Coastal Engineering 48 (2003) 189 195

[Link]/locate/coastaleng

The drag coefficient, bottom roughness,

and wave-breaking in the nearshore
Falk Feddersen a,*, E.L. Gallagher b, R.T. Guza c, Steve Elgar a
a
Woods Hole Oceanographic Institution, AOPE MS#12, Woods Hole, MA 02543, USA
Department of Oceanography, Naval Postgraduate School, Monterrey, CA 93943-5122, USA
c
Scripps Institution of Oceanography, University of California, La Jolla, CA 92093-0209, USA
b

Received 18 February 2002; accepted 14 April 2003

Abstract
The bottom drag coefficient in the nearshore has been suggested to depend on bottom roughness (bedforms) or alternatively
on wave breaking. The hypothesis that bottom drag coefficient depends on bottom roughness is tested with 2 months of field
observations collected on a sandy ocean beach during the Duck94 field experiment. Both the drag coefficient (estimated from
alongshore momentum balances) and bottom roughness (estimated from fixed altimeters) are larger within the surfzone than in
the region farther seaward. Although the drag coefficient increases with roughness seaward of the surfzone, no relationship was
found between the drag coefficient and roughness-related quantities within the surfzone. These results suggest that breakingwave generated turbulence increases the surfzone drag coefficient.
D 2003 Elsevier Science B.V. All rights reserved.
Keywords: Nearshore; Bed roughness; Drag coefficient; Circulation

1. Introduction
The mean (time-averaged) bottom stress is an
important component of nearshore circulation and
sediment transport dynamics. In depth-integrated circulation models, the mean alongshore bottom stress sby
often is written as
!
1
sby qcd hAuAvi
* Corresponding author. Scripps Institution of Oceanography,
University of California, San Diego, 9500 Gilman Dr. La Jolla, CA
92093-0209, USA.
E-mail addresses: falk@[Link] (F. Feddersen),
egallagh@[Link] (E.L. Gallagher), rguza@[Link]
(R.T. Guza), elgar@[Link] (S. Elgar).

where q is the water density, cd is the non-dimensional

drag coefficient, and hi represents a time average
over many wave periods. The horizontal velocity
!
vector u and the alongshore velocity v include both
mean and wave components, above the bottom boun!
dary layer. In nearshore circulation models, hAuAvi
can be represented well with low-order moments of
the velocity field (Feddersen et al., 2000), and thus
accurate parameterizations of cd are required to model
the bottom stress.
The bottom stress is equal to the turbulent vertical
flux of horizontal momentum into a viscous bottom
boundary layer, i.e., for the alongshore bottom stress,
syb qhv Vw Vi

0378-3839/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0378-3839(03)00026-7

190

F. Feddersen et al. / Coastal Engineering 48 (2003) 189195

where v V and w V are the turbulent alongshore and

vertical velocities, respectively. Therefore, cd depends
!
on the turbulence, and for constant hAuAvi , cd
increases with increased turbulence levels. Both of
the two primary sources of nearshore turbulence,
shear in the bottom boundary layer (e.g., Grant and
Madsen, 1979), and the breaking of surface gravity
waves (e.g., Svendsen, 1987), have been proposed to
affect cd (e.g., Fredsoe and Deigaard, 1992; Van Rijn,
1993). For simplicity, many nearshore circulation
models have assumed a spatially constant cd (Lon zkan-Haller and Kirby, 1999,
guet-Higgins, 1970; O
and many others), with the value of cd usually
determined by fitting to observations.
On the continental shelf, breaking wave-generated
turbulence does not reach the lower part of the water
column and thus does not influence the bottom
boundary layer. Grant and Madsen (1979) generalized
the Prandtl Karman law of the wall to the continental
shelf bottom boundary layer in the presence of waveorbital velocities and bottom roughness, i.e.,


v
z
v z * log
;
3
j
za
where z is the height above the bottom, za is the
apparent roughness height that depends on waves and
bottom roughness, and j is von Karmans constant.
The current friction velocity v* is defined so that
sby qv*2 :

Garcez-Faria et al. (1998) used this model to

estimate the alongshore bottom stress in the nearshore
(depths < 4 m) by fitting alongshore current observations that spanned much of the water column to a log
profile (3), and solved for cd using Eqs. (1) and (4).
Garcez-Faria et al. (1998) found that cd was related to
the root-mean-square bottom roughness normalized
by water depth krms/h with correlation r = 0.63, and
that cd was inversely proportional to the percentage of
waves breaking. In contrast, Fredsoe and Deigaard
(1992) and Church and Thornton (1993) hypothesized
that differences in cd within and seaward of the
surfzone are caused by differences in breaking-wave
turbulence levels, with increased breaking resulting in
larger cd. Feddersen et al. (1998) found larger (by
factor of 3) cd within the surfzone relative to seaward
of the surfzone, but it was unclear whether this cd

variation resulted from differences in bottom roughness or wave breaking.

Here, the dependence of nearshore cd on bottom
roughness and wave breaking is examined further
using 2 months of observations acquired on a sandy
ocean beach. Bottom roughness observations (Gallagher et al., 1998a) obtained concurrently with wave
and current observations (Feddersen et al., 1998) are
used to estimate spatial averages of roughness quantities (krms and krms/h) and cd, as described in Section 2.
Although cd and roughness variables are consistently
larger within the surfzone than seaward of the surfzone,
no relationship is observed between cd and roughness
variables within the surfzone (Section 3). This suggests
that in these observations the elevated surfzone cd
likely is influenced more strongly by breaking-wave
generated turbulence than by elevated bottom roughness, consistent with the hypothesis of Fredsoe and
Deigaard (1992) and Church and Thornton (1993).
Reasons these results differ from those of Garcez-Faria
et al. (1998), in particular the limitations of the methods
used here and of the log-profile approach (3) in the
surfzone, are discussed in Section 4.

2. Observations and methods

Observations were obtained during the Duck94
field experiment (September October, 1994) near
Duck, North Carolina. Pressure sensors, bidirectional
current meters, and altimeters, sampled at 2 Hz, were
deployed on a cross-shore transect (Fig. 1) extending

Fig. 1. Depth versus distance from the shoreline (solid curve) on

August 25, 1994 and locations of current meters (circles) and
altimeters (crosses). An additional current meter and the pressure
sensor array were located 750 m from the shoreline in approximately 8 m water depth.

F. Feddersen et al. / Coastal Engineering 48 (2003) 189195

191

750 m from near the shoreline to 8-m water depth

(Gallagher et al., 1998a; Feddersen et al., 1998).
2.1. Roughness estimates
The altimeters measure acoustically the distance to
the seafloor from a fixed frame. At each altimeter, the
2 Hz data were processed into 32-s bed-location
estimates (e.g., Fig. 2) (Gallagher et al., 1996) that
were demeaned and detrended to produce 24-h rootmean-square (rms) bed roughness krms estimates. Bed
roughness normalized by water depth krms/h also was
estimated every 24 h, using the 24-h average water
depth h.
This krms estimation method assumes that the bedforms migrate under the altimeter so that time variability approximates spatial variability. If either the
roughness field is frozen or if the bed erodes or
accretes uniformly in space (at time scales not
removed by detrending), this method fails. The krms
estimates are believed accurate for the following
reasons. First, coherently migrating bedforms were
observed 60% of the time under a 1.4 1.4 m
altimeter array co-deployed 70 m from the shoreline
(Gallagher et al., 1998b), supporting the assumption
that time variability approximates spatial variability.
For most of the remaining 40% of the time, krms was
small. Second, the magnitude and variability of krms in
the cross-shore (Fig. 3) is consistent with spatialseries based krms observations occasionally collected
during this experiment (Thornton et al., 1998). The
mean and variability of krms are largest within 100 m

Fig. 2. Demeaned 32-s bed elevation versus time for a 48-h period.
The krms is 6.9 and 4.5 cm for the first and second 24-h periods,
respectively.

Fig. 3. Mean (diamonds) and standard deviation (vertical bars) of

krms versus distance from the shoreline. There are between 53 and
58 24-h observations at each location. The mean for all 511 24-h
observations is 2.9 cm, the standard deviation is 2.2 cm, and the
largest observed krms is 10.7 cm.

of the shoreline (where the surfzone usually is

located) and decay farther offshore. Third, towed
and fixed altimeter krms estimates agree well within
and seaward of the surfzone at the same beach during
an experiment 3 years later (not shown).
Spatially weighted (i.e., integral) averages of krms
and krms/h were calculated on the transect within and
seaward of the surfzone (Fig. 1). The cross-shore
extent of the surfzone (averaged over 24 h) was
estimated heuristically based on energy flux relative
to the flux in 8 m depth, the local energy flux gradient,
and time-lapsed video images (R.A. Holman, personal
communication, 1996), as described in Feddersen et
al. (1998). The 24-h krms are significantly (>98%)
correlated at cross-shore lags up to 125 m, indicating
that the cross-shore krms variability is not dominated
by unresolved short spatial scales that would cause
errors in the spatial averages. The 24-h averaged
roughness quantities (krms and krms/h) for the fulltransect, the surfzone, and seaward of the surfzone
were averaged into 48-h estimates for comparison
with cd estimates.
Bedform lengthscales and orientation cannot be
determined from these altimeter observations. During
the same experiment, the dominant horizontal spatial
scales of bed variability were in the range 1 5 m
(Thornton et al., 1998; Gallagher et al., 1998b), similar
to the observed wave-orbital diameters (1 4 m). The
altimeter-estimated krms are assumed to correspond to
the bedforms with these lengthscales. The orientation
of long-crested bedforms relative to a steady current

192

F. Feddersen et al. / Coastal Engineering 48 (2003) 189195

can have a significant effect on the bottom stress

(Barrantes and Madsen, 2000). However, the effect
of bedform orientation and lengthscales in the surfzone
with combined wave-current flows is not understood.
Thus, as in Garcez-Faria et al. (1998), the relationship
between bedform height and cd is investigated, and the
effects of bedform lengthscales and orientation while
potentially significant are not considered.

Drag coefficient estimates are based on a 1-D

alongshore momentum balance between wind and
wave forcing, bottom stress, and lateral mixing given
by
dSxy
dMxy
!
qcd hAuAvi
;
dx
dx

qcd2

x8m

!
hAuAvidx;

xb

2.2. Drag coefficient estimates

swind

y

assumed spatially constant within each the surfzone

and seaward of the surfzone regions, and cd is passed
through the integrals. With a spatially constant wind
stress, the cross-shore integral of Eq. (5) becomes,
Z xb
!
swind

x

S
j

qc
hAuAvidx
8m
xy
x
d1
8m
y

where swind
is the alongshore ( y) wind stress. Crossy
shore (x) derivatives of Sxy and Mxy, components of
the radiation and depth-integrated lateral Reynolds
stresses (e.g., Svendsen and Putrevu, 1994), respectively, are difficult to estimate with observations.
However, the observations can be used to estimate
cross-shore integrals of terms in Eq. (5), from which
spatially averaged surfzone and seaward of the surfzone cd can be calculated.
The cd within and seaward of the surfzone are
calculated by integrating Eq. (5) over the entire 750m long current meter transect from approximately the
shoreline (x = 0) to 8-m water depth (x = x8m) (Feddersen et al., 1998). The cross-shore integration is
separated into two components, one spanning the
surfzone and one seaward of the surfzone. In 8-m
water depth, usually well seaward of the surfzone,
pressure array data and linear theory are used to
estimate Sxy, and the Reynolds stress Mxy is assumed
negligible. At x = 0, the location of the most shoreward
instrument, Sxy and Mxy are set equal to zero. Swash
processes onshore of the most shoreward current meter
are thus assumed to contribute negligibly to the crossshore integrated, alongshore momentum balance. This
assumption is consistent with standard models for
radiation (based on depth-limited wave breaking) and
Reynolds stresses, and is supported by the closure of an
integrated alongshore momentum balance that neglects
the swash region (Feddersen et al., 1998). Although cd
may vary continuously in the cross-shore, cd is

where xb is the location of the breakpoint. The

unknown drag coefficients within and seaward of
the surfzone are represented by cd1 and cd2, respectively. Observed 2-Hz velocity time series are used to
!
estimate hA uAvi at each current meter. Hourly
estimates of the total forcing, and surfzone and
!
seaward integrated hA uAvi are linearly regressed to
calculate 48-h values of best-fit cd within and seaward
of the surfzone (see Feddersen et al., 1998 for details).
The selected 48-h time interval is a compromise
between long-time intervals needed for statistical
stability of the cd estimates, and short-time intervals
that better resolve temporal variability of krms, cd, and
the cross-shore extent of the surfzone. The results
were similar using 24- and 48-h averaging intervals.
!
Successive hourly estimates of integrated hA uAvi
(and of the total forcing) are not independent (Feddersen et al., 1998), and thus, the effective degrees of
freedom and confidence limits for each 48-h cd
estimate cannot be determined. To eliminate inaccurate estimates, 48-h cd values were rejected if the
regression had poor skill (defined as skill < 0.4) or if
cd was negative.

3. Relationship between the drag coefficient,

roughness, and wave breaking
Estimates of cd, krms, and krms/h are consistently
larger in the surfzone than seaward of the surfzone
(compare circles with crosses in Fig. 4). The average
surfzone cd is significantly (>99% confidence) larger
than the average cd seaward of the surfzone, consistent
with cd estimated using a single regression for the
entire 2-month period (dashed lines in Fig. 4), but
roughly a factor of two less than the equivalent cd
derived from friction factors cited by Nielsen et al.
(2001).

F. Feddersen et al. / Coastal Engineering 48 (2003) 189195

A relationship between krms and cd (Fig. 4a), and

krms/h and cd (Fig. 4b) is apparent when the regions
within and seaward of the surfzone are considered
together. However, this is misleading because considering both regions together does not control for other
factors that could effect cd such as breaking-wave
generated turbulence.

193

Table 1
Correlation r between drag coefficient cd and both bottom roughness
krms and normalized bottom roughness krms/h
Surfzone
krms
krms/h

N
17
13

Seaward of the surfzone

r
0.27
0.23

N
15
11

r
0.33
0.47y

The number of data points in each correlation is N.

y
Correlation significant ( p 0) with 90% confidence.

To isolate the effect of enhanced turbulence due to

bottom roughness from turbulence due to breaking
waves, the relationship between roughness quantities
and cd is examined separately within and seaward of
the surfzone. Within the surfzone, no relationship is
observed between krms and cd (circles in Fig. 4a), nor
between krms/h and cd (circles in Fig. 4b). The
correlations (Table 1) are not significant at the 90%
confidence level. The lack of a detectable cd dependence on spatially averaged krms or krms/h suggests that
roughness quantities are not the critical factor determining the surfzone cd.
Seaward of the surfzone, the correlation between
krms and cd (r = 0.33, Table 1, crosses in Fig. 4a) is
increased relative to the correlation within the surfzone, but is not significant at the 90% level, suggesting
that krms alone is not responsible for the cd variation
seaward of the surfzone. However, the correlation
(r = 0.47, Table 1) between seaward of the surfzone
krms/h and cd is significant at the 90% level (crosses in
Fig. 4b). This is consistent with the hypothesis that cd
seaward of the surfzone depends on the depth-normalized apparent roughness ka/h because ka is a function of
the physical roughness (Grant and Madsen, 1979).

4. Discussion

Fig. 4. (a) cd versus krms and (b) cd versus krms/h for the surfzone
(circles) and seaward of the surfzone (crosses). The upper and lower
dashed lines in each panel are the surfzone and seaward of the
surfzone 2-month best-fit cd, respectively (Feddersen et al., 1998).
Statistics are presented in Table 1.

The consistently elevated surfzone cd over a broad

range of roughness (Fig. 4) implies that other surfzone
processes, such as wave breaking, are important to cd.
However, this conclusion is tentative due to limitations of the data and analysis methods. These limitations include the inability to estimate bed roughness
more frequently than every 24 h because bedforms
migrate slowly past the altimeter (Fig. 2). Using 24-h
averaged roughness quantities and breakpoint locations can degrade roughness estimates within and

194

F. Feddersen et al. / Coastal Engineering 48 (2003) 189195

seaward of the surfzone by including observations

from locations sometimes outside each region when
there is a rapid change in surfzone width. Also, the
instantaneous bed roughness is spatially patchy (Gallagher et al., 2003), which is obscured with a 24-h
averaged krms. The potential effect of variable bedform lengths and orientations (Barrantes and Madsen,
2000) were not taken into account. In addition, the
estimated correlation between cd and roughness will
be reduced below the true correlation if cd and roughness vary over the 48-h averaging time necessary to
obtain statistical stability.
The result that cd does not depend on roughness in
the surfzone differs from the (log profile based) result
of Garcez-Faria et al. (1998). A requirement of logprofile models (Eq. (3)) is that the bottom boundary
layer is a constant stress layer where shear production
balances turbulent dissipation e, yielding a dissipation
scaling
v 3
e *
jz

where e decreases with height z above the bed. This

dissipation scaling (Eq. (7)) is consistent with measurements (for example) on the Northern California
continental shelf (Grant et al., 1984) and the Hudson
River Estuary (Trowbridge et al., 1999).
In the shallow water of the surfzone, breakingwave-generated turbulence can penetrate the entire
water column. For example, laboratory observations
show increased turbulence associated with wave
breaking within 0.3 cm of the bed (Cox and Kobayashi, 2000). Surfzone dissipation observed in the field
(George et al., 1994) is 102 103 times larger than the
near-bottom dissipation observed in a tidal estuary
(Trowbridge et al., 1999), and does not decay as z 1
(Eq. (7)), but increases with height above the bed.
Laboratory measurements of surfzone turbulent kinetic
energy q show a maximum at mid-water column (Ting
and Kirby, 1994). If e f q3/2, as commonly is assumed
in the surfzone (e.g., Svendsen, 1987), the laboratory
surfzone e also increases with height above the bed.
These (and other) laboratory and natural surfzone
observations suggest that the dissipation scaling (Eq.
(7)) likely is not applicable in the surfzone. Thus,
mean bottom stress (and therefore cd) inferred by
fitting mean currents to a log profile may be incorrect.

Similarly, the concept of an apparent roughness ka

(e.g., Grant and Madsen, 1979) also is not applicable
to the surfzone, and the relationship between ka and cd
therefore is not investigated.

5. Conclusions
Observations along a cross-shore transect of current
meters and altimeters extending 750 m from the shoreline to 8-m water depth show that the drag coefficient
cd and bottom roughness krms are larger inside the
surfzone than outside the surfzone. No dependence of
cd on krms or krms/h is found within the surfzone, nor
between cd and krms seaward of the surfzone. There is a
weak relationship between cd and krms/h seaward of the
surfzone, consistent with the hypothesis that krms/h
influences cd when waves are not breaking. Although
the data and methods have limitations, the lack of an
observed relationship between cd and roughness,
together with the existing evidence of increased surfzone turbulence dissipation associated with breaking
waves, suggest that breaking-wave generated turbulence leads to increased surfzone cd.
Acknowledgements
The instrument array was deployed and maintained
by staff from the Center for Coastal Studies. Tom
Herbers and Britt Raubenheimer helped collect and
process the data. Excellent logistical support and the
8-m depth pressure sensor array data were provided
by the U.S. Army Corps of Engineers Field Research
Facility. Steve Lentz provided the wind stress data.
We have benefited from discussions with John
Trowbridge, Tony Bowen, Rob Holman, and an
anonymous reviewer. This research was funded by
the Office of Naval Research, the National Science
Foundation, the National Ocean Partnership Program,
and the Army Research Office. Woods Hole Oceanographic Institution contribution 10286.
References
Barrantes, A.I., Madsen, O.S., 2000. Near-bottom flow and flow
resistance for currents obliquely incident to two-dimensional
roughness elements. J. Geophys. Res. 105, 26253 26264.

F. Feddersen et al. / Coastal Engineering 48 (2003) 189195

Church, J.C., Thornton, E.B., 1993. Effects of breaking wave induced turbulence within a longshore current model. Coast. Eng.
20, 1 28.
Cox, D.T., Kobayashi, N., 2000. Identification of intense, intermittent coherent motions under shoaling and breaking waves.
J. Geophys. Res. 105, 14223 14236.
Feddersen, F., Guza, R.T., Elgar, S., Herbers, T.H.C., 1998. Alongshore momentum balances in the nearshore. J. Geophys. Res.
103, 15667 15676.
Feddersen, F., Guza, R.T., Elgar, S., Herbers, T.H.C., 2000. Velocity
moments in alongshore bottom stress parameterizations. J. Geophys. Res. 105, 8673 8686.
Fredsoe, J., Deigaard, R., 1992. Mechanics of Coastal Sediment
Transport. World Sci., River Edge, NJ.
Gallagher, E.L., Boyd, W., Elgar, S., Guza, R.T., Woodward, B.,
1996. Performance of a sonar altimeter in the nearshore. Mar.
Geol. 133, 241 248.
Gallagher, E.L., Elgar, S., Guza, R.T., 1998a. Observations of
sand bar evolution on a natural beach. J. Geophys. Res. 103,
3203 3215.
Gallagher, E.L., Elgar, S., Thornton, E.B., 1998b. Megaripple migration in a natural surf zone. Nature 394, 165 168.
Gallagher, E.L., Thornton, E.B., Stanton, T.P., 2003. Sand bed
roughness in the nearshore. J. Geophys. Res. 108, in press.
Garcez-Faria, A.F., Thornton, E.B., Stanton, T.P., Soares, C.M.,
Lippmann, T.C., 1998. Vertical profiles of longshore currents
and related bed shear stress and bottom roughness. J. Geophys.
Res. 103, 3217 3232.
George, R., Flick, R.E., Guza, R.T., 1994. Observations of turbulence in the surf zone. J. Geophys. Res. 99, 801 810.

195

Grant, W.D., Madsen, O.S., 1979. Combined wave and current interaction with a rough bottom. J. Geophys. Res. 84, 1797 1808.
Grant, W.D., Williams, A.J., Glenn, S.M., 1984. Bottom stress
estimates and their prediction on the Northern California continental shelf during CODE-1: the importance of wave current
turbulence. J. Phys. Oceanogr. 14, 506 527.
Longuet-Higgins, M.S., 1970. Longshore currents generated by obliquely incident sea waves 1. J. Geophys. Res. 75, 6790 6801.
Nielsen, P., Brander, R.W., Hughes, M.G., 2001. Rip currents: observations of hydraulic gradients, friction factors, and wave
pump efficiency. Proc. Coastal Dynamics 01, Lund, Sweden.
Am. Soc. Civil Eng., New York, p. 489.
zkan-Haller, H.T., Kirby, J.T., 1999. Nonlinear evolution of shear
O
instabilities of the longshore current: a comparison of observations and computations. J. Geophys. Res. 104, 25953 25984.
Svendsen, I.A., 1987. Analysis of surf zone turbulence. J. Geophys.
Res. 92, 5115 5124.
Svendsen, I.A., Putrevu, U., 1994. Nearshore mixing and dispersion. Proc. Royal Soc. Lond. A 445, 561 576.
Thornton, E.B., Swayne, J.L., Dingler, J.R., 1998. Small-scale morphology across the surf zone. Mar. Geol. 145, 173 196.
Ting, F.C.K., Kirby, J.T., 1994. Observations of undertow and turbulence in a laboratory surfzone. Coast. Eng. 24, 51 80.
Trowbridge, J.H., Geyer, W.R., Bowen, M.M., Williams, A.J.,
1999. Near-bottom turbulence measurements in a partially
mixed estuary: turbulent energy balance, velocity structure,
and along-channel momentum balance. J. Phys. Oceanogr. 29,
3056 3072.
Van Rijn, L.C., 1993. Principles of Sediment Transport in Rivers,
Estuaries, and Coastal Seas Aqua Publications, Netherlands.