Continental Shelf Research

Transcription

1 Continental Shelf Research ] (]]]]) ]]] ]]] Contents lists available at SciVerse ScienceDirect Continental Shelf Research journal homepage: Research papers Variation in the Hatteras Front density and velocity structure Part 1: High resolution transects from three seasons in Dana K. Savidge a,n, Jay A. Austin b, Brian O. Blanton c a Skidaway Institute of Oceanography, 10 Ocean Science Circle, Savannah, GA 31411, United States b Large Lakes Observatory, University of Minnesota, Duluth, United States c Renaissance Computing Institute, North Carolina, United States article info Article history: Received 27 January 2012 Received in revised form 5 November 2012 Accepted 7 November 2012 Keywords: Coastal circulation Seasonal variability Cross-shelf transport Buoyancy effects Coastal fronts abstract On the continental shelf near Cape Hatteras, cool fresh Mid-Atlantic Bight and warm salty South Atlantic Bight shelf waters converge alongshelf 90% of the time, causing strong alongshelf gradients in temperature, salinity, and density known as the Hatteras Front. Mechanisms of shoreward transport in this region have long been a topic of interest, since many commercially important species spawn on the outer shelf, but utilize the adjacent Albemarle and Pamlico Sounds for nurseries, requiring some physical transport mechanism to move the eggs and larvae from the outer shelf to these nursery areas. One mechanism providing such shoreward transport is strong shoreward velocity along the cross-shelf oriented nose of the Hatteras Front. The Frontal Interactions near Cape Hatteras (FINCH) project used shipboard ADCP and a towed undulating CTD to examine Hatteras Front property, density and velocity fields in August 2004, January 2005, and July Strong property gradients were encountered across the nose of the Hatteras Front in all cases, but the density gradient evolved in time, and along with it the dynamic height gradient driving the observed along-front cross-shelf velocities in the nose of the Front. In August and January FINCH data, MAB shelf waters on the north side of the Hatteras Front are less dense than SAB shelf waters, driving shoreward velocities along the Hatteras Front. By July, MAB shelf waters are slightly more dense than SAB shelf waters, with areas of weak seaward and shoreward velocities within the Hatteras Front. As Part 1 of a pair of contributions, this article focuses on FINCH data to illustrate the range of density gradients encountered and resulting cross-shelf velocities. Whether these observations are typical of variability in the Hatteras Front is explored in a second article, Part 2. & 2012 Elsevier Ltd. All rights reserved. 1. Background On the continental shelf and slope near Cape Hatteras, Mid- Atlantic Bight (MAB) and South Atlantic Bight (SAB) shelf waters converge alongshelf 90% of the time in daily alongshelf transports (Savidge and Bane, 2001). Since shelf waters derived from north and south of Cape Hatteras have large differences in temperature (T) and salinity (S) characteristics, this convergence supports a strong alongshelf gradient in T and S expressed across the cross-shelf oriented Hatteras Front (Stefansson et al., 1971; Pietrafesa et al., 1994; Berger et al., 1995). Such convergence requires offshelf export of shelf waters through continuity, examined for the MAB shelf water component in several field studies, including the Shelf Edge Exchange Projects (SEEP and SEEP-II) and the Ocean Margins Project (OMP) (Walsh et al., 1988; Biscaye et al., 1994; Verity et al., 2002). Mechanisms of shoreward transport in this region have also been a topic of interest (Checkley et al., 1988; Shanks, 1988; n Corresponding author. Tel.: þ ; fax: þ address: dana.savidge@skio.usg.edu (D.K. Savidge). Stegmann and Yoder, 1996; Quinlan et al., 1999). Many commercially important species spawn on the outer shelf, but utilize the adjacent Albemarle and Pamlico Sounds for nurseries, requiring some physical transport mechanism to move the eggs and larvae from the outer shelf to these nursery areas. Persistent interest in gas and oil deposits on the shelf and slope raises the question of how any pollution resulting from such activities might reach and affect the ecologically and economically important sounds and beaches of North Carolina. Velocities along the Hatteras Front provide one effective shoreward conduit in winter, first demonstrated by Savidge (2002) using mooring records (Fig. 1). In that study, the T and S gradients across the Hatteras Front did not completely compensate, such that cold, relatively fresh MAB shelf water was less dense than warmer, saltier SAB shelf water. The resulting dynamic height gradient across the Front was estimated to be of sufficient magnitude to drive observed shoreward alongfront velocities. The Frontal Interactions near Cape Hatteras (FINCH) project was designed and carried out to investigate this shoreward transport mechanism. Using shipboard ADCP combined with a towed undulating CTD, the circulation and density fields /$ - see front matter & 2012 Elsevier Ltd. All rights reserved.

2 2 Fig. 1. Cape Hatteras region and measurement locations. Panel A: Cape Hatteras (CH) MMS field study site (March, 1992 February, 1994), with mooring locations (black diamonds) along three cross-shelf lines (lines A C in panel A) and two additional shelf-edge moorings (between lines A and B). The hydrographic data shown in Figs. 6 8 were collected on cross-shelf transects along the lines defined by the A moorings and the C moorings. A schematic Hatteras Front (HF) shows both the cross-shelf oriented nose of the Front and its more along-shelf oriented seaward flank. The approximate mean position and width of the Gulf Stream is also shown. The lower panels show the FINCH transects in August 2004, January and July 2005 used within this paper. Red lines show locations of transects in Fig. 3. Together with the black lines, they indicate the locations of all transects included in the stream coordinate means shown in Fig. 5. (For interpretation of the references to color in this figure caption, the reader is referred to the web version of this article.) associated with the cross-shelf nose of the Hatteras Front were intensively sampled in three field seasons: August 8 11, 2004, three 2 3 day forays in late January 2005, and July 19 21, The August data are described in detail in Savidge and Austin (2007), verifying strong property (T and S) and density gradients across the Front, and strong geostrophic shoreward velocities along its nose. MAB shelf waters are known to undergo strong seasonal T variability, in both T ranges and strength of vertical stratification (Lentz, 2010; Castelao et al., 2008). SAB shelf water T also varies with season, with more modest stratification changes, as described in Blanton et al. (2003). Salinity evolution is much less seasonal than T in either SAB or MAB shelf waters, but both MAB and SAB T and S are subject to interannual variability (Blanton et al., 2003; Mountain, 2003; Castelao et al., 2008). Such seasonal or interannual variability may effect the density gradient across the Hatteras Front, and therefore the geostrophically driven shoreward velocities along its nose. In this article, data collected in the January and July FINCH field seasons are compared to the August data to examine how

3 3 density and velocity fields in the Hatteras Front changed with the changing property fields in the converging MAB and SAB shelf waters. Several archived shipboard transects taken near Cape Hatteras are also examined as examples of property and density evolution on both sides of the Hatteras Front through spring, summer and fall. These snapshots imply both a potentially repeatable seasonal evolution of density and velocity within the Hatteras Front and the possibility of interannual variability in density contrast across the Front. For that reason, this article constitutes Part 1 of a pair of contributions. In the companion article, Savidge et al. (this issue), hereinafter Part 2, several longterm archived data sets are examined to determine whether the density evolution described here in Part 1 is characteristic of typical seasonal variability within the Hatteras Front. Seasonal or interannual variability could have important ramifications for species that are dependent upon Hatteras Front associated circulation for cross-shelf transport. 2. Data and methods The August 2004 and July 2005 FINCH data from the nose of Hatteras Front were collected from the R/V Fay Slover, a 55 ft dayboat owned and operated by Old Dominion University, while the January 2005 FINCH data were collected from the R/V Savannah, a 92 ft UNOLS vessel owned and operated by Skidaway Institute of Oceanography (Fig. 1 shows transect locations discussed herein). On all cruises, CTD data were collected with a SeaBird SBE-19plus at high vertical and horizontal resolution using an undulating towed vehicle, the SeaSciences Acrobat. The SBE-19p samples at 4 Hz, and was equipped with a Wetlabs C- Star transmissometer and Wet-Star fluorometer. The Acrobat profiled within 3 m of the bottom and surface between the 15 and 58 m isobaths, with 0.25 km between up and down casts along the zigzag flight path. The R/V Slover is also equipped with a hull-mounted 600 khz RDI ADCP, while the R/V Savannah is equipped with a hull-mounted 300 khz RDI ADCP. The data shown here are in 1 m vertical bins, and have been averaged over 20 s, for an alongtrack resolution of 60 m. The shipboard ADCP data were detided using alongtrack tidal predictions from the ADCIRC model (Luettich and Westerink, 1992; Westerink et al., 2008). Tides and inertial motions are relatively small near Cape Hatteras, with M2 magnitudes typically less than 10 cm/s (Pietrafesa et al., 1985; Lentz et al., 2001; Berger et al., 1995; Savidge et al., 2007). Preliminary plots with undetided data showed similar features to those examined below in the detided data that is, the tidal variability is sufficiently small that it does not mask the Hatteras Front associated circulation. The CTD data were linearly interpolated onto a uniform grid coincident with the ADCP data grid for plotting purposes. Wind information during FINCH was collected from the NDBC Cape Lookout CMAN station (CLKN7, Fig. 2). More relevant CMAN data from Diamond Shoals Fig. 2. Wind records from CMAN station CLKN7 for FINCH field seasons. Northward winds are aligned with the positive y-axis. Sticks represent the direction towards which the wind was blowing, in the oceanographic conventional sense.

4 4 off Cape Hatteras (previously tower station DSLN7, replaced by buoy in 2003) was repeatedly disrupted by instrument and buoy failure during FINCH. An additional valuable dataset near Cape Hatteras comes from a two-year-long mooring, drifter, and hydrographic study funded by the Minerals Management Service (MMS) (Berger et al., 1995). Moorings were maintained from March, 1992 to February, 1994, with seasonal hydrographic surveys at three to four month intervals during the experiment. Shipboard hydrography from transects along the A and C moorings lines (Fig. 1A) are discussed herein. To emphasize the density space contrasts illustrated by the FINCH undulating CTD data, T-S diagrams are presented as frequency plots. Frequencies were calculated by sorting the CTD data into regularly spaced bins (S from 28 to 38 psu by 0.1 psu, T from 7 to 29 1C by 0.1 1C) and then counting the total number of values within each bin. Frequency distributions have been smoothed over S and then T bins with a 3 point Hanning filter, and actual numbers of data points in each smoothed bin were then normalized by the overall maximum. The natural log (zeros excluded) of the results were then contoured. Normalized data frequency o5% is not contoured for August and January, and o2% is not contoured for July. The presentation of the August data differs slightly between Savidge and Austin (2007) and this paper, since the values in that article were not smoothed or normalized and small quantity bins were not eliminated from the contouring. The contours of the highest frequency samples clearly match between versions (their Fig. 6 and Fig. 4 herein). The darkest bands indicate the T S space that was sampled most frequently. No colorbars are provided, as the actual numbers of samples, which depend on the sampling rate and time spent in and out of the water masses of interest, is not relevant to the definition of the T S ranges and density characteristics of the target water masses. To further illustrate the features discussed below in individual sections, stream coordinate mean density and velocity fields were calculated from FINCH transects for August, January and July. Methods are described in Savidge and Austin (2007). Briefly, to estimate a stream coordinate mean from a sequence of oblique transects across a front in motion, a local coordinate system for each transect needs to be defined that is perpendicular to the orientation of the front, and whose origin is consistently located relative to some common feature within the transects. Here the orientation of the Front relative to each transect is taken as the direction of the (spatial) mean ADCP velocity in each transect within a density range that approximately maximizes the magnitude of that velocity. ADCP velocity sections are then rotated into along-front and across-front components. The origin is defined as the location of a specific density contour at a specific depth along the transect (Table 1). The realized transect density and ADCP velocity data are then projected onto a perpendicular to the front s orientation, aligned across the Front to the selected origin, smoothed slightly with an optimal interpolation, and averaged. The density ranges used to define the orientation of the front and the origin definition changed between August and January, and between the three January forays (Table 1). With only weak Table 1 Values used to define cross-front direction and origin for stream coordinate means. FINCH foray # sections s T range Contour ðs T Þ Depth (m) August January July NA velocities measured in July, a default frontal orientation was assumed to be east to west, and transect ADCP velocities and densities were projected onto south to north oriented crossfront lines. The first objective in each FINCH excursion was to determine the alongshelf position of the Hatteras Front for sampling. Though convergent in the mean, SAB and MAB shelf water alongshelf transport variability is correlated with alongshelf wind, so that the location of the Hatteras Front shifts alongshelf with alongshelf wind forcing (Savidge and Bane, 2001; Savidge, 2002). While temperature contrast between the SAB and MAB shelf waters in satellite SST imagery can be useful for locating the Front, cloud cover at Cape Hatteras often obscures the coastal ocean there. On cloud-free days, the Front can also be obscured by the warm surface layer in the MAB that develops in summer, or by the presence of a fresh cap of Chesapeake Plume waters, which carries a T signature of its own. In early August 2004, the Hatteras Front propagated rapidly south past Cape Hatteras under the influence of Hurricane Alex, which skirted the barrier islands there on August 3. The strong winds (Fig. 2) mixed the warm surface layer MAB shelf water down into the cold bottom layer, making the T contrast between MAB and SAB shelf waters immediately apparent in satellite SST imagery. In January, cloud cover obscured the Cape Hatteras shelf in satellite imagery almost continuously. A sequence of strong wind events affected the Front s location, making shipboard CTD sections the most reliable method of locating the Front. Cloud cover also obscured satellite detected SST during July sampling, so shipboard CTD sections were again useful for locating the Front. The primary goal of the July field work was to recover several moorings, so cross-front sampling with undulating CTD and shipboard ADCP was accomplished only 1 2 times on each of three consecutive days (July 19 21). 3. Results Undulating CTD sections across the Hatteras Front in August 2004, and January and July 2005 show large S and T gradients across the Front in all sampled seasons. From August through January, and again from January through July, large temperature changes in the MAB and SAB shelf waters shifted the density space in which both shelf waters resided. These shifts led to an unanticipated evolution in the density contrast between the MAB and SAB shelf waters. Along with the density gradient change through time, the resulting dynamic height gradient and alongfront velocity in the nose of the Front changed between the three field seasons. In both August and January, strong shoreward velocities were encountered across the nose of the Hatteras Front, while in July, much weaker shoreward and seaward velocities were measured. In the following, the August 2004 results of Savidge and Austin (2007) are reviewed, then January and July measurements are compared to the August results. To sample a representative range of densities resulting from the seasonal T evolution in MAB and SAB shelf waters, several MMS-study shipboard transects from late spring, summer and fall are also presented August 2004 August FINCH measurements occurred just after Hurricane Alex had travelled through the study site, vertically mixing both SAB and MAB shelf water on the mid-shelf. The relatively uniform surface to bottom mixed MAB shelf water was cooler, fresher, and less dense than the warmer saltier SAB shelf water, so that a density gradient existed across the Front. Several example crossfront transects were discussed in Savidge and Austin (2007). One

5 5 Fig. 3. Property, density and velocity (positive shoreward) transects across the Hatteras Front in August 2004, January 2005 and July Transect locations are shown in red on the maps in Fig. 1. Left panels: August 9, 2004, middle panels: January , and right panels: July From top to bottom, the panels are: salinity (contour interval 0.5 psu); temperature (contour interval 0.5 1C); density (contour interval 0.25 sy units); cross-shelf velocities (positive shoreward in orange, contour interval 0.05 m/s, bold line is 0.0 contour). The bold gray dashed lines are schematic boundaries of the jet within the Front, traced roughly along the 15.0 cm/s contour in the bottom panels for August and January, and repeated on the upper three panels. No such jet exists for July. example is repeated in the left panels of Fig. 3 (transect location is shown as the red line in Fig. 1B), which shows strong S, T and density fronts. While the strong narrow T gradient located about 6 km along the section is co-located with a relatively strong locally narrow S gradient, in fact the strongest S and density gradients are co-located somewhat farther along near the middle of the transect, with fresher MAB water situated on the lighter side of the density front. The T gradient is also strong near the middle of the transect, but in both subregions is compensated for by the S gradient. At the left, the T and S almost completely compensate, and the density gradient is small. But in the middle of the transect, the S gradient controls the sign of the density gradient, despite the strong opposing T gradient there. Strong shoreward alongfront velocities were recorded in the Hatteras Front with the shipboard ADCP in August, reaching magnitudes of over 30 cm/s. Estimated dynamic height relative to 20 m depth increased by several centimeters crossing the Hatteras Front from the SAB waters into the MAB waters over approximately 10 km. This was of sufficient magnitude to account for the observed alongfront velocities. This centrally located density gradient is collocated with strong surface intensified shoreward velocities. A schematic boundary of this jet within the Front has been sketched along the position of the 15.0 cm/s contour in the bottom panel, which is repeated on the upper three panels as well. In August 2004, the fastest velocities existed within the fresh and cool MAB shelf water side of the density and property fronts, and thus preferentially carried MAB shelf water shoreward. The density contrast across the Hatteras Front in the August 2004 FINCH data is illustrated in a composite T S plot (Fig. 4A), showing the strong MAB-SAB T and S contrasts, with density ranging from 22 to 23.5 sy units across the Front. The slightly cooler, much fresher MAB shelf water is significantly lighter than the SAB shelf water because of the S contrast, and in spite of the difference in T January 2005 In late January 2005, three separate forays of two to three days each were completed between several strong wind forcing events (Fig. 2 shows vector winds). One transect collected during the first foray on January (Fig. 3, transect location is shown as the red line in Fig. 1C) illustrates the strong T and S contrasts across the Hatteras Front in January. SAB and MAB shelf waters were

6 6 Fig. 4. Upper panels: T S frequency plots from the FINCH project, taken across the nose of the Hatteras Front with the undulating CTD. The darkest grey bands encompass the properties of the MAB shelf water at the cold fresh end, and the SAB shelf water at the salty warm end. Lower panels: T S diagrams from the MMS study hydrographic sections in 1992 and 1993, taken from the 5th CTD stations seaward along the A line and C line density sections shown in Figs. 6 8, located near moorings A2 and C2 in Fig. 1. August, November and May data are represented by circles, squares, and triangles, respectively. SAB samples are shown as dark gray filled symbols, while upper and lower MAB water column samples are shown as open and filled light gray symbols, respectively. May 1992 MAB water column was well mixed at this location so open and closed triangles overlay one another. Boundaries of different water masses defined for the Hatteras region by Flagg et al. (2002) (their Fig. 3) are plotted as gray boxes in the upper right panel. Delineated regions are labeled Virginia Coastal Water (VCW, summer and winter), Mid Atlantic Bight Water (MABW, summer and winter), South Atlantic Bight Water (SABW, summer and winter), and Gulf Stream Water (GSW, summer and winter). Their upper and deep slope water categories have not been included in this figure. 10 1C cooler than in August, consistent with the magnitude of seasonal cooling documented in Blanton et al. (2003) for SAB shelf waters and in Lentz (2010) and Castelao et al. (2008) for MAB shelf waters. No large change from August to January in salinity range covered by either MAB or SAB shelf water component appears (T S diagram in Fig. 4A). This is consistent with relatively small seasonal salinity signals documented by Blanton et al. (2003) for SAB shelf waters and by Mountain (2003) and Castelao et al. (2008) for MAB shelf waters. Density across the Hatteras Front in January, as in August, was not entirely compensated, with density ranging from 24.5 to 26.5 s y units across the Front. As in August, MAB shelf water was lighter than SAB shelf water, due to the fresher MAB S, despite cooler MAB T. Strong alongfront velocities were also measured in January, directed shoreward along the portion of the Hatteras Front oriented cross-shelf (Fig. 3). As with the August data, estimated dynamic height relative to 20 m depth increased by several centimeters crossing the Hatteras Front from the SAB waters into the MAB waters over approximately 10 km. This was of sufficient magnitude to account for the observed alongfront velocities. The schematic boundary within the Front sketched along the position of the 15.0 cm/s contour shows that in the January case, SAB shelf waters reside in the portion of the Front where the highest shoreward velocities occur. This is in contrast to the August 2004 case, when the strongest velocities in the Front transported MAB waters shoreward. Whether this represents a seasonal progression or storm response is unknown from the limited data. However, it does demonstrate that there are occasions when either shelf water component may be transported shoreward within the Front. The Front was repeatedly sampled on the second and third January forays also (January and January 31 February 2), showing strong property and density gradients across it during those periods, as during the first January 19 20th foray. Properties and densities on both sides of the Front evolved from each January foray to the next, under the strong storm forcing between forays. The three bands of decreasing temperature distributions in Fig. 4A are from the three separate January forays, with decreasing temperatures with each foray. Along-Front velocities were apparent in the first and second forays, but were not demonstrated convincingly from the relatively few transects accomplished during the third foray July 2005 In July 2005, the Hatteras Front was again located and sampled over a three day period. At this time, the Front was positioned over the shelf north of Cape Hatteras, consistent with the apparent more northward position of the Front in summer surmised by Savidge (2002). One example July transect shows quite cold MAB shelf water on the north (righthand) side of the transect contrasting with much warmer SAB shelf water on the left (Fig. 3, right column,

7 7 upper two panels, location shown as the red line in Fig. 1D). The T S diagram for July indicates that middle and lower water column MAB shelf water had warmed only slightly over values found in January, while the SAB water had warmed to values more typical of summertime (Fig. 4B). Near surface water sampled in these transects was of fairly uniform T overlying both the warm and cold subsurface shelf waters. However, the near surface water was significantly fresher than either warm or cold underlying MAB or SAB shelf water, suggesting a significant contribution from Chesapeake Bay water. The strong property gradients in July were again not entirely compensated in density. However, in contrast to the August 2004 and January 2005 measurements, in the July data the middle and lower water column density gradient was reversed, with cold MAB middle and lower layer shelf water now denser than SAB shelf water by about one s y unit across the Front (Fig. 3), despite lower MAB salinity. In the near surface water, T and S gradients were weak, the density gradient is also small, and in the opposite direction to that of the deeper water column. There is some question whether the warm subsurface water encountered on the shelf in July was actually SAB shelf water, or simply Gulf Stream water stranded on the shelf, due to the low number of transects (4) and patchy, cloud-contaminated satellite imagery. In the transect shown, the warm salty bolus in the center of the transect is especially suspicious. However, nowhere within it does S exceed psu, and nearby 20 m isobath moorings suggest the warmer saltier water measured along the July transects was predominantly SAB shelf water. Part 2 lends further evidence that the July FINCH transects sampled SAB shelf water, showing that FINCH data are consistent with SAB and MAB shelf water densities and contrasts from the NODC long term shipboard hydrographic database. Dynamic height estimates from the lower layer densities (at 10 m depth relative to 20 m) indicates only slightly lower dynamic height over MAB water than over SAB water, amounting to a few mm along the 23 km long transect. This implies at most extremely weak seaward alongfront velocity, while the slope of the isopycnals would imply shoreward thermal shear (decreasing seaward flow with increasing depth). Dynamic height at the surface relative to 20 m reference is dominated by the warm salty bolus in the center of the transect and the warm fresh layer near the surface. The horizontal gradient is small relative to the values seen in August or January, reaching only cm over the 10 km region surrounding the warm bolus, and otherwise remaining lower than a fraction of a cm. Integration through the warm fresh surface layer reverses the slight dynamic height gradient of the lower layer, so that slightly higher total water column dynamic heights occur over MAB shelf water. Diagnosing geostrophic alongfront velocities from small density gradients that reverse with depth is obviously subject to more error than estimates from large gradients with similar sign throughout the water column, as found in August and January. Cross-shelf velocities measured by ADCP along the July transects are slight, and modestly shoreward in the lower water column, from 0 to 10 cm/s (Fig. 3, bottom right panel), not seaward as implied by the slight density gradient in the lower layer. Seaward flow is evident in the upper water column and in the warm salty central bolus in the transect shown (discussed below). Instead of resembling the velocity field implied by the density field, this is actually more consistent with a two layer Ekman response to the northeastward winds measured for this timeframe (Fig. 2). On the other hand, alongshelf velocities are equatorward (not shown). With only four transects and an inability to define the orientation of the Hatteras Front in July, a clear diagnosis of these velocities has not been established. At the least, these measurements demonstrate a change in the density gradients observed from that in August and January, and the absence of strong shoreward velocities within the Front. Both FINCH and the NODC archive (in Part 2) illustrate the possibility that the density contrast across the Front may diminish or reverse in spring of any given year Stream coordinate mean structure The stream coordinate averages constructed from the snapshots show strong shoreward velocities associated with the strong density gradients across the Hatteras Front in August 2004 and January 2005 (Fig. 5). January horizontal density, dynamic height and sea surface gradients were of the same order as those measured in August The January shoreward velocities were not surface intensified, as they were in August, but are situated in the middle of the water column, within the denser part of the horizontal density gradient. The July stream coordinate density average over four transects illustrates the reversed density gradient, especially over the deep water column (Fig. 5, right panels). Note the left edge has very high standard deviation in density, relative to the average values at the left edge. The stream coordinate velocity section from ADCP data illustrates the absence in July of the strong shoreward velocities in the Front that were evident in August and January. The measured velocities are not seaward, as expected from a dynamic height estimate for the lower water column. An interesting feature is the wedge of seaward velocity in the middle and upper water slightly north of the center of the transect. This is consistent with diverging isopleths in the density stream coordinate field, and is located approximately where the largest salinities and temperatures appeared in the sample July transect shown (Fig. 3). As discussed in Section 3.3, S within that patch is o35:85 psu, below typical Gulf Stream S from Flagg et al. (2002) or Pietrafesa et al. (1994), and T is closer to that measured at the moorings in SAB water than to at least surface Gulf Stream temperature. Such a bolus could result from recirculation behind the Hatteras Front, similar to that discussed in Savidge and Austin (2007) (their Figs ) Summertime stratification in MAB The range of density contrasts across the Hatteras Front observed during FINCH does not span the entire range of possibilities, as the strong summertime stratification known to occur in MAB shelf water appears only weakly in the July FINCH measurements. An August 1993 transect along the northern A line from the MMS shipboard hydrographic data illustrates that the well-known strong seasonal stratification in MAB shelf water extends to the southern end of the MAB (Fig. 6, panels A C). Vertical stratification along the C line August 1993 (Fig. 6D), is typical of the much weaker (than in the MAB) summertime stratification observed for SAB shelf water (Blanton et al., 2003). By the time the October 1993 ship section was taken, the strong T-based seasonal MAB stratification seen along the A-line in Fig. 6 had been eroded, replaced by a weaker stratification more typical of winter (Fig. 7, panels A C). Stratification in late October was also weak in SAB shelf water along the C line farther southward (Fig. 7D). The density contrast between the MAB and SAB shelf waters before and after the fall transition in these MMS sections is illustrated in composite T S diagrams (Fig. 4C) of data from the 5th hydrographic station seaward along both the A and C lines in August and October In August 1993, the strong contrasts in T, S and density between upper MAB, SAB, and lower MAB shelf waters are evident, with density ranging from 20.5 to 24.4 s y from upper to lower MAB water, with SAB waters of intermediate density of s y. The coolest, freshest component is upper

8 8 Fig. 5. Stream-coordinate means and standard deviations of Hatteras Front along-front velocity (positive shoreward) and density fields from transects taken during the three FINCH field seasons. Transects are oriented such that SAB shelf water appears at the left of the panels, transitioning into MAB shelf water at the right side of the panels. Left column of panels: August 9, 2004, middle panels: January , and right panels: July From top to bottom, the panels in each column are: mean ADCP along-front velocities; mean density; standard deviations of the ADCP along-front velocities; standard deviations of the densities. Contour interval for velocity means and standard deviations is 0.05 m/s, bold line is 0.0 contour, with positive shoreward in orange. Contour interval for the density means and standard deviations is 0.2 s y units). Maps of transect locations included in the stream coordinate means are shown in Fig. 1, panels B D. Notice that the horizontal scale is shorter for August than for January and July, and that the density ranges change for each month in the second row of plots. MAB shelf water, the warmest and saltiest component is SAB shelf water, and lower MAB shelf water is cooler and fresher, and of similar magnitude density to the SAB shelf water. By October, the light upper and dense lower MAB shelf waters have mixed so that the MAB shelf water became lighter than the SAB shelf water (Fig. 4C). Density of the SAB shelf water did not change much between August and October, indicating that little seasonal cooling had occurred (Fig. 4C). Note the strong resemblance between the October 1993 T S diagram and the August 2004 FINCH T S diagram (Fig. 4A), taken immediately after Hurricane Alex passed close by. In May 1992, a hydrographic section from an MMS shipboard transect along the A line shows conditions after seasonal warming near Cape Hatteras had commenced, but before strong vertical stratification had developed in the MAB shelf water. The SAB shelf waters along line C had warmed somewhat at all depths, but the MAB upper, middle and lower layer waters remained below 10 1C (Fig. 8). In these sections the cool fresh MAB shelf waters along line A are very nearly the same density as the warm salty SAB shelf waters along line C, illustrated in TS plots from the 5th station seaward along both lines (Fig. 4D). Note the strong similarity of this May 1992 T S diagram to that from the July 2005 FINCH transects (Fig. 4B). 4. Synthesis The snapshots of density contrast and associated velocity fields in the nose of the Hatteras Front suggest an evolution that may be seasonal, and therefore annually repeatable. In the following, the suggested seasonality is outlined, relative to the MMS and FINCH data analyses to date. Seasonal aspects of density

9 9 Fig. 6. Summer (August 2 11, 1993) cross-shelf sections from the MMS project, collected during Seaward Explorer cruise SE9309. Top three panels show data collected along the A mooring line: panel A: temperature, panel B: salinity and panel C: density. Panel D shows density collected along the C mooring line. Sigma-T values of 23 and lower are plotted in black, of and higher are plotted in gray. Temperature contour interval is 1 1C, salinity contour interval is 0.25, density contour interval is 0.25 kg/m 3. evolution include (1) the fall transition from strong summer stratification to lower vertical stratification in winter; (2) cooling through winter, (3) spring warming and (4) summer restratification. Whether these results are consistent with long-term seasonal evolution is then examined in detail in Part 2 using archived datasets Fall transition The destruction of summer stratification with increasing winds and decreasing surface heat fluxes into the coastal ocean in fall is well known, and has been examined in both MAB and SAB settings. Figs. 6 and 7 show the strong contrast in stratification in particularly MAB water, but also in SAB shelf water before and after the fall transition in The contribution of this mixing of especially the MAB shelf water is illustrated quite clearly in panel C of Fig. 4. Before the transition in the August 1993 MMS data, upper layer MAB shelf water is much lighter than either upper or lower layer SAB shelf water. Lower layer MAB shelf water does not occupy significantly different density space than SAB shelf water, and is roughly equivalent to upper MAB water in S, but not in T or density. After mixing, as seen in the November 1993 MMS data in Fig. 4C, MAB upper and lower layer shelf waters are less dense than SAB shelf water, due to the mixing of the light upper MAB water into the denser lower MAB water. That is, by mixing away the significant density difference between upper and lower MAB shelf water that is due to T, the strong S contrast between mixed MAB and mixed SAB shelf water results in a strong density contrast between them. The August 2004 FINCH data of either Fig. 3 or Fig. 4A shows well mixed conditions on both the MAB and SAB shelf water sides

10 10 Fig. 7. Fall (October 28 November 9, 1993) cross-shelf sections from the MMS project, collected during Seaward Explorer cruise SE9316. Top three panels show data collected along the A mooring line: panel A: temperature, panel B: salinity and panel C: density. Panel D shows density collected along the C mooring line. Sigma-T values of 23 and lower are plotted in black, of and higher are plotted in gray. Temperature contour interval is 1 1C, salinity contour interval is 0.25, density contour interval is 0.25 kg/m 3. of the Hatteras Front. The density contrast between these shelf waters is large and is due to the S difference, as the lighter MAB component is slightly cooler than the SAB component. The measured shoreward along front velocities of August FINCH are consequently not representative of summer conditions, but rather of conditions prevailing after the fall transition. The August 2004 mixing of the MAB waters by Hurricane Alex during FINCH may have occurred earlier and affected a more limited geographical area than mixing from a series of region-wide fall wind events would have. It is unknown whether MAB shelf water restratified after mixing by Alex. However, MAB mixing does occur every fall, and will alter the density space occupied by the MAB shelf waters, and therefore its contrast to the density range occupied by SAB shelf water in fall Winter cooling Equally as well known as the fall transition is the seasonal cooling of shelf water in the SAB and MAB through fall and winter. This is shown in T time series in Savidge et al. (2007) (their Fig. 3) from the two year MMS shelf moorings north and south of Cape Hatteras, and is quite evident in the T contrast between August and January FINCH values shown in Fig. 4A. The density contrast between MAB and SAB shelf water across the Hatteras Front persisted through fall cooling and the sequence of storms that punctuated the January FINCH sampling. It is conceivable that the integrated winter-long effect of cooling in the more northern MAB could exceed that in the SAB sufficiently to erode the S-based density discrepancy between

11 11 Fig. 8. Spring (April 29 May 6, 1992) cross-shelf sections from the MMS project, collected during Cape Henlopen cruise CH9222. Top three panels show data collected along the A mooring line: panel A: temperature, panel B: salinity and panel C: density. Panel D shows density collected along the C mooring line. Sigma-T values of 23 and lower are plotted in black, of and higher are plotted in gray. Temperature contour interval is 1 1C, salinity contour interval is 0.25, density contour interval is 0.25 kg/m 3. MAB and SAB shelf water. Indeed the August 2004 January 2005 cooling appears larger in MAB water than in SAB water (Fig. 4A). However, for MAB salinities near 32, density contours become increasingly vertical with cooler temperatures in T S diagrams. In that case, large changes in T are necessary to result in small changes in density, so that differential cooling between SAB and MAB shelf waters would have to be large to overcompensate for the large S difference contribution to density. Episodes of strong shoreward velocities attributed to Hatteras Front density contrasts persist through April and arguably into May in the 1992 and 1993 mooring data examined by Savidge (2002), their Fig. 5. In those winters at least, sufficiently large cooling in MAB water compared to SAB water apparently did not occur Spring warming Warming of coastal waters in spring and summer is also a well known aspect of the seasonal cycle. If SAB warming precedes or exceeds MAB warming, which is conceivable due to its more southern exposure, there is opportunity to reduce the S-based density discrepancy between SAB and MAB shelf waters in spring. For SAB salinities near 35 36, density contours become less vertical with warmer temperatures in T S diagrams, so that relatively smaller changes in T result in larger changes in density than for fresher MAB water. In Fig. 4B, it appears that July 2005 upper and lower SAB waters have warmed to T characteristic of summer, and that lower layer MAB shelf water remains cold.

12 12 T time series from the MMS shelf moorings shown in Savidge et al. (2007) (their Fig. 3) suggests this is typical, since springtime warming commenced in April in the SAB, and cold lower layer MAB T persisted through summer. In 2005 sampling, this resulted in slightly denser MAB water in the lower layer. In other years, whether the wintertime density gradient in the lower layer is eliminated, not quite eliminated, or reversed would be subject to the magnitude and timing of both heat and freshwater fluxes in both shelf waters in spring. This suggests the potential for interannual variability in the density gradient and resulting velocities associated with the springtime Hatteras Front Summer restratification Upper layer MAB shelf water does not stay cool throughout the summer, of course, another seasonal aspect of the temperature evolution that is well described. Upper layer MAB water warming in the southern MAB commences by late April in 1993, but apparently not until June in 1992, in the two year MMS T mooring data shown by Savidge et al. (2007). Upper layer MAB shelf water was not frequently sampled in July FINCH, as indicated by the absence of an upper layer MAB component in the frequency plot of Fig. 4B. However, a warm fresh light layer does appear in the individual transects for July 2005 (see the example shown in Fig. 3), which includes a strong thermocline above the lower layer cold MAB water. Strongly stratified MAB conditions were in place by May in 1996, as described from OMP data by Flagg et al. (2002). Their diagnosis indicated that the strong summertime T stratification was triggered by initial S stratification, which then facilitated rapid strong warming of the upper layer MAB shelf water. The T and depth of the surface warm layer presumably both increase over the summer, though details of that evolution in these data sets is limited by the sparse vertical coverage typical of most mooring or shipboard data. The July 2005 warm layer is only 5 10 m thick, whereas the late summer example from MMS August 1993 shows an upper layer that fills half the water column across the middle MAB shelf. With the limited summertime FINCH sampling accomplished, it is unknown whether the warm layer development is typical for mid-july, or if summer stratification became established somewhat later in 2005 than observed in other field programs. In any case, the measurements shown suggest that interannual variability in the timing of stratification onset, its eventual establishment and ultimate strength may depend on interannual wind, S and T variability in both MAB and SAB settings. The further issue is that the July FINCH data and the August MMS data illustrate a density gradient across the Hatteras Front that is not of consistent sign in the upper and lower water columns. As described above, more complicated pressure gradient fields result, with correspondingly more difficult to diagnose geostrophic velocities along the Front in spring and through summer. The July 2005 ADCP data suggest that complexity, but otherwise FINCH, MMS and the OMP data are insufficient to detail the summer evolution of alongfront velocities. 5. Conclusions The August and January FINCH shipboard undulating CTD and ADCP sections across the nose of the Hatteras Front document strong property and density gradients across a relatively narrow front, and strong shoreward velocities along the Front. Densest waters reside on the SAB shelf water side, where the warm and salty waters are several s y units denser than the MAB shelf water. Dynamic height gradients across the Front estimated from the August and January density fields are sufficient to support geostrophic velocities of the order measured in the Front, and are taken to be the cause. During the August FINCH observations, the shelf water delivered shoreward by the alongfront surface intensified jet was primarily MAB shelf water, while in the January FINCH sections, the strong shoreward directed flow along the Front carried primarily SAB shelf water (Fig. 3). Since larval assemblages have been shown to be quite disparate across the Hatteras Front (Grothues and Cowen, 1999), this implies the Front may preferentially deliver MAB or SAB shelfwater larval assemblages shoreward at different times, depending on the evolving density field or wind forcing, in ways which are presently undetermined. The results from August and January FINCH are of interest regardless of whether the process is primarily a seasonal progression or storm response. In either case, Savidge (2002) has demonstrated that strong alongfront shore directed velocities are a frequent occurrence in winter on the Hatteras shelf. Seasonal variability in T and its effects on the evolving density suggests that the density evolution may be repeated on a yearly basis. This hypothesis is examined in Part 2. The July 2005 FINCH sections indicate that for this particular year, MAB shelf water density exceeded that of SAB shelf water. The resulting density gradient across the Front was of very low magnitude, and reversed in direction from the August and January cases. Velocities measured along the Front were weakly shoreward in the lower layer, opposite that predicted from dynamic height estimates from the density gradient. However, the discrepancy is small, given the low magnitudes of both predicted and measured velocities. The near equivalence of densities in the July FINCH and the MMS May 1992 data demonstrates the possibility of reduced or reversed cross-shelf transport in the nose of the Front in spring and summer. Interannual variability in heat and freshwater fluxes within the MAB or SAB may determine the sign and magnitude of the density gradient in spring and summer. The strong property gradients measured across the Hatteras Front in all FINCH excursions are consistent with the mooring data from the earlier MMS field program, and are to be expected, based on the significantly different life histories of the shelf waters impinging on Cape Hatteras from the north and south. What had not been anticipated by prior work on the Hatteras Front was that the density contrast across the Front where these shelf waters meet might change significantly, as it did between the three FINCH excursions. In Part 2, several archived data sets are examined. That study illustrates that the density evolution in MAB and SAB shelf waters seen in these FINCH and MMS hydrographic sections is characteristic of the climatological evolution of density with seasonal T variability in MAB and SAB shelf waters. The large density contrast in fall and winter defined here applies in the climatology throughout the winter, from the fall mixing of MAB shelf waters until springtime warming begins to occur. Spring and summer density contrast between cold lower layer MAB shelf water and SAB upper and lower shelf waters in the climatology is relatively weaker. This is consistent with the possibility of interannual variability in the strength and direction of the density gradient between MAB and SAB shelf waters in spring, due to interannual variability in S or T over either the MAB or SAB shelf. Cross-shelf transport in the nose of the Hatteras Front implied by the density fields of Part 2 is consistent with transport snapshots shown here, with strong shoreward along Front velocities in winter, diminished and perhaps reversed in spring. Summertime velocities in the Front are still undetermined in sign or vertical shear, and will vary with the differing contrasts between lower MAB and SAB shelf water and upper MAB and SAB shelf water. Both the implied robust shoreward transport in