130 days after formation. Also as an example, significant interaction occurred with a small ring, Ring 82-A, within a week of its demise.

The idealized results of Wang (1992) were used to merge the seemingly disparate descriptions of this study's streamers. In Wang (1992), a series of models generate results with which to examine the interaction of an eddy with topography resembling a continental slope. The first model is based on a step in topography (i.e., small escarpment) and point vortex under quasigeostrophic conditions, which is then expanded to a finite step and vortex. The fluid in the model is assumed to be barotropic and frictionless in a rotating coordinate system with a rigid lid. Contour dynamics are used in the solutions to the model, giving similar principal results in both cases: 1) ageostrophic nonlinear advection of fluid across the escarpment, 2) formation of a cyclonic eddy containing shelf water offshore, and 3) excitation of waves trapped against the escarpment. The second model, which incorporates a semispectral primitive equation model developed by Haidvogel et al. (1991), uses the same basic equations, including bottom friction, but with three different cross-shelf exponential topographies, in order to generate results on the f and -planes. Wang (1992) concluded that there were "no fundamental differences in shelf/slope responses and cross-slope exchanges between the results of these two models." The initial tongue of entrained shelf water with high potential vorticity evolves into a narrow streamer and eddy (Wang 1992). The eddy at the head of the streamer is formed from fluid with cyclonic vorticity continually entering after being compressed between the vortex and the anticyclonic field north of the tongue. The saddle point between the eddy and the rest of the streamer is formed at the point of greatest compression just inshore of the eddy. For example, Ring 82-B during June has a deep pool of low-salinity water at the southern end, and a narrowing toward the continental shelf (Figure 6). The cyclonic circulation of the eddy then continues constricting the streamer, eventually detaching as in Ring 82-A (Figure 4).

Cyclonic eddies located within entrained shelf water have been interpreted from satellite imagery (e.g., Evans et al. 1985; Ramp 1986), and those located northeast of WCRs have been interpreted from ADCP data (Kennelly et al. 1985). This study's data and the modeling by Wang (1992) provide evidence that not only were curvilinear streamers and cyclonic eddies dynamically linked with entrainment, but also those eddies northeast of the ring. For example, the second lobe of shelf water to the northeast of Ring 82-B during April (Figure 5) corresponds to the evolution of a cyclonic eddy.

The extent to which the assumption of a barotropic fluid by Wang (1992) describes the observed entrainment could be problematic, depending on the vertical scale chosen for the comparison. To illustrate, consider Ring 82-B; it was distinguishable at a depth of 2000 m during June after moving along the continental slope for about 4 mo (e.g., Kennelly et al. 1985). Figure 16 shows the geostrophic velocity for the 0-300 m depth range, relative to approximately 1000 m, for Stations 6-11 (Figure 2c and Figure 6a) across the streamer associated with Ring 82-B during June. Within the depth range corresponding to shelf water, 65 m (Figure 16b), the ring could be considered radially barotropic (i.e., little vertical shear at a given distance from the center). Although the velocity is somewhat intensified near the surface for the region with shelf water, below 100 m, all velocities are <10 cm sec^-1, and vertical shear in the velocity field above 1000 m is small. Therefore, by choosing a vertical scale of approximately 300 m, qualitative comparisons for the interaction between a barotropic vortex and shelf water seems reasonable, if not complete, for the level of comparison presented here.

CONCLUSIONS

Data have been analyzed to describe the interaction between WCRs (i.e., 81-E, 82-A, and 82-B during April and June) and shelf water, leading to entrainment of shelf water along the trailing edge of WCRs as they move generally southwest in slope water. In-situ data of this resolution have not been described before.

The short-term estimated volumes of entrained shelf water (i.e., salinities < 34 PSU) were 58-290 km³. An underestimate of 10-15% was likely for all but the nearly enclosed bowl of shelf water associated with Ring 82-A, since the entrainment was not circumscriptively sampled in other cases.

Geostrophically calculated transports of shelf water in the direction of the anticyclonic circulation of an associated WCR were 0-488 x 10³ m³ sec^-1 (mean = 165 x 10³ m³ sec^-1; standard deviation = 149 x 10³ m³ sec^-1). Transports in the opposite direction were 0-345 x 10³ m³ sec^-1 (mean = 62 x 10³ m³ sec^-1; standard deviation = 90 x 10³ m³ sec^-1). These transport estimates are thought to be reasonable, based on a comparison between geostrophic calculations and ADCP/"tow-yo" CTD measurements taken concurrently during a two-ship crossing of an active streamer during June 1982 associated with Ring 82-B. Changes in the direction of flow are probably a consequence of cyclonic eddies that develop within entrained water.

Analysis of temperature and salinity sections indicates that the shelf water supplying streamers comes primarily from the outer edge of the continental shelf, seaward of the cold band. Since the source of shelf water for streamers is distributed along the shelfbreak for considerable distances rather than being limited to an area nearby, the majority of the continental shelf is, therefore, isolated from direct influence by passage of rings. This finding confirms results of prior studies which measured in detail the temperature and salinity structure and currents across the shelf as WCRs passed.

If the mean seaward extent of the cold band is the 100-m contour (Houghton et al. 1982), and if the mean position of the shelf-slope front is anchored at the 100-m contour and intersects the surface 40 km seaward (Beardsley et al. 1985), then the ratio of the volume (km³) of shelf water in a streamer to the length (km) of shelfbreak supplying water to the streamer is 2:1. This ratio results in a range of 29-145 km along the shelfbreak for supplying shelf water to the streamers in this study. It therefore seems plausible to use calculated volumes of streamers to determine the lengths of shelfbreak affected by entrainment.

Being able to consistently and generally interpret the data with modeling is an encouraging, but problematic, accomplishment, given the ideal assumptions of the model -- particularly a barotropic ocean -- and the variability of the strictly nonsynoptic hydrographic data. One puzzling aspect, at least, remains: the streamer associated with Ring 81-E during September 1981 (Figure 13a). Since the presence of an adjacent continental shelf was apparently not required to initiate this streamer, the model by Wang (1992) cannot directly apply in this case. Therefore, other processes should be important in explaining the initiation of entrainment by WCRs at different stages in the history of those rings.

One possible explanation for this "continental-shelf-free" initiation of entrainment are growing frontal waves which are due to horizontal-shear instability on the northern periphery of a WCR, and which then "break" to subsequently become entrained by that ring (Ramp et al. 1983). Such frontal waves were observed for a number of days on the periphery of Ring 82-B based on AVHRR images, but the final fate of those waves was not observed due to clouds. The position of the horizontal-shear instability was close to the anticyclonic advection of an eddy on the periphery of Ring 82-B, as interpreted from sequential satellite images (Kennelly et al. 1985, Figure 8b). Also, the AVHRR image for Ring 82-B on 30 April 1982 (Figure 5b) supports those observations.

Conceptually, a combination of curvilinear streamers on the periphery, and cyclonic eddies which may remain coherent for 2-3 wk, can describe the observed structure for entrainment by WCRs. Evidence from this study's data and from the modeling supports this concept. Clearly, further investigation is needed to understand the initiation and evolution of these streamers.

A major goal of this study was to examine the possibility that passage of WCRs could affect recruitment of commercially important fish species during the larval stage. Distributions of larval fish have been portrayed by many authors for the continental shelf between Cape Hatteras and Nova Scotia (e.g., Grosslein and Azarovitz 1982; Morse et al. 1987). These portrayals show that the maxima in the general distributions of larvae are well inshore of the 200-m isobath, although bias may occur due to limited sampling near the offshore boundaries. Since the direct measurements and analyses of water masses in this study indicate that the onshore influence of WCRs is about the 100-m isobath, then the potential for significant mortality caused by entrainment of shelf water, as a single process, seems unlikely. Earlier modeling by Flierl and Wroblewski (1985) and statistical analysis by Myers and Drinkwater (1989) had linked WCRs with reduced recruitment of fish. Flierl and Wroblewski (1985) recognized, however, that the results of their model, averaged across the shelf and also vertically, would not be applicable if entrainment were limited near the shelf break.

Detailed analysis of the biological samples collected concurrently during this study's series of cruises was not possible except for the analysis by Le Blanc (1986). Le Blanc (1986) reported that larvae of fish species normally spawned on the continental shelf were sampled in the entrained shelf water associated with Ring 82-B during June. A quantitative assessment of the impact upon those species' populations was not attempted, though. Unfortunately, other biological samples for the series of cruises were lost.

These data were originally collected as part of the Warm-Core Experiment in 1981 and 1982. Thanks go to all who participated during the series of cruises, particularly the officers and crew of the Albatross IV. Their dedication to the success of data collection made these analyses possible.

Bisagni, J.J. 1983. Lagrangian current measurements within the eastern margin of a warm-core Gulf Stream ring. J. Phys. Oceanogr. 13:709-715.

Beardsley, R.C.; Chapman, D.C.; Brink, K.H.; Ramp, S.R.; Schlitz, R. 1985. The Nantucket Shoals Flux Experiment (NSFE79). Part I: a basic description of the current and temperature variability. J. Phys. Oceanogr. 15:713-748.

Celone, P.J.; Price, C.A. 1985. Anticyclonic warm-core Gulf Stream rings off the northeastern United States during 1982. Ann. Biol. 39:19-23.

Chamberlin, J.L. 1981. Application of satellite infrared data to analysis of ocean frontal movements and water mass interactions off the Northeast coast. NAFO [Northwest Atl. Fish. Org.] SCR [Sci. Counc. Res.] Doc. 81/IX/123; 15 p.

Churchill, J.H.; Cornillon, P.C.; Milkowski, G.W. 1986. A cyclonic eddy and shelf-slope water exchange associated with a Gulf Stream warm-core ring. J. Geophys. Res. 91:9615-9623.

Evans, R.; Baker, K.; Brown, O.; Smith, R.; Hooker, S.; Olson, D.; Warm Core Rings Program Service Office. 1984. Satellite images of warm core ring 82-B sea surface temperature and a chronological record of major physical events affecting ring structure. Unpublished document. Available from: R.J. Schlitz, National Marine Fisheries Serv., Woods Hole, MA.

Evans, R.H.; Baker, K.S.; Brown, O.B.; Smith, R.C. 1985. Chronology of warm-core ring 82B. J. Geophys. Res. 90:8803-8811.

Fitzgerald, J.L.; Chamberlin, J.L. 1984. Anticyclonic warm-core Gulf Stream rings off the northeastern United States in 1981. Ann. Biol. 38:29-33.

Flierl, G.R.; Wroblewski, J.S. 1985. The possible influence of warm core Gulf Stream rings upon shelf water larval fish distribution. Fish. Bull. (Wash., D.C.) 83:313-330.

Fornshell, J.A.; Criess, W.A. 1979. Anticyclonic eddy observations in the slope water aboard CGC Evergreen. J. Phys. Oceanogr. 9:992-1000.

Garfield, N., III; Evans, D.L. 1987. Shelf water entrainment by Gulf Stream warm-core rings. J. Geophys. Res. 92:13003-13012.

Garvine, R.W.; Wong, K.-C.; Gawarkiewicz, G.G.; McCarthy, R.K. 1988. The morphology of shelfbreak eddies. J. Geophys. Res. 93:15593-15607.

Grosslein, M.D.; Azarovitz, T.R. 1982. Fish distribution. MESA (Mar. Ecosyst. Anal.) N.Y. Bight Atlas Monogr. 15; 182 p.

Haidvogel, D.B.; Wilkin, J.L.; Young, R.E. 1991. A semi-spectral primitive equation ocean circulation model using vertical sigma and orthogonal curvilinear horizontal coordinates. J. Comput. Phys. 94:151-185.

Houghton, R.W.; Olson, D.B.; Celone, P.J. 1986. Observations of an anticyclonic eddy near the continental shelf break south of New England. J. Phys. Oceanogr. 16:60-71.

Houghton, R.W.; Schlitz, R.; Beardsley, R.C.; Butman, B.; Chamberlin, J.L. 1982. The Middle Atlantic Bight cold pool: evolution of the temperature structure during summer 1979. J. Phys. Oceanogr. 12:1019-1029.

Iles, T.D.; Sinclair, M. 1982. Atlantic herring: stock discreteness and abundance. Science 215:627-633.

Joyce, T.M. 1985. Gulf Stream warm-core ring collection: an introduction. J. Geophys. Res. 90:8801-8802.

Joyce, T.; Backus, R.; Baker, K.; Blackwelder, P.; Brown, O.; Cowles, T.; Evans, R.; Fryxell, G.; Mountain, D.; Olson, D.; Schlitz, R.; Schmitt, R.; Smith, P.; Smith, R.; Wiebe, P. 1984. Rapid evolution of a Gulf Stream warm-core ring. Nature 308:837-840.

Joyce, T.M.; Bishop, J.K.B.; Brown, O.B. 1992. Observations of off-shore shelf-water transport induced by a warm-core ring. Deep-Sea Res. 39(suppl.):S97-S113.

Joyce, T.; Wiebe, P. 1983. Warm-core rings of the Gulf Stream. Oceanus 26:34-44.

Kennelly, M.A.; Evans, R.H.; Joyce, T.M. 1985. Small-scale cyclones on the periphery of a Gulf Stream warm-core ring. J. Geophys. Res. 90:8845-8857.

Le Blanc, P.R. 1986. The distribution and abundance of larval and juvenile fishes in shelf water entrained by warm-core ring 82-B and contiguous water masses. [M.S. thesis.] Dartmouth, MA: Southeastern Massachusetts Univ.; 143 p.

Morgan, C.W.; Bishop, J.M. 1977. An example of Gulf Stream eddy-induced water exchange in the Mid-Atlantic Bight. J. Phys. Oceanogr. 7:472-479.

Morse, W.W.; Fahay, M.P.; Smith, W.G. 1987. MARMAP surveys of the continental shelf from Cape Hatteras, North Carolina, to Cape Sable, Nova Scotia (1977-1984). Atlas No. 2. Annual distribution patterns of fish larvae. NOAA [Natl. Ocean. Atmos. Admin.] Tech. Memo. NMFS [Natl. Mar. Fish. Serv.]-F/NEC-47; 215 p.

Myers, R.A.; Drinkwater, K. 1989. The influence of Gulf Stream warm core rings on recruitment of fish in the Northwest Atlantic. J. Mar. Res. 47:635-656.

Ramp, S.R.. 1986. The interaction of warm core rings with the shelf water and shelf/slope front south of New England. [Ph.D. thesis.] Narragansett, RI: Univ. of Rhode Island; 137 p.

Ramp, S.R. 1989. Moored observations of current and temperature on the shelf and upper slope near ring 82B. J. Geophys. Res. 94:18071-18087.

Ramp, S.R.; Beardsley, R.C.; Legekis, R. 1983. An observation of frontal wave development on a shelf-slope/warm core ring front near the shelf break south of New England. J. Phys. Oceanogr. 13:907-912.

Richardson, P.L. 1983. Gulf Stream rings. In: Robinson, A.R., ed. Eddies in marine science. New York, NY: Springer-Verlag; p. 19-64.

Schlitz, R.J.; Manning, J.P.; Smith, K.W. 2001. Structure and transport of alongshelf currents across the southern flank of Georges Bank during late summer, 1982. Deep-Sea Res. II 48:341-372.

Schmitt, R.W.; Olson, D.B. 1985. Wintertime convection in warm-core rings: thermocline ventilation and the formation of mesoscale lenses. J. Geophys. Res. 90:8823-8837.

Smith, P.C. 1978. Low-frequency fluxes of momentum, heat, salt, and nutrients at the edge of the Scotian Shelf. J. Geophys. Res. 83:4079-4096.

Walford, L.A. 1938. Effects of currents on distribution and survival of the eggs and larvae of the haddock (Melanogrammus aeglefinus) on Georges Bank. Bull. U.S. Bur. Fish. 49; 73 p.

Wang, X. 1992. Interaction of an eddy with a continental slope. [Ph.D. thesis.] Woods Hole, MA: MIT/WHOI [Mass. Inst. Technol./Woods Hole Oceanogr. Inst.] Joint Program in Oceanography; 216 p.

Warm-Core Rings Executive Committee. 1982. Multidisciplinary program to study warm core rings. Eos 63:834-835.