Click
Here
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112, C07006, doi:10.1029/2006JC003933, 2007
for
Full
Article
The Hatteras Front: August 2004 velocity and density
structure
Dana K. Savidge1 and Jay A. Austin2
Received 12 September 2006; revised 2 March 2007; accepted 4 April 2007; published 6 July 2007.
[1] The Hatteras Front is a persistent mesoscale cross-shelf oriented front off Cape
Hatteras, North Carolina. It is the boundary between relatively cool, fresh Mid-Atlantic
Bight shelf waters and warmer, saltier shelf waters of the South Atlantic Bight, which both
converge along-shelf upon Cape Hatteras year round. The Frontal Interaction Near
Cape Hatteras (FINCH) project was conducted in 2004–2005 to intensively sample the
Hatteras Front with shipboard ADCP and undulating towed CTD. This paper documents
velocity and density structures associated with the cross-shelf oriented zone of Hatteras
Front during the August 2004 field season. Property gradients across the Hatteras
Front are large, with temperature (T) and salinity (S) differences of 4–6°C, 2–5 psu,
respectively over distances of 1–2 km. The T and S are not completely compensating, and
a strong density (r) gradient also exists, with Dr of 2 kg/m3 across a gentler 10 km wide
front. The density gradient results in a steric sea-level height gradient of 1–2 cm
across the Front, which is in approximate geostrophic balance with a surface intensified
jet, directed shoreward along the cross-shelf oriented Front. The velocity is sheared with
depth at 3.0 102 to 5.0 102 s1 in the upper 5 m of the jet; a rate consistent with the
density gradient according to the thermal wind relationship. Shoreward transport of
4.8 104 m3/s results from the surface intensified jet. The structure of the velocity field
associated with the Hatteras Front resembles that of a slope-controlled buoyant plume, as
described by Lentz and Helfrich (2002). Velocity and density structures are similar during
both advancing (southwestward) and retreating (northeastward) motion of the Front.
Citation: Savidge, D. K., and J. A. Austin (2007), The Hatteras Front: August 2004 velocity and density structure, J. Geophys. Res.,
112, C07006, doi:10.1029/2006JC003933.
1. Introduction and Background
[2] On the continental shelf and slope near Cape Hatteras,
Mid-Atlantic Bight (MAB) shelf water and South Atlantic
Bight (SAB) shelf water converges along-shelf 90% of the
time, carrying water of significantly different origin with
large differences in temperature (T) and salinity (S) characteristics [Savidge and Bane, 2001]. This surprisingly
robust convergence supports a strong along-shelf gradient
in temperature, salinity, and density known as the ‘‘Hatteras
Front’’ [Pietrafesa et al., 1994; Savidge, 2002; Berger et al.,
1995]. The Front apparently exists in all seasons, due to
the significant climatological T-S differences between the
two shelf water masses and the consistent convergence of
both water masses on Cape Hatteras. The existence of
the Hatteras Front has been recognized for many years
[Stefansson et al., 1971].
[3] Such large along-shelf convergence requires significant offshelf export from the shelf to the open sea, which
has been reported previously for the region. See Churchill
1
Skidaway Institute of Oceanography, Savannah, Georgia, USA.
Large Lakes Observatory, Department of Physics, University of
Minnesota at Duluth, Duluth, Minnesota, USA.
2
Copyright 2007 by the American Geophysical Union.
0148-0227/07/2006JC003933$09.00
and Berger [1998] for a summary of MAB shelf water
export observations and estimates at Cape Hatteras. This
region is also known as a location where substantial
transport of shelf-derived and terrestrial particulate matter
to continental slope depocenters occurs [Biscaye et al.,
1994; Lee et al., 1991]. Recently, the Hatteras Front has
been implicated as an important conduit for offshelf export
near Cape Hatteras, with significant quantities of shelf water
advected seaward along the ‘seaward flank’ of the Front
(see Figure 1) [Churchill and Berger, 1998].
[4] Despite the net seaward export here, mechanisms of
suspected shoreward transport in this region have been a
topic of interest for some time. 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 between the two regimes [see, e.g., Checkley et al.,
1988]. Transport of pollutants from the open ocean into the
delicate Albemarle and Pamlico Sound ecosystems is also
of concern, owing to the possibility of outer shelf commercial drilling for oil and gas, or from potential shipwrecks in
the treacherous waters off Cape Hatteras.
[5] Using 2-year mooring records from a Minerals Management Service Study (February 1992 to February 1994)
Savidge [2002] investigated strong shoreward velocities
associated with the southern cross-shelf oriented portion
C07006
1 of 16
C07006
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
C07006
Figure 1. Map of the study region near Cape Hatteras, North Carolina. (left) Schematic Hatteras Front
(HF) and Gulf Stream and Mid-Shelf Front (MSF). The approximate location of the Shelf-Break Front is
illustrated by a drifter track from Lozier and Gawarkiewicz [2001] (dots at daily intervals). The two
export regions described by Churchill and Berger [1998] are shown with grey bars bracketing the limits.
Duck (DK), Cape Hatteras (CH), Ocracoke Inlet (OK), and Diamond Shoals Light C-MAN station (DSL)
are indicated. (right) Sample cruise tracks for two ships to concurrently sample the nose and seaward
flank zones of the Hatteras Front.
of the Front (Figure 1). As the Front moved along-shelf
under wind and Gulf Stream forcing, it repeatedly crossed
over the locations of several moorings in water depths of
20– 60 m, with strong shoreward velocities recorded during
periods of frontal passage, especially in the middle to upper
water column. These shoreward currents are of sufficient
magnitude to be relevant to the shoreward transport of eggs,
larvae, or pollutants, but details of the circulation are
difficult to obtain from mooring data. While moorings can
detect any unusual frontal structure that fortuitously sweeps
by (in some indeterminant direction at some unknown
frontal orientation angle and translation speed), they are
not ideal for defining horizontal structure at high resolution.
Nor were these particular moorings well suited to measuring
the vertical structure, as they were instrumented with only
three pointwise current meters and conductivity and temperature (CT) sensors on each taut line.
[6] To examine the circulation and density fields associated with the Hatteras Front, a field program was designed
and carried out in 2004– 2005. This project, Frontal Interactions near Cape Hatteras (FINCH), consisted of three
intensive sampling sessions: two weeks in August 2004,
three weeks in January 2005, and a final one week foray in
July 2005. The primary observational strategy was to deploy
two research vessels equipped with ADCP, undulatingCTD sensors, and surface mapping CT sensors, to acquire
repeated high-resolution vertical sections and horizontal
surface fields of velocity and density across both the
cross-shelf oriented limb (the ‘nose’), and the approximately
along-shelf oriented limb (the ‘seaward flank’) of the
Hatteras Front. The final July 2005 cruise used only one
ship concentrating on the nose of the Hatteras Front. Such
an approach is ideal for locating the mobile front, and
measuring its horizontally compact (cross-front) density
and velocity fields at adequate vertical and horizontal
scales, and for mapping their orientation and evolution
under meteorological forcing. In this paper, the velocity
and density structures of the nose of the front in August
2004 are examined. The seaward flank, the seasonal variability of both the nose and the seaward flank, and frontal
interactions and dynamics will be addressed in future
articles.
2. Data
[7] The August 2004 FINCH data from the nose of
Hatteras Front were collected from the R/V Fay Slover, a
55-foot dayboat owned and operated by Old Dominion
University. CTD data were collected with a SeaBird SBE19plus 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 at
water depths between about 15 and 58 m depth, with
0.25 km between up and down ‘casts’ along the zigzag
flight path. The R/V Slover is also equipped with a hullmounted 600 kHz RDI ADCP. The data shown here are at
1 m bin spacing, and have been averaged over 20 s, for an
along-track resolution of 60 m. These data were detided
using along-track tidal predictions from the ADCIRC model
[Luettich and Westerink, 1992]. Tides and inertial motions
are relatively small south of Cape Hatteras [Berger et al.,
1995; Savidge et al., 2007], and 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.
2 of 16
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
C07006
C07006
Figure 2. Hourly wind vectors from the CLKN7 CMAN station at Cape Lookout. Vectors oriented
along the positive y axis represent northward winds, those oriented along the positive x axis represent
eastward winds (oceanographic convention. (top) Winds from 17 July through 28 August 2004. (bottom)
A blowup of the 6 – 15 August timeframe. Note change in y axis limits between plots.
[8] The CTD data were linearly interpolated onto a
uniform grid coincident with the ADCP data grid for
plotting purposes. Shipboard surface mapping was also
available, using a flowthrough system equipped with a
SBE-45 CTD, sampling every 4 s. Wind information was
collected from the NDBC Cape Lookout CMAN station
(CLKN7). More relevant data from the Diamond Shoals
CMAN records (previously tower DSLN7, now buoy
41025) was unavailable after the passing of Hurricane Alex
on 4 August 2004.
3. Setting
[9] Winds near Cape Hatteras in August are predominantly upwelling favorable, and steadier in July and August
than in the rest of the year (see Savidge [2002, Figure 5] for
a two year example). However, the North Carolina coastline
is frequently subjected to the influence of passing hurricanes. In August 2004, the field program was immediately
preceded by the passage of Hurricane Alex directly through
the study site. Strong winds from Hurricanes Bonnie and
Charley were also felt at the study site, and necessitated
a slightly earlier end to field activities than planned:
12 August transit to home port instead of 14 August (Figure 2,
top). One favorable effect of Hurricane Alex was on the sea
surface temperature (SST) field, as observed from NOAA’s
AVHRR satellite sensors. Satellite imagery of SST along the
U. S. east coast is typically fairly featureless in summertime,
as solar insolation warms the surface layer of the ocean and
obscures any subsurface temperature variation that may exist.
A large temperature contrast between subsurface SAB shelf
water and MAB shelf water is apparent in archived subsurface
mooring data (summer and winter), shown quite well in the
7-year near surface salinity time series of Pietrafesa et al.
[1994]. The contrast also appears in winter satellite SST
imagery [Savidge, 2002], but is usually not evident in
summertime satellite SST imagery. Hurricane Alex mixed
away the surface warm layer near Cape Hatteras, exposing
the temperature contrast across the Hatteras Front between
the cold MAB shelf water and warmer SAB water. As a
result, the Hatteras Front was visible in SST, and easily
located for sampling with shipboard instrumentation. The
Front moved southward of Cape Hatteras during or shortly
after the passage of Hurricane Alex, where it remained
through the end of the August sampling period (Figure 3).
During sampling from 7 to 11 August, the Front moved first
3 of 16
C07006
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
Figure 3. A sequence of NOAA satellite AVHRR sea surface temperature for the Cape Hatteras region,
generated from data obtained from the Beaufort NC NOAA Coastwatch website. The bars perpendicular
to the coastline south of Cape Hatteras are markers to measure the along-shelf position of the Hatteras
Front against over time.
Figure 4. Planviews of shipboard ADCP data from 8 August transects. (top left) Shipboard ADCP
velocity data from 3.2 m below the surface. (bottom left) Data from 16.2 m below the surface. Color track
in both left plots is near surface salinity from the shipboard surface mapper, at 4-s intervals. The color
map used is the same as that used for the surface track and the salinity section shown in Figures 5a and 5d.
(right) A satellite SST image for 7 August 22:31 UTC, with vectors from the top left plot superimposed,
subsampled every fourth vector for clarity.
4 of 16
C07006
C07006
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
C07006
Figure 5. High-resolution hydrography and velocity vertical sections across the Hatteras Front for
section 6 on 8 August 2004, from shipboard ADCP and a towed undulating CTD. (a) Day-long shiptrack
with surface mapper salinity, (b) density, (c, d) CTD temperature and salinity (interpolated to the ADCP
grid points), and (e, f) shipboard ADCP across and along-shelf velocity (rotated as discussed in the text).
Positive across-shelf velocities are seaward, positive along-shelf velocities are northeastward. Figure 5a
color map is the same as in Figure 5d. Start (circle) and end (cross) points for section 6 are shown in
Figure 5a. Section 6 started at the north end and proceeded southward, so that the colder, fresher, MAB
shelf water appears at the left on the hydrographic and velocity sections, Figures 5b– 5f. Density contour
interval is 0.2 sT, velocity contour interval is 0.05 m/s (dashed line, negative; solid line, positive; bold
black line, = 0). The towed undulating flight path is shown by a dotted line on Figures 5b, 5c, and 5d.
southwestward until 9 August, and then retreated toward
Cape Hatteras to the northeast between 9 and 11 August.
Motion of the Front was more complicated than suggested
by the panels in Figure 3, as indicated in available clear
imagery for the period. However some of the variability in
the imagery may reflect only wind-blown near-surface layer
water, and not the motion of the underlying water masses.
The position and shape of the nose of the Front appeared to
be influenced by both wind and Gulf Stream variability.
Shipboard sampling verified movement of the Front, to be
discussed in the following. Winds were strong and to the
south during 6– 7 August, and weak and variable during
sampling on 8 – 11 August (Figure 2, bottom).
4. Velocity and Density Sections
[10] Plan views of the velocity field associated with the
Hatteras Front on 8 August (Figure 4) illustrate clear
shoreward directed flow associated with the Hatteras Front,
shown here for two levels: near surface (3.2 m) and in the
lower water column (16.2 m). In satellite imagery, the nose
of the Hatteras Front was advancing southwestward alongshelf, under the influence of downwelling favorable winds.
In situ data from 8 – 9 August support this, discussed below.
Five cross-front transects were accomplished on this day,
each showing along-front velocities wrapping anticyclonically around the nose of the front.
[11] Each of the transects accomplished on 8 August
crossed remarkably strong T and S fronts, with 4 – 6°C,
2– 5 psu contrast across extremely short horizontal distances
of 1 – 2 km. For example, along a N-S transect from
8 August (Figure 5a), strong T and S gradients are evident
(Figures 5c and 5d). Complex structure in the S and T fields
is apparent, which may facilitate mixing across the front. A
strong density front also exists (Figure 5b), but is not
colocated with the strongest property (S and T) fronts,
5 of 16
C07006
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
Figure 6. T-S frequency plot for entire August 2004 data set. Darker grays indicate higher frequency of
sampling. The dark band 34sT unit wide, extending from about 28°C, 36 psu to 21°C, 32 psu indicates
the T-S space that was sampled most frequently (>300 samples in each 0.1 psu 0.1°C bin) and
encompasses the properties of the targeted water masses on either side of the Hatteras Front: the MAB
shelf water at the cold fresh end, and the SAB shelf water at the salty warm end, labeled and highlighted
with ellipses in the figure. The high-salinity band at the right is Gulf Stream water (labeled ‘GS’) and the
fresh band at the left is from relatively near shore and is Chesapeake Bay Plume water (labeled ‘ChBP’),
diluted with saltier shelf water. No color bar is 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.
Figure 7. Planviews of shipboard ADCP data from 9 August transects. Plots are as described for
Figure 4. Satellite SST for 9 August 02:13 UTC.
6 of 16
C07006
C07006
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
C07006
Figure 8. High-resolution hydrography and velocity vertical sections across the Hatteras Front for
section 7 on 9 August 2004, from shipboard ADCP and a towed undulating CTD. Plots are as described
in Figure 5. Section 7 started at the northeast end and proceeded southwestward, so that the colder,
fresher, MAB shelf water appears at the left on the hydrographic and velocity sections, Figures 8b– 8f.
offset from the property fronts in the direction of the lighter
MAB shelf water. The density front is wider, with 2 sT
unit variation across a distance of about 10 km. Note that
the S and T ‘fingers’ evident in Figures 5c and 5d extend
along constant density surfaces.
[12] T-S mapped into density space from the August
transects, shown here as a frequency T-S diagram from all
August 2004 data (Figure 6), illustrates the density characteristics of the SAB and MAB water masses. The dark gray
band showing the highest number of CTD samples highlights the contrasting waters across the Hatteras Front that
the FINCH cruises were designed to examine. Warm salty
SAB shelf water (near 28°C, 36 psu) exists at the top right
end of the band, while cooler fresher MAB shelf water (near
21°C, 32 psu) inhabits the lower left extent. A precise T-S
definition for either body of shelf water is not possible, as
the characteristics change in time with precipitation, evaporation, heating and cooling. Consistent with the mooring
data described by Savidge [2002], the cold fresh MAB shelf
water is less dense than the warm salty SAB shelf water.
[13] ADCP velocity sections (Figures 5e and 5f) are
rotated into along and across-shelf components. Rotation
of axes 64.25° clockwise was determined from the principal
axis of archived mid-water-column mooring 48-hour low-
passed velocities in this location [Savidge and Bane, 2001].
These velocities are surface intensified, and oriented along
the Front throughout much of the water column, with the
lighter MAB shelf water on the right looking downstream in
the direction of the flow. Large velocities are evident though
much of the upper water column, up to 50 cm/s near the
surface, and exceeding 30 cm/s through much of the cooler
water. The largest velocities in this section and many of the
others are associated with the MAB shelf water, on the less
dense side of the Hatteras Front.
[14] On 9 August the Hatteras Front was again sampled
repeatedly. Strong along-front shoreward velocities are
again evident in the plan view, extending through much
of the water column (Figure 7). Notice the offshore reversal
at the northern end of the shorewardmost transect. This is
consistent with the SST imagery, which suggests a recirculation zone behind the Hatteras Front. The analytical model
of Lentz and Helfrich [2002] for circulation near the nose of
a buoyant plume over a sloping bottom in a rotating
reference frame may be appropriate for interpreting the
Hatteras Front observations. Their model predicts both
shoreward velocities along the nose of an advancing plume,
and a recirculation zone behind the nose of the plume, if the
plume is more slope-controlled than surface-trapped. The
7 of 16
C07006
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
C07006
Figure 9. High-resolution hydrography and velocity vertical sections across the Hatteras Front for
section 11 on 9 August 2004, from shipboard ADCP and a towed undulating CTD. Plots are as described
in Figure 5. Section 11 started at the northwest end and proceeded southeastward, so that the colder,
fresher, MAB shelf water appears at the left on the hydrographic and velocity sections, Figures 9b– 9f.
Hatteras Front does not appear to bound a surface trapped
plume; this will be discussed more fully in section 5.
[15] Vertical sections from two of the 9 August transects
are shown, one oriented along-shelf across the nose of the
Hatteras Front, and another oriented cross-shore, across the
along-shelf oriented seaward flank of the Hatteras Front
(Figures 8 and 9). Similar features appear in both transects,
suggesting continuity of flow along the length of the front.
Detailed S, T structure is again evident, as on 8 August, with
strong property fronts, and a strong, gentler density front
displaced from the strongest gradients in either S or T
(Figures 8b, 8c, 8d, 9b, 9c, and 9d). As on 8 August, the
density front is approximately 10 km wide, and the velocity
sections illustrate strongest along-front flow in the lighter
water (evident in the cross-shelf component in the alongshelf section, and in the along-shelf component of velocity
in the cross-shelf transect (Figures 8e, 8f, 9e, and 9f).
[16] Satellite SST and in situ measurements indicate that
the Front occupied its southwestward-most position (during
the cruise period) on 9 August. The furthest equatorward
position in satellite SST imagery appeared on the morning
of 9 August at 04:30 Eastern Daylight Time (EDT). By the
morning of 10 August the Front had retreated northeastward
past its position on 8 August.
[17] Motion of the Front was estimated in situ from the
observed translation of the density field along transects that
were repeated on more than one day. This is illustrated in
Figure 10, where the locations of the 23.1 sT isopycnal at
7.5 m depth along each cross-front transect are shown for
8– 11 August. As discussed in section 5, this intersection
serves as a useful marker for Hatteras Front location. The
Hatteras Front translated southwestward between 8 and
9 August, and retreated to the northeast from 9 to 11 August.
Neither the SST nor the in situ data are taken at sufficient
temporal resolution to determine exactly when the retreat
began, but transects taken throughout the day on 9 August
do not indicate rapid motion to the northeast. However, by
the morning of 10 August the Front had retreated northeastward past its position on 8 August, and continued
northward at least until the 11 August measurements. Tidal
excursions are much smaller than this, typically about 1 km
cross-shore and less than half a kilometer alongshore owing
to the dominant M2 tide [Savidge et al., 2007].
[18] In Figure 10, the northernmost limits of the more
northern transects mark the southern edge of a designated
‘danger zone’ on the navigational maps for the region. The
danger zone was heavily mined during World War II, and is
still designated as a no-trawl, no-mooring, no-anchor zone.
8 of 16
C07006
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
Figure 10. Map of Acrobat transects and Hatteras Front locations along those transects. Transects are
thin gray lines. Black asterisks mark the location of the 23.1 sT isopycnal at 7.5 m depth along each
transect, and are each labeled with the August date. Thick dark gray lines are drawn between same-day
asterisks to indicate approximate frontal location between transects.
Figure 11. Planviews of shipboard ADCP data from 11 August transects. Panels are as described for
Figure 4. Satellite SST for 11 August 03:08 UTC.
9 of 16
C07006
C07006
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
C07006
Figure 12. High-resolution hydrography and velocity vertical sections across the Hatteras Front for
section 15 on 11 August 2004, from shipboard ADCP and a towed undulating CTD. Plots are as
described in Figure 5. Section 15 started at the southeast end and proceeded northwestward, so that
the colder, fresher, MAB shelf water appears at the right on the hydrographic and velocity sections,
Figures 12b – 12f. Small arrows along bottom axes mark boundaries of the transect subsections referred to
in the text.
The captain of the R/V Slover was reluctant to tow the
Acrobat undulating CTD system through this area, so
all transects were conducted completely outside of it. By
11 August, the nose of the Hatteras Front had retreated
toward Cape Hatteras almost into the danger zone. The first
transects of the day south of 35°N (bottom left triangular
path segment in Figure 11, left) barely sampled the southern
extent of the nose region. T and S structure was complex,
with alternating patches of SAB and MAB water along the
northern parts of the transects (not shown). The ship then
traveled northeastward; the curving leg of the towing path in
Figure 11 shows the boundary of the danger zone. Along
the first 10 km of this arc, heading E-SE, the Acrobat passed
from the edge of MAB water into SAB shelf water. The
patchiness of the T and S fields evident in earlier sections of
the day also appears to a lesser extent along this transect
segment (Figures 12c and 12d). Along the next 15 km or so
(from 10 to 25 km, marked with arrows above the x axis),
heading E-NE to NE, the Hatteras Front was crossed again
at an oblique angle, moving from SAB to MAB water
across strong property and density gradients (Figures 12a,
12b, 12c, and 12d). Shoreward velocities are evident along
this section (Figures 12e and 12f), indicating that shoreward
velocities are associated with the retreating nose of the
Hatteras Front, as well as with the advancing Front sampled
on 8 – 9 August. This is in contrast to the findings of
Churchill and Berger [1998] for the seaward limb of the
Front north of Cape Hatteras. Using mooring data, they
showed that the direction of flow along the Front depended
on whether the Hatteras Front seaward limb was moving
shoreward or seaward.
[19] One additional interesting aspect of the 11 August
measurements is seen in the northeastern-most sections,
taken over Diamond Shoals directly eastward of Cape
Hatteras (the two straight line transects forming a ‘V’ north
of 35° on the eastern edge of the domain (Figure 11, left)).
Strong seaward flow of relatively warm water was observed
behind (north of) the nose of the Hatteras Front. The
satellite SST (Figure 11, right) shows slightly warmer water
wrapped clockwise around behind the colder water in the
nose of the Hatteras Front, advecting seaward in the
location where the strong seaward velocities are measured.
This may be a more extreme example of the recirculation
behind the Front seen first in the 9 August data. Whether
10 of 16
C07006
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
C07006
Figure 13. High-resolution hydrography and velocity vertical sections across the Hatteras Front for
section 17 on 11 August 2004, from shipboard ADCP and a towed undulating CTD. Plots are as
described in Figure 5. Section 17 started at the southeast end and proceeded northwestward, so that the
colder, fresher, MAB shelf water appears at the right on the hydrographic and velocity sections,
Figures 13b– 13f.
this recirculation and export is somehow enhanced or driven
by the retreating motion of the nose of the Hatteras Front is
an interesting question.
[20] Vertical shear in this export zone is in the ‘veering’
sense (clockwise with height in the water column), consistent with warm advection [Bluestein, 1992]. However,
vertical T and S sections from this zone show that the
temperature contrast is slight, and that the ‘warm’ water
moving seaward is only slightly warmer than the ambient
MAB shelf water, and not significantly more saline
(Figure 13). This water is apparently some mixture of
SAB and MAB shelf water, or simply MAB water that has
warmed more than water from deeper regions under
similar solar insolation. Since the water advected most
effectively in the nose of the Hatteras Front in the August
sections was the cooler MAB shelf water component, it is
likely that the recirculated water behind the Front would
be primarily MAB water also.
5. Stream-Coordinate Averages
[21] With a velocity jet that moves laterally (perpendicular to the jet velocities) in space and time, calculating mean
structure relative to a coordinate system that moves with the
jet, known as a stream-coordinate mean, can be quite useful
[Rossby and Gottlieb, 1998; Rossby and Zhang, 2001;
Fratantoni et al., 2001; Flagg et al., 2006]. Several preliminary steps, represented in the schematic (Figure 14), are
required to calculate a stream coordinate average from a
series of oblique transects across the curved Hatteras Front
as it moves along shelf, changing shape and orientation. In
Figure 14 (left), velocities along the Front point in two
different directions at the locations of two transect crossings. An along-front direction can be defined at each
transect crossing by the mean direction of currents along
that transect. In Figure 14 (right), a perpendicular has been
dropped to the mean direction of velocities at each transect
location (black lines), along which cross-front distance can
be measured. The dot at the southern end of each perpendicular represents the origin from which distance across the
front is measured, suitably chosen to be a southern edge of
the Front that is reasonably consistent between individual
transects.
[22] Frontal orientation can be defined in a variety of
ways. Herein it is taken to be the average direction of the jet
velocities, in the spirit of Rossby and Gottlieb [1998], who
used maximum velocity along each Oleander transect to
define Gulf Stream jet direction. As mentioned in section 4,
11 of 16
C07006
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
C07006
Figure 14. Schematic of stream-coordinate preliminary steps. White band represents the Hatteras Front,
small black arrows show along-front-oriented velocities, and gray lines indicate two contiguous transects
across the Front, each crossing being not perpendicular to the Front. The mean direction of the velocities
defines frontal orientation, as described in the text. The black lines are perpendiculars to the velocities,
and thus to the Front at the transect locations, used for measuring cross-front distance. The black dots are
the zero points along that line, with distance increasing to the right of the dot, looking downstream along
the velocity vectors.
the strongest frontal velocities were associated with the
fresher water, so a density range was defined, and all
velocities associated with water in that density range (between sT = 22.1 and 23.1) were averaged to determine the
along-front direction across each transect. The direction
estimate varied with the density range chosen, with results
varying by a few degrees depending on the upper bound of
the density range chosen. Results were relatively insensitive
to the definition of the lower density bound. The selected
upper bound was chosen to maximize stream-coordinate
mean along-front transport; higher bounds increased alongfront transport for the outer-shelf transects and decreased
transport for the shallower transects (where the shelf water
was less dense in general), while lower upper bounds
decreased along-front transport across the deeper sections
and increased transport across the transects in shallower
water nearer shore. Velocities across each transect were
rotated into along and across-front directions. The transect
line was projected onto a line perpendicular to the alongshelf direction defined, to assess cross-shelf distance along
each transect. The origin of each transect was selected as the
position where the sT = 23.1 surface crossed the 7.5 m
depth. As the lower bound of the density range defining the
jet, this seems a good definition of the edge of the Front.
The depth of 7.5 m was chosen somewhat arbitrarily as a
balance between isolating the selection from near surface
effects, while remaining in the lighter MAB waters nearer
the surface where the strongest jet velocities reside. Then, to
simplify the averaging process, the velocities and densities
were optimally interpolated onto a vertical cross-front
oriented grid with 1 km horizontal spacing, and 1 m vertical
spacing using a Gaussian correlation function with a decay
scale of 2 (meters in vertical, kilometers in horizontal).
Resulting transects were very slightly smoothed versions of
the original transects, retaining most of the significant
structure and appropriate peak magnitudes of density and
velocity. Extrapolated regions outside the domain of the
measurements were eliminated. Stream-coordinate mean
density and velocity sections were then calculated for grid
points where data existed for 5 or more of the 10 transects
from 8 and 9 August. There were 7 or more transects with
data between cross-shelf locations of 8 and +6 km for
velocity sections. For density, there was a slightly narrower
range of 6 to +4 km with 7 or more transects.
[23] The resulting stream coordinate means illustrate a
well defined surface intensified jet (Figure 15, top left).
Near-surface along-front velocity approaches 0.4 m/s, and
exceeds 0.2 m/s to 10 m depth. The jet axis tilts with
increasing depth. Relative vorticity is high in the upper
5 m of the jet itself. Left of the jet maximum, looking
downstream, the horizontal shear of the along-front component exceeds 6.0 105 s1, and is less than 3.0 105 s1 to the right of the jet axis. Vertical shear values
range from 3.0 102 to 5.0 102 s1 in the upper 5 m
of the jet. Resulting downstream mean transport is 4.8 104 m3/s, which constitutes a significant quantity of water
moved shoreward along the nose of the Front. For comparison, combined discharge from the Mississippi and Atchafalaya Rivers averages 1.8 104 m3/s [Milliman and
Meade, 1983].
[24] In the cross-stream velocity component, divergence
across the front is indicated, though the standard deviations
of the across front velocities are almost as large as the
means themselves (Figure 15, middle). The divergence
suggested is credible, however, as it also appears in the
vector plan view plots (Figures 4 and 7). Since water
depth decreases in the direction of the along-front velocities,
this is not unreasonable. Estimating divergence from the
0.1 m/s difference across a distance of 10 km evident in
Figure 15 gives 104 s1, quite similar to the careful
estimate for the Shelf Break Front calculated by Fratantoni
et al. [2001].
[25] The stream coordinate mean density field indicates a
2 sT unit difference across the Hatteras Front, of order
10 km wide. The density contrast accounts for the measured
velocities, both near the surface, where they geostrophically
balance a SSH gradient across the Front caused by the steric
set up, and in the magnitude of the vertical shear through the
water column, due to a diminishing horizontal pressure
gradient with depth. To illustrate this, the dynamic height
relative to 20 m depth was calculated from the streamcoordinate average density field, using the UNESCO seawater routines encapsulated in the CSIRO seawater matlab
toolbox [Morgan, 1994]. A sea level setup of a few cm
12 of 16
C07006
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
C07006
Figure 15. (left) Stream-coordinate averages and (right) standard deviations. (top) Along-front
component, positive looking downstream (into the page). (middle) Cross-front velocities, positive to the
right of the downstream direction. (bottom) Density, with denser SAB shelf water at the lower left, and
lighter MAB water to the upper right. The horizontal axis on all plots is distance across the Front
increasing to the right looking downstream along the Front (into the page), vertical axis is water depth.
Averages were calculated where 5 or more transects of the 10 total had data.
across the 10 km Front is evident (Figure 16, top). This is
consistent with similar calculations made for all individual
cross-Front transects taken in August 2004 (not shown). As
estimated [Savidge, 2002], this sea surface height gradient is
of the appropriate magnitude to account for the stream
coordinate along-front velocities of the magnitude observed
in the Hatteras Front. Geostrophic velocities were also
calculated, using the dynamic method (using tools from
CSIRO), referenced to 20 m. The density field was first
smoothed in the horizontal with a 5 point triangular filter,
which diminishes the horizontal gradients somewhat, but is
necessary to estimate gradients in the density field at
horizontal scales consistent with the velocity field structure.
The predicted near-surface velocities are of nearly the same
magnitude as those observed, diminishing with depth at
approximately the same rate as observed, owing to a
diminishing pressure gradient with depth (Figure 16). Geo-
strophic velocities and shear estimated from unsmoothed
density are significantly noisier, but are as large as the
stream coordinate mean observed velocities (not shown).
6. Plume Character
[26] Finally, it was suggested in section 4 that the observed velocities in the nose of the Hatteras Front, and in an
apparent recirculation zone north of the Front, were consistent with those predicted for a slope-controlled buoyant
gravity current in a rotating reference frame, as developed
by Lentz and Helfrich [2002]. Their key nondimensional
parameter is
13 of 16
cw =ca ¼
ð2Qg0 f Þ1=4
;
ag0 =f
ð1Þ
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
C07006
C07006
and narrower Chesapeake Bay Plume examined by Lentz et
al. [2003] behaves more like a surface trapped plume, with
an estimated cw/ca 0.2.
7. Discussion
Figure 16. Cross-front SSH and geostrophic velocity
calculated from the stream-coordinate average density field
using the dynamic method, referenced to the 20-m stream
coordinate mean ADCP along-front velocity. Contour
interval is 0.05 m/s, and thick black contour is zero line.
where cw is the nose propagation speed in the wall limit
(steep bottom slope), ca is the nose propagation speed in the
slope-controlled limit (small bottom slope), Q is plume
transport, g0 is reduced gravity, g rMABr rSAB , f is the
SAB
Coriolis parameter, and a is bottom slope. The value of this
parameter will indicate whether a buoyant plume will
propagate as a slope-controlled (cw/ca ! 1) or surfacetrapped (cw/ca ! 0) feature. Estimating roughly using
f ¼ 104 =s;
a¼
60 m
;
60 km * 1000 m=km
g0 ¼ 9:8 m=s * 2=1024;
Q ¼ 48000 m3 =s;
yields
cw =ca 3:4:
Compared to the range of cw/ca values examined by Lentz
and Helfrich [2002] in laboratory experiments, this value is
sufficiently large to indicate a buoyant feature that is slopecontrolled in character. By contrast, the significantly fresher
[27] The primary result of this paper is that strong
velocities and density gradients are associated with the
Hatteras Front, with very strong shoreward directed crossshelf flow resulting in the nose of the Front. This result was
indicated in the prior sporadic sampling by the moorings
reported on by Savidge [2002]; here that somewhat sketchy
result has been verified and elaborated substantially. Property gradients are large, with DT and DS 4– 6°C, 2 – 5 psu,
respectively, and exist over quite narrow fronts, of order 1 –
2 km. The T and S are not completely compensating, and a
strong density gradient also exists, with Dr of 2 kg/m3
across a gentler 10 km wide front. The density front is not
precisely colocated with the narrower property fronts, and is
displaced toward the cooler fresher MAB shelf water side of
the Front. The vertical and horizontal T and S structure is
quite detailed and interesting. Fingers of SAB or MAB
water can be seen extending across the sections along
constant density surfaces.
[28] Shoreward velocities exceed 0.2 m/s in the upper 5 –
10 m of the jet, and are of sufficient magnitude to transport
passive particles or pollutants across the 60 km wide Cape
Hatteras shelf from the shelf edge to the nearshore in less
than two days. The velocity jet is surface intensified, and
geostrophically balances a cross-front sea surface height
gradient that results from the steric setup of the lighter MAB
water over the denser SAB shelf water. The vertical shear
evident is also consistent with that estimated from the
thermal wind relationship and the measured density fields.
The strong cross-shelf oriented flows in the nose of the
Hatteras Front are present regardless of whether the Front is
translating poleward or equatorward.
[29] Without more information from closer to shore and
farther north or south of the Front, the fate of the water
delivered cross-shelf along the nose of the Hatteras Front is
unclear. If the plume bounded by the Hatteras Front is
slope-controlled, as indicated by the section 6 estimate of
the Lentz and Helfrich [2002] parameter, the circulation
behind the Front might be expected to resemble that of their
idealized slope-controlled plume, with transport along the
nose acting to fill in a rather quiescent region behind (north
of) the nose, in a coordinate system moving with the nose of
the Front. This appears to be consistent with the ‘swirl’ of
water behind the Front evident in Figure 11, and the evidence
noted in section 4 and Figure 13 that the water advecting
seaward behind the retreating Front is likely modified MAB
water. However, the plume passes over quite shallow local
bathymetry at Diamond Shoals, either retreating or
advancing, which will likely alter the details of circulation
north of the Front from the idealized case, and affect the
ultimate fate and mixing properties of the plume waters.
[30] An important aspect of the circulation at Cape
Hatteras is that, whatever the wind and Gulf Stream
variability may be, and whatever the details of the shelf
circulation, the along-shelf currents deliver more water to
the area than they remove 90% of the time [Savidge and
Bane, 2001]. Prior mooring and drifter deployments show
14 of 16
C07006
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
substantial offshore export here, with very short timeframes
required [Berger et al., 1995], consistent with the robust
along-shelf convergence. Evidence for shoreward crossshelf transport has been more elusive. Regional modeling
efforts (for example those in the South Atlantic Bight
Recruitment Experiment, SABRE) have not, to date, included the substantial horizontal density gradients found
near Cape Hatteras, and have failed to demonstrate substantial shoreward directed currents near Cape Hatteras
[Werner et al., 1999]. Velocities associated with features
like the Hatteras Front are only captured in mooring data
when the Front moves past one particular mooring or
another, and their signatures are unlikely to be impressed
upon averages at specific locations. These factors have all
combined to successfully mask the existence of distinct
shoreward directed, Hatteras Front-associated velocities
here. During FINCH, it was possible to locate and intensively sample velocities across the Hatteras Front with a
mobile platform (a ship), and the strong velocities associated with the Front became clear.
[31] There remains the question of whether velocities of
this magnitude exist all the time, or only during and in the
aftermath of extreme events, such as Hurricane Alex, which
immediately preceded the August FINCH cruises. The
durability argument is supported by the constancy of the
presumed forcing; the density contrast and steric height
difference must routinely exist at the boundary between the
climatologically dissimilar SAB and MAB shelf water
masses that converge upon Cape Hatteras year round. The
two year mooring data set from 1992 – 1994 shows evidence
throughout the experiment that the Front did exist, sweeping
past one mooring or another frequently. It was accompanied
in all seasons by large shoreward directed velocities, despite
the large shelf-water export to the sea required by the alongshelf convergence. On the other hand, the width of the Front
may vary, such that the SSH and density gradients might be
reduced under less extreme wind forcing. One cannot rule
out the possibility that these results, gathered in the immediate aftermath of Hurricane Alex, are a measure of hurricane response, not of the typical frontally associated
velocity and density structures. The same remarks apply
to Hatteras Front data collected in January 2005, which will
be discussed in a future article. In winter, the Cape Hatteras
region is subject to strong recurrent northeasterly winds
alternating with periods of somewhat relaxed winds from
other compass points. The winds of winter of 2005 were no
exception, and the FINCH cruises were conducted in the
relative calms following periods of extreme winds. Whether
the present results are to be viewed as representative of
usual or unusual circumstances on the shelf depends on
whether the Cape Hatteras region is event-dominated or not.
Extreme wind events are not unusual at Cape Hatteras at
any time of year.
[32] While persistent mesoscale fronts oriented across
isobaths are unusual, other examples do exist, in regions
where coastal water masses converge along shelf. For
example, on the east coast of South America, northward
flowing shelf water of subantarctic origin and southward
flowing shelf water of subtropical origin converge near
35°S, resulting in large along-shelf gradients in temperature and salinity across the ‘‘Subtropical Shelf Front’’ [see
Piola et al., 2000, Figures 6 and 7]. These authors further
C07006
suggest the along-shelf convergence on the shelf may be the
shoreward extension of the convergence of the western
boundary currents adjacent to the shelf, the Brazil/Malvinas
Current Confluence. The Hatteras Front also coincides with
a region of deep ocean convergence seaward of the continental slope (the Gulf Stream flowing northward and the
nearshore limb of the Slope Sea Gyre flowing southward
[Csanady and Hamilton, 1988]). From this one may reasonably speculate that along-shelf convergence on the shelf
could potentially occur shoreward of other western boundary current convergence zones or separation points, subject
to the details of local wind and buoyancy forcing on the
shelf, making mesoscale cross-shelf oriented fronts like the
Hatteras Front, and the energetic velocities associated with
it, of possible significance in a variety of locations.
[33] Acknowledgments. Support for this study came from NSF OCE
grant 0406543. We appreciate the advice and cooperation of our collaborator Glen Gawarkiewicz, who read and commented on the manuscript.
Comments of two anonymous reviewers and our JGR Editor were also
helpful. Logistical support was provided by the Center for Coastal Physical
Oceanography (CCPO), Old Dominion University, Norfolk, Virginia.
Assistance from Chris Powell, Captain Richard Cox, and First Mate Laura
Gibson of the R/V Slover (Old Dominion University, Norfolk, Virginia) is
greatly appreciated. We especially thank Trent Moore for technical and
cruise support, Mark Santana and Hyun Yoon for assisting as science crew
on the August 2004 cruise, and the graduate students of CCPO for their
timely assistance in a variety of roles.
References
Berger, T. J., P. Hamilton, R. J. Wayland, J. O. Blanton, W. C. Boicourt,
J. H. Churchill, and D. R. Watts (1995), A physical oceanographic field
program offshore North Carolina, Tech. Rep. OCS Study MMS 94-0047,
Miner. Manage. Serv., U.S. Dep. of the Inter., New Orleans, La.
Biscaye, P. E., C. N. Flagg, and P. G. Falkowski (1994), The Shelf Edge
Exchange Processes Experiment, SEEP-II: An introduction to hypotheses, results and conclusions, Deep Sea Res., Part II, 41, 231 – 252.
Bluestein, H. B. (1992), Synoptic-Dynamic Meteorology in Midlatitudes,
vol. 1, Principles of Kinematics and Dynamics, 431 pp., Oxford Univ.
Press, New York.
Checkley, D. M., S. Raman, G. L. Maillet, and K. M. Mason (1988), Winter
storm effects on the spawning and larval drift of a pelagic fish, Nature,
335, 346 – 348.
Churchill, J. H., and T. J. Berger (1998), Transport of Middle Atlantic Bight
shelf water to the Gulf Stream near Cape Hatteras, J. Geophys. Res., 103,
30,605 – 30,622.
Csanady, G. T., and P. Hamilton (1988), Circulation of Slope Water, Cont.
Shelf Res., 8, 565 – 624.
Flagg, C. N., M. Dunn, D.-P. Wang, H. T. Rossby, and R. L. Benway
(2006), A study of the currents of the outer shelf and upper slope from
a decade of shipboard ADCP observations in the Middle Atlantic Bight,
J. Geophys. Res., 111, C06003, doi:10.1029/2005JC003116.
Fratantoni, P. S., R. S. Pickart, D. J. Torres, and A. Scotti (2001), Mean
structure and dynamics of the shelfbreak jet in the Middle Atlantic Bight
during fall and winter, J. Phys. Oceanogr., 31, 2135 – 2156.
Lee, T. N., J. A. Yoder, and L. P. Atkinson (1991), Gulf Stream frontal eddy
influence on productivity of the southeast U.S. continental shelf, J. Geophys. Res., 96, 22,191 – 22,205.
Lentz, S. J., and K. R. Helfrich (2002), Buoyant gravity currents along a
sloping bottom in a rotating fluid, J. Fluid Mech., 464, 251 – 278.
Lentz, S. J., S. Elgar, and R. T. Guza (2003), Observations of the flow field
near the nose of a buoyant coastal current, J. Phys. Oceanogr., 33, 933 –
943.
Lozier, M. S., and G. Gawarkiewicz (2001), Cross-frontal exchange in the
Middle Atlantic Bight as evidenced by surface drifters, J. Phys. Oceanogr., 31, 2498 – 2510.
Luettich, R., and J. Westerink (1992), A three dimensional circulation
model using a direct stress solution over the vertical, in Computational
Methods in Water Resources IX, edited by T. Russell et al., Comput.
Mech. Publ., Southhampton, U. K.
Milliman, J. D., and R. H. Meade (1983), World-wide delivery of river
sediment to the oceans, J. Geol., 91, 1 – 21.
Morgan, P. P. (1994), SEAWATER: A library of MATLAB computational
routines for the properties of seawater, Tech. Rep. 222, Commonw. Sci.
and Ind. Res. Org. Mar. Lab., Hobart, Tasmania, Australia.
15 of 16
C07006
SAVIDGE AND AUSTIN: HATTERAS FRONT STRUCTURE
Pietrafesa, L. J., J. M. Morrison, M. P. McCann, J. Churchill, E. Bohm, and
R. W. Houghton (1994), Water mass linkages between the Middle and
South Atlantic Bights, Deep Sea Res., Part II, 41, 365 – 389.
Piola, A. R., E. J. D. Campos, O. O. Moller, M. Charo, and C. Martinez
(2000), Subtropical shelf front of eastern South America, J. Geophys.
Res., 105, 6565 – 6578.
Rossby, T., and E. Gottlieb (1998), The Oleander project: Monitoring the
variability of Gulf Stream and adjacent waters between New Jersey and
Bermuda, Bull. Am. Meteorol. Soc., 79, 5 – 18.
Rossby, T., and H.-M. Zhang (2001), The near-surface velocity and potential vorticity structure of the Gulf Stream, J. Mar. Res., 59, 949 – 975.
Savidge, D. K. (2002), Wintertime shoreward near-surface currents south of
Cape Hatteras, J. Geophys. Res., 107(C11), 3205, doi:10.1029/
2001JC001193.
Savidge, D. K., and J. M. Bane (2001), Wind and Gulf Stream influences
on along-shelf transport and off-shelf export at Cape Hatteras, North
Carolina, J. Geophys. Res., 106, 11,505 – 11,527.
C07006
Savidge, D. K., C. R. Edwards, and M. Santana (2007), Baroclinic effects
and tides on the Cape Hatteras continental shelf, J. Geophys. Res.,
doi:10.1029/2006JC003832, in press.
Stefansson, L. P. Atkinson, and D. F. Bumpus (1971), Hydrographic properties and circulation of the North Carolina shelf and slope waters, Deep
Shelf Res., 18, 383 – 420.
Werner, F., B. Blanton, J. A. Quinlan, and R. Luettich (1999), Physical
oceanography of the North Carolina continental shelf during the fall and
winter seasons: Implications for the transport of larval menhaden, Fish.
Oceanogr., 8, suppl. 2, 7 – 21.
J. A. Austin, Large Lakes Observatory, Department of Physics,
University of Minnesota at Duluth, Duluth, MN 55812, USA. (jaustin@
d.umn.edu)
D. K. Savidge, Skidaway Institute of Oceanography, 10 Ocean Science
Circle, Savannah, GA 31411, USA. (dana.savidge@skio.usg.edu)
16 of 16