Journal of Coastal Research 20 3 853861 West Palm Beach, Florida Summer 2004

Process Based Modeling of Total Longshore Sediment

Transport
Kevin A. Haas and Daniel M. Hanes
Georgia Tech Savannah
210 Technology Circle
Savannah, GA 31407, U.S.A.
[email protected]
Civil and Coastal
Engineering Department
University of Florida
Gainesville, FL 32611 and
Pacic Science Center
Coastal and Marine Geology
Program, USGS
400 Natural Bridges Drive
Santa Cruz, CA 95060, U.S.A.
[email protected]
ABSTRACT
HAAS, K.A. and HANES, D.M., 2004. Process based modeling of total longshore sediment transport. Journal of Coastal
Research, 20(3), 853861. West Palm Beach (Florida), ISSN 0749-0208.
Waves, currents, and longshore sand transport are calculated locally as a function of position in the nearshore region
using process based numerical models. The resultant longshore sand transport is then integrated across the nearshore
to provide predictions of the total longshore transport of sand due to waves and longshore currents. Model results are
in close agreement with the I
l
P
l
correlation described by Komar and Inman (1970) and the CERC (1984) formula.
Model results also indicate that the proportionality constant in the I
l
P
l
formula depends weakly upon the sediment
size, the shape of the beach prole, and the particular local sediment ux formula that is employed. Model results
indicate that the various effects and inuences of sediment size tend to cancel out, resulting in little overall depen-
dence on sediment size.
ADDITIONAL INDEX WORDS: Waves, currents, sediment transport, coastal evolution, numerical models.
INTRODUCTION
One of the most challenging problems currently attracting
considerable effort in the coastal engineering and nearshore
processes communities is the prediction of waves, currents,
sediment transport, and morphological evolution using nu-
merical models based upon physical principles. In this work
we address a very simplied subset of this challenge: the pre-
diction of total longshore sand transport resulting frommono-
chromatic waves incident upon on a two dimensional beach.
While this idealized problem is greatly simplied relative to
natural phenomenon, it represents an important benchmark
in developing the capability to model typically more complex
nearshore regions.
One difculty that arises immediately is that the spatial
distribution of the longshore transport of sand on real beach-
es is a rarely and poorly measured quantity. It is therefore
difcult to evaluate the accuracy of predictive models. For
example, BAYRAM et al. (2001) evaluate local models for long-
shore sand transport using eld measurements obtained at
the U.S. Army Corps of Engineers Field Research Facility. As
a measure of accuracy, they describe the percent of cases in
which the measured values lie within the range of 1/5 to 5
times the model predictions. This rather broad indication of
success is actually an indication of the large uncertainties
03-0038RR received 6 May 2003; accepted in revision 2 August 2003.
in both eld measurements and model predictions. Another
example, DAVIES et al. (2002), compare several research and
practical sand transport models with each other and also to
eld observations. They nd differences of a factor of 10 to
100 to be common, and excellent agreement is considered a
factor of two.
The total (gross) longshore transport of sand, I
l
, has been
measured in the eld on many occasions using a variety of
measurement techniques (e.g., KOMAR, 1998). While the un-
certainties in these measurements are still signicant, the
body of measurements of total longshore sand transport cov-
ers approximately three order of magnitude. As described by
KOMAR and INMAN (1970) and the CERC (1984) formula, there
is strong evidence that the total longshore transport is relat-
ed to specic characteristics of the breaking waves. In par-
ticular, the product of S
xy
, the onshore ux of longshore mo-
mentum, and C, the wave celerity, both evaluated at the
break point, is well correlated with the total longshore sand
transport. This product, with the symbol P
l
in Eq (2) below,
is sometimes called the longshore (component of) wave power.
We regard the I
l
P
l
correlation as perhaps the most reliable
feature of longshore sand transport on beaches, and we there-
fore have chosen to focus our model prediction upon these
quantities.
Our approach is to apply models that predict local waves,
hydrodynamics, and sediment ux, and from these predic-
854 Haas and Hanes
Journal of Coastal Research, Vol. 20, No. 3, 2004
tions calculate the total longshore sand transport and the P
l
factor. We regard the ability to reproduce the known I
l
P
l
correlation as a general verication of the component models.
As will be seen, however, the correlation turns out to be so
robust that nearly all local sediment transport models we uti-
lized were able to reproduce it adequately, so in the future
more detailed verications will also be required.
Komar and Inman or CERC Formula
The KOMAR and INMAN (1970) or so-called CERC (1984) for-
mula is given as
I
l
KP
l
(1)
where I
l
is the immersed-weight sediment transport rate, K
is a constant and P
l
is given by
P
l
(ECn cos sin )
b
(2)
where E is the wave energy, C is the wave celerity, n is the
ratio of wave group speed to wave celerity, is the wave
angle and the subscript b indicates the breakpoint. The vol-
umetric sediment transport rate is given by
I
l
Q (3)
l
(s 1)g(1 p)
where is the water density, s is the relative density of the
sediment, g is gravity, and p is the porosity of the bed.
The value of the coefcient K has been the subject of much
debate. Based upon a variety of measurements KOMAR (1998)
suggests a value of 0.70, which is slightly smaller than the
value of 0.77 previously suggested by KOMAR and INMAN
(1970). In an analysis of existing eld data, SCHOONES and
THERON (1994) nd K to be around 0.4. However, KOMAR
(1998) remarks that when the data consisting of suspended
load and excluding bed load is dropped from the analysis, the
coefcient turns out to be around 0.82. WANG and KRAUS
(1999) nd K to be in the range of 0.044 K 0.541, albeit
with large error bars.
The effect of the grain size upon the value of K has also
been examined because the CERC formula contains no other
dependence upon grain size. Based on a few eld measure-
ments, DEAN and DALRYMPLE (2002) show that K decreases
with increasing grain size. This is also shown by KAMPHUIS
et al. (1986) and DEL VALLE et al. (1993). On the other hand,
KOMAR (1988) nds no correlation of K with grain size or bed
slope, although he remarks that the quality and quantity of
data is probably insufcient and there may be counteracting
effects of grain size and beach slope.
MODEL FORMULATIONS
This section briey describes the models which are used in
the present study. This includes both the hydrodynamic mod-
els as well as the sediment transport models. In addition the
methods for calculating shear stresses, bottom friction coef-
cients and the bathymetry are also described.
Hydrodynamics
The hydrodynamic model system consists of a short-wave
transformation component and a short-wave averaged model,
interacting simultaneously to simulate short and long wave
motions in nearshore regions. The short-wave model REF/
DIF (KIRBY and DALRYMPLE, 1994) is used as the wave driver
accounting for the combined effects of bottom induced refrac-
tion-diffraction, current induced refraction and wave break-
ing dissipation. The nearshore circulation model used is
SHORECIRC as described in SVENDSEN et al. (2002). The
model determines the ow pattern by solving the quasi 3-D
short-wave averaged hydrodynamic equations.
A Generic Sediment Transport Formula
We rst utilize a simple generic local longshore sediment
transport relation corresponding to the processes mentioned
in the introduction. The time averaged local sediment trans-
port rate is given as
C f
1 w
2
q (u ) V (4)
HH my
g
with

u (t) u (t) V (t) (5)

w b
where C
1
is a constant,
w
is the friction factor, g is gravity,
V
my
is the depth averaged longshore current, u
w
is the near
bottom wave orbital velocity, V

b
is the near bottom current
velocity and the overbar represents time-averaging over a
short wave period.
Using the denition of the bottom shear stress,
1
f u u (6)
w
2
equation (4) can be written in terms of the wave-average of
the magnitude of the bottom shear stress as follows,
2C
1
q V (7)
HH my
g
This formula can be interpreted as the product of a sedi-
ment load, characterized by the time average bed shear stress
magnitude, and an advective velocity, characterized by the
depth averaged longshore current velocity. By comparing the
model results to be shown later with the value for K (in equa-
tion 1) of 0.7 we determined the value of C
1
to be 1.3. This
model could be extended by allowing C
1
to vary with sediment
characteristics, bedforms or morphology, but such complica-
tions are not justied for the present work. Henceforth this
model will be referred to as HH.
Bailard, Bowen and Bagnold Formula
A commonly used sediment transport model comes from
the energetics approach attributed to BAGNOLD (1966), BOWEN
(1980) and BAILARD (1981). The formula for the local im-
mersed-weight longshore transport rate is given as
(1 ) f f
s b w b w
3 2
i u u u u (8)
BBB y y
W tan
o
where W
o
is the fall velocity, u
y
is the longshore component
of the instantaneous bottom velocity dened by (5),
s
is the
suspended load efciency factor (typically 0.01),
b
is the bed
855 Modeling Longshore Transport
Journal of Coastal Research, Vol. 20, No. 3, 2004
load efciency factor (typically 0.1) and tan is the angle of
internal friction (typically 0.6). This formula includes com-
ponents of suspended load, the rst term, and bed load, the
second term. This model will be referred to as BBB.
Watanabe Formula
A sediment transport formula based on the power or energy
dissipation concept is used by WATANABE (1992). The rate of
longshore sediment transport in this model is dened as
max

b cr
q A V (9)
w c my
g
where A
c
is a constant taken to be 2.0, is the maximum
max
b
instantaneous bottom shear stress for the combined wave and
current ow and
cr
is the critical bottom shear stress for the
onset of sediment motion based upon the critical Shields pa-
rameter. This model is quite similar to the HH model with
the main difference being the use of the maximum bottom
shear stress rather than the wave-time average of the mag-
nitude of the bottom shear stress. This model will be referred
to as W.
Ribberink Formula
A bedload sheet-ow model by RIBBERINK (1998) relates the
transport to the effective shear stress. This wave-averaged
longshore transport is given as
3 n
q m(s 1)gd ( ) ( /) (10)
R 50 cr y
with

and (11)
(s 1)gd
50

y
(12)
y
(s 1)gd
50
where d
50
is the median grain size and m and n are constant
coefcients equal to 11 and 1.65 respectively. This model will
be referred to as R.
Friction Factor
The calculation of the shear stress acting upon the bed is
sensitive to the value of the friction factor. Because of the
uncertainty involved in determining the friction factor, it is
estimated using two different methods. Model runs are done
using both friction factors. The rst method uses the formula
from SWART (1974) to estimate the friction factor f
s
w
0.191
r
s
f exp 5.213 5.977 (13)
w
[ ] a
o
where a
o
is the amplitude of the bottom orbital excursion. The
bottom roughness, r, in this formulation is taken to be 2.5
d
50
. This friction factor varies with the wave conditions in the
cross-shore direction. The second method comes from NIEL-
SEN (1992). The bottom roughness is assumed to be due to
bedload transport, and is calculated by using
r 170d 0.05 (14)
50 2.5
where
2.5
is determined by
s 2
0.5 f (a )
w o
. (15)
2.5
(s 1)gd
50
is the wave frequency and f is the bottom friction factor
s
w
determined using the Swart formula with a roughness of 2.5
d
50
. The difference between the two methods for determining
the friction factor is in the bottom roughness. The method by
NIELSEN (1992) results in a bottom roughness nearly 100
times larger than d
50
, resulting in a much larger friction fac-
tor.
Bathymetry
Two types of bathymetry are used in this study: 1) a typical
cross-shore prole shape we will refer to as the average beach
prole (ABP), and 2) a plane beach. The depth for the ABP
is given by the expression
23
h Ax 0.05 m (16)
where x is the cross-shore distance from the shoreline and A
is taken from DEAN and DALRYMPLE (2002) based on the sed-
iment size. In order to avoid numerical singularities at the
shoreline, we specify a minimum depth of 5 cm which has a
negligible effect upon the results.
The slopes of the plane beaches are 1/40, 1/20 and 1/10.
These slopes roughly correspond to the ABP by using the dis-
tance from the shoreline to the 1 m depth contour. This re-
sults in a plane beach slope equal to approximately A
3/2
RESULTS
A matrix of model runs were conducted using a variety of
input conditions: the breaking wave heights were 0.25, 0.5 1,
1.5, 2, 2.5 and 3 m, the breaking wave angles were 3, 6, 9,
12 and 15 degrees and the wave period was 10 s. Additional
tests with wave periods of 6 and 8 s were also performed but
will not be presented here because the results were similar.
Three grain sizes (d
50
) were also used: 0.2, 0.4 and 0.8 mm
for both types of beach proles and both friction factors. This
resulted in a total of 420 simulations.
The cross-shore depth prole is a function of the grain size
through the parameter A in Eq (16). Figure 1 shows the cross-
shore proles for the plane and ABPs for the three sediment
sizes. The beach clearly becomes steeper with the larger sed-
iment size. The model domain is longshore uniform, and the
cross-shore extent is sufcient to fully resolve the longshore
transport for the largest wave conditions.
First, the hydrodynamics are calculated for all combina-
tions of parameters and then the local longshore sediment
transport is calculated at each cross-shore grid point (with a
grid spacing of 1 m) using each of the four transport equa-
tions. Figures 2 and 3 show the cross-shore variation of the
wave height, longshore current, and longshore sediment
transport for the same case on the ABP and plane beach,
respectively. The hydrodynamic conditions are virtually iden-
tical on the two types of beaches. The resulting longshore
sediment transports are also quite similar.
The total longshore sediment transport is found by inte-
grating across the cross-shore. Figure 4 shows a log-log plot
of the total longshore transport on the ABP as a function of
856 Haas and Hanes
Journal of Coastal Research, Vol. 20, No. 3, 2004
Figure 1. Cross-shore proles of the plane and ABP for the 3 grain sizes. Figure 3. Cross-shore proles wave height, longshore current and long-
shore transport for the case on the plane beach with 1 m wave height, 3
degree wave angle and 0.2 mm grain size using the HH transport formula
and the larger friction factor, Eq (14).
Figure 2. Cross-shore proles wave height, longshore current and long-
shore transport for the case on the ABP with 1 m wave height, 3 degree
wave angle and 0.2 mm grain size using the HH transport formula and
the larger friction factor, Eq (14).
Figure 4. Total longshore transport on the ABP versus wave power us-
ing the HH transport formula and the larger friction factor, Eq (14), for
d 0.2 mm.
P
l
, Eq. (2), for the case with d
50
0.2 mm using the HH
transport equation. The circles represent a number of eld
and laboratory measurements, as summarized by KOMAR
(1998), the xs are the results using the model, and the dark
line is the CERC formula with K 0.7. The model results
closely reproduce the CERC formula.
Using the HH model as an example we dene the ratio R as
K P
HH l
R (17)
I
HH
where K
HH
is the best t for K in (1) for the HH model, and
I
HH
is the immersed weight sediment ux calculated by the
HH model. The deviations from 1.0 of this ratio are mini-
mized using a least squares method in order to nd the best
t K
HH
. This method allows us to t a straight line to the log-
log plot of Q vs P
l
. The line with the best t K
HH
is shown in
Figure 4 as the light line, which is difcult to distinguish
from the dark line (CERC). In this case the best t K is 0.71,
virtually identical to the value for K suggested by KOMAR
(1998). The measured data have a large amount of scatter,
far larger in fact than the model results. The scatter of the
model results is quantied by the variance of the ratio R for
857 Modeling Longshore Transport
Journal of Coastal Research, Vol. 20, No. 3, 2004
Figure 5. Total longshore transport on the plane beach versus wave
power using the HH transport formula and the larger friction factor, Eq
(14), for d 0.2 mm.
Table 2. Variance of the ratio R, Eq (18) for each type of transport equa-
tion on the ABP. Larger values represent increased scatter in the model
results.
f
w
d
50
(mm)
Transport Model
HH BBB W R
Eq 14
(large)
0.2
0.4
0.8
0.11
0.049
0.036
0.040
0.043
0.039
0.067
0.028
0.036
0.075
0.022
0.026
Eq 13
(small)
0.2
0.4
0.8
0.56
0.28
0.13
0.022
0.063
0.056
0.25
0.096
0.068
0.68
0.29
0.14
Table 1. Estimate of K in Eq (1) using each type of transport equation on
the ABP.
f
w
d
50
(mm)
Transport Model
HH BBB W R
Eq 14
(large)
0.2
0.4
0.8
0.71
0.77
0.87
0.64
0.39
0.31
1.16
1.31
1.57
0.19
0.22
0.28
Eq 13
(small)
0.2
0.4
0.8
0.56
0.67
0.81
0.49
0.34
0.30
0.75
0.90
1.15
0.081
0.10
0.13
Table 3. Estimate of K in Eq (1) using each type of transport equation
for the plane beach.
f
w
d
50
(mm)
Transport Model
HH BBB W R
Eq 14
(large)
0.2
0.4
0.8
0.73
0.86
1.0
0.68
0.46
0.41
1.20
1.51
1.97
0.20
0.26
0.36
Eq 13
(small)
0.2
0.4
0.8
0.64
0.82
0.93
0.59
0.45
0.40
0.84
1.13
1.15
0.10
0.14
0.17
the 35 simulations for each grain size. In this case the vari-
ance of R is 0.11.
Figure 5 is for the same case as Figure 4, but on the plane
beach. The best t K in this case is 0.73 and the variance is
0.034. The primary difference between the plane and ABP
model results is that there is slightly more scatter on the
ABP.
Next, the local sediment transport is calculated using each
of the other predictive formulae following the same proce-
dures. They all result in plots similar to Figures 4 and 5, but
with different values of the predicted K and different
amounts of scatter. In order to facilitate the analysis, the K
values calculated from all the model runs with the ABP beach
are tabulated in Table 1 and with the plane beach in Table
3. The scatter (R variance) of the results from the models
with the ABP beaches are documented in Table 2 and with
the plane beaches in Table 4.
In general the larger friction factor, Eq (14), results in larg-
er values of K. Also, the larger friction factor produce less
scatter in the model results using the HH, W and R transport
formulae but only slightly less scatter for the BBB formula.
The W formula tends to give the largest total transport
while R gives the least. The other two formulae give total
transports somewhere in between. The sediment transport on
the plane beach tends to be larger than on the ABP. All the
formulae, except BBB, have constant coefcients which are
set to their standard values. These coefcients could be ad-
justed in which case each could generate values of K to es-
sentially match the CERC formula. Note that the BBB for-
mula has efciencies which are set to standard values, but
are actually quite uncertain in the surfzone.
All of the transport formulae are somewhat sensitive to
sediment size. The HH, W and R formulae tend to increase
transport with larger sediment while BBB tends to decrease
transport with larger sediment size. The scatter of the model
results tends to decrease a little with larger sediment sizes,
although not consistently. Overall, the sensitivity of the total
longshore sediment transport to sediment size is remarkably
small. To understand why this is the case, we have to ex-
amine all the effects contained in the models which are inu-
enced by the sediment size.
The hydrodynamics, especially the longshore current, is
quite sensitive to the grain size through the beach slope and
the friction factor, under the assumption that the bed is rea-
sonably at and bedforms dont contribute signicantly to the
ow resistance. As the grain size increases so does the beach
slope. As the beach gets steeper, the waves break closer to
the shoreline. For any given wave condition, the waves break
over a shorter distance on a steeper beach. This results in a
stronger radiation stress gradient and hence stronger long-
shore currents.
We isolate the effect of the bottom slope by keeping the
bottom roughness constant but varying the slope for a given
breaking wave condition. In this case, the radiation stress
forcing is signicantly larger on the steeper slope because the
surf zone width is signicantly narrower. The top two panels
of Figure 6 show the longshore current and longshore bottom
858 Haas and Hanes
Journal of Coastal Research, Vol. 20, No. 3, 2004
Table 4. Variance of the ratio R Eq (18) for each type of transport equation
for the plane beach. Larger values represent increased scatter in the model
results.
f
w
d
50
(mm)
Transport Model
HH BBB W R
Eq 14
(large)
0.2
0.4
0.8
0.034
0.013
0.010
0.028
0.041
0.014
0.030
0.018
0.0077
0.013
0.0081
0.0070
Eq 13
(small)
0.2
0.4
0.8
0.25
0.090
0.068
0.070
0.024
0.098
0.087
0.027
0.068
0.24
0.086
0.058
Figure 7. Cross-shore proles of longshore current, bottom shear stress
and longshore transport based on HH, BBB, W and R for d
50
0.2 mm
(solid) and d
50
0.8 mm (dashed) where the slope is constant but the
friction factor varies. The case is on the ABP, with a breaking wave height
of 1m, a wave angle of 9 degrees and the larger friction factor, Eq (14).
Figure 6. Cross-shore proles of longshore current, bottom shear stress
and longshore transport based on HH, BBB, W and R for d
50
0.2 mm
(solid) and d
50
0.8 mm (dashed) where only the slope varies, but not
the friction factor. The case is on the ABP, with a breaking wave height
of 1m, a wave angle of 9 degrees and the larger friction factor, Eq (14).
shear stress for this case. For the steeper beach the peaks of
the current and stress are shifted shoreward and are much
larger. This is a direct result of the increased radiation stress
forcing. The longshore sediment transport shown in the bot-
tom four panels show a signicant increase in transport due
to the increase in longshore current as well as the increased
shear stress.
Counteracting the increase in radiation stress forcing is an
increase in the bottom friction coefcient through an increase
in bottom roughness (due to larger sediment). Isolating the
effect of grain size on the longshore sediment transport via
the roughness is accomplished by keeping the slope constant
but changing the friction factor. Under such a case, the ra-
diation stress forcing will be identical, and hence, ideally the
bottom shear stress will be the same since this is the primary
momentum balance. However, the longshore current will de-
crease with the increased roughness because the friction fac-
tor is larger.
The upper two panels of Figure 7 show the resulting long-
shore current and longshore bottom shear stress for these
conditions. The longshore current indeed decreases for the
larger sediment size. However, the bottom shear stress in-
creases slightly. This is because the momentum balance is
not as simple as the ideal situation and the convective and
mixing terms are different due to the weaker longshore cur-
rents.
The bottom four panels in Figure 7 show the longshore sed-
iment transport based on the four formulae. The decrease in
the longshore current decreases transport in all the formulae,
however, HH, W and R are based on the shear stress and
859 Modeling Longshore Transport
Journal of Coastal Research, Vol. 20, No. 3, 2004
Figure 8. Cross-shore proles of longshore current, bottom shear stress
and longshore transport based on HH, BBB, W and R for d
50
0.2 mm
(solid) and d
50
0.8 mm (dashed). The case is on the ABP, with a break-
ing wave height of 1m, a wave angle of 9 degrees and the larger friction
factor, Eq (14).
hence the transport increases slightly. BBB is highly sensi-
tive to reduction in the longshore current, however, and
transport decreases signicantly.
The combined effect of the steeper beach and the increased
drag on the longshore current is seen in the rst panel of
Figure 8. Clearly the longshore current extends over a much
larger region on the milder sloping beach due to the larger
width of the surf zone. However, the magnitude of the peak
of longshore current is similar for the two cases. This is be-
cause the bottom shear stress shown in the second panel of
Figure 8 is much larger due to the increased friction factor
resulting from the increased roughness associated with the
larger sediment size. The overall effect of grain size is to de-
crease the total longshore current with increasing grain size,
which decreases the total longshore sediment transport.
Looking at each of the longshore sediment transport for-
mulae in detail, we see they are all functions of the grain
size. Each of them are a linear function of the friction factor

w
except R, which is a nonlinear function of
w
. As shown
earlier, when the friction factor increases corresponding to a
larger grain size, the shear stress increases and each trans-
port formula predicts more longshore transport.
The HH transport formula only has grain size dependence
via the shear stress. The variation of the longshore sediment
transport using HH for two sediment sizes is shown in the
third panel of Figure 8. Even though the magnitude of the
peak of the longshore current is similar for the two grain
sizes, the magnitude of the peak of the longshore transport
is larger for the bigger grain size. The increase in transport
due to the increase in bottom shear stress is larger than the
decrease due to the hydrodynamics, such that the total trans-
port increases slightly with grain size.
In addition to the friction factor, the BBB formula is also
a function of the fall velocity W
o
. As the fall velocity increases
due to larger grain sizes, the suspended sediment transport
decreases. The cross-shore variation of the longshore trans-
port is shown in the fourth panel of Figure 8. With this for-
mula, the longshore transport is clearly decreased for the
larger sediment case. The decrease due to the fall velocity
and the longshore current is much more prevalent than the
increase due to the bottom shear stress.
In W, the critical shear stress
cr
is dependent on the grain
size. Therefore, a larger grain size leads to a larger critical
shear stress required to initiate grain movement, and there-
fore slightly less transport. This model is a function of the
bottom shear stress and because the longshore currents for
the different grain sizes are similar in magnitude, the bottom
shear stress is larger due to the bigger friction factor for the
larger grain size resulting in more longshore transport. The
decrease due to the larger critical shear stress is much less
than the increase due to the friction factor resulting in an
increase in transport as seen in the fth panel of Figure 8.
The R formula is dependent on d
50
directly such that larger
grain sizes result in less transport in this formulation. As
seen in the bottom panel of Figure 8, the longshore transport
decreases with larger grain sizes. This indicates that the di-
rect dependence on the grain size has a smaller effect than
the bottom shear stress.
Even though the different transport formulae contain sev-
eral mechanisms, they result in only a weak dependence (less
than 25% variations in K) on the grain size. The reason is
that all the models have several mechanisms which cause
increases or decreases in transport such that they counteract
each other. This is perhaps the reason why the CERC for-
mula works robustly without sediment size dependence.
DISCUSSION
It has often been noted that the I
l
P
l
correlation is re-
markably robust, which is particularly surprising because it
does not contain many of the parameters which are expected
to be important. In shallow water, the P
l
factor is only sen-
sitive to two factors, the breaking wave height and breaking
wave angle. Foremost amongst the omitted parameters are
the beach slope and sediment size.
Explanations of this apparent deciency are in part pro-
vided by KOMAR (1998), and will be expanded upon here. We
base the following upon the simplied model that the sedi-
ment is suspended by the bed shear stress and is advected
860 Haas and Hanes
Journal of Coastal Research, Vol. 20, No. 3, 2004
by the longshore current. The ow is assumed to consist of a
cross-shore oscillatory component due to waves and a steady
longshore component due to the longshore current: u (t) u
o
sin(t) V
my
j .
Combining equations 5 and 6 and the assumption stated
above, the time average of the magnitude of the bottom shear
stress is:
1 1
2 2 2 2
f u sin (t) V f u f , (18)
w o my w o 1
2 2
where f
1
. In the limit of small longshore
2 2 2
sin (t) V /u
my o
currents relative to the wave orbital current speed, f
1
2/.
Using linear long-wave theory and assuming H/h we
then take gh
2
/4 as the value of u in the surf zone such that
2
o
1
2
f ghf . (19)
w 1
8
For bathymetries with no longshore gradients, the long-
shore current is related to the gradient in the radiation stress
component S
xy
(e.g., LONGUET-HIGGINS, 1970), given approxi-
mately by
5g
sin
b
V h (20)
my

8 f h w
b
Here we have assumed a plane beach (h x), a wave
breaking criterion H h, C gh, and weak longshore
currents (f
1
2/). Using these approximations and the sed-
iment ux formula from equation (7), the local longshore sed-
iment transport is then proportional to
sin
b
1.5 3 2
i(x) g h , (21)

h
b
where we have omitted dimensionless constants such as po-
rosity and specic density. Integrating across the surf zone
yields the total longshore sediment transport
1.5 3 2.5
I g sin( )h . (22)
l b b
Under the same assumptions we nd
1
3/ 2 2 5/ 2
P (ECn cos sin ) g h n cos sin . (23)
1 b b b b b
8
If we assume n
b
1, cos
b
1, then the ratio of these
two expressions is proportional only to the factor . The suc-
cess of the I
l
P
l
correlation (on a log-log plot) implies that the
factor , the sediment porosity, and the sediment density do
not vary by more than a order of magnitude in nature, which
is in fact the case. Note that the inuence of both the drag
coefcient and the beach slope have cancelled in this deri-
vation.
CONCLUSIONS
The well known I
l
P
l
correlation for the total longshore
transport of sand has been reproduced through the use of
numerical models that predict the local waves, currents, and
sediment ux. Although the proportionality coefcient K was
found to vary somewhat depending upon the specic model
components, the sediment size, and the shape of the beach
prole, the overall correlation was found to be remarkably
robust. When quantifying the effects of grain size on long-
shore transport, we found that many grain-size related ef-
fects counteract each other, resulting in limited overall grain
size dependence. Ironically, because all four sediment trans-
port models showed similar skill, these specic results cannot
be used to comment on the validity of the particular transport
formula.
ACKNOWLEDGMENTS
This work was sponsored by the US Ofce of Naval Re-
search and the National Science Foundation through the Na-
tional Oceanographic Partnership Program. KAH was also
supported by the U.S. Army Research Ofce. During the lat-
ter stages of this work DMH was supported by the Coastal
and Marine Geology Program of the U.S. Geological Survey.
Internal reviews by Curt Storlazzi and Peter Ruggiero are
appreciated.
LITERATURE CITED
BAGNOLD, R.A., 1966. An approach to the sediment transport prob-
lem from general physics. Professional paper 422-I, US Geological
Survey.
BAILARD, J.A., 1981. An energetics total load sediment transport
model for a plane sloping beach. Journal of Geophysical Research,
86, 1093810954.
BAYRAM, A.; LARSON, M.; MILLER, H.C., and KRAUS, N.C., 2001. Cross-
shore distribution of longshore sediment transport: comparison be-
tween predictive formulas and eld measurements. Coastal En-
gineering, 44, 7999.
BOWEN, A.J., 1980. Simple models of nearshore sedimentation: Beach
proles and longshore bars. In Coastline of Canada, pages 111,
Halifax. Geological Survey of Canada.
CERC, 1984. Shore Protection Manual. Vicksburg, Mississippi: U.S.
Army Coastal Engineering Research Center, Corps of Engineers.
DAVIES, A.G.; VAN RIJN, L.C.; DAMGAARD, J.S.; VANDE GRAAFF, J., and
RIBBERINK, J. S., 2002. Intercomparison of research and practical
sand transport models. Coastal Engineering, 46, 123.
DEAN, R.G. and DALRYMPLE, R.A., 2002. Coastal Processes with En-
gineering Applications. Cambridge, UK: Cambridge University
Press.
DEL VALLE, R.; MEDINA, R., and LOSADA, M. A., 1993. Dependence of
the coefcient K on the grain size. J. of Waterway, Port, Coastal
and Ocean Engineering., 118, 417432.
KAMPHUIS, J.W.; DAVIES, M.H.; NAIRN, R.B., and SAYAO, O.J., 1986.
Calculation of littoral sand transport rate. Coastal Engineering,
10, 121.
KIRBY, J. and DALRYMPLE, R., 1994. Combined refraction/diffraction
model REF/DIF 1, version 2.5. Technical Report CACR-9422, Cen-
ter for Applied Coastal Research, University of Delaware.
KOMAR, P., 1988. Environmental controls on littoral sand transport.
Proceedings 21
st
Coastal Engineering Conference (ASCE), pp. 1238
1252.
KOMAR, P.D., 1998. Beach Processes and Sedimentation. Englewood
Cliffs, New Jersey: Prentice Hall.
KOMAR, P.D. and INMAN, D.L., 1970. Longshore sand transport on
beaches. Journal of Geophysical Research, 75, 55145527.
NIELSEN, P., 1992. Coastal Bottom Boundary Layers and Sediment
Transport. World Scientic.
RIBBERINK, J.S., 1998. Bed-load transport for steady ows and un-
steady oscillatory ows. Coastal Engineering, 34, 5982.
SCHOONES, J.S. and THERON, A.K., 1994. Accuracy and applicability
of the SPM longshore transport formula. Proceedings 24th Coastal
Engineering Conference (ASCE), Volume 3, pp.25952609.
SVENDSEN, I.A.; HAAS, K.A., and ZHAO, Q., 2002. Quasi-3D nearshore
861 Modeling Longshore Transport
Journal of Coastal Research, Vol. 20, No. 3, 2004
circulation model SHORECIRC, Report # 200201, Center for Ap-
plied Coastal Research, University of Delaware.
SWART, D.H., 1974. Offshore sediment transport and equilibrium
beach proles. Laboratory Publication No. 131, Delft Hydraulics.
WANG, P. and KRAUS, N.C., 1999. Longshore sediment transport rate
measured by short-term impoundment. Journal of Waterway, Port,
Coastal and Ocean Engineering, 125, 118126.
WATANABE, A., 1992. Total rate and distribution of longshore sand
transport. Proc. 23
rd
Coastal Engineering Conference (ASCE), Vol-
ume 3, pp. 25282541.