Deformation Up To Breaking of Periodic Waves On A Beach
Deformation Up To Breaking of Periodic Waves On A Beach
Deformation Up To Breaking of Periodic Waves On A Beach
ABSTRACT
An experimental description is presented for 'the transformation of
periodic waves which approach breaking on a gently sloping beach. The
data include the variation of wave height, phase velocity, wave surface
profiles, and the maximum value of the wave height to water depth ratio
(H/h)max around the breaking point.
The results are compared with the theories of sinusoidal and cnoidal
wave shoaling, and the latter is shown in most cases to agree remark-
ably well when the laminar energy loss along the walls and bottom of
the wave tank is included.
An empirical relation is established between wave length to water
depth ratio L/h at the breaking point and the deep water wave steep-
ness H0/L0. Also the maximum wave height to water depth ratio at
breaking shows considerably less scattering than found previously,
when plotted versus S = hx L/h, hx being bottom slope.
1. INTRODUCTION
The literature shows a considerable number of experimental investi-
gations of the slow transformation of waves on a sloping bottom, which
is denoted shoaling, in particular, data for the variation of the wave
height have been reported.
Most of these results, however, do not confirm each other. Thus no
definitive conclusion has been obtained so far neither about the real
variation of the wave height nor as to which theory will predict the
variation sufficiently accurately.
Iversen (1952)' presented experimental data which showed that the
height of periodic waves on a sloping bottom grows much faster than
predicted by the sinusoidal wave theory, and Brink-Kjaer and Jonsson
(1973) showed that actually the variation resemble a cnoidal wave
shoaling.
Similar experiments were made by Ippen and Eagleson (1955) , and
Eagleson (1956) arrived at the same conclusion though the pattern was
477
478 COASTAL ENGINEERING-1976
(as in most cases) grows far beyond 30 or 40, which is about the limit
for which a higher order Stokes theory is applicable.*
In each case will be discussed outcome of the comparison, and an
analysis will be attempted of the possible reason for discrepancies.
.PISTON TYPE
/WAVE GENERATOR
COMMAND WAVE
-SIGNAL DATA DATA IN DIGITAL FORM:
GENERATOR ANALYSER H,T,f|,c, surface profiles
PDP 8/M PDP8/E
MINICOMR MINICOMR
mm lH
80
70
60
50 m/s
2.0
40
1.5
30
1.0
20
10 0.5
0.08 h/L0
24 22 20 18 16 14 12 10 8
Figs. 4 and 5 show the recorded variation of the wave height for
deep water wave steepnesses ranging from 0.0039 to 0.064. Both dimen-
sionless wave period TVg/h, wave height to water depth ratio H/h in the
constant depth part of the flume, and the theoretically determined deep
water wave steepness H_/L are given in each figure.
There are two experimental curves in each figure. One represent
each individually measured wave height, the other a moving average
over the length of the reflection pattern.
Though the curves for the individual wave heights seem to show a
continuous variation they are actually step-curves. This is because
the' carriage with the wave transducer moves 2 - 4 cm (depending on the
wave period) along the wave flume during one wave period, i.e. between
each new result for the wave height.
In the following the origin of the theoretical curves is described
and discussed, but first we consider the effect of energy loss due to
friction.
mm iH
50 a
40 . L.-l,^-.r„'',n'j'i-*Vn -r^T^*^'if** •"''-"" "•"' -*-W
^fV —•*—'^ •»...,,.
10 •
0.332
mrruH
60
50
40
V^vMftte/^V ^ A. t^AiO "^A'
30 H0/L0= 0.0099 ;T>/g7h = 8.70
H/h= 0.106
20
10
00829
o'
mm, H
80
70 SINUSOIDAL
CN0IDAL
60 MOVING AVE. EXPER.
50
40
10
0.0576
0
24 22 20 18 16 14 12 10 8
mmiH"
SINUSOIDAL
110 — CNOIDAL Ty/g7F=5.22
MOVING AVE. EXPER.
100
90
H0/L0 = 0.064
80
H/h = 0.258
70
v^tf fruM^V
60 H0/L0 = 0.0458
50
H/h = 0.185
30
H0/L0= 0.0270
H/h = 0.109
20
10
0.230
24 22 16 14
the measurement as h/LQ decreased towards the value 0.10. Fig. 6 shows
an example of this where HQ/LO is 3.58%, and the linear theory yields
results up to 8% higher than the measurements {i.e. a minimum value of
H/H0 = 0.913 against 0.85 measured).
It has turned out that the major reason for this discrepancy is that
friction has been neglected in Brink-Kjaer and Jonsson's calculation of
the theoretical curves. In particular in Iversen's case, with a wave
flume only 30 cm wide and a horizontal bottom depth of 77.8 cm in the
case considered, the friction along the side walls has a considerable
effect. Taking this into account brings the theoretical minimum value
of H/H down to 0.869 in the case shown in Fig. 6, and this must be
considered in fair agreement with the measured value. The theoretical
variation with friction included is shown in the figure. It may also
be noticed that the wave heights measured by Iversen are most likely
influenced by the fact that the first 4.6' (= 1.40 m) of Iversen's
slope are steeper (1:5.75 = 0.174) than the value 1:13.8 = 0.072 re-
ferred to as the slope for the experiment.
H
\-h"'
-£=0.97
h
\\ V
V\ .SI »IUS0IDAL
CN0I0A
Fig. 6 NUSOIDAL W
v^ > >> FRICTION
The effect of friction > h
losses on Iversen's
results
For the experiment shown in Fig. 4 a, H0/L0 is almost the same (3.57%)
and here the measured minimum value of H/HQ is 0.875 against the calcu-
lated value (including friction) of 0.889.
From Fig. 5a-c, however, we see that if the deep water steepness
increases, the wave height to water depth ratio will grow to large val-
ues already outside the cnoidal region. In the case of H0/LQ =6.4%
(Fig. 5c) the wave actually breaks at h/LQ =* 0.10, so that the entire
shoaling process has been determined by the linear theory, tod quite
obviously, linear theory cannot handle the large values of H/h.
Since we here at the breaking point have Omax ~ 45 (H/h ~ 0.71 and
L/h ~ 8) it seems likely that the problem could be overcome by using a
second or third order Stokes shoaling theory.
solved and tabulated the variation of the wave height. A more direct
presentation of the results can be found in Skovgaard et al. (1974).
Perhaps it should be mentioned that this theory is based on (8) (see
Sect. 6). A slightly different version will appear if (10) (in which
I'I + A H/h is substituted by 1 + ^AH/h) is used, and other differences
of similar nature are possible too. Formally all these versions (as
e.g. Shuto (1974) and Ostrovskiy and Pelinovskiy (1970) are equal in
that they only differ in the way the small terms are handled. For
practical applications, however, where H/h is not really as small as
envisaged in the theory they result in considerable differences in the
numerical results for e.g. the wave height variation, in particular as
we approach the breaking point. In our numerical calculations we have
found that the best fit to the measurements is obtained by using the
version developed by Svendsen and Brink-Kjaer.
As appears from Fig. 4 b and c, the combined linear-cnoidal shoaling
model fits the experimental data surprisingly well in those cases where
the H/h-ratio remains small for h/LQ > 0.10. The predictions even fol-
low the development all the way to the breaking point, although the
theory should not be applicable there.
It should be emphasized, however, that essentially this only indi-
cates that the relationship between cnoidal energy flux and wave height
shows a realistic variation with water depth. The absolute value of
the energy flux is determined from the wave height in the constant depth
part of flume and may not be correct (and other cnoidal wave properties
as e.g. the position of the mean water level may be even rather inaccu-
rately predicted by the same theory). This must be recalled in those
experiments were h/LQ ~ 0.10 in the constant depth part of the flume.
Then linear wave theory is applied for h/LQ > 0.10, and at h/LQ = 0.10
the theoretical result must be matched with the cnoidal shoaling, which
is used shoreward of that point.* Svendsen and Brink-Kjaer (1972)
matched the two theories by assuming continuity in energy flux. How-
ever logical this approach seems it results in a discontinuity in wave
height at the matching point.
Since, however, neither of the two theories yields the exact energy
flux for a given wave height it may be argued that it is equally cor-
rect to match the wave heights, which we know are continuous, and ac-
cept a discontinuity in the theoretically determined energy flux (which
is approximate anyway) at the matching point. This is actually the
method chosen here. Finally is mentioned that in the numerical evalua-
tions the still water depth has been corrected for wave set-down.
Disaussion
As mentioned the figures show a reasonable agreement, though discrep-
ancies up to 6 - 8% in wave height develop close to the breaking point.
This cannot surprise, however, since the energy flux used for the cal-
culations was based on the assumptions that H « h, that the horizontal
velocity u is constant over the water depth, and that the excess pres-
sure p due to the wave is constant too, and proportional to the local
*In principle any point shoreward of h/LQ ~ 0.10 could be chosen as the
matching point between the two theories.
DEFORMATION TO BREAKING 487
K being the complete elliptic integral of the first kind, m its parame-
ter, and 8 a function of m. ri'1) is the above mentioned first approxi-
mation to the effect of the sloping bottom, n'1) is given by
kVH
Fig. 7 Wave profiles for wave with Ho/L0 = 0.0165 on slope hx = 1:35
T)/H
0.5 \ e^o
t
Fig. 8 Wave profiles for wave with HQ/L0 = 0.0114 on slope hx = 1:35
DEFORMATION TO BREAKING 489
6. PHASE VELOCITY
The measurements of the phase velocity c were obtained by measuring
the time (in milliseconds) it took the wave crest (identified digitally
by the computer) to travel the distance between two wave gauges placed
20.0 cm apart in the direction of wave travel.
The results obtained in this way are rather sensitive to small
changes in the shape of the wave crest between the two wave gauges.
As a consequence, the individual measurements show a considerable
scattering (± 10 - 25%) . This is particularly pronounced for the very
small wave steepnesses. Consequently the scattering is much reduced
when the waves steepen on the slope.
The results for c presented in Fig. 9 a-d represent a moving average
over a number of waves. The results have further been confirmed by a
different method based on measuring electrically the time it takes the
wave to travel between two pointed metal-rods placed 20 cm apart. The
mean value of the measurements obtained in this way confirmed that the
results obtained from the computer when the pointed ends of the metal-
rods were placed at a level close to the wave crest.
It can be mentioned that one of the reasons for the large scattering
in the experimental results is that small free second harmonic waves
still exist in the flume. Such disturbances result in phase velocities
which are constant in time, but vary from point to point. Hence the
tendency mentioned in Sect. 2, that even the irregularities are repro-
duced when an experiment is repeated.
The measurements are compared with linear and cnoidal results for
the phase velocity. From linear theory we have
490 COASTAL ENGINEERING-1976
m/sic~
SINUSOIDAL
CNOIDAL
MOVING AVE. EXPER.
24 22 20 18 16 14 12 10
c = 2. tanh kh (7)
which is used for all values of h/LQ > 0.10. As the cnoidal result for
c is used (for h/LQ < 0.10)
1 2
h(1 + 4)Hll / (8)
A = 1
m
- 1 - mif
K
(9)
where E is the complete elliptic integral of the second kind.
Since cnoidal theory assumes H/h « 1, (8) may also be written
2
C = ^h(l + iA|) +0(f} (10)
which is equally valid. In the analogy with the wave height variation
we realize that for waves near breaking the numerical results obtained
for (8) and (10) differ appreciably. It turns out that the results ob-
tained from (8) fit the measurements better. In (8) the theoretically
determined wave height (i.e. from the shoaling process) has been used.
In general-the conclusion is positive. The two theories predict the
phase velocity to within a few per cent, the linear theory for h/LQ >
0.10, the cnoidal shoreward of that point. The only exception is close
to breaking, where the cnoidal theory overestimates the finite amplitude
effect and yields results somewhat above the measurements. In Fig. 9 c
is for comparison given the linear curve even though h/LQ < 0.10 every-
where .
7. WAVE BREAKING
The last topic to be discussed in this paper is the characteristics
of the waves at the breaking point, including the prediction of the po-
sition of this point, e.g. in terms of the water depth where breaking
is initiated.
Even though cnoidal theory seems to predict the wave height varia-
tion reasonably well, no information can be deduced from that theory
(or any other known theory) about where the breaking occurs. In that
question we must rely entirely upon empirical data.
One of the problems is to define exactly where the breaking has
started. Often breaking is defined to start 'where the front of the
wave becomes vertical', though in the case of a spilling breaker there
is no such point. Also the initiation of foam may be a very uncertain
definition in small scale experiments where the surface tension will
cause scale effects for the foam production.
In consequence of these arguments we have chosen to define the break-
ing point as the point where H/h is maximum. Since the wave height has
a maximum close to the point where the energy dissipation starts and h
is decreasing, H/h appears to have a rather sharp maximum. In the eval-
uation of h the set-down is incorporated. The choice of H/h to iden-
tify the breaking point has the advantage that from an engineering point
of view the maximum of H/h is one of the primary information about the
wave breaking.
492 COASTAL ENGINEERING-1976
*The table required should have the entries L/h and H0/L0 and yield
values of h/LQ.
DEFORMATION TO BREAKING 493
auu
M\
» 1/10 I o 1/10 1
° 1/20 Iwac aki & Sakai (1976) * 1/20
>lversen (1952)
-« 1/30' J
» 1/50 J
"^ o 1/35 ISVA
++
/I"1
•
s
' ^L.i»v«— LJ*
**
^L"
"°^iji'
v\ EqJI
, , ,,, i , . .. . 'io
0.005 0.01
wo.
0 1/35 ISVA
le 1°<»vg a. » e
S. Wh
>*4» *
s-- KVh
8. CONCLUSION
Linear ('sinusoidal') and cnoidal wave theories are compared with
experimental results obtained with waves without free second harmonic
disturbance on a plane slope h = 1:35. It is shown that:
(a) Linear theory can predict the shoaling as long as the wave height
to water depth ratio H/h is small (Fig. 4 a).
(b) Cnoidal theory, which can only be used for h/L0 < 0.10 (L0 being
deep water wave length), predicts the variation of the wave height
quite well even close to breaking (Fig. 4 b and c) .
(c) LinSar theory is used for h/LQ > 0.10, and wave height is matched
with cnoidal theory at that point. For waves with large deep water
steepness H0/L0 (> 3 - 4%) the value of H/h is not small for h/LQ >
0.10. Hence linear theory fails (Fig. 5 b and c) . Second or higher
order Stokes theory is recommended in this case for h/L > 0.10.
•
(d) The skew shape of the wave profiles is well predicted by a theory
taking into account the effect of the bottom slope (Figs. 7 and 8).
The theory cannot predict breaking.
(e) Both the linear and the cnoidal formulae for phase velocity c fit
remarkably well to the data for h/LQ > 0.10 and < 0.10, respectively
(Fig. 9).
(f) At the breaking point the wave length to water depth ratio (L/h)R
appears to be independent of bottom slope, for bottom slopes less
than 1:10, i.e. (L/h)B = f (HQ/L0) , Fig. 10, whereas (H/h)B is pri-
marily a function of the slope parameter S = hx(L/h)B (Fig. 11).
DEFORMATION TO BREAKING 495
ACKNOWLEDGMENT
The authors wish to acknowledge help and guidance from J0rgen
Christensen and Paul Prescott in the experimental work, and from
Ivar G. Jonsson in the calculations of friction losses.
APPENDIX
Given a wave with period T = 12 s, and H0/L0 = 0.01.
Find the wave height H at h = 7 m.
Since LQ = g T2/2ir. = 225 m, we have h/LQ = 0.031 > 0.10 so that
cnoidal wave theory is appropriate to use for'this wave at h = 7 m.
Continuity in energy flux at the matching point, Table 3 in Skov-
gaard et al., yields directly (using h/LQ and H0/LQ as entry) that
H/HQ = 1.133, or H = 2.55 m, since H0 = 2.25 m.
Continuity in wave height at the matching point requires that we
stage the calculation through that point. Linear theory yields (1^
denoting the depth at the matching point)
hm/L0 = 0.10 =» H/H0 = 0.933 H = 0.933 • 0.01 • 225 = 2.10 m
1^ = 0.10 • 225 = 22.5 m
At this point the table for linear (i.e. sinusoidal) waves yields
Hsin/HQ = 0.933, whereas the cnoidal result for H0/L0 = 0.01 is Hcn/HQ=
0.867. If we now require that at the matching point Hcn = Hs^n, we find
that in the cnoidal calculation we must formally use a deep water wave
height HQ cn = HQ • 0.933/0.867 = 2.43 m, i.e. in the cnoidal computation
the deep water steepness must be HQ cn/LQ = 0.01076 ~ 0.0108. The wave
height at h = 7 m can then be found from Table 3 (Skovgaard et al.)
using h/LQ = 0.031 and HQ/L0 = 0.0108 as entry. We get
H/HQ =1.140 H = 1.140 • 2.43 = 2.78 m
against 2.55 m obtained by the other matching procedure.
REFERENCES
Battjes, J.A. (1974) Computation of set-up, longshore currents, run-up
and overtopping due to wind-generated waves.
Brink-Kjar, O., and I.G. Jonsson (1973) Verification of cnoidal shoal-
ing: Putnam and Chinn's experiments. Progr. Rep. 28, pp. 19-23,
Inst. Hydrodyn. and Hydraulic Engrg., Tech. Univ. Denmark.
Buhr Hansen, J., and I.A. Svendsen (1974) Laboratory generation of
waves of constant form. Proc. 14th Coastal Engrg. Conf., Copenhagen,
Chap. 17, pp. 321-339. (Identical with Part I in Buhr Hansen et al.
(1975)).
Buhr Hansen, J., P. Schlolten, and I.A. Svendsen (1975) Laboratory gen-
eration of waves of constant form. Series Paper 9, Inst. Hydrodyn.
and Hydraulic Engrg., Tech. Univ. Denmark.
Camfield, F.E., and R.L. Street (1969) Shoaling of solitary waves on
small slopes. Proc. ASCE, J. Waterways and Harbors Div., 95, WWl,
pp. 1-22.
496 COASTAL ENGINEERING-1976