The shallow water wave hindcast model HISWA Part 11: test cases
N. Booij
L.H. Holthuijsen
T.H.C. Herbers Report No. 7-85
•
•
Delft University of Technology Department of Civil Engineering Group of Fluid Mechanics
•
•
The shallow water wave hindcast model HISWA Part 11: test cases
N. Booij
L.H. Holthuijsen
T.H.C. Herbers Report No. 7-85
•
PROJECT REPORT Delft University of Technology
Department of Civil Engineering
Group of Fluid Mechanics
•
Project title GEOMOR wave model (HISWA)
•
Project description Development of a two-dimensional model to
hindcast spectral wave parameters in an estuary with tidal flats on the basis of bottomtopography, current and wind data
•
Customer Rijkswaterstaat
Deltadienst, Afdeling Kustonderzoek
van Alkemadelaan 400
2597 AT THE HAGUE, the Netherlands
•
represented by J. v , Marle
•
Project leader dr.ir. L.H. Holthuijsen
•
work carried out by d rsLr , N. Booijdr.ir. L.H. Holthuijsen ir. T.H.C. Herbers G. Marangoni
•
Conclusion Computational results of the HISWA modelwith default parameter values are
compared with field observations in the Haringvliet and over the Galgenplaat. The results indicate that some tuning of the parameter values is required but not extensively.
•
•
Status of report confidential, final report, part 11•
City/date: Delft, February 15, 1985released for unlimited
•
Contents
•
pagel. Introduction 1
•
2. Test in the Haringvliet estuary 43. Test in the Galgenplaat region 20
•
4. Computer capacity 23 5. Conclusions 24•
6. References 25•
•
•
•
•
•
1. Introduction
•
•
To demonstrate that the basic concepts of the HISWA model and the
numerical implementation thereof provides realistic information in
complex geophysical conditions, two tests have been carried out
without turning the HISWA model. The following test should therefore
be considered as "blind" tests in the sense that all coefficients in
the HISWA model were either taken from literature or from arbitrary
choices and that no feed-back was used from the observations in these
tests to the model.
The first test was carried out
branches in the river Rhine
for the
delta).
Haringvliet (one of the
This situation can be
•
characterized as non-locally generated waves passing over a shoalwhere breaking and refraction (and perhaps diffraction) are dominant.
Local wave generation was relevant only far behind the shoal and
currents were taken to be zero.
•
•
The second test was carried out for the Galgenplaat (on extensive
shoal surrounded by deep channels in one of the branches of the river
Rhine delta). This situation can be characterized as local wave
generation on a current with considerable current refraction. Over the
top of the shoal wave breaking is dominant and behind the shoal bottom
refraction is dominant.
•
•
•
•
-1-•
•
2. Test in the Haringvliet estuary
•
In the mouth of the Haringvliet estuary (see fig. 1) a comprehensivemeasurement campaign has been undertaken which is described by
Dingemans (1983). These measurements were carried out to verify the
refraction-diffraction computations with the CREDIZ-model (Booij and
Radder, 1981).
•
The computations have been carried out for a situation with low
current velocities which occurred during the measurement period on
October 1982 at 22.00 (M.E.T). The bottom topography on a large scale
is shown in fig. land the location of the area of computation is
shown in fig. 2. A more detailed topography in the computation area is
shown in fig. 3. The results of the computations are given in
figs. 4-7.
•
•
The wave situation can be characterized as one in which waves have
been generated by wind in deeper water and which penetrate the area
under consideration. There the waves break on a shoal with a minimum
depth of about 2 m. The wave height reduction over the shoal is
•
considerable: from
behind the shoal.
breaking over the
around the shoal.
about 3 m in front of the shoal to about 0.5 m
The situation is obviously dominated by wave
shoal and by refraction and possibly diffraction
•
At seven locations measurements were taken (see figs. 2 and 3):•
a pitch-and-roll buoy providing not only the
significant wave height and mean wave period but also
the main direction and the directional energy
spreading. These measurements served as input at the
upwind boundary of the model
WR1, WR2, WR3 waverider buoys in front of the shoal giving
WA
WR4, WR5, WR6 E75
significant wave height and mean wave period waverider buoys behind the shoal
a wavestaff far behind the shoai
•
•
•
•
The results
(1983). They
of the measurements are taken from Dingemans
are given in figs. 3 through 15 together with relevant
•
HISWA results. It should be noted that two sets of analysis were
available: one labelled "Hellevoetsluis" and the other ..D.I.V ....The
former is based on a statistical analysis of the observed time series,
the latter is the result of aspectral analysis. Since the HISWA model
is based on energy (or action) considerations the "D.I.V." results are
the relevant observation results. The computations were carried out
for time 22.00 hr. (M.E.T.), the observations were taken mostly at
another time. The two nearest observations (nearest in time) are
indicated in the figures.
•
Additional information regarding the wind and the computation
•
parameters are given on the next page.For convenience of comparison the following list gives the
time-interpreted observations at 22.00 hours M.E.T. and the
•
corresponding HISWA results.•
location D.I.V. observation HISWA
H Tmean Hs Tmean s (m) (s) (s) (m) WA 3.23 T = 8.3 input input p WR-1 3.30 6.3 3.02 7.8 WR-2 2.36 6.3 2.58 7.0 WR-3 2.56 5.9 2.52 7.0 WR-4 0.61 2.6 0.46 3.7 WR-5 0.99 3.7 1.04 4.2 WR-6 1.47 5.1 1.34 4.6 E-75 0.89 2.8 0.77 3.6
•
•
•
•
-3-•
•
TEST:
HARINGVLIET OCT. 14, 1982,
at22.00
bottomgrid:
21750
m x29000
m87
x116
computational grid:15000
m x15000
m x120
0225
x60
x15
•
boundary values incoming waves:•
H
=
3.285
m, s per8.3
s, dir.5.5
0, 2 cos distr..
'
wind:UlO
=
16.5 mis dir•
breaking on/~ frequency on/~ bottom friction on/~ frequency on/~
diffraction off
numeric diffusion on/~ (upstream scheme) wave blocking on/~
•
notes: in wind, breaking and friction only standard coefficients were used.•
•
•
•
-4-•
•
•
•
•
•
•
•
•
•
•
•
Fig. 1 Bathymetry of Haringvliet estuary, contourlines in m, interval is 2.0 m.
-5-•
•
I
•
r======"\\
"
~\\ ~A
wR3
\\
~Rl
1\\
~R2
~R4
\\
\\
~~:5
Et~~
"
~\\
\
•
2000
m ~•
•
•
•
•
•
•
Fig. 2 Location of computation ar :
HaringvIi ea in
et area sh
with buoy 1 own in fig. 1
ocations.
•
•
•
1000 m land•
•
•
•
•
•
•
Fig. 3 Detailed bathymetry in area of
computation. Contourlines in m,
interval is 2.0 m.
-7-•
•
•
•
•
•
•
•
•
•
•
•
land L() cf.-75
o
Fig. 4 Significant waveheight in m,
contourline interval is O~5 m.
-8-•
•
•
•
•
landtlR3
tlR2
•
~-75
0 0.
.
v CC c•
,....
•
•
•
•
•
Fig. 5 Mean wave period in s,
contourline interval is 1.0 s.
-9-•
•
•
•
•
•
•
•
•
•
•
~ !!JO I!l' ~..
.-~R3
SI' !!JO lP I!l'
..
..
-tlR
1ïlR4
SI'
,...
l\1R2
-
9'....--
-~ ç;::
-
....
===;;:
_-
ilRS
!!JO !!JO ".
"""::IP'--tlR6
lP ~ BI ~ -land 3 2 m /s t----+~-7S
Fig. 6 Energy transport vectors in 3
m Is.
-10-•
•
land '-'•
10-3 m2/s2_.
~--
!-E>
•
WR3
_-
~IIR
1
,WR4
-
,WR2
-•
<,--
ilR5-.-~-75
I---WR6
•
-
.__--
<,-,
.
---
-•
--\.-
-•
•
•
2 2Fig. 7 Wave induced stress in m /s •
-ll-•
SENSOR
WA
14 October
1982
•
811014 :!20C.U\,;!l IlVEt. ID energy spectrum IST
If~E : 22.00
•
Hs (m)
3.23
'(:0 l!:: .1. ~ flll 014 2:!crj l''':!l lSe
q .2 .a.'
.'J 16 .(1;2!main direct ion
•
1112
•
.1 .•1 '.J .4 .5
&.!1 014 2:!0C. UiJR IIflt,
•
12 0~
.
12 6 .6 directional spreading lHl! .~_'--~~--~~ .i .2 .l .' .j.•
2•
Hs
=
significant
wave height
Tp
=
peak period
SPR=
.
standard
deviation
of direct
.
ional
energy
distribution
•
•
Fig. 8 Measurements at location WA.-12-•
SENSOR
WR-l
14-0ctober-1982
•
MEASUREMENTS
:
lIf:1:Q 12 ~~...
lil 1I•
•
,
1)2 101 4 21 00 UUR HWIII WST: 30 Wil: 330 WSN= 12 HS: 249 TH5: '.40Time:
21.00
Hellevoetsluis
D.I.V •
Hs(m)
3.22
Tm(s)
6.1
Hs(m)
2.49
Tm(s)
6.0
•
•
Time:
22.30
Hellevoetsluis
D.I.V.
Hs(m)
Tm(s)
Hs(m)
Tm(s)
4.81
6.5
3.71+
6 .+
•
.4+from Hellevoetsluis measurements with
same ratios as at 21.00
•
•
HISWA
Time:
22.00
Fig. 9 Measurements at locationWR-l.
Hs(m)
3.02
Tm(s)
7.8
•
•
Hs
=
significant wave height
Tm
=
mean wave period
-13-SENSOR
WR-2
•
MEASUREMENTS
~Z1014 2100 UUR HWftlt WST: 30 WR: 330 WSN: U HS= U7 TH!;'; '1.40"
•
•
6 2•
L-~ .1__ ~~.2 .3 .' F~!.5•
1E1 i(i 12 ~ -u.. lil"
•
,
•
2 2 101 4 2230 UUR HWft2 liST: 120 Wit: 300' WSN_ IS HS: 240 THS: '1.70•
•
•
HISWA
Time:
22.00
Hs(m)
2.58
Tm(s)
7.0
•
•
Hs
=
significant wave height
Tm
=
mean wave period
I
.
ITime:
14-0ctober-1982
Hellevoetsluis
21.00
Hs(m)
2.89
Time:
T
m(s)
6.2
Hellevoetsluis
22.30
Hs(m)
3
.
57
Tm(s)
6.5
O. I.V •
Hs(m)
2.27
'
Hs(m)
2.40
O.l.V.
Tm(s)
6.1
Tm(s)
6.4
Fig. 10 Measurements at location WR-2.
-14-•
•
•
•
•
•
•
•
•
•
•
•
SENSOR
WR-3
MEASUREMENTS
11(4 ::;; 12 ~ 8;!101422::;0 UUR HIII!3' o/:>T- I~O IIR- $00 W~N· I ~ H~· ~II)I') TH:;- 7,70 IB 2HISWA
Time:
22.00
Hs(m)
Tm(s)
2.52
7.0
Hs
=
significant wave height
Tm
=
mean wave period
14-0ctober-1982
Time:
21.20
Hellevoetsluis
O.l.V.
Hs(m)
2.54
Tm(s)
5.9
Hs(m)
2 5
• j~ +
Time:
22.50
Hellevoetsluis
O.l.V.
Hs(m)
2.61
Tm(s)
5.8
Hs (m)
2.60
+from Hellevoetsluis measurements with
same ratios as at 22.50
Tm(s)
5.9
+
Tm(s)
5.8
Fig. 11 Measurements at location WR-3.
-15-•
SENSOR
WR-4
•
MEASUREMENTS
•
1E2 ~ 8210142140 UU!! ~ HNU 31 "Sh 60-
.lIh 320...
~ IISN: 14 25 KSo: SB TKS: 2.20 21•
15I'
A
5 F !HZ!•
.1 .2 .3.4 .5 .G.7 .8 .91.1•
1E2 ;;; .1~
1121014 USO UUII Hf.R4 31 - liST: ISO Wil: 300 IISN:: IS ·.. S= 64 THSo: 3.110•
•
•
HISWA
Time:
22.00
Hs{m)
Tm{s)
•
0.46
3.7
•
Hs
=
significant wave height
Tm
=
rneanwave period
•
14-0ctober-1982
Time:
21.40
He11evoets1uis
Hs
(m)
0.58
Tm{s)
2.1
Time:
22.50
Hellevoetsluis
Hs{m)
0.61
Tm{s)
3.2
n.i.v.
Hs{m)
0.58
Hs{m)
0.64
0.1. V.Tm{s)
2.1
Tm{s)
3.2
Fig. 12 Measurements at location WR-4.
-16-•
•
•
II{') ~ 38§
~ lilt 25 2t 15 11 5•
•
•
•
•
•
•
•
SENSOR
WR-5
MEASUREMENTS
1121014 2140 UUR HII"5 IISI: 60 1111: 320 IIS"= 14 HSc 87 THh 4.110 .1 .2 .3 .~ .5 821014 2230 UUR UII'!S liST: 120 IIR: 300 liS"" 15 HS: 112 T"S: 5.10 ·25 2t 15 11 5HISWA
Time:
22.00
Hs(m)
Tm(s)
1.04
4.2
Hs
=
significant wave height
Tm
=
mean wave period
14-0ctober-1982
Time:
21.40
Hellevoetsluis
Hs(m)
0.86
Time:
Tm(s)
3.6
Hellevoetsluis
22.30
Hs(m)
1.10
Tm(s)
3.6
0.1. V.Hs(m)
0.87
Hs(m)
1.12
O.l.V.
Tm(s)
3.6
Tm(s)
3.8
Fig. 13 Measurements at location WR-S.
-17-•
SENSOR
~R-6•
MEASUREMENTS
•
!¬ ii~. , . ~ 821014 2HO' UUR HWR6 S - W5T: 60 Wil,. 320 W6W: 14 "5= 131 T"5:6.80Time:
3Hs(m)
1.36
14-0ctober-1982
Hellevoetsluis
21.40
2•
•
821014 2110uu~ HWR6 liST: 185 WR= 300 IISN: 14 H5= 179 T"5= 1.60Time:
-...
l1iTm(s)
5.1
Hellevoetsluis
23.10
Hs(m)
1. 79
•
2•
•
HISWA
Time:
22.00
Hs(m)
Tm(s)
•
1.34
4.6
•
Hs
=
significant wave height
Tm
=
mean wave period
Tm(s)
5.7
D.I.
V.Hs(m)
1.37
Hs(m)
1. 79
0.1. V.Tm(s)
5.1
Tm(s)
5.7
Fig. 14 Measurements at location WR-6.
•
I
-18-•
SENSOR
E-75
14-0ctober-1982
•
MEASUREMENTS
•
IEl ëQ 821014 2200 UUR§
E15 . 12 N$Ta: iOu:
IIh 300Time:
22.00
~..
IISH:"
Ha. 89
Hellevoetsluis
O.l.V.
•
•
,
Hs(m)
Tm(s)
Hs(m)
Tm(s)
0.90
2.8
0.89
2.8
2 F D!Z!•
.1 .2 .3••
.5,
•
Time:
Hellevoetsluis
O.l.V.
Hs(m)
Tm(s)
Hs(m)
Tm(s)
•
•
•
HISWA
Time:
22.00
Hs(m)
Tm(s)
•
0.77
3.6
•
Hs
=
significant wave height
Tm
=
mean wave period
Fig. 15 Measurements at location E-75.
-19-•
3. Test in the Galgenplaat region
•
The Galgenplaat is located in the Oosterschelde estuary. TheGalgen-plaat is a shoal surrounded by deep channels, with depths up to 50 m.
•
In contrast with the Haringvliet test, measurements are scarce. Wavegauges were instalIed in two places, indicated in fig.
16
as 'GALA'and 'GAHO'. The test was carried out for a situation on March 30,
1983, at 13:00 (M.E.T.).
•
•
The wind was coming from the South-West, so the waves were primarily
locally generated. Thus wave frequencies are rather high, and the
velocities of the currents around the Galgenplaat cannot be neglected
in this test. Current velocities were determined by the 2-dimensional
unsteady flow model' WAQUA. The station can be characterized as local
wave generation on a current upwind from the shoal, wave breaking over
the shoal and refraction around the shoal. At the location of
observation the situation seems to be dominated by the breaking of
locally generated waves.
•
•
Details of the parameters of the computations are on the next page.
The following figures are provided:
fig.
16:
Location of the area.fig. 17: Depths and current vectors in parts of the computational
region.
fig. 18: Significant wave height and energy transport vector in the
same region.
fig. 19: Mean wave period in the same region.
•
•
The results of the mean wave period were unreliable in part of the
area (the shaded region in fig. 19). It is the region downwave of a
dry part of the shoal, where the action densities are small; the mean
wave period is there obtained as the ratio of two small quantities, so
that small errors deteriorate the results.
•
Because the wave gauge 'GARO' was not flooded at the time of
observation, comparison with measurement can only be based on the
gauge 'GALA'. These measurements are: H
=
0.33 m, and Tmean=
•
s1.82 sec; the computation gave: H
=
0.15 mand T=
2.2 sec.s mean
-20-•
•
The differences are considerable, but it must be stressed that
1) the model has been run with default parameter values which have not
been tuned;
2) the current field was not observed but has been computed, thus
introducing differences between the actual situation and the input
current field for the HISWA model;
3) the significant wave height and period are very low in the
measurements. The effects of instrument noise and the reliability
of these measurements is not known to the present authors. Some
caution seems to be called for considering the quality of some of
the observations in the tests described in the previous section
(Har Lngv l.fe t )•
•
•
•
To estimate the effect of the spatial resolution of the computation
grid, the computations were repeated for a sub-region of the region
considered above with a mesh of 50 m x 124 mand with a directional
resottition of 8°. The result at the location GALA is for the
signifi-cant wave height 0.15 mand for the mean wave period 2.6 s. This is
slightly better than the results of the large grid computations
compared with the observations. Possible causes for the discrepancies
between observations and computational results are a) wave generation on a current is too slow
b) wave dissipation on a current is too high
c) up-stream boundary effects penetrate deep into the area under
consideration
d) breaking at the shoal is overestimated e) wave set-up is not taken into account
•
•
•
•
•
•
-21-•
•
TEST:
GALGENPLAAT MARCH 14, 1983
•
bottomgrid:7950
m x159
x7950
m x 100 x12400
m248
12400
m x50
x computational grid:•
boundary valuesincoming waves: absent
•
H=
s
o
m, per=
s, dir.=
dspr.=
•
wind: South-West
UlO
=
9. mis dirbreaking on/-&# frequency on/~
•
bottom friction on/~ frequency on/~ diffraction off
numeric diffusion on/~ wave blocking on/~
•
notes: in wind, breaking and friction only standard coefficients were used.•
•
•
•
-22-•
•
•
•
N
•
The Netherlands
•
Oostersehelde ~•
•
•
DuiveLand
•
Noord
a
e
veland
Thol
e
n
•
Zuid
a.ve-land
•
Fig. 16 Location of the Galgenplaat shoal in the Oostersehelde tidal basin.
-23-•
•
•
/
1 mis 1000 m ~•
land\
\
\
\
•
\"
\<,
,
,
\.
<, 'I. \.•
•
•
•
•
•
•
•
F;i.g.17 Bathymetry and current pattern in the area of
the Galgenplaat. Contourlines are in m, interval is 10.0 m.
•
wind 9 mis 1000 m 0.01 m /s3 ) ~•
"
...
-
-land-
-
-
-
-
-•
0o
-
-
--
....
-
--
<,"
.....
... \,
•
I0
..
.,..•
.....
"•
•
•
..
-..
-•
-
--
--land
-•
Fig. 18 Contourlines of significant wave height and
energy transport vectors in the Galgenplaat
area; values are in m, interval is 0.1 m. Wind
is south~west, 9 mis; direct ion as indicated
above.
-25-I
.
•
•
•
•
•
•
•
•
•
•
Fig. ]9 Contourlines of the average wave pe~, values are in seconds, the interval 1S
0.5 s. In the shaded area the results are truncated at 3.0 s.
-26-•
4. Computer capacity
•
The computer capacity required for running HISWA is estimated fromsome runs with the Galgenplaat model.
•
As a general rule it can be stated that the amount of CPU-timerequired for a HISWA computation is roughly proportional to the total
number of points in the computational grid, i.e. the product of N ,
x the number of points in
in x-direction,
•
the number of points
y-direction, and Ne'
illustrated by table 2.
the number of
N ,
y
points in e-direction. This is
•
Nx
*
Ny*
Ne CPU-time UNIVAC 1100part of region 6000 44 sec
whole Galgenplaat
region (coarse grid) 40000 319 sec
•
Table 2. Relation of gridsize and computer time for the Galgenplaattest.
•
The computer storage required is a more complicated matter; this
depends on the number of points in the bottom grid and on the number
of points
points in
preparation
in the computational grid. If M
*
M is the number ofx
ythe bottom grid, the number of data to be stored during the
phase is a few thousand
+
3*
M*
M if currents arex y
•
taken into account, or M*
M if currents are absent. During thex
ycomputational phase the number of data is roughly 30
*
Ny Ne; Nxdoes not influence the required storage because a stepping procedure
is employed in x-direction.
•
•
•
-27-•
•
5. Conclusions
•
The HISWA model has been run in two regions: the Haringvliet and theGalgenplaat without any tuning.
•
The agreement between the observations and the computational resultsis fair for the Haringvliet situation with differences in significant
wave height of typically 5%-20% (HISWA results too low) and in mean
period of 10%-30% (HISWA results too high) over a range of 0.5-3.0 m
and 6-2 s respectively.
•
•
The is -50%agreement between the observations and the computational results
not good for the Galgenplaat situation. The difference is about
for the observed significant wave height of 0.33 mand about +20%
for the observed mean wave period of 1.82 s.
•
Other wave parameters produced by the HISWA model, such as the
directional energy distribution or the radiation stress gradient have
not been observed and the HISWA results could therefore not be
validated for these parameters.
•
The main conclusion is that the HISWA model produces realistic valuesfor the wave field in complex geophysical situations but some tuning
is required especially in the presence of a current.
•
•
•
•
-28-•
•
6. References
•
Booij, N., Gravity waves on water with non-uniform depth and current.Dissertation, Delft University of Technology, 1981.
Booij, N. and A.C. Radder, CREDIZ, a refraction-diffraction model for
sea waves. DIVISIE, Data Processing Division of Rijkswaterstaat,
1981.
•
Dingemans, M.W., Verification of numerical wave propagation modelswith field measurements; CREDIZ verification Haringvliet. Delft
Hydraulics Lab., report W488 part 1, 1983.
Radder, A.C., On the parabolic equation method for water wave
propa-gation. J. Fluid Mech. vol. 95, 1979, 159-176.
•
Stelling, G.S., On the construction of computational methods for
shallow water flow problems. Dissertation, Delft University of
Technology, 1983.