Professional Documents
Culture Documents
www.elsevier.com/locate/coastaleng
Abstract
A laboratory study of the flow over a bar with a single rip channel has been performed. First, the well-known pattern of a bar
circulation cell with a strong offshore-directed current out through the rip channel and a weaker onshore-directed return flow over
the bar is documented. Then measurements of the three-dimensional structure of the flow in the area where the rip channel, the
bar and the trough meet and well inside the rip channel are presented. These measurements reveal that 3D effects play an
important role, and that a depth-integrated viewpoint may not always be sufficient for predicting the flow in the near bed region.
Particle-tracking experiments illustrate the near bed flow pattern over the entire area. These demonstrate how the overall
trajectory pattern changes as a function of the distance of wave breaking from the bar crest: For some conditions, the rip current is
fed from the trough and for other conditions it is fed directly from the bar. Both the 3D measurements and the trajectory tests show
the existence of a weaker onshore-directed near-bed drift in the area where the rip current ceases. Finally, in a series of sensitivity
tests, measurements of the rip current intensity for different wave climate and water level conditions reveal a strong correlation
between the rip current intensity and the wave height (both normalized). D 2002 Elsevier Science B.V. All rights reserved.
Keywords: Rip channels; Rip current; Three-dimensional flow; Near-shore circulation; Coastal processes
0378-3839/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 - 3 8 3 9 ( 0 2 ) 0 0 0 3 5 - 2
224 N. Drønen et al. / Coastal Engineering 45 (2002) 223–238
over the bar and into the rip channel, but it cannot
reproduce the tendencies for the rip to form a mean-
der-like pattern offshore of the rip channel, as ob-
served in the experiments of Haller et al. (1997).
In summary, the present paper presents measure-
ments of the three-dimensional flow pattern over a
schematized bar/rip channel topography in a wave-
flume (normal wave incidence). To illustrate specifi-
cally the near bed flow pattern, bed particle trajecto-
ries over the topography are included. Finally,
changes in flow pattern and rip current intensity for
different wave climates and water levels are presented.
The experiments were performed in a 4-m wide Fig. 2. Non-dimensional spectrum of surface elevation (field data)
and 30-m long wave tank at ISVA’s laboratory. A applied for generation of irregular waves.
model of a bar with a rip channel was set up, and the
bed topography, which is sketched in Fig. 1, was kept
represent a real topography, but rather to schematize
constant throughout the experiment. The slope of the
some typical features associated with bar/rip channel
offshore side of the bar was 1/27, and the beach slope
systems. With respect to the along-shore scaling this
was 1/17. The bar height (height of crest over trough
geometry is very similar to the one used in Haller et
level) was 0.13 m and the width was 4.8 m. The width
al. (1997). The dark areas in Fig. 1 represent sloping
of the trough was 1.9 m. The width of the rip channel
parts placed at the different junctions to avoid any
was 1 m. These scales have not been selected to
sharp corners in the topography.
As stated earlier, the model can be interpreted as a
section of a longshore bar periodically interrupted in
the along-shore direction by rip channels. The model
thus corresponds to half a wave length (in the along-
shore direction) of such a bar/rip channel system, and
the side-walls in the cross-shore direction correspond
to symmetry lines normal to the coast. In this set-up
the rip current is, due to the side-wall, not allowed to
fluctuate in the longshore direction. This may explain
why the low-frequency motion observed in Haller et
al. (1997), where the rip was free to oscillate, was not
observed in the present study.
The wave generator produced regular or irregular
waves that were unidirectional and normal to the
coastline. Irregular wave generation was based on a
wave recording from the field. The bandwidth was
rather broad, corresponding to a bandwidth measure q
of 0.35, where q is defined as
Fig. 1. Sketch of physical model. Top: side view of bed topography. rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
m1 m1
Bottom: plan view of bed topography. Hatched areas are sloping q¼ 1
parts. m0 m2
226 N. Drønen et al. / Coastal Engineering 45 (2002) 223–238
Fig. 3. (a) Plan view of the measurement locations for tests 1a – c and 3a – k. (b) Plan view of the measurement locations for tests 1d – e and 3*.
ities were measured 2 cm from the bed, at five velocity at one third of the depth above bed level in
different cross-shore positions (separated by 25 cm) test 1a, are plotted.
in the rip channel. For each cross-shore position, In Fig. 4a, the distribution of wave height over the
measurements were made at four different longshore domain is given as isolines (in centimeters). Waves
positions; in this way, horizontal profiles of cross- coming in from the wave maker (located outside the
shore near-bed flow down the rip-channel were figure to the right) are seen to be homogeneously
obtained. The locations of the measurements are given distributed along-shore before they reach the bar/rip
in Fig. 3b. channel. As the waves travel over the bar along-shore
For tests 1b –e, samples were made over a period gradients in the wave height are observed. The decay
of 50 waves at 33.3 Hz. pattern of the wave height over the bar/rip channel
The purpose of tests 3a – k was to measure the clearly shows that the waves are generally breaking
maximum velocity in the rip channel for different test more over the bar than in the rip channel (where there
conditions. Before choosing a strategy for measuring hardly is any breaking). It is furthermore noticed that
the maximum velocity, a pilot test, test 3*, was the waves over the bar are not breaking at the same
performed, where the longshore profile of the cross- distance to shore despite the fact that the topography
shore velocity in the rip channel was measured. On is locally along-shore uniform. In fact, the breaking
the basis of test 3*, the following measurement seems to be larger close to the wall at y = 4 m
strategy was determined. Tests 3a –k concern meas- compared to the middle part of the bar ( y = 2– 2.5
urements at one third of the depth from the bed, 10 cm m). In the proximity of the rip channel (say around
from the wall. The cross-shore position was shifted y = 1.5 m) the wave decay also is larger compared to
down the channel (see Fig. 3a) until a maximum in the the middle part. This effect of along-shore variations
mean velocity was identified. This procedure was in the wave field not directly supported by the under-
necessary because the position of the maximum rip lying topography, may be due to the presence of
current velocity to a high degree depended on the test standing edge waves (standing waves normal to the
conditions. The sampling frequency in all tests was 50 direction of wave propagation). Standing synchronous
per peak period (for both wave and velocity measure- edge waves in a wave flume were observed and
ments), and the total number of waves was 50. described by Bowen and Inman (1969) for an along-
From velocity time series the overall time-averaged shore uniform plane beach topography, and a standing
(mean) velocities were found. wave system may possibly also exist over a more
complex topography like the present one. On the other
3.3. Particle tracking hand, if standing waves were present in the present
case, the effect was certainly much smaller compared
To visualize the flow for different test conditions, to what — for some parameter settings — could be
particle-tracking tests were performed. A ball (approx- observed for the corresponding along-shore uniform
imately 2 cm in diameter) with a relative density case. In preliminary tests, before the rip channel was
around 1.05 was released at selected positions, and introduced in the bar (i.e., a uniform barred topog-
its movement over the bed was taped on video. The raphy), standing waves were for some parameter
trajectories were then found by identifying the posi- settings seen to be generated. In those cases, they
tion of the ball relative to a grid on the bed at regular could easily be identified by just looking at the water
time intervals. surface and the breaker line (amplitudes in the order
centimeters). In the case of a non-uniform topography
like the one treated here, standing waves were on the
4. Results and discussion other hand not detectable by eye for any of the
parameter settings used in the study. Thus, the pres-
4.1. Horizontal circulation ence of the rip channel was assumed to dampen the
effect of standing waves compared to the uniform
In Fig. 4a – c, the wave height distribution, the case. Furthermore, in order to reduce the possible
mean surface elevation and the mean horizontal effect of standing waves in the present specific test,
N. Drønen et al. / Coastal Engineering 45 (2002) 223–238 229
Fig. 4. (a) Wave height distribution in cm for test 1a. (b) Mean surface elevation in cm for test 1a. (c) Mean horizontal velocity distribution at
one third of the depth from the bed, test 1b. The waves are traveling from right to left.
230 N. Drønen et al. / Coastal Engineering 45 (2002) 223–238
the wave parameters were chosen so that standing because the Eulerian wave drift is not present in the
waves in the corresponding uniform case did not measurements. The Eulerian wave drift exists above
appear. wave trough level and gives an onshore mass-flux
In Fig. 4b, the mean surface elevation in centi- contribution, which balances out the offshore-directed
meters is given. Comparison between the wave height mass-flux below the wave trough.
distribution (Fig. 4a) and the mean surface elevation From visual observation of dye movement on the
distribution (Fig. 4b) indicates that the onshore-direc- beach, indications of a second circulation cell rotating
ted wave forcing induced by wave breaking, i.e. in the opposite (clockwise) direction on the beach
gradients in the shore normal radiation stress (Lon- were found. This second cell is not directly repre-
guet-Higgins and Stewart, 1964), locally tend to be sented in the laser measurements (no measurements
balanced by gradients in the mean surface elevation were made on the beach), but in the vicinity of the
(set-up). Because of the along-shore non-uniformity in side-wall ‘intersecting’ the bar ( y = 4 m) a strong
the wave field, the set-up pattern is relatively com- offshore-directed flow is observed close to the beach.
plex. Notice first the strong offshore (directed) gra- The strength of this ‘jet’ is in the same order of
dient in the rip channel around x = 0 m. This gradient magnitude as the strength of the rip current, and
is by far the strongest gradient in the entire domain not seems to have some influence on the dynamics on
balanced by radiation stress gradients (which is also the bar part just offshore. The influence of this current
reflected in location of the maximum flow velocity in on the rip current is believed to be of secondary
Fig. 4c). This gradient is not due to wave breaking importance. A beach circulation cell rotating in the
(there are hardly any breaking in the rip channel). It opposite direction of the bar/rip channel circulation
must be supported by some other mechanism. Close was also found in Haller et al. (1997).
to the transition between the rip channel and the bar,
the set-up over the bar is slightly larger than the set-up 4.2. Three-dimensional flow distribution
in the rip channel, corresponding to more waves
breaking over the bar compared to the rip channel The three-dimensional distribution of mean hori-
(almost no wave breaking). This creates — locally — zontal flow velocities from test no 1c is shown in Fig. 5.
along-shore gradients in the surface elevation between The profiles cover the water column below the wave
the rip channel and the bar, eventually driving a flow trough only. Thus, the wave drift profile from around
from the bar toward the rip channel as depicted in Fig. the wave trough to the wave crest is not resolved. From
4c, hence supporting the strong rip channel gradient. this plot, several three-dimensional features can be
Fig. 4c shows the horizontal velocity field at a seen. In the rip channel, the strong outward-directed
given level — one third of the depth above bed level. current is seen to reduce rapidly seawards after having
First of all, the significant offshore-directed current (in reached some maximum. In fact, a small onshore-
the order of magnitude 0.3 m/s) out through the rip directed drift (strongest close to the bed), is observed
channel is noticed. in the outer region of the channel where the offshore-
In contrast to the classic conceptual idea where the directed rip current has disappeared. The offshore-
set-up created over the bar induce feeder-currents in directed rip flow has a tendency to be stronger closer
the trough, the present test shows a lateral flow going to the bed, but as a first approximation the profile is
from the bar more directly into the trough. The driving relatively uniform especially around the location where
agent for this lateral flow is the local gradient in the the current is maximum. On the bar crest and around
mean surface elevation in the proximity of the rip the point where the bar, the trough and the rip channel
channel. The effect of the rip current being fed meet, the orientation of the flow is seen to be very
directly from the bar is discussed later for the particle dependent on the vertical position, suggesting that
trajectory tests, where flow patterns for different wave three-dimensional effects like helical motion and/or
heights and water levels are investigated. undertow are important for the dynamics.
Notice that integration of the cross-shore velocities Fig. 6 shows the development of the vertical
in Fig. 4c over a cross-section of the flume, say at profiles of cross-shore mean velocity (at 30 cm from
x = 1.0 m, yields a seaward-directed net flux. This is the side-wall) down the rip channel for test 1d (note
N. Drønen et al. / Coastal Engineering 45 (2002) 223–238 231
that the vertical scale in the figure is grossly exagger- channel, but as the offshore-directed current weakens,
ated). The same tendencies as in test 1c are recognized it is ‘‘separated’’ from the bed by an onshore-directed
in this test. The rip channel flow has a tendency to be flow component. The observed onshore flow in the
slightly stronger near the bed in the start of the
Fig. 6. Cross-shore velocity profiles at several locations in rip Fig. 7. Cross-shore mean velocity 2 cm from bed in rip channel,
channel, test 1d. test 1e.
232 N. Drønen et al. / Coastal Engineering 45 (2002) 223–238
outer region is likely to be connected to drift and/or depth from the bar into the channel. As the lateral
streaming in the wave boundary layer. The effect of flow moves closer to the ‘imaginary’ symmetry line of
the rip current being slightly stronger near the bed in the rip channel (the wall), a relatively higher pressure
the start of the channel, may possibly be due to the will thus be build up in the upper part of the water
following three-dimensional effect. In the start of the column in order to curve the along-shore streamlines
channel, the rip current is being built up by the lateral downwards. The lower part of the rip current therefore
(along-shore) flow from the bar. This lateral flow is is accelerated slightly which results in a rip current
not uniformly distributed over the vertical, but stron- profile with relatively higher velocities in the lower
gest in the upper part of the water column. We expect part. This effect is, e.g. analogous to the mechanism
this vertical non-uniformity in the lateral velocity that generates a horseshoe vortex around a vertical
profile mainly because of the sudden increase in water pile placed in a steady current.
Fig. 8. Particle trajectories, tests 2a – d. (a) Large waves — regular waves. (b) Small waves — regular waves. (c) Low water level — irregular
waves. (d) High water level — irregular waves.
N. Drønen et al. / Coastal Engineering 45 (2002) 223–238 233
Fig. 8 (continued ).
4.3. Near-bed flow in rip channel in plan view can feel the rip current becomes increasingly narrow
seawards. The results for the near bed velocities
Fig. 7 is a plan view of the cross-shore mean combined with the ‘‘lift’’ or ‘‘separation’’ of the rip
velocity at 2 cm from the bed in the rip channel, test current from the bed (Fig. 6), leave the impression of a
1e. The figure shows that the offshore-directed veloc- highly three-dimensional flow structure in this region.
ity in the rip channel has its largest value close to the The near bed flow seems to converge seawards, but at
wall. The figure also shows how the onshore-directed the same time the flow (taken at some distance from
flow is ‘competing’ with the offshore-directed rip the bed) also creates a circulation pattern in the
current. As the mean offshore-directed rip current horizontal plane, where the flow after having traveled
weaken seawards, the onshore-directed flow close to some distance in the rip channel tends to move back
the bed becomes more dominant. The separation line into the bar region again (Fig. 4c). Hence, the picture
between the two flow components is indicated by the is that in the region where the rip current strength
dotted line, and shows how the region where the bed decreases the near-bed flow tends to converge, but the
234 N. Drønen et al. / Coastal Engineering 45 (2002) 223–238
‘‘separated’’ upper flow tends to diverge in the hori- illustrated by the bed particle trajectories, Fig. 8a– d)
zontal plane. some considerations had to be made on what defini-
tion of ‘rip current intensity’ should be used, that is,
4.4. Particle trajectories where in the rip channel to measure the current.
For a given test configuration, the strength of the
Fig. 8a –d shows the results from the four particle rip current was found to increase in the seaward
trajectory tests (tests 2a –d). Fig. 8a and b shows the direction from where the rip channel meets the trough
particle trajectories corresponding to tests 2a and 2b to some position down the channel where the current
(regular waves, for different wave heights). Fig. 8c reached a maximum before ceasing further offshore
and d shows the trajectories corresponding to tests 2c (see Figs. 4c, 5 and 6). For the following tests it was
and 2d (irregular waves, for different water levels). found natural to define ‘the intensity’ of the rip
In all four cases, it is seen how the trajectories of current as the value of the rip current velocity at its
the ball released from a variety of positions on the bar maximum. In test 3*, the cross-shore position of the
tend to merge as they leave the bar (either going into maximum rip current velocity was determined, and at
the rip channel directly or passing the trough first). this position (x = 1 m) the along-shore variation of the
Notice furthermore how all trajectories are trapped in cross-shore velocity was measured (at one third of the
a stagnation area down the channel. This area repre- depth over the bed). The result is plotted in Fig. 9. It is
sents a position where the mean cross-shore current in seen how the cross-shore velocity increases almost
the rip channel was observed to ‘‘separate’’ from the linearly towards the side-wall ( y = 0 m). This test
bed, as seen in previous tests. indicated that measurements taken at 10 cm from
From Fig. 8a and b, it is seen how particles the side-wall would give a good measure of the real
released on the bar, for the low wave height (Fig. maximum velocity (which could be located even
8b), tend to move from the bar into the trough and closer to the wall) for other test conditions as well.
along the shore in the trough, before entering the rip Assuming the flow distribution close to the wall for all
channel. For higher waves (Fig. 8a), the particle other test conditions to be similar to test 3*, the
moves directly into the rip channel from the bar. deviation from the real value would be less than
From Fig. 8c and d, it is seen that the pattern for 20% (Fig. 9) in all cases.
the low water level is similar to the case of high wave Notice that the largest velocities also were found
height in the regular test (the trajectories tending not close to the side-wall in the near bed case, test 1e (Fig.
to reach the trough, but to enter the rip channel 6), but here the variation, due to the proximity of the bed
directly from the bar, Fig. 8b), and that the pattern (z = 2 cm), was not as linear as in the present test case.
for the high water level is similar to the low wave
height in the regular test (the trajectories going into
the trough before reaching the rip channel, Fig. 8a).
These observations show that the flow pattern is
very much dependent on how far from the bar crest
the waves break. The set-up difference, and thereby
the forcing, between bar and rip channel is controlled
by the cross-shore distribution and the intensity of the
wave breaking. The farther from the shore the waves
start to break, the larger water masses are driven into
the rip channel directly from the bar.
Fig. 10 shows the cross-shore rip current velocity tributed over a larger cross-shore distance compared
at one third of the depth above the bed as a function of with regular waves.
cross-shore position for different wave heights (10 cm To summarize, the tests 3a –k produced a rather
from the side-wall). The maximum cross-shore rip extensive data set where the maximum rip current
current velocity for each condition is easily identified. intensity Vmax was determined for many different
The position of the peak shifted in the offshore wave conditions and water levels. These data were
direction for increasing wave height, and that the then used as the basis for a dimensional analysis.
maximum velocity increased for increasing wave
height (except the largest wave heights where the 4.6. Dimensional analysis
velocities are the same). Similarly, for a decrease in
water level it was found that the rip channel velocity In the case of a fixed topography, the magnitude of
peaked further offshore, and that the magnitude of the the maximum rip-current velocity Vmax depends in
maximum velocity increased. The shift corresponds to principle on four variables: a wave height (the deep
the findings in the particle tracking tests, i.e. as the water wave height H0 was chosen), a characteristic
waves on the bar break farther offshore, also the flow wave length (here the deep water wave length L0), a
is forced further offshore. Different (peak) periods did characteristic water level (here the still water depth at
not seem to have a significant effect on the intensity of the bar crest Dc) and the acceleration of gravity g.
the rip current. These variables are grouped into three dimensionless
Irregular and regular waves gave similar results for quantities
the maximum velocity with respect to increasing wave Vmax H0 Dc
height and decreasing water level, although the cross- pffiffiffiffiffiffiffiffi , and
gDc Dc L0
shore variation of the current in the rip-channel for
irregular waves, was different from the regular wave
The first parameter is in the form of a Froude number
case. Due to the individual waves in a wave train
(although the velocity Vmax and the depth Dc are not
breaking at different distances from land, the phase-
found at the same position), the second is a normal-
averaged forcing and thereby the velocity was dis-
ized wave height and the third is depth to (deep water)
wave length ratio. The deep water wave length is
given as L0 = g/2pT 2.
The normalized velocity is plotted against the deep
water wave height to water depth ratio for different
wave steepness in Fig. 11. In the case of irregular waves,
H0 in the plot is the root mean square water height. The
plot shows that the points tend to group around a sin-
gle monotonic curve independent of the crest depth
to wave length ratio. The following functional rela-
tionship was found to fit the measurements well
Vmax H0 =Dc H0
pffiffiffiffiffiffiffiffi ¼ 0:30ln for 0:4 < <3
gDc 0:4 Dc
Fig. 11. Non-dimensional maximum rip current velocity as function of non-dimensional wave height.
N. Drønen et al. / Coastal Engineering 45 (2002) 223–238 237
expect the curve to depend on the degree of irregu- bar crest was important for the trajectory pattern. When
larity of the incoming waves. the waves broke close to the bar crest, the water tended
The second aspect to notice is the relative insig- to move over the bar and into the trough, then flow
nificance of the crest depth to deep water wave length along the shore before it reached the rip channel. For
on the rip current intensity, and thus of the forcing. waves breaking at some greater distance from the bar
This is an expected result, because the normal onshore crest, the flow pattern changed. In this case, the on-
radiation stress component for a shallow water wave shore flow over the bar tended not to reach the trough
in the surf zone does not depend on the wave period, but enter the rip channel more directly from the bar.
but only on the wave height and the local depth. Finally, the variation of the rip current intensity
Although the use of a deep water wave height has with the water level and wave conditions was studied.
proven to be the dominant quantity describing the rip Different test conditions were examined. It was found
current intensity, a very small dependency of Dc/L0 that the position of the rip current velocity peak moved
can be seen in Fig. 11. The larger the Dc/L0, the more offshore and that the intensity of the current increased
the rip current intensity tends to decrease. This minor with increasing wave height and decreasing water
tendency is believed to be due to the deep water level, corresponding with the findings in the trajectory
waves shoaling before breaking, but the effect is so tests. A dimensional analysis revealed a monotonic
pffiffiffiffiffiffiffiffi
small compared to the main trend (and the scatter) that relationship between Vmax = gDc and H0/Dc.
it can be neglected.
Acknowledgements
5. Conclusion
This study has been partially supported by the
A series of laboratory tests has been performed for project SASME (Surf and Swash Zone Mechanics)
a physical model of a bar/rip-channel system attacked funded by the Commission of the European Union,
by regular and irregular normally incident waves. Directorate General for Science, Research and Devel-
Tests showed that a circulation cell on the bar opment under contract No. MAS3 CT97-0081, by the
appeared due to the along-shore non-uniformity of project Intermediate Scale Coastal Behaviour: Meas-
the bar. The cell was found to be driven by differences urement, Modeling and Prediction in the NICOP
in set-up over the bar and in the channel. programme sponsored by the US Navy, Office of
Dye injections indicated a second cell on the Naval Research — Grant No. N00014-97-1-0792, by
beach, rotating in the opposite direction. the programme ‘‘Coasts and Tidal Inlets’’ funded by
The distribution of mean velocity in 3D space was the Danish Technical Research Council (STVF) and
examined, and several three-dimensional character- by the European Commission Research Directorates’
istics were found. Around the bar crest, the flow FP5 specific programme ‘‘Energy, Environment and
direction was very dependent on the vertical position, Sustainable Development’’ Contract No. EVK3-CT-
whereas the flow in the strong rip current-region in the 2000-00041, Environmental Design of Low Crested
start of the rip channel was more uniformly distributed Coastal Defence Structures (DELOS).
over the water column. Farther down (offshore) the rip
channel measurements showed that an onshore-direc- References
ted drift separated the rip current from the bed as the
rip current weakened. The structure of this phenom- Bowen, A.J., Inman, D.L., 1969. Rip currents: 2. Laboratory and
enon was furthermore measured in the near bed field observations. J. Geophys. Res. 74 (23), 5479 – 5490.
region. Chen, Q., Dalrymple, R.A., Kirby, J.T., Kennedy, A.B., Haller,
Ball trajectories illustrated the flow from a Lagran- M.C., 1999. Boussinesq modeling of a rip current system. J.
gian point of view. The trajectories were trapped in the Geophys. Res. 104 (20), 617 – 637.
Haas, K.A., Svendsen, I.A., Haller, M.C., 1998. Numerical model-
rip channel where the mean rip current started to cease ing of nearshore circulation on a barred beach with rip channels.
offshore. The effect of different wave conditions indi- Proc. 24th Int. Conf. on Coastal Eng., ASCE, 801 – 814.
cated that the distance from where the waves break to Haller, M.C., Dalrymple, R.A., Svendsen, I.A., 1997. Rip-channels
238 N. Drønen et al. / Coastal Engineering 45 (2002) 223–238
and nearshore circulation: experiments. Proc. Coastal Dyn. ’97, a process of geological importance. Journal of Geology 49,
ASCE, 594 – 603. 337 – 367.
Longuet Higgins, M.S., Stewart, R.W., 1964. Radiation stress in Sonu, C.J., 1972. Field observation of nearshore circulation and
water waves; a physical discussion with applications. Deep- meandering currents. Journal of Geophysical Research 77 (18),
Sea Res. 11, 529 – 562. 3232 – 3247.
McKenzie, P., 1958. Rip-current systems. J. Geol. 66, 103 – 113. Sumer, B.M., Fredsøe, J., 1997. Hydrodynamics around Cylindrical
Shepard, F.P., Emery, K.O., Lafond, E.C., 1941. Rip currents: Structures. World Scientific, Singapore.