Professional Documents
Culture Documents
ManagingEditors:
Malcolm J. Bowman Richard T. Barber
ChristopherN.K. Mooers John A.
Coastal
and Estuarine Studies
40
ChristopherN.K. Mooers
Ocean ProcessAnalysisLaboratory
Institutefor the Studyof the Earth, Oceansand Space
Universityof New Hampshire
Durham, N.H. 03824-3525, USA
John A. Raven
Dept. of BiologicalSciences,Dundee University
Dundee, DD1 4HN, Scotland
Editor
David Prandle
ProudmanOceanographicLaboratory
BidstonObservatory,Birkenhead
MerseysideL43 7RA, United Kingdom
ISBN 0-87590-254-5
Contributors List
Part I. Introduction
Introduction
D Prandle ......................................................... 3
A topographically
inducedinternalwaveandmixingin theTamarEstuary
D R Sturleyand K R Dyer .............................................. 57
Turbulenceandshearinducedmixingprocesses in estuaries
E J Darbyshire& J R West ............................................. 75
Dynamically-active
modelsin the predictionof estuarinestratification
J H SimpsonandJ Sharples...................................... 101
Thestratified
hydrodynamics
of thePalmiet- a prototypical
bar-built
estuary
J L Largier,J H SlingerandS Talijaard ...................................
viii
Salinitystructureof a shallow,tributaryestuary
W W Schroeder, S P Dinnel and W J Wiseman Jr ............................. 155
On meteorologically
inducedsubtidalmotionin HangzhouBay
J L Su and W Chen .................................................. 173
Observations of fine-sedimentconcentrations
and transport
in the turbiditymaximumregionof an estuary
R J Uncles,J A Stephensand M L Barton .................................. 255
Acousticmeasurements
of suspended sedimentover sandwaves
P D Thorne,R L Soulsbyand P J Hardcastle................................ 335
Someconsiderations
on mathematical
modellingof morphological
processes
in tidalregions
Z B Wang ......................................................... 467
A three-dimensional
transportmodelfor dissolvedandsuspended
matter
in estuaries and coastal seas
G C van Dam and R A Louwersheimer .................................... 481
An estuarineand coastalsandtransportmodel
B A O'Connor and J Nicholson .................................... 507
On thedistribution
of suspended
matterandthe densitydrivencirculation
in the Dutch coastal area
M Visser ......................................................... 551
Coastaldynamicsalonga ruggedcoastline
B King and E Wolanski ............................................. .. 577
Interdisciplinary
studyon the tidalfxontin theBungoChannel,Japan
T Yanagi,O Matsuda,S TanabeandS Uye ................................. 617
Hydrodynamic
modellingfor a tidalpowerproject
T L Shaw ...................................
Acknowledgements
The organisingcommitteefor the GregynogConferencewas as follows:
David Prandle
ProudmanOceanographic Laboratory
BidstonObservatory,
Birkenhead
Merseyside
L43 7RA, United
List of Contributors
D G Aubrey P D Craig
225 Clark Lab CSIRO Division of Oceanography
WoodsHole Oceanographic
Institute GPO Box 1538
Woods Hole Hobart Tas. 7(}01
MA O2543 Australia
USA
G C van Dam
J O Backhaus Rijkswaterstaat
Instiut fuer Meereskunde Tidal Waters Division
Germany
E J Darbyshire
M L Barton Schoolof Civil Engineering
Departmentof Civil Engineering Universityof Birmingham
Universityof Birmingham BirminghamB 15 2TT
BirminghamB 15 2TT United Kingdom
United Kingdom
J Darbyshire
V Casulli Unit for Coastal and Estuarine Studies
W Chen A G Davies
SecondInstituteof Oceanography School of Ocean Sciences
PO Box 1207 UniversityCollegeNorth Wales
Hangzhou Menai Bridge
Zheijiang310012 GwyneddLL59 5EY
China United Kingdom
R T Cheng W P M De Ruijter
US GeologicalSurvey Universityof Utrecht
Menlo Park Instituteof MeteorologyandOceanography
California Princetonplein5
USA 3584 CC Utrecht
The
xii
S P Dinnel F C Groenendijk
Center for Marine Science Rijkswaterstaat
Universityof SouthernMississippi Tidal Waters Division
StennisSpaceCenter Koningskade4
MS 39529 2696 AA Den Haag
USA The Netherlands
K R Dyer P J Hardcastle
Institute of Marine Studies ProudmanOceanographic
Laboratory
PolytechnicSouthWest BidstonObservatory
Drake Circus Birkenhead
A J Elliott P E Holloway
Unit for Coastal and Esmarine Studies Departmentof Geographyand
Marine Science Laboratories Oceanography
Menai Bridge AustralianDefenceForceAcademy
GwyneddLL59 5EY CampbellACT 2600
United Kingdom Australia
C T Friedrichs F Jiang
MIT-Woods Hole Ocean Institute Coastaland Oceanographic
Woods Hole Ocean Institute EngineeringDepartment
Woods Hole, MA 02543 Universityof Florida
USA Gainesville
Florida 32611
A Van der Giessen USA
National Institute of Public Health
and Environmental Protection B King
(RIVM), Centre for MathematicalMethods Australian Institute of Marine Science
PO Box 1 PrivateMail Bag No 3
3720 BA Bilthoven Townsville Mail Centre 4810
The Netherlands Australia
P A Gillibrand R Kirby
Departmentof Agriculture,Fisheries Ravensrodd Consultants Ltd
and Food 6 QueensDrive
Marine Laboratories Taunton, Somerset TA1 4XW
PO Box 101 United
Victoria Road
Aberdeen AB9 8TB
United Kingdom
J van de Kreeke A J Mehta
Rosensfiel School of Marine and Coastaland Oceanographic
AtmosphericScience EngineeringDepartment
Division of AppliedMarine Physics Universityof Florida
4600 RickenbackerCauseway Gainesville
A Y Kuo J Nicholson
School of Marine Science Departmentof Civil Engineering
Virginia Instituteof Marine Science Universityof Liverpool
Collegeof William and Mary Brownlow Street
Gloucester Point VA 23062 PO Box 147
USA LiverpoolL69 3BX
United Kingdom
J L Largier
Center for Coastal Studies, 0209 B A O'Connor
Japan ProudmanOceanographic
Laboratory
BidstonObservatory
Birkenhead
MerseysideL43 7RA
United Kingdom
xiv
C E Sasser R Smith
CoastalEcologyInstitute Loughborough
Universityof Technology
Center for Wetland Resources Loughborough
LouisianaStateUniversity Leicestershire LEll 3TU
J H Simpson
School of Ocean Sciences E M Swenson
Marine Science Laboratories CoastalEcologyInstitute
Menai Bridge Center for Wetland Resources
Instituteof Meteorology
andOceanography
S Tanabe Universityof Utrecht
Departmentof EnvironmentConservation Princetonplein5
EhimeUniversity 3584 CC Utrecht
Matsuyama790 The Netherlands
ZB Wang
K T Tee Delft HydraulicsLaboratory
Physicaland Chemical ScienceBranch PO Box 177
Departmentof Fisheriesand Oceans 2600 MH Delft
BedfordInstituteof Oceanography The Netherlands
Dartmouth, Nova Scotia
Canada J R West
R J Uncles T Yanagi
PlymouthMarineLaboratory EhimeUniversity
PlymouthPL1 3DH Matsuyama790
United Kingdom Japan
Part I
Introduction
Dynamics and Exchanges in
Coastal and Estuarine Studies Estuaries and the Coastal Zone Vol. 40
Introduction
D Prandle
4 D Prandle
Part HI - Circulation
Part IV - SedimentDynamics
D Prandle 5
Part V - SedimentModelling
Thesixpapers
in thissection
emphasise thediversityof regimes,
withobservations
fromfour
continents
andfroman estuarywith (almost)the world'slargesttidesto onewheretidal
exchanges
aresecondary. De Ruijteret al describe
thecoastal
plumefromtheRhinewhich
persists
for up to 80kmalongshore
andfor 25kmoffshore;boththelocalisedstratification
and
theoverallgeographic
configurationareshownto varystronglywithbothwindandfiver
discharge.
Visserdevelops
a linearised
analytical
modelof thecross-shore
circulation
within
theRhinePlume,thereby
explainingsurface
observations
of a minimum in suspended
sediment
concentrations
alongitsoffshore
edge.KingandWolanskiusea vertically-averaged
6 D Prandle
1.3 Conclusions
Baroclinic Dynamics
Dynamics and Exchanges in
Coastal and Estuarine Studies Estuaries and the Coastal Zone Vol. 40
Abstract
2.1 Introduction
Crookhaven Bight
35ø00'S Hare Bay
&Green
&Blue
(Isson
Bay
obart
ired
100
,range
Metoo
'ological
s/
ß Currentmetermooring
It Tide gaugemooring
0 CTD station
Winds duringthe measurement period were light. The region is subjectto a distinct
sea-breeze
cycle,and yet the maximumrecordedhourly-averaged wind speedwaslessthan
10ms"With diurnalvariations
filteredout,the maximum
windspeedis about7 ms"
(Figure2). Thewindsignalin Figure2 hasbeentreatedwitha least-squares
filterof thetype
describexl
by Thompson(1983)anddesigned, in thiscase,to suppress
oscillationswithperiods
lessthan1 day,andpassthosegreaterthan2 days.
10'0
•I1"
IIIIIIIIIIIIIIII[IIIII1IIIII (a) Wind
5.0
0.0
360
I I•,[• I I I I I I I I I.i i i i i i i i i I i I''11 I;• I I
ß .
o
ß•- 180
- \
,r-- 9O
;
0.10 I I I I I I I I I I I I I I I I I I I I I I [ I I I I I
• (c) Modelled
current
"-' 0.05
0,00
0.10
0.05
-IIIl.IIIIIIIIIIIIII[IIIIIIIIIII
--
(d) •easured current
0.00 I I I I I I I I • IV
8 10 15 20 25 3031 1 5 7
DEC., 1988 JAN 1989
Figure2: Windspeed
(a)anddirection
(b)forthemeasurement
period,
together
withsimulated
(c)and
measured
(d) currentspeeds
from6.5 m depthat theRedstation(Records
arefilteredto removediurnal
and higherfrequencies).
oscillatory
periodof 7 to 10 days.Fig. 4 alsoshows,for comparison,
filteredrecordsof the
sea-surfaceelevationfrom the Black station,and watertemperatures
from 11 m at Black and
20 m at Yellow.The correspondence betweenthefourcurvesis obvious.The surfaceelevation
signalwasrecorded by a submerged pressure
sensor,and so doesnot containthe "inverse
barometer"signal.Therangeof surface movement indicated
in thesignalis approximately
0.2
m. The Yellow thermistoris 9 m deeperthat that at Black, and consistently recording
temperaturesabout1ø cooler.On thisbasis,the temperatureoscillations
of around2ø canbe
convertedto a verticalisothermmovementof about20 m, or approximately 100 timesthe
amplitudeof thesurfaceoscillation.
Theisotherm movement is out-of-phasewiththesurface:
as the surfacerises,the isothermsfall, characteristic
of baroclinicmotion.The atmospheric
pressure
lagsthesurfaceelevationslightly,with a delayof approximately
2 days.
1030.0
1 I I I I I I I I I I I I I I I I I I I I i I I I I I I I' I I I I I I I I /
1020.0
1010.0
----' 1000.0
990.0
980.0
-
IIIIIIIIIIItIIIIIIIIIIII!IIIiiii!II!
6 10 15
DEC 1988
20 25 3031 1
JAN 1989
5 10 12
Figure3: Atmospheric
pressure
recordfor 6 Decemberto 12 January
P D CraigandP E Holloway 15
Themostlikelyexplanation
for thecorrelation
evidentin Figure4 is thatit is indicative
of coastally
trapped
waveactivity
overtheshelfandslope region external
toJervisBay.
DuringtheAustralian
Coastal
Experiment in 1983/84,
threemooring linesweredeployed
acrossthesouth-eastern
Australian
continentalshelfandslopetodocument thepropagation
ofcoastally
trapped
waves
(Freelandetal.,1986;Church etal.,1986a,b;Huyeretal.,1988).
The firstthreewavemodeswerein factfoundto account for approximately
half the
sub-inertial
currentenergyalongthispartof thecoast.The correlation
betweensurfaceand
isotherm
displacement,
asseenin Figure4, is anestablished
property
of thewaves, andthe
displacement
ratioof 1:100iscompatible
withtheory (e.g.GillandClarke,1974).
| J I I i I I I I I i i I I I J J J J I i i I i i i i I i i i I I J J J I
1030
._.1020
• lOlO
17.1
(b) Blacksurfaceelevation
17.0
16.9
19.0 -
18.0 -
17.0 -
16.0 -
15.0 -
14.0 -
(e) northwardwind
5.0
o.o
-5.O
ß I ! ! ! I I I ! I I ! I J I I I f I I f I I ! ! I I I ! I I ! ! I I I I
6 10 15 20 25 30 I 5 10
DEC 88 JAN 89
Figure
4:Filtered
records
of(a)atmospheric
pressure,
(b)surface
elevation
from
theBlack
station,
(c)
watertemperature
at 11 •n depth
fromBlack,(d) watertemperature
at 20 m fromYellowand(e)
northwardwindcomponent.
I I t I I [ I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
22.0 - - 22.0
21.0 - - 21.0
19.0 ß ,,•.•h.,•..." /
- / ,•, - 19.0
18.0 _ /•-
'"'Bl.u..e...
/:":i •r ,on•'/',"'"'/'
/'• / "-'"
17.0 /.,.z
-, • .... •
....... ";,,..• //
- t•,, x, '- Z'"'..'",.,.-"/ - 17.0
/,: Blac,
k
16.0 _ /.:, ----J - 16.0
,
ß
o
15.0 - - 15.0
14.0 - - 14.0
I6 I I I I10 I I I I 11
5I I I I I
20 I I I I I
25 I I I I 1301
I1 I I I I5 I I I I I10 I
DEC 88 JAN 89
Figure5' Filteredtemperature
recordsfroInBlackat 11 m depth,Blueat 11.5m, Red at 10.5m and
Green at 9.5 m.
P D CraigandP E Holloway 17
i I I I I I I J I '1 I I I I I I I I t i I I I I I i i I i I I I I I I I I
2.2.0- -
21.0
20.0
19.0
18.0
17.0
16.0
15.0
14.0
I ! I I I ! I I I I I I I I I 1 [ t I I ! ! J J I I I I t I I I I I t I
6 10 15 20 25 30 I 5 10
DEC 88 JAN 89
] I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
0.10 - - 19
0.05 -
Red r• 18'5m
_18
18.5m 17
E 0.00
"---6
5•16
-0.05 -
-0.10 - 15
0.10 -
0.05 _ 20
Gre 5.4m
•7
19
0.00
-0.05 18
-0.10
0.10
0.05
-
-
• •_•11.5m
_20
18
v
E 0.00 -""-"• 115
-0.O5
-0.10 •6
I I I I • I I I I I I I I ! I I I I I I I I I I I I I ! I • • • • • I I
6 10 15 20 25 30 1 5 9
DEC 88 JAN 89
2.4 A quasigeostrophic
model
In this section,we examinea simplemodel to determinewhetherthe observedcurrentand
temperature oscillations
arecompatible
withthehypothesis thatJervisBayresponds principally
to longshorewindsas a resultof remoteforcingthroughthe mouthof the bay. The simplest
setof equationsthatcouldreasonablybe expectedto describethestratifiedcirculationin Jervis
Bay is linear,hydrostatic,Boussinesq
and frictionless:
u, - fv - -P=, (2.1)
New' 0 , (2.3)
and
w = •, (2.6)
whereg istheacceleration
duetogravity,and• isthesurface
displacement.
Using(2.3),(2.5)
and (2.6) togethergive
P•- /•:
8
Patz-O. (2.7)
0 km
I 2.0 I I
in which•o2hasbeenneglected
by comparison
withif. Thesurface
boundary
condition
onP
is (2.8), and the bottomboundarycondition,by (2.3), is
P,•-Oonz'-H. (2.10)
and
to
(-7-
P•P•)' (2.12)
By the hydrostatic
approximation,
implicitin (2.3), therelationship
betweenthe pressure
P and temperatureperturbations
T is
P•wg
otT , (2.13)
p=Q(x,y)cos
mtz
H '
(2.14)
wheren is an integer,representing
the modenumber.Taking n = 1, the equationfor Q is
where
k = (an (2.17)
NH
and
(2.18)
NH
The expressions
for k and l• are derivedusingthe requirementthat v, givenby (2.12), is
zeroonthecoast,andagainmakeuseof theassumption that•0:< <2e. Thewavesdescribed
by (2.16)to (2.18)havevelocities strictlyparallelto thecoast.Fortypicalvalues,•0= 10'5s'•,
[fl = 10"*s'l, N = 10-:s'l, H = 20 m, theoffshore length-scale1/Itis about1 km,whilethe
longshorewavelength,2•:/k, is about40 km. Accordingto (2.15), the formal pictureof the
density-drivendynamicsin JervisBay is that of a wave propagatinginto the bay alongthe
southcoastof the inlet channel(Figure8), clockwisearoundthe bay, andthenoutalongthe
f O•+
Q•,
=0 (2.20)
on the walls at x = 0 and x = X, and
(2.21)
O,,-ikQ- -•kexp
•(y•- y) (2.22)
on x = X, y• < y < y:. After solvingthe systemfor Q, the total solutioncanbe scaledto give
the correctamplitudeat the Black station.
As we havealreadynoted,JervisBaywasstronglystratifiedduringthemeasurement program.
Figure9 showstemperature profriescollectedbeforeand after the programat the central
stationshownin Figure 1. We will concentrate on the December5 profile.The average
buoyancyfrequency,calculated by fittinga straightline betweenthe surfaceand bottom
Temperature (øC)
I i 1
14-
18-
Jan 1989
22-
26
Figure9: Temperature
profiles
collected
in thecentralbayon5 December
1988and12January
1989.
temperatures,
is0.022s'•. Themaximum
value,calculated
forthetemperature
gradient
in the
centre of the water column is close to double this value.
a ( • az)+•z..o
dz t•(z) dz '
with
P D CraigandP E Holloway 25
dZ
Oil z-0,-H.
•
•:=--=7.1m -Is.
NH
For theactualtemperature
profile(Figure9), theeigenvalue
(calculated
numerically)
is •:
= 4.6 m4s.Thisis thesameeigenvalueaswouldoccurfor a constantbuoyancyfrequency of
N= • = 0.034s-•
•:H
If P hastheverticalstructure
givenby (2.14) then,by (2.13), the temperature
at theBlack
stationwill be givenby
T._Tosianz
H
p---H
•
g• To
cOS
•HZ
A full list of the parametervaluesusedin the modelrunsis shownin Table2. The results
of themodelsimulation
withN = 0.034s'• areshownin Figures
11and12.Figure11shows
the (normalised) pressure
field bothin-phaseandout-of-phase with the temperatureforcing;
that is, at the time of maximumtemperature at Black, and quarterof a periodlater.For this
value of N, the offshoredecayscale(baroclinicradiusof deformation),l/g, is 2 km and the
longshorewavelengthis 136 km, approximatelythreetimesthe circumference of the basin.
Figure 12 showsthe surfacevelocity vectorssimultaneous with the pressurefields in
Figure 11. By (2.11), (2.12), and(2.14), thevelocityvariesverticallyascosrcz/H.
Thusbottom
velocities,for example,will havethe samemagnitudebut oppositedirectionto thoseshown
in Figure 12.
1.0 - //•lacktemperature •
0.0
-1.0 -
l I I I I I I I I I I I I I I I .I I I I I I I I I I I I I I I I I I I I I
6 10 15 20 25 30 1 5 10
DEC 88 JAN 89
Figure10: Comparison
betweenthe linearlydelxended
temperature
signalfrom Black,andthebest-fit
harmonic,
havingamplitude
0.56øCandfrequency
9 x 10-6s-•
0.800
(a) (b)
Figure11: Themodelled
(normalised)
surface
pressure
fieldfor N = 0.034s-l, (a) in-phase
and(b)
out-of-phase
with the specifiedtemperature
at Black.
Fortheweaker
density
structure,
N = 0.022s'l, thesolution
structure
is notdramatically
altered(Figure13). Becausethe horizontaldecayscaleis shorterin thiscasethen,for the
sameamplitude temperature
forcing,currentsnearto thecoastaremarginallystronger,
butthe
decayis morerapidtowardsthecentreof thebay.
., . ,,
4 ,
'k ß
', 'k
(a) (b)
5 m/s
Figure12:Modelled
surface
current
fieldsforN = 0.034s-1,(a)in-phase
and(b)out-of-phase
withthe
specifiedtemperature
at Black.
2.6 Discussion
To enablea quantitative
comparison
with themodelresults,themeasured
time-series
were
eachleast-squares
fittedwitha single
harmonic
having
a fixedfrequency
to- 9 x 10'6s'l, as
describedfor theBlacktemperature in theprevioussection.The time-series
analysedin this
way werethosefromstationdepthsshownin Figs5 and7, exceptthat,for theRedstation,
the 14.5m temperature(Fig. 6) andthe6.5 m currentwereused.The analysis determinesan
amplitudeandphasefor eachtime-series.
(a) (b)
5 m/s
Results for both the model and measurementsare summarisedin Tables 3 and 4. Table 3
listsamplitudesof bothtemperaturesandcurrents. Thecunentamplitudes quotedarethose
in the directionof the semi-majoraxisof the currentellipse.The modelamplitudes are
maximaoverthewatercolumn,thatis, mid-watervaluesfor temperature andsurfacevalues
for currents.
Theagreement betweenthemodelandthemeasurements indicated
by Table3
is well within a factor of 2, exceptat the Red station.The measuredvelocitiesand,
particularly,
temperatures at Redsuggestlittle,if any,decayin signalstrength
withdistance
from the coastof the bay.
Station Temperature
(øC) Current(ms-1)
Model Measured Model Measured
Modelled Measured
An obviousshortcoming of the modelis the formulationof the forcingat the mouthof the
bay. As we found,the prescriptionof a temperatureoscillationat the nominalBlack station
(Figure 8) leads to an internal Kelvin wave propagatingclockwisearoundthe bay. For
consistency,to ensurethatenergydoesnot accumulate withinthe bay, it is thennecessary
to
allow the wave to exit, by specificationof the radiationcondition(2.22) at the mouth.This
formulationignoresall complicationassociatedwith matchingthe bay motionto the larger
scalemotionover the shell As just discussed, it also leadsto inconsistent phasebehaviour
across the mouth.
Retentionof the to/f termsin (2.11) and (2.12), however,allowsfor internalKelvin waves,
and indicatesthat someaspects,at least,of thephasestructurein the bay are consistent with
wave propagation.Since,in reality,the wavedoesnot appearto propagatein andout through
the mouth,the model needs,at the next stage,to incorporatedissipationto ensurethat the
wave energyin the bay doesnot grow withoutbound.
Acknowledgements
We wishto thank:GrahamSymondsandRick NunesVaz for theircontributions to thiswork;
Marc JeffreyandJohnMathiasfor theftassistance
with the field work;JohnAndrewarthafor
preparation
of thedata;JohnChurchfor hiscomments on themanuscript;and
CathyFlanagan
andNikki Pullenfor seeingthe manuscript
into typescript.
Thisprojectwaspartiallyfundedby theAustralian
Department of Defence,asa component
of a baselinestudysupervisedby the CSIRO Divisionof Fisheries.
2.7 References
Church,J. A., H. J. Freeland and R. L. Smith, 1986a: Coastal-trappedwaves on the east Australian
continentalshelf.Part I: propagationof modes.Journalof PhysicalOceanography,16, 1929-1943.
Holloway, P. E., G. Symonds,R. Nunes Vaz, M. Jeffrey and J. Mathias, 1989: Oceanographic
measurements in JervisBay, December1988 to January1989.Workingpaper1989/2,Departmentof
GeographyandOceanography, AustralianDefenceForceAcademy,44 pp.
Huyer, A., R. L. Smith, P. J. Stabeno,J. A. Church and N.J. White, 1988: Currentsoff south-eastern
Australia:resultsfromtheAustralian
CoastalExperiment.
AustralianJournalofMarineandFreshwater
Research, 39, 245-288.
Middleton,J. F. andF. Viera: 1991: The forcingof low frequencymotionswithin BassStrait.J. Phys.
Oceanogr,21, 695-708.
Abstract
3.1 Introduction
-[ 1i i i i i i i I I i I I I I I I [.•."L.I•I
IIIIIIIi
LOCH SUNART
5 10 15 20
I,
+ >:+ ' .
100
5 10 15 20 25 30
Distance from mouth (•)
Figure2: Observedsalinitydistributions
(psu):(a) August14-15(days226-227),Co)August24-25 (days
236-237).
A
2-
vvvvv'vvvvvvvvvvv
212 214 216 218 220 212 214 216 218 220
?IHE IN DAYS ?IHE IN DAYS
15.00 B
14.75' ø
14 50'
14 25'
1400'
13 75'
13 50'
13 25' -30
212 214 216 218 220 212 214 216 218 220
TII•IE IN DAYS TI[VlE IN DAYS
Table 1 showsthe resultsof a tidal analysisof the elevationand currentdata. If the tide
has the characterof a progressivewave as it entersthe inner basinwe would expectthe
elevationand currentsto be in phase. However,Table 1 showsthat the surfacecurrentled
the elevationby a phaseof about 55ø at the M,. frequencywhile the S,. currentled the
elevationby nearly80ø.
Elevation
H G
M2 136.6 178.0
S2 54.8 232.5
M4 5.4 201.8
MS4 4.7 331.6
5m
This suggests
thata significantamountof tidalenergywasbeingreflectedfromthehead
of the loch,causingthe tidal waveto havesomeof the characterof a standingwave. Both
surfaceandbottomM,. currentswererecfilinearanddirectedtowards11ø (with respectto
magneticeast)which impliesa true bearingof about23ø anti-clockwise from east. This
agreeswell with the orientationof thebasinaxis(Figure1). A strikingfeatureof the tidal
analysis
is thenearly180øphase
difference
between
theM•_currents
nearthesurface
and
bottom
(Table
1). Figure
4 shows
a short
portion
ofthedata,
taken
fromaperiod
ofspring
tides,thatillustrates
thephase
difference
in thecurrents
andthenon-linear
character
of the
currentsand temperaturesignal.
30
20'
14.6
14.4
14
I • I • I •
Figure
4:Current
andtemperature
details
during
aperiod
ofspring
tides
(August
2-3):(a)5m(solid
curve)and50m(dashed
curve)currents,
(b)5mtemperature.
Figure5: Depth-time
contour
plotof theADCPcurrentprofile,ebbflow is shownby shading(thedata
showncoverthe periodfrom 1900on August18 (day230) to 2(X)0on August19 (day231)).
(3.1)
U- 1 dW
ß (3.2)
k dz
If thesecond
derivative
in (3.1)is writtenin finitedifference
formthena matrixequation
for the valuesof W at eachdepth canbe obtainedin whichthe unknownwavenumber
scale
k is relatedto the eigenvalues of thematrix. A moreintuitivesolution
canbe foundby
writing(3.1)in finitedifference
form,setting
W(0)=0andthenmarching thesolution
forW(z)
from the surfaceto the bottom. If W(-H)=0 thena modalsolutionhasbeendetermined,
otherwisea new valuemustbe selected for k andtheprocedure
repeated.The methodcan
be madeto convergerapidlyby carefulchoiceof the parameterk, the numberof zero
crossingsof the solutiondeterminingthe orderof the mode.
VERTICAL MODES
20' 20'
40' 40'
50' $0'
60 6(
Figure6 shows
a representative
profileof sigma-t
fromthemooring position
duringthe
survey
of August
14-15.Thiswasusedtocalculate
theN2(z)profilethatappears
in (3.1)with
to setto the frequencyof the M2 fide. The figurealsoshowsthe verticalstructure
of the
amplitude
of thehorizontalcurrent,
thesolidcurveshowing
thefirstmodeandthesecond
modebeingshown by thedashedcurve.Thefirstandsecond
modes hadhorizontal
length
scalesof27kinand12kin,respectively
(Table2). Bothmodes
show theobserved
180øphase
difference
betweenthedepths
of 5mand50m.It shouldbenotedthatthe27kmlengthscale
of the firstM2 modeis aboutfour timesthelengthof theinnerbasinandthata resonant
response
maytherefore
exist.Thecalculations
werealsomadeforthefrequency
of theM4
tidalconstituent
andtheresultsare again summarized
in Table2.
Calculated
mode 1 mode 2
Observed
3.5 0.9
[u_,l
Equationof continuity
u +(wa9-
0 (3.3)
(3.4)
Momentumequation
(3.5)
+8Bigtl+ pdz
•- 0
cgx po Ox
Salt conservation
(3.6)
Equationof state
p - po(4•+ ps) (3.7)
wherex and z are the longitudinaland verticalcoordinates; u and w are the corresponding
velocitycomponents; B is the width;•1 is the surfaceelevation;H is the meanwaterdepth;
CDis thebottomfrictioncoefficient;K•, N• are thelongitudinaldiffusivityandviscosity;and
I•, Nzaretheverticaldiffusivityandviscosity;
P0is thedensityof freshwater,and[• (7.6 x
10'4)represents
thedependence of density
on thesalinity,s.
Nz3u/3z= windstress
at thesurface
andN• 3u/3z= I u I at thebottom.Thecondition
thatthereshouldbe no verticalflux of saltat the surfaceandbottomis satisfiedprovidingthat
I• 3s/3z= 0 at z = •1 andz = -H.
At the headof the loch the fiver flow is specifiedand the salt flux is set to zero, while at
the mouththe sea level elevationis specified. When inflow occursat the mouth the ocean
salinity is specified,otherwisean advectivebalanceis assumed and horizontaldiffusionis
neglectedon outflow.
(c) The hybrid advectionschemeof James(1986) has been used to treat the horizontal
advectionof salt. A purelycenteredschemegenerated
negativesalinitiesin theregionsof
strongadvectionnear the head of the loch.
Threeformulations
for theverticaleddycoefficients
were tested(in cgsunits):
(i) Wang(1983)allowedtheturbulent
fluxesto dependondepthby usinga Munk-Anderson
parameterization:
inwhich
Ri isthelocal
Richardson
number
given
by
(3.10)
I u3 u3
i U3 US U3 US U3
I
U
l 1• UI.•S UIS UlS i
i
z--,o
ui
!
!
s ui s
-- I
I
I ui S ui S ui $ u• •..,•u • s
I U3 U3
I
I
U3 U3 US US •3 I
u• S u• s u• s u• s Ul S Ui S UI S
i
i
]./)//U3/ / /Us
.//./U$-=•
Us ui S ui ui
us
S ui
us
S
i
i
I
I
U3 I
u! s Ui Ui S Ul S
,
i
Figure
7:Grid
scheme
used
inthe
model,
the
surface
isallowed
tomove
vertically
through
the
grid
(ii)The
model
ofBowden
and
Hamilton
(1975)
wasunstable
whenthemixing
coefficients
were
depth
dependent
sothey
related
theeddy
coefficients
toanoverall
Richardson
number
defined
in terms
ofthetoptobottom
current
anddensity
differences.
However,
with
the
present
integration
scheme
their
formulation
remains
stable
when
theRichardson
number
is
made
depth
dependent,
consequently
thefollowing
expressions
were
tested
inwhich
Riwas
givenby (3.10):
N•nlul.(1+7
N,' No+ 10 R0-• (3.11)
aHlu
Kz' Ko+ 10 l.(1+R0_7 (3.12)
It should
benoted
thatthese
expressions
areempirical
andwerederived
fromdatacollected
in theMersey
Estuary
(Bowden
andGilligan,
1971).
(iii) Using
theresults
of KentandPritchard
(1959),
Blumberg
(1977)
represented
the
dependence
of K• ontheRichardson
number
as
(3.13)
where
Ricisacritical
value
oftheRichardson
numberabovewhichmixing
issuppressed.
By
assuming
thatthegradient
Richardson
number
(Ri)andthefluxRichardson
number
(Rf)are
relatedby
Rf. 1
Ri l+Ri
he derivedanexpression
for N,.of theform
N• = Kz(I+Ri)forO<Ri<Ri• (3.14)
andN=- K=- 0 whenR•• R/•
Figure
8 shows
theexternal
forcing
thatwasused
when
simulating
theobserved
salinity
section
shown
inFigure
2b. Themodel
wasspun-up
for3 tidalcycles
byspecifying
and
holding
constant
thesalinity
observed
onAugust
14-15
(Figure
2a)while
forcing
with
thetidal
elevation
atthemouth,windstress
andrun-off.Theintegration
wasthenstepped
forward
for21tidalcycles,
over
theperiod
August
14-25,
withthesalinity
distribution
beingupdated
dynamically
using
(3.6).Thefinaltidal
cycle
wasthen
averaged
toproduce
thedistributions
showninFigure9. Following
thestrong
run-off
eventonAugust23theobservations
showed
averythin(less
thanlmthick)
surface
layerwithasalinity
ofabout
5psu.Thislayer
could
notberesolvedbythe2mgridspacingused
inthemodel;thelowest
salinities
predicted
by
49
A J Elliott,P A Gillibrand& W R Tunell
themodel
inthesurface
layeroftheinner
basin
wereoftheorder
of 10psu.
August 1989
14 15 16 17 18 19 20 21 22 23 24 25
-1
-2
20
10
-10
-20
280
240
200
160
120
80
40
Figure
8: External
forcing
used
todrive
themodel
when
simulating
thesalinity
distribution
observed
on August24-25.
TheMunk-Anderson
formulae
givenby (3.8)-(3.9)
produced
goodagreement
withthe
observations
(Figure
9a),bothinthenear-surface
layers
oftheinner
basin
andinthedeep
waterof themainbasin.However,
thelowvalues
of thecoefficients
thatweren•essaryto
reproduce
thesharp
near-surface
gradients
intheinner
basin
resulted
inreduced
mixing
at
mid-depth
inthemiddle
basin.
Forexample,
the33psu
contour
wasreproduced
atadepth
of
about
15mwhich canbecomparedtotheobserved
valueofabout25m(Figure
2b).Forlarge
values
ofRi,theMunk-Anderson
formulaehave
theeff•t ofreducing
theeddycoefficients
tovalues
lessthanthose
givenbytheothermethods.Thebestsimulations
wereobtained
20-
- - i.•]:5.!'i•...:
' "-:--33.5----!,. •:• •
40'
60"
80'
100-
5 10 15 20 25 30
---• Distance from mouth (km)
0.4 m/s
• 4o . •'" t ......
'' - 'm
!:•i....
- •.'.-i"•-;'.½-tl.' '"• - -
1oo' 5 10 15 20
;--•--.'
25 30
Distance from mouth (km)
0.4 •/s
Figure9: Salinitysimulations
for different
mixingparameterizations:
(a) Wang(1983), (b) Bowden
andHamilton(1975), (c) Blumberg(1977).
The formulae (3.11)-(3.12), adaptedfrom Bowden and Hamilton (1975), allow the
maximumvaluesof the mixingcoefficients to dependon waterdepthandthe instantaneous
currentspeed.The eddydiffusivity,Kz,givenby (3.11)-(3.12)for a chosen Ri, is higherthan
the valueobtainedusing(3.8)-(3.9) due to the strongtidal currentswithinthe loch. In the
innerbasinthishasthe effectof reducingthe salinityvaluesin the deepwater,althoughthe
overallagreementwiththeobserved salinitiesis good(Figure9b). Thebestsimulations were
obtainedusingNo = Ko = 0.05, N• = 0.20 and K• = 0.10.
Realistic salinitiesin the inner basin could not be reproducedwhen using expressions
(3.13)-(3.14) due to Blumberg (1977). Although depth resolvingmodels are usually
insensitive
to thevalueschosen
forK• andN• provided
theyarelessthan10s - 10• cm2 s-1,
themodelwouldnotremainstableunlesstheywerebothsetto 10 cm2 s4 when(3.14)was
used. The schemewas also unstableunlessthe criticalRichardsonnumber,Rio was set to a
valueof 5, eventhoughBlumberg(1977) suggests a valueof around10 for the upperlimit.
With theserestrictions,the near-surfacesalinitiesin the inner basin reacheda minimum value
of the order of 20 psu and the observed strongnear-surfacestratificationcould not be
reproduced (Figure9c). This suggests thatthe formulationgivenby (3.14), whichproduced
realistic salinitiesin the weekly stratifiedwaters of the ChesapeakeBay, may not be
appropriatein highly stratifiedfjord regions. However,in contrastto the surfacewaters,
reasonablesalinityvalueswere computedfor the deepwater of the outer basins. The best
simulationswere obtainedusingKo = 0.05, k• = 0.035 and Ri• = 2.
Figure 10 showsthe resultsfrom a seriesof testsin which the sill at the entranceto the
innerbasinwassetat differentdepthsduringhighrun-offconditionsthatgeneratesignificant
near-surface stratification.To maketheseteststhefiver flow wassetat 65 m3 s4, a realistic
initialsalinitydistribution
wasassigned to thegridpointsandthentheoceansalinitywasheld
fixed while the model was spun-upfor 40 tidal cycleswithouttidal forcing- and withouta
sill at the entrance to the inner basin. This allowed the internal circulation to establish and
for realisticsalinitystratificationto develop. At the endof the 40 tidal cycles,by which time
thesolutionwasapproaching steadystate,the tidal forcingwasappliedat themouth,a sill was
inserted at the entrance to the inner basin and the simulation was continued for a further 7
tidal cycles,the last two of which were usedto generatethe figures. (The resultsof the 40
tidal cyclesspin-upwas storedon disk so that the final 7 tidal cyclesof simulationcould
continuewithoutduplicatingthe initial calculations.)With the sill at 52m (Figure 10a) the
simulatedtidal cunenmat the mooringpositionare fairly barotropicwith maximumvaluesat
springtidesbeingreduced
to around10cms". Thesurface
flowebbsthroughout
thetidal
cycle becausethe tidal currentsduringflood are weakerthan the river inducedsurface
velocity. Raisingthe sill to 20m and 14m(Figure10b,c)createsa jet centeredat 10mon the
flood tide and introducesthe observedverticalphasedifference. In Figure 10d are shown
resultsfrom the simulationwith the sill at its actualdepthof 8m, and the strongverticalphase
signalis in goodqualitativeagreement
with theADCP velocitiesshownin Figure5.
HW LW HW LW LW
2.0
•
D--'•'
:"
•'•;.•:•i•"-'
'::•
•. ':'•'•'-•
•-'•.!-•
........
•'•:•;
1
..
10
*'•:•i•-.'•.
.... • - ''•"•'"
'•:':
"•'
•2o .20
30
4O
50' , , , i , , ,
Figure 10: The influenceof sill depthon the tidalcurrentsin the innerbasinduringa periodof high
run-off: (a) 52m, (b) 20m, (c) 14m, (d) 8m.
• 02 0.2•
•0•
t
s0•
0.5 1.0 1.5 2.0 0.5 1.0 1.5 2
Figure11' Predictedcurrents,shear,stratification
andRichardson
numberduringa periodof low run-off:
(a) velocity,(b) 104.(au/az)
2,(c) -g/p(ap/az).104,
(d) Richardson
number.
3.4 Discussion
HW LW HW
40
50
0.5 1.0 5 2.0 0.5 1.0 1.5 2.0
lO
•20
• •o
40
50•o 2•
. • , , , ---0.2,--. •n , ,• , , 50
0 5 1.0 1.5 2 0 0.5 1.0 1.5 2.0
Tidal Cycles Tidal Cycles
Figure12: Predicted
currents,
shear,stratification
andRichardson
numberduringa periodof highrun-
off: (a) velocity,(b) 104.(3u/3z)
2,(c) -g/p(3p/3z).104,
(d)Richardson
number.
Thehighlynon-linear character
of thelochdynamics limitsthesizeof thetimestepthat
canbe usedwith theimplicitmodel. Thisis in contrast to theshallow,dissipativeregions
suchas theChesapeake Bay, wherethe modelhadbeenpreviously applied.The mixing
parameterizations of BowdenandHamilton(1975)andBlumberg (1977),whichhadbeen
developed for applicationto the MerseyandPotomac estuaries,
respectively,werenot
successful in simulatingthe strongnear-surface
stratification
observed in the loch. This
suggests thatthemodelcouldbeusefullyextended through theuseof a closure scheme for
theprediction of theturbulent mixing(SmithandTakhar,1981). A Richardson number
probe,consisting of twoS4 currentmetersseparated
by a CTD wasprofiledverticallynear
thesillduringthefieldworkandthedataarenowbeinganalysed to deriveestimates of the
verticalshearandRichardson numberthroughout
thetidal cycle. Thisshouldleadto a better
understanding of theshearinducedmixingandenableits effectiveparameterization in future
modellingstudies.
Acknowledgements
We thankourcolleagues
in theHydrography
Sectionof theMarineLaboratory,
Aberdeen,
in
particular
R D AdamsandR Payne.We alsothankthecaptainandcrewof theRV SeolMara
for their help duringthe field work.
3.5 References
Blumberg,
A.F., 1977:Numerical
modelof estuarine
circulation.
J.Hydraul.Div. ASCE,103,295-310.
Cushman-Roisin,
B., andH. Svendsen,1983: Internalgravitywavesin sill fjords:verticalmodes,ray
theoryand comparison
with observations.CoastalOceanography,
H.G. Gade,A. EdwardsandH.
Svendsen,Eds., Plenum Press, 373-396.
Hamilton,
P., 1975: A numerical
modelof theverticalcirculation
of tidalestuaries
andits application
to theRotterdamwaterway.Geophys.J. Roy. Astron.Soc.,40, 1-21.
Hamilton,P., 1977:Onthenumericalformulation
of a time-dependent
multi-level
modelof anestuary,
withparticularreference
toboundary
conditions.
Estuarine Processes
II, M. Wiley,Ed.,Academic
Press,
347-364.
Kent,R.E.,andD.W. Pritchard,
1959: A testof mixinglengththeories
in a coastal
planeestuary.J.
Mar. Res., 8, 67-72.
Wang,D.-P.,1979:Wind-driven
circulation
in theChesapeake
Bay,winter1975.J.Phys.Oceanogr.,
9, 564-572.
Wang,D.-P.,1983:Two-dimensional
branching
saltintrusion
model.J.Waterways,
Port,Coastal
and
OceanEng. ASCE, 109, 103-112.
Abstract
The onsetof vigorousverticalmixing in the seawardend of the wave, after the time of
maximum barotropiccurrent,has indicatedthat this hydraulicphenomenonmay be an
importantcontributorto overallestuarinemixing. Theseobservations are the first of theft
kind for a smallpartiallystratifiedestuary,andmay be evidenceof an importanttransitional
phenomenon in estuarinemixing.
4.1 Introduction
58 D R Sturleyand K R Dyer
Estuaryobservedintense
mixingperiods,
presumably produced by internalwaves,occupied
only20% of thetidalcycle,butaccounted
for 50% of theverticalturbulent
saltflux.
4.2 Survey
The timeof predictedhighwaterat Cargreen, on 6 November1989,was 10 a.m. A launch
was mooredat Station"45" (seeFigure 1) from an hourand twentyminutesafter predicted
high waterHW+l'20 to HW+5:40, takingverticalprofilesof salinity,temperature, current
velocityanddirection,at twentyminuteintervals. Theseprofilesarepresented in Figures2
to 6. The temperature (T) and salinity(S) measurementswere madewith an MC-5 in-situ
T-S bridge(accurateto +_0.1 salinityunitsand_+0.1øC). The currentvelocityanddirection
were measured with an NBA (DNC-3) impellercurrentmeter(accurateto + 2% in velocity,
and+ 10ø in direction). The depthwascalculated from wire lengthout andangle,with a
typicalerrorof 5% to 10%.
D R Sturleyand K R Dyer 59
METRES
TQ--
FIX
MARKS
Figure1: The depressionin theTamarEstuarybedat Cargreen.The Tamarformsa naturalboundary
betweenDevonandCornwall. It is nominallypartiallymixedandhasa meanneaptidalrangeof 2.3m.
Cargreenis 20km fi'om the tidal limit and 10km from the mouthof the estuary. The isobathsare
plottedevery lm referencedto chartdatum. Station"45" was wherethe Eulerianmeasurements
were
taken.
60 D R SturleyandK R Dyer
station"45" wasselected
to represent
the "upstream"
conditions
of theflow but,asFigures
2 to 5 show, the internal wave distorted the level of the thermocline at this location. The
velocityprofilesshowa similar,butmoresubtleeffect. Figure6 showsthedepthaveraged
currentspeed,•, at "45". It depictsan ebbstreamwith a doublepeakin velocity. This
behaviouris commonin themiddleandupperTamar(George,1975).
o - 30 ppt
om
lm
2m
3m
4m
11:20 11=40 12:00 12=20 12:40 13:00 13:20 13:40 14:00 14:20 14:40 15:00 15:20 15:40
Time in
houre (GIRT)
10 - 14'c
Om
D•pth
in
mettee
11:20 11:40 12:00 12:20 12:40 13:00 13:20 13:40 14:00 14:20 14:40 15:00 15:20 15:40
Time in
houre (GMT)
At thecentreof thedepression
theorientationof themainchannelwasestimated
to be 142ø
fromChartNorth.Thevelocity measuredat"45"wasdividedintovector
components, along
the orientation
of the mainchannelat "45"(longitudinal
velocity-positive
velocitiesare
seaward in Figure4), andin the cross-channel
direction
(transversevelocity- positive
velocities
aretowardtheCornishbankin Figure5). Figure4 indicates
maximum seaward
flow at mid-depthat, or slightlyabove,the level of the thermocline. The level of this
maximum flowevolvedin timein exactlythesamewayasshown by theT andS profiles.
Figure5 showsa strong
transverse
component oftheflowin theupper
layerswithnegligible
flow below the thermocline. The level at which the transverse
flow becamezero coincided
with thebottomof the thermocline,
andevolvedwith it in time. The transverse
flow reversed
D R SturleyandK R Dyer 61
o - I m/s
om
I
Depth
in
metres
.......
. . .
11120 11140 12:00 12120 12140 13100 13120 13:40 14100 14=20 14140 15100 15:20 15:40•
Time in
hours (GNU)
o - 1 m/s
Om
Depth
in
metres
11:20 11:40 12:00 12:20 12:40 13:00 13:20 13:40 14:00 14:20 14:40 15:00 15:20 15:40
Time in
hours (GNT)
62
D R SturleyandK R Dyer
Depth
averaged
velocity
in m/s
0'6
0'5-
0'4 -
0'3-
0'2-
O' 1 -
0'0 ' I I I I 1
11 12 13 14 15 16 17
Figure6: Themagnitude
of thedepthaveraged
velocity,
fi, at "45"on6 November
1989. Thetime
of predicted
highwaterwas10:00(GMT).
4.4 Echograms
andSparData
TheSpardataandsynchronous echogramspresentedin Figures
7 to 12 weremadeon
longitudinal
transects
in theseaward
direction.Theyareinsequence
andareannotated
with
fixedpositions
(yachtbuoysandlandmarks).Table1 shows
someof thedistances
between
fixes in metres.
D R Sturleyand K R Dyer 63
J BR GP 45 VI $W WP Clt VA
0m
Landward Seaward
64 D R Sturleyand K R Dyer
I-• .tu D g
P
T
2u H
t t t t t t t t t - 3u
j DR GP 45 SW WP CH VA RH
0•
2•
........ 4m E
P
T
H
6m
45
La ndwa rd Seaward
HW + 4:45
BLQ I.• J BR GP 45 vI SW WP CH VA
0m
8m
Landward
Seaward
D R SturleyandK R Dyer 65
HW + 5:30
0m
2m
4m
,• TU BLQ 45 VII -
Landward Seaward
HW + 6:00
Om
lm
2m
- 3m
TU BLQ LQ J BR GP 45 Vl SW WP CH VA RH
0m
4m
TU
I
Landward
Seaward
66 D R Sturleyand K R Dyer
HW + 6:25
-- ,• ,ldwm
,, lm
2m
TU• S DQd SR GP 45 VI SW C VA
0m
6m
8m
BLQ
Landward Seaward
4.5 Results
D R Sturleyand K R Dyer 67
68 D R SturleyandK R Dyer
Free surface
. u
(b). Before
maximum
stream
(c). Maximum
stream
( u
< u
(e). After
maximum
stream
•-- u
(f) . No tidal
stream
ß u
Figure13:Thestages
whichformtheconceptual
modelofleewaveformation.
(a),(b),(c): Thewave
growsoverthelm•dward
slopeof thebed&pression
in theaccelerating
tidalstream.(d): After
maximum tidethewavesteepens
nonlinearly.(d),(e): Thewavepropagates
intotheslackening
stream.
(e),(f): Thewavemayfissionintosolitary
waves through a balance
between
dispersion
andnonlinear
steepening.
At HW+6:45thethermocline
stillappeared
to be displaced
by about1.3m- 1.4m,butthe
wave structureand the level of the undisturbedthermoclinewere difficult to define.
Theerrorincurredin measuring
thesizeof thewavetroughis estimated to beof theorder
+0.2m. This erroris roughlyequivalent
to a 10% uncertaintyin the largestthermocline
D R SturleyandK R Dyer 69
displacement,
andis a resultof makingessentially
subjective
judgements
aboutthelevelof
thethermocline.
It ignores
instrumental
errors,
whicharefelttobesmallin comparison.
Maximum
thermocline
displacement
in metres
&
Depth averaged
velocity in m/s
2.5
1 '5'
ECHOGRAMS•
T•HERtdALitdAPS
_
0'5 -
11 12 13 14 15 16 17
Thetotalwaveenergy,
perunitlength
inthey (across
estuary)
direction,
Er,wascalculated
from (Phillips,1977):
70 D R Sturleyand K R Dyer
Er-f •(x)'X
pg•x
In makingthecalculations
it wasnecessaryto definea fixedhorizontal
domainoverwhich
to calculatetheenergy. Sincethesurveysdid notextendfar enoughin a landwarddirection
to definecompletelytheupstream thermoclineposition,"BLQ"waschosenas thelandward
limit of the waveform. In the cases when the transectdid not extend to "BLQ" the
thermocline
levelwasextrapolated
fromthemostlandwardpointof thetransect.In all cases
"VA" was more seaward than the seaward end of the wave. The undisturbed thermocline
levelwasidentifiedin theway alreadydescribed.Theintegration
wasperformed in sections,
usingthe fixed stationpositions,and measuringa meanthermoclinedisplacementbetween
section
ends.Apwasassumed
as16kgm
-a,whichwasthemeandifference
in layerdensities
over the periodHW+2:45 to HW+6:45.
This definitionof the boundariesof the wave would,certainlyfor the latter slagesof the
study,be restrictiveof the landwardedgeof the wave. However,it is anticipated thatthe
effects of this are not severe, and that the largest uncertaintyis in the thermocline
displacement,
(_+0.2m
or +10%), leadingto an estimated
errorof about20-25%in thelargest
energyestimates.
Thetotalwaveenergyisrelatively
high,around160- 180kJm '•, atHW+2.45,beforefalling
to 120-140kJm
-• at HW+3:00in linewitha sharpincrease
in theebbstream.Thereason for
thisis unclear. Otherwise,the energyincreased
with an increasing
ebbstream,andcontinued
to increasefor a while when the ebb slackened,presumablydue to the nonlinearsteepening
mechanismdiscussedearlier. The energy diminishedwhen the ebb streamdiminished
significantly,especiallyafter HW+4:45.
D R SturleyandK R Dyer 71
Total wave
energy in
Kilojoules
per metre
&
Depth averaged
velocity in cm/s
250 -,
ECHOGRAIdS
200 -
150 -
100 -
TH•:RIdAL:MAP<•
50-
0 I I I I I
11 1:2 1• 14 1• 1• 17
Figure
15'Theobserved
total
wave
energy
inkilojoules
permetre,
with
thedepth
averaged
speed
at
"45"incmpersecond.
Thetimeofpredicted
highwater
was10:00
(GMT). Theobservations
are
derived
frombothechograms
andthermal
maps.
4.5.3 Mixing
Between
HW+4:45
andHW+6:45radical
changes
inthevertical
distribution
ofdensity
took
place
at"RH"and"VA". Thedistortion
ofthethermocline
duetothewave presence
enhanced
theshear
intheseaward
endofthewave,
near"RH",
causing
fresher
water
tobe
mixed
from
above
thethermocline
down intoit. Water
wasthen
redistributed
vertically
in
thethermocline
through
mixing.Thethermocline
thickened,
expanding
downwards,
atthe
expense
ofthelowerlayer.Withtimethisthermocline
thickening
wasmanifest
alongthe
72 D R Sturleyand K R Dyer
4.6 Discussion
Nonlinearsteepening
in the slackeningtidal streamappearedto play a largerole in the
wave formation and evolution. There is clear evidence for this role in theoretical studies:
Lee & Beardsley(1974), Pingteeet al (1983) and Sturley(1990). In particular,non linear
advectiveaccelerationof internal waves in a slackeningtide can producesteepeningand
growth. Twice, whenthebarotropic tidal streamslackened,
aftera maximum,the wavegrew
in heightas a response. The heightof the wave, in relationto the total waterdepth,itself
suggeststhatnonlinearities mightbe importantin thebehaviourof thebaroclinicresponse of
the estuaryat this location.
A reductionin the wave growthoccurredas the wave height approached2m, and this
appearedto be a limitingheight. Wave energywaslostthroughmixinganddissipation and
this causeda reductionin wave height. The positionat which mixing in, or across,the
thermoclinewas first observed,was in the downstream(seaward)end of the wave. This was
also the positionat whichthe shearacrossthe thermoclinewould havebeenmostenhanced.
The regionof increased mixingmigratedupstream(landward)to meetthe troughof thewave,
suggesting that it had a largerpropagationvelocitythanthe wave itself.
The transferof energyto mixing was almostcertainlythe causeof the lee wavesdecline
in heightandenergy,in the latterstagesof the tide (Figures9 to 12). Futurestudiesaim at
improvingthe quantificationof thisprocess.
4.7 References
D R Sturleyand K R Dyer 73
Lee, C., and R.C. Beardsley,1974: The generationof long nonlinearinternalwavesin a weakly
stratifiedshearflow. J. Geophys,Res., 79, 3, 453-462.
Maxworthy,T., 1979. A noteon the internalsolitarywavesproducedby tidal flow overa 3-D ridge.
J. Geophys.Res., 84, C1-6, 338-346.
New, A.L., K.R. Dyer, andR.E. Lewis, 1986: Predictionsof thegeneration andpropagation
of internal
wavesand mixingin a partiallystratifiedestuary. Est. Coast.SheffSci., 22, 199-214.
Pietrzak,J.D., C. Kranenburg,
andG. Abraham,1990: Resonantinternalwavesin fluid flow. Nature,
344, 6269, 844-847.
Pingree,R.D., D.K. Griffiths,andG.T. Mardell, 1983: The structureof the internaltide at the Celtic
Sea shelf break. J. Mar. Biol. Assoc. U.K., 64, 99-113.
74 D R SturleyandK R Dyer
Posmentier, E.S., 1977: The generationof salinity finestructureby vertical diffusion. J. Phys.
Oceanogr.,7, 298-300.
Abstract
During the summerof 1988 datawerecollectedin the upperreachesof the Tamar estuaryto
investigateverticalturbulentmixing of soluteduringneaptides. Measurements were made
of the vertical variationof turbulentmean longitudinalvelocityand salinity and of the
turbulentperturbations of the verticalandlongitudinalcomponentsof velocityand salinity.
On theearlyebbtidea largeverticalsalinitygradientdeveloped,themeasured turbulentfluxes
were found to be small and an extensive internal wave field was observed. Consideration of
the temporalevolutionof the verticalsalinityprofilessuggested
thatthemeasuredfluxeswere
an underestimateof therequiredverticalexchange andit wasconcludedthateitherundetected
smallscaletransportassociated with internalwavebreakingor Wansport due to a transverse
secondarycirculationwas takingplace.
5.1 Introduction
The movementof water and solutesin estuariesis governedby the turbulentmean flow
structureandturbulentdiffusionprocesses.Themeanflow is causedby fluvial,tidalandwind
influencesandcontrolledin detailmainlyby turbulentdiffusionprocesses. The production
of turbulencewhich causesthe diffusionis causedby bed drag characteristics, the mean
velocityanddensitygradients,internalandsurfacewavecharacteristics
andacceleration.Thus
thereis strongfeedback
betweentheturbulence
andmeanfields. Giventhatthebathymetry
of estuarinechannelsis usuallycomplexit may be anticipatedthat esmarineflows and the
associatedscalarfieldsare complexfunctionsof spaceandtime.
In thelasttwo decades
extensive
measurements
havebeenmadein boundary
layersin air
and water in laboratories, some measurementshave been made at sea and a few have been
madein estuaries.The paucityof thelatterdatais surprisingconsidering theanthropogenic
pressures
on natureleadingtonumerous mathematicalmodelling studiesof esmarinetransport
processes.
Thereare,however,considerable physicaldifficultiesin obtaininggoodfield data
76 E J DarbyshireandJ R West
The verticalturbulenttransport
of momentumor solutemay be relatedto the gradientof
the meanflow via the mixing lengthgivenby
(5.2)
for momentum,and by
(5.3)
E J Darbyshireand J R West 77
Theoretical
relationships
relatingtheturbulent
mixinglengthsin homogeneous
andstratified
flows to Ri have been proposedby Rossby& Montgomery(1935) and Kent & Pritchard
(1959) with
and
78 E I DarbyshireandI R West
I I ! I
16.00 ca
--1
(b.) Velocity profiles O. 75 ms
I I
11. O0 12. O0 13. O0 14. O0 15. O0
I I I
11. O0 12. O0 13. O0 14. O0 15. O0 16 O0
Time
E J Darbyshireand J R West 79
depthvariesfrom 1.0m to 5.0m dependingon the stateof the tide. The measurements were
takenon the 4th and 8th of July 1988 whenthe daily meanfluvial dischargesof 6.8 cumecs
and 5.6 cumecsrespectivelywere aboveaveragefor the time of year but low comparedwith
winter valueswhich can exceed100 cumecs. The 8th was a neap tide and the 4th was an
intermediatebetweenneapand spring.
•_
(a.)Depth
o Velocity
meanvalues
z• Salinity _
•-- I I I I I I I I I I I
c5 13.O0 14.O0 15.O0 16.O0 17.O0 18.O0
-1
(b.) Velocity profiles o. 75 ms
I
[ I I I
13. O0 14. O0 15. O0 17.00 18.00
I I I
13. O0 15. O0 16. O0 18. O0
Time
80 E J DarbyshireandJ R West
d50 pp•t
Figure3: Profiles
andverticalgradients
of salinity
computed
fromfittedquadratics.
Turbulent
meanvelocityprofilesweremeasured usingeightBraystoke
impellerheads
mountedona 2.0mmastat intervals
of 0.25m. Themeanvalueswerecomputed over50s.
Turbulentmeansalinity
profiles
weredeterminedfromwatersamplespumpedtothesurface
via pipeswhoseinletswereattachedto the samemast. The salinitiesweremeasured
with an
MC5 salinometer.
Turbulent
perturbations
oflongitudinal
andvertical
velocity
weremeasured
using5.5x10'2m
diameter
electromagnetic
flowmeters(Coinbrook
Instrument
Development
Ltd.)andthoseof
salinity using an inductivesalinometerwith four sensorheads. These instrumentswere
mounted
on a streamlined
mastin sucha way thatthehydraulic
interference
betweenthe
sensorheadswasconsideredminimal. The flowmeterheadsweresituatedat 0.40m and 1.12m
withthesalinometer
heads
at0.56m,1.24m,
2.06mand3.07mabove
thebed.Theseparation
of theflowmeterandsalinometer
headsis of relevanceto thelaterdiscussion.
The mastwas
mounted on a weighted
bedframefittedwithorthogonal inclinometers
to ensurethatit was
level on the bed.
Turbulent
meanvelocitiesandsalinities
areshownasverticalprofiles
in figs.1 & 2 forthe
4thand8thofJulyrespectively
together
withthetemporalvariation
ofthedepth meanvalues.
Thevelocity
profiles
areplotted
ona logarithmic
scale.Thepointsdefiningeachprofilehave
E J Darbyshireand J R West 81
00.79
00.66
oO.
6J
o0.71
00.97
ol.
95
o15.50
oO.
/
JJo0.45
00.54
o0.
7,«o,.16
04.56/
o0..•1
o0.50
o0.?• o1.05 /
o.
II
Figure4: GradientRichardsonNumbers,4/7/88.
82 E J DarbyshireandJ R West
Height,abovet,hebed (m)
0.00 0.50 1.00 1.50 P. 00 •.50 3.00 3.50 4.00
I I I I
o21.29
o5.18
ol.
82o1•
,'•o0.9<3
o0.95
027.68
ø2.
31/•9000.6400.58
00.63
oo.
00.78
o0.
• 00.42
0.•
00.47
o0• 00.43 00.42 o 0.•
Figure5: Gradient
Richardson
Nmnbers,
8/7/88.
Therecordsof theturbulent
perturbations
of thevelocity
andsalinity
records
weredivided
intorunsof 13.6minutesanddigitized
at 10s
-•. Largespikeswerethenremoved anda four
point
smoothing
applied
togivetmeffective
digitization
rateof2.5s
'•. Inorder
toeliminate
longer
period
non-turbulent
contributions
tothefluxes
therunswerethenfurther
divided
into
sub-sections
of 51.2sandthe lineartrendremovedfromeachrecord. Eachsub-section
was
thenFourier
analysed
using
a standard
F'FFroutine
andthevalues
of theturbulent
intensifies
andturbulentfluxesdetermined
for eachone. High andlow frequency
cut-offerrorswere
E J Darbyshireand J R West 83
ß ß ß
o L! v W .-I-
/ \ z=O.40m
W t 9:15
4/?/88
z=l. 11•m
8/7/88
z=O.40m
HSW at 13:15
i I I
141O0 ' 151O0 161O0 19. O0 61
8/7/88
z= 1. l•'m
HSW at 13:15
i i I i
14.00 15.00 16.00 I• O0 16.00
Time
Figure6: Turbulentintensities
of velocityandsalinity.
84
E J DarbyshireandJ R West
4/?/•
z=O.40m
H$•r at 9:15
4/?/•
z=l.
H$•r at 9:15
•/?/•
z=O.40m
HSW at 13:15
z--1.118rn
HSW at 13:15
__
I i i i
241oo 2•I oo z61oo 27. oo •doo •
Time
Figure7: Shearstress
andturbulent
fluxesof salt.
Computation
oftheturbulent
meansalinity
fromtheinductive
salinometer
data
provided
anadditional
representation
ofthevertical
density
structure.
Although
thisviewisless
detailed
inaspatial
sensethe
temporal
resolution
ismuchbetter
thanforthe
pumped
sample
profiles.
Profiles
computed
fromthese
data
areshown
infig.3 with
quadratic
curves
fitted
tothefour
points.
Also
shownisthetemporal
variation
ofthevertical
gradient
evaluated
E J Darbyshireand J R West 85
from the quadraticsat z=0.56m and z=l.24m. Each inductivesalinometerhead had been
calibratedagainsta large MC5 salinometerfield data set which includedthe data which is
under considerationhere. The correlationcoefficientsfor this linear calibration were 0.9894,
0.9873, 0.9899 & 0.8825 for heads1, 2, 3 & 4 respectivelywherehead 1 is nearestthe bed.
The rootmeansquarevalues(turbulentintensities)
of thevelocity(u',w') andof thesalinity
(s') perturbations
aboutthe lineartrendin eachrecordare shownin fig. 6 at two depthsfor
theearlyebbtideof eachday whensalinitywaspresent.The associated valuesof shearstress
andverticaland longitudinalturbulentsaltfluxesare shownin fig. 7.
5.4 Observations
Significantverticalgradients
about1.0kgm 4 are presentat the commencement of data
collection and theseare observedto increase,particularlynearto the bed, up to about
8.0kgm'4. Fig. 3 showsthe character
of the profilesto changefrom 'concave',
where
ß S/dz2>0to'convex',
where•S/dz2<0,overtheperiod of datacollection
passing
through
a
periodwhenthe profile is nearlinear.
86 E J DarbyshireandJ R West
5.5 Discussion
(5.6)
E J Darbyshireand J R West 87
$W I (5.7)
2dt•z
The differencebetween the flux recordedby the sensorsand the values of the flux are
shownin column5 of table 1 wherea positivevalue is shownif the predictedflux exceeds
that which was observed.Althoughthe numberof positivevaluesis not muchgreaterthan
the numberof negativeonestheir magnitudes are generallygreaterleadingto the conclusion
that the verticalturbulentflux is frequentlybeingunderestimated.There are threepossible
reasons for this observation.
88 E J Darbyshireand J R West
is not observed.
• z= 3. 07111
•_ z= 2. o6m
s• z= 1. 24m
• o •o •o •o •o • •o •o •o •o •
Time (seconds)
Figure 8: Salinityperturbations
at four heights.
E J Darbyshireand J R West 89
0 2•2]c4_2G•D2]c
3+(G•2D2+ •2 •g2
_•2 )•2) •.2_2G•
07g
2]C
+G•
2•g' 0 (5.8)
which canbe solvednumericallyfor k, the wavenumberwhereD is the flow depthand o• is
the observedwave frequency.
Nun
9 •
Time
Da •e 4
• Slope
o!
s•roigh•
lines=-5/$
z=O. 56m o
z= 1. •4m
z=•. 06m +
z=3. OTm
Log•o Fre q u en cy
Figure9: Normalizedsalinityvariancespectra.
On solvingfor k it is foundthattworealrootsexist,onerepresenting
propagation
with the
flow andoneagainst.The resulting phasevelocities
aresuchthattheupstream
propagating
solutioncannotpropagateagainstthe flow and is thereforenot realistic. The waves are
thereforeassumedto be propagatingin the samedirectionas the meanflow.
Theparticlevelocities
associated
withthewavepropagation arecomputedbyassumingthat
theobserved
varianceof thesalinitysignalis dueto a singlefrequency
sinusoidally
varying
wave. Thedetailsof assumptions
andcomputation aregivenin appendix1 andtheresulting
maximumlongitudinal andverticalparticlevelocitiesare givenby
90 E J DarbyshireandJ R West
1¾
o - (SS/&)
-• 2xa$o (5.9)
and
(5.10)
where/5"
= to2(N2- to2)/lc
2. Theresults
of thesecomputations
foroneof thesensor
headsare
shownin table2 where7-o,So,U0 andW0aretheinternalwaveverticaldisplacement, salinity,
longitudinaland vertical velocity amplitudes. Referenceto fig. 8 showsthat while the
sinusoidalassumption is not very realisticfor the z= 1.24mtime seriesit alsoshowsthat the
predictedvalueof So(table2; run 9) is not a largeoverestimate and is sometimes exceeded
on timescalesof a few wave periods.
Table 2: Salinity,DisplacementandVelocityamplitudes,
z=l.24m, 4/7/88.
E J Darbyshireand J R West 91
Vertical
Velocity
Longitudinal Velocity
Salinity
Time (seconds)
[] []
o
o 0
[] A
% [] A 0
0 0 /x /• rn•
0.56m o
1.24m /x
2.06m []
3.07m v
Figure11' Salinityspectrum
decayexponent
vs.verticalsalinitygradient.
92 E J Darbyshireand J R West
Thefactthatperturbations
of frequencies
lessthanaboutls'• aresuppressed
canbe seen
from examinationof the salinityspectra.A regionwherethe spectrumdeclinesmoresteeply
than-5/3 canbeseenextending
fromfrequencies
0.0ls'• to 1.0s
'•. Thereis a suggestion
that
abovethese frequencies the rateof decrease becomeslesssteepalthoughthereare ratherfew
pointsuponwhichto basethisspeculation.The magnitudes of theseslopeshavebeenplotted
againstthe verticalsalinitygradientin fig. 11. There is a clear trendto steeperslopesat
steepergradientsalthoughthereis considerable scatter.The observedvalueswhichare <5/3
are generallyconfinedto periodsof very low salinitywhenthepresence of randomnoisemay
have affectedthe spectra.
0.56m o
1.26m /x
2.06m []
3.07m v
[]
D •x [] •
0 V zx 0 o o
zxO• [] 0
o •
E J DarbyshireandJ R West 93
5.6 Conclusions
94 E J DarbyshireandJ R West
6. It is conceivable
thatthesecondary circulationcausedby a transverse
densitygradientmay
contributeto the vertical fluxes. In somecasesonly a weak circulationis neededfor a
significantcontribution.Sucha motionis not physicallyunrealisticdespitethe presenceof
steepdensitygradientsandthe absence of surfacedebrisaccumulatedduringthefloodtide is
circumstantial evidence for its existence.
Notation.
d Layer thickness.
D Flow depth.
g Accelerationdue to gravity.
k Wave number.
1 Mixing length.
N Buoyancyor Brunt-Vaisalafrequency.
P Pressureor Spectrumdecayexponent.
Ri Local gradientRichardsonNumber.
s Perturbationof salinityfrom the turbulentmeanvalue.
s' Turbulentintensityof salinity.
s Salinity.
s Turbulentmean salinity.
So Internalwave Salinityamplitude.
t Time.
U,V,W Perturbations
of longitudinal,transverse
and
vertical velocities from the turbulent mean values.
E J Darbyshire
andJ R West 95
Acknowledgements
The authorsgratefullyacknowledgethe financialsupportof the Scienceand Engineering
Research Council and the Natural Environment Research Council and the assistance of the
membersof the Tidal WatersResearchGroup,Schoolof Civil Engineering,in the collection
of data.
5.7 References
96 E J Darbyshire
andJ R West
au aw
• + = 0 (5.A1)
ax az
ap+waPO
.0 (5.A2)
at az
E J DarbyshireandJ R West 97
au ap (5.A3)
øø3•' ax
aw.
at - Op
Po Oz_ pg (5.A4)
Here, P0 is the fluid densityat the equilibriumposition. Eliminatingp and p from these
equationsgivesthe followingequationfor the verticalvelocity,w
ilw N2
i5[iS2w
Oaw
ax: N
at• az• s'2• )+ •v
+
ax2
- 0 (5.A5)
•vhereN2= (-g/p0)0Po/bZ)
radss'•. If it is assumed thattheflowcanbe approximatedto a
layerof constant
buoyancyfrequency, i.e./}Na/bz-•,corresponding
toa density
profilewhich
is nearlinearover the waterdepth,thensolutionsexistof the form,
w- F,•z)e
•(•'•-'00 (5.A6)
W- W.ß-satt2$
sin(•z) (5.A8)
98 E J DarbyshireandJ R West
a•z - (5.A10)
•2Dak4
- 2•o•DZk
3+(½ozDz+•ff•a-D2Hz)Icz-2•o•2k+½o:•
'0 (5.All)
To calculateparticlevelocities'itis necessary
to returnto equation5.A8 andconsiderthevalue
of W, whichfrom now on will be referredto asWo, the verticalvelocityamplitude.For the
purposes
offurther
analysis
it isassumed
thate'•:saz
= 1 since
bothN andz aresmall.If
changesin time of the meansalinityare sufficientlysmallto be ignoredthen,
= (5.A12)
At At 0z
Thisallowscomputation
of thewavedisplacement
amplitudein theverticaldirection,Zo= So
(•}SfOz)
'•. If a singlefrequency
is considered,
thenequation
5.A12canbe written
dS
• = w• (5.A13)
dt
(5.A14)
giving,
(5.A15)
thusallowingevaluationof w.
E J Darbyshireand J R West 99
(5.A16)
W• Wo•l(kz-•øt) (5.A17)
(5.A18)
u- -• wo-•tm(/5•)
giving,
(5.A19)
•rø'• tan(O•)
whereUo is the amplitudeof the longitudinalvelocity.
Abstract
6.1 Introduction
6.2 Prescriptivemodel
In two previouspapers(Simpsonet al. 1990,Simpson,Sharples
andRippeth(1991))we have
identifiedtheprincipalprocesses
operating topromoteanddestroystratification
andattempted
to representthese processesin simplemodelsof the time developmentof stratification.
Closelysimilarapproaches havebeenusedby NunesVaz et al. (1989)andvanAken(1986),
Thesestratifyinginfluences
areopposedby stirringdueto the tidalflow andwindstress
with additionaleffectsfromconvective
coolingduringthewintermonths.
In the prescriptive
model(Simpson,SharplesandRippeth),we specifiedtheseprocesses
usinganalyticalsolutions for thetidalflow (Prandle(1982))anddensitydrivenflow (Hansen
andRattray(1965) or Heaps(1972))andusedenergetics conditionsto describethemixingby
tidalandwind-stirring.The modelis forcedby a depthuniformdensitygradientandsurface
slopeswhichareconstrained by a no net flux condition(in the downgradientdirection).The
justificationfor the strongassumptionof a depth-independent densitygradient,which
represents an importantsimplification of theproblem,hasto be foundmainlya posteriorior
in reference to observations for the area concerned.
6.3 A dynamicallyactivemodel
We aimfor thesimplest
profilemodelthatcanadequately
represent
theseinteractions
between
the density-drivenand tidal flows.
wherewe useconventional
notationwith z positiveupwardsandz=Oat theseabedandz=h
at the surface.
The pressure
p is hydrostatic
andrepresented
by
(6.3)
•xx' p ax ox
au (6.4)
• - -Jr,paz
wherethe turbulenteddy viscosityN. is to be determinedfrom the closurescheme.At the
bottomboundarythe stressis setequalto
ap . -u +• (6.8)
at Ox cgz
(6.9)
(6.10)
(6.11)
Rf- • A*Ri - . Ri - 4Rf•R
I I I (6.12)
•. Kz(l_z/h)• (6.13)
whichbehaveslike Kz near the bottomboundary(K =von Karmansk) and falls off rapidly
near the surface.
(6.14)
a,,
a'•- -fu-xay-•$
(n-o a,,'
-pay +az[ •az,
andtheforcingincludestwo components of densitygradientandtidalpressure
gradient.We
retainthecondition
of no-net-fluxnormalto theisopycnals
andagainsetthebottomstress in
termsof thequadraticdraglaw now in the form:
x/•- -kp
UuI ; U- (Ul
2+v12)
m (6.16)
x• - -kp UvI
6.4 Results
6.4.1 Non-rotatingmodel
_iVVVVVVVVVVVVV
1.5
1.0
(b)
0.5
0.0
0 2 4 6 8 10 12 14 16 18 20 22 24 26
1.5
1.0
(c)
0.5
0.0
0 2 4 6 8 10 12 14 16 113 20 22 24 26
Figure1' Comparisonbetween
turbulence
closure
andprescriptive
models forsinglecomponent
turbulent
closuremodel.(a) Depthmeantidalcurrent
(m s-1)versus
time(days).(b) Turbulence
closure
model:
bottom-surface
density
difference
(kgm-3)versus
time.(c)Prescriptive
model:
bottom-surface
density
difference
m $-1. versus
time.
(N=--0.08
mes-l).
1D •x=2.5
x10
-8m-1,
depth
=40m,
surface
wind
speed
=5.2
A twocomponent
versionof themodelhasalsobeendeveloped
usingessentially
thesame
numerical
scheme
andtestedby comparing
a constant
eddyviscosity
formagainst
theresults
of theanalytical
calculation
of Heaps(1972)
for thecaseof density
gradient
drivingonly.
(a) (b)
; I I I I
' 50 , :50
, 25
1
5
0 0
When run with theserealisticinputs,the model reproduces mostof the featuresof the
observations.The 8 dayperiodof enduringstratification
evidentin themodelresults(fig 3c)
closelyparallelstheobservations
(fig 3b) with goodagreementin thetiming andintensityof
11o J H Simpson
andJ Sharples
themaximumstratification.
Thevariation
inthemagnitude
ofthe semi-diurnal
oscillations
overthetwoweekperiodisalsoreasonably
well-represented
inthemodeloutput.
(a) 0
-1
(b) 1
0 1 2 5 4 5 6 7 8 9 10 11 12 15 14
2
(c) 1
0 1 2 5 4 5 6 7 8 9 10 11 12 13 14
1%
xlO-e lO
5
(d)
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Figure
3:Turbulence
closure
model
comparison
with
Liverpool
Baymooring
data.
(a)Depth
mean
tidal
current
(ms'l)versus
time
(days).
(b)Observed
bottom-surface
density
difference
(kgm'3)versus
time.
(c)Modelledbotto•n-surface
density
difference
versus
time.(d)Observed
variation
ofthedensity
gradient
1P •x(m-')
Table 1: Parameters
usedby the turbulence
closuremodelin figure3(c).
Tidalpressure
gradient
amplitudes
m s'2:
Constituent x-component phase y-component phase
M2 8.469x
10'• 2.415 7.113x10'• 3.749
S2 3.114x10
'• 0.383 2.364x10-• 1.669
N2 1.497x10
's 0.329 1.21lx10-• 1.529
Mean depth= 36.0m.
Depth variationamplitudes:
Constituent amplitude(m) phase
M2 2.52 1.977
S2 0.83 6.127
N2 0.45 6.082
Quadraticfrictioncoefficientk• = 0.003
Mean surfaceslopefactory = 0.55
Wind speedfrom meteorologicalobservations
Numberof depthbins= 10
Time step= 0.005 hr.
6.5 Discussion
It also appearson the basisof the model trials to date that it is importantto allow for
variationsin thehorizontaldensitygradientwhich,in previoussimplifiedapproaches, we had
treatedasconstantin timeas well asdepth. Estimatesfrom shipCTD sectionsandinferences
Acknowledgement
We are gratefulto JohnHamrickof theVirginiaInstituteof MarineSciencewho generously
providedguidanceandadvicein theapplicationof turbulentclosureschemesduringour stay
at VIMS.
6.6 References
van Aken H.M. (1986) The onsetof seasonalstratificationin shelf seasdue to differential
advectionin the presenceof a salinitygradient.ContinentalShelf Research,5 (4), 475-485.
Haas L.W. (1977) The effect of the spring-neaptidal cycle on the verticalsalinitystructure
of the James,York, and Rappahannockrivers. Estuarineand Coastal Marine Science,5,
485-496.
HeapsN.S. (1972) Estimationof densitycurrentsin the Liveqx)olBay areaof the Irish Sea.
GeophysicalJournalof the Royal AstronomicalSociety,30, 415-432.
Circulation
Dynamics and Exchanges in
Coastal and Estuarine Studies Estuaries and the Coastal Zone Vol. 40
Abstract
7..1 Introduction
....
i::::
:;4I•SIDENTIAL
CANAL
DEVELOPMENT
I SECtiON
OF
I•SIDUAL
FLOW
MEASUREMENTS
• SECTION
OFWAIER-
AND
SAI.I
FLUXMEASUI•EMENTS
0
[]•'OME'r•
mESSU•
.....
Figure2: Circulation
ill Estuarywifll MultipleMouths;PlanView
7.2 Physicalcharacteristics
Watersin thestudyareaare very shallowwith typicaldepthsrangingbetween2 m and5 m.
As a resultof dredging,NaplesBay hassomedeeperareasup to 8 m.
The primarywatermovementin the bay is forcedby the tide in the Gulf of Mexico. A
typicalrecordof waterlevelsobserved at a tidestationat thesouthernboundary of DollarBay
is presented
in Fig. 3. For the locationof the tide station,seeFig. 1. The fidesare of the
mixedtype. The averagetidal rangeand the averagespringtidal rangeoff GordonPass
respectively
are0.64 m and0.85 m. (NOAA tidetables).Whengoingtowardstheheadof
NaplesBay,tidalrangesincrease slightlyandreacha maximum in themiddleof NaplesBay,
andafterthatslightlydecrease (Simpson 1979). Thispatternis a resultof a sudden change
in geometryhalfway betweenGordonPassand the head.
•" J/'Vv',VV/V//V/•
!v
-500vV•VV-lV/•l
1989
v' •VV/•/VV
MaJ3 M•6
'!1'M•9 Mg12 •15 M218 M'21
• M'2a
• M• 23 M• 30
•
Apr
Figure3: Observed
Waterlevels
at Tide Stationin DollarBay
Duringflood,waterentersGordonPassandflowsintoNaplesBayandDollarBay. During
ebb thispatternreverses.The tidal maximumof the currentspeedin GordonPassvaries
between 0.6 ms4 and1 ms4 andin NaplesBay,DollarBay,andtheIntra-Coastal Waterway
variesbetween 0.2 ms4 and0.5 ms'l, theexactvaluedepending onthephaseof thetide. For
averagetide conditions, a particleenteringGordonPassat the beginningof the floodcould
travelup to the middleof NaplesBay or whenenteringDollar Bay couldreachthe southern
boundaryof this bay.
Processesresponsiblefor the salt intrusionand, for that matter, the transportof any
dissolvedconstituent,are dispersionand advection. The dispersionis characterizedby a
dispersioncoefficient. Its valuewill be estimatedusingthe observedsalinitydistributionand
knownfreshwaterinflow. Advectionis associated with the residualflow. For NaplesBay
properthe residualflow equalsthe freshwaterdischarge.For GordonPass/DollarBay/ICW,
the residualflow is not a priori known. Its valuewill be estimatedfrom simultaneous
current
and waterlevelmeasurements carriedout over a periodof one monthin cross-section I, see
Fig. 1.
7.3.1 Definitions
in which
Q = rate of flow
A = cross-sectional
area(m2)
V = axialvelocityobservedby a currentmeterin thecross-section
(ms')
t = time
Writing
in which
Ao = cross-sectional
area below mean sealevel
W = width of the cross-section
q = watefievel variation relative to Still Water Level
and substituting
in Eq. (7.1), it follows
<O> <qV>
(7.3)
25
2O
34-
'" 32-
DECEMBER
28,
1976
'" 313-
;>
>'- 28-
::d
@ 12' 26-
z
<•: 24-
>-z
• 22-
z
_o 20-
18-
• 16-
14-
7.3.2 Measurements
Figure6: Measurement
SectionI; Locationof CurrentMetersandTide Gauge
7.3.3 Results
1oo
oo
-10.0
150
-15o
1.0
o.o
-1.0
10.0
-10.0
!
i i i i i i i i i i i
Mat 3 Max 6 Mar 9 Mar 12 Mar 15 Max 18 Mar 21 Mar 24 Mar 27 Mar 30 Apr '
1989
Figure
7. FromToptoBottom,
Eulerian
Me,'u•
Current,
Water
Level,
Stokes
DriftandTransport
Velocity.Positive
direction
is towards
Naples
Bay.
series
fortheEulerian
mean
current
obtained
bylowpass
filtering
V(t)ispresented
inFig.
7. In thesame
figure,
thelowpass
filtered
timeseries
forthewaterlevel
inthecross-section,
theStokes
velocityandthetransport
velocityarepresented.
All fourtimeseries
exhibit
vm'iations
witha period
of5-10days.Timeseriesaretooshort
todelineate
possible
longer
period
variations
inresiduals.
Becausetidalvariations
in waterlevel
andcurrent
speedare
x,- <V>
M
•' (7.4)
where<V> is thetidallyaveraged
basinvolumeandM (kg/sec)is thestrength
of thesource.
(van de Kreeke, 1983).
< V• So-S
(7.5)
O•T s•
in whichQsisthefreshwater
discharge,
T isthetidalperiod,
Soisthesaltconcentration
of the
receiving
watersand<V> (So-S-)/So
represents
thevolumeof freshwaterin theestuary.From
the longitudinalsalinity distributionobservedon December28, 1976, see Fig. 5, and
informationon thebathymetrypresented in Simpson(1979), a valueof thetidallyandspatially
averaged
valueof • = 30.6pptwascalculated.
With<V> = 8.73 10• m3 (= tidallyaveraged
volumeof NaplesBay and GordonPass),T = 44,712 secand So-- 34 ppt it followedthat xr
= 19.3 tidal periodsor approximately10 days.
7.0
-7.0
-10.0
i i ß i i i i i ' ' ßi i i i
Mar 3 Mnr 6 Mar 9 Mar 12 Mar 15 M• 18 Max 21 Max 24 M• 27 Mar 30 Apr '
1989
<O>c - AD d•c
dx
=K (7.6)
in which
Figure9: Link-NodeSchematization.
Arrowindicates
positivex-direction.
in which
Making use of the conservationof water and constituentat each node, the valuesof the
concentrations
at the nodesfollow from Eq. (7.7). Valuesof concentrationsin the Gulf of
Mexico,node1, andin RookeryBay, node7, are setequalto zero.
Concentrating
on a constituentdischargeat node5 it followsthata changein residualflow
from-2.2 m3/s
-1to 6 m3/s
'• haslitfieeffecton concentration
distributions
andresidence
times;
see Table 3. However, the distribution of the constituent fluxes over the Intra-Coastal
WaterwayandGordonPassdiffersconsiderably;
seeTable4. For a residualrate of flow of
-2.2 m3s-1theconstituent
flux in theIntra-CoastalWaterwayis 0.02kg s'l andfor a residual
rateof flowof 6m3s
'l theconstituent fluxis 0.18kg s'l. In bothcases
thefluxesaretowards
RookeryBay. Similar conclusions
holdfor a constituentdischargeat node2. Concentrations
andresidencetimesare notvery sensitiveto theresidualflow in theIntra-Coastal
Waterway;
seeTable 5. However,the constituent flux towardsRookeryBay stronglydependson the
residualflow; seeTable 6. Comparisonof the concentrationdistributionsin Tables3 and 4
showsthat a constituentdischargeat node 2 leads to much smallerconcentrations and
residencetimesthana constituentdischargeat node5.
Table3: Concentrations
andResidence
Times;Sourceof 1 kg s4 at node5.
5 0.944 0.943
4 0.189 0.188
3 0.068 0.067
2 0.031 10.24days 0.030 10.26days
1 0.000 0.000
6 0.024 0.030
7 0.000 0.000
,
Concentrations
arein kg m-3
Table4: Constituent
Fluxes;Sourceof 1 kg s4 at node5.
Table5: Concentrations
andResidence
Times;Sourceof 1 kg s4 at node2.
5 0.002 0.002
4 0.025 0.025
3 0.029 0.029
2 0.030 2.36 days 0.030 2.40 days
1 0.000 0.000
6 0.024 0.030
7 0.000 0.000
concentrations
arein kg m4
Table6: Constituent
Fluxes;Source
of 1 kg s-• at node2.
7.5 Discussion
The success
of measuringresidualflow with a singlecurrentmeterdependson theuniformity
of the cross-sectional distributions of the tidal velocities and Eulerian mean current.
Experimental
dataon thedistribution
of thesevelocitiesin a homogeneous
tidalchannelcould
not be found. Theoretical studieson the vertical distribution of tidal velocities, Eulerian mean
current,Stokesvelocityand transportvelocityfor a sealevelcanalare presentedby Huanget
al. (1986) and Wang (1989). Huanget al. (1986) usinga 2DV numericalmodel calculated
the tidal- and mean currentsfor the Chesapeakeand Delaware canal. Wang (1989) presents
an analyticalsolutionfor the mean currents. The resultsof both studiessuggestthat for a
typicalSouthFloridacoastalchannelwith a lengthon theorderof 10 km anda depthof 2-3
m, the mean currentsdo not reversedirectionin the vertical.
In thepresentstudy,thevelocitymeasurementscarriedoutwiththehandheldcurrentmeters
showa practicallyuniformvelocitydistribution
overthe measurementinterval,i.e., between
the free surface -30 cm and the bottom +30 cm. Velocities of the stations at +35 rn and +45m
differ less than 5%. Velocities of the stations at +20 rn and +60 rn are 10-15% lower than
thoseof the two stationsin the middle part of the cross-section.Values of the mean of the
axialcurrentspeedfor theperiodMarch3-7, 1989are6.5 cms'• for currentmeter#350and
5.8 cms'1for currentmeter#348. Themeansof theaxialcurrentspeedfor theperiodMarch
16-21, 1989 are 3.1 cm s'• for currentmeter#350 and 2.3 cm s'• for currentmeter#348.
Althoughdefinitelynotproof,theforegoingtheoreticalandexperimental resultsprovidesome
supportfor theassumption of a relativeuniformvelocitydistribution
andat leastsuggestthat
the orderof magnitudeof the observedresidualflow and the periodof its fluctuationsare
correct.
7.6 Conclusions
The contribution of the Eulerian mean current to the residual rate of flow is about five
timesthecontribution
of the Stokesvelocity.
7.7 References
Huang, P.S., D.P. Wang, and T.O. Najarian, 1986. Analysis of Residual Currentsusing a Two-
DimensionalModel In: Physicsof ShallowEstuariesandBays,J. van de Kreeke(ed). Lecturenoteson
Coastal and Estuarine Studies No. 16, 71-80.
van de Kreeke,J. 1982. Salt and Water Flux in NaplesBay, Florida. UniversityMiami Rosenstiel
Schoolof Marine and AtmosphericScience. TechnicalReportTR82-2 pp. 28.
Wong, K.C., 1989. Tidally GeneratedResidualCurrentsin a Sea Level Canal or Tidal Strait with
ConstantBreadthand Depth. Journalof GeophysicalResearch,Vol. 94, No. 6, 8179-8192.
Abstract
8.1 Introduction
The Palmiet Estuaryis a small, drownedriver valley on the southcoastof Africa. Its
mouthis a narrowchannelthrougha largewave-builtsandspit(Figure1). The heightof the
sill increasesandthe widthof the contraction decreases throughthe dry summerseason,both
beingscouredopenduringthe winterrains. The strongfiver flow duringwinterstormsalso
scoursdeeperregionsin the bendsof the upperestuary(Figure 1). Althoughthe mouthis
usuallyopen,on averageit closesfor abouta monthat the end of everyothersummer.As a
resultof its small size and owing to its undisturbed,
naturalcharacter,this estuaryhasbeen
the siteof a numberof investigations in ecology(BranchandDay, 1984;Taljaardet al., 1986;
CSIR, 1987) and hydrodynamics (Largier, 1986; Slingerand Largier, 1989; Largier and
Taljaard, 1990). The basinis small (Figure 1) with immediateconnections to the sea and
fiver. The fiver entersthe estuaryas a fast-flowingmountainstream,low in suspended and
dissolved matter.
riverflowsstrongly
(e.g.,80m3s
4 on14August
1986)andthePalmiet
Estuary
iscompletely
flushedof anysalt. As thefiver flow decreases
and/orthetideincreases,
newseawater may
entertheestuary
onthefloodtide(Figure2), asit didon 19August(riverflow30 m3s
'•, tidal
range1.7 m in the estuary). Approaching
the subsequent
neaptide (tidal range0.5 m), the
riverflowincreased
againto 40 m3s
4 on26 August
andall theseawater
wasremoved.Of
interesthereis that periodbetween19 and 26 Augustduringwhichthe estuarywas salt-
stratified.
We discuss threeprocesses:
thetidalintrusion
front(plungingflow);thegravity-
drivenbottomdensitycurrent(saltwedge),anderosionof thehaloclineby ebbtidaloutflow.
/•r3D 2
•0•0
Station No.
0 1 2 3 4 5
I I I I I
Figure2: A longitudinal
sectionof thesalinitydistribution
in thewaterchannel
duringfloodtideon
19 August1986 (seeFigure 1 for stationlocations).
1.0
• 1625
(m)
1610 •1310 ;- 1320
1335
..-r 1445
'•u•.
1355/17•0
(1445)
<•
IL.... (1730)
½ l
-1245
M.S.L.
--'-•1•5)•
<-•
-I.0
uthl
'
1320
-2O
l•O
-3 0
.40L , M• I
0.2m/s 30%o
I i I I I Stla'
31 I [ st:'
4I • I I
Even after the tide hasturnedat the mouth,and after the seawaterintrusionis halted,this
basalunderflowcontinuesto draininto thedeepestsections
of theestuary.The saltcontent
of thewatercolumnat station4 reaches a maximumbetween1600and1700,by whichtime
thesurface
layeris flowingoutat0.3 m s4 (Figure3) - upward
entrainment
intotheturbulent
surfacelayer beginsto removesomeof this salt. Richardsonnumberswere near-unitybut
supercritical,consistentwith the observationthat the haloclinedoesnot breakdownbut is
rathereroded
slowlyfromabove.Thelocalentrainment
velocityat station
4 is 4.2 x 10'5m
s4 on themorningebbtide(between
0930and1330),witha maximum of 8.3 x 10'5 m s'1
obsewedbetween0958 and 1008. In the afternoon(between1700 and 1900) the entrainment
velocity
is 3.5x 10'5m s'1. Following
Turner(1973),theobserved bulkRichardsonnumbers
of about3 correspondto entrainmentvelocities
of about4 x 10'5m s4. Thisentrainment
in
theupperestuary,wherethesurfaceflow is containedandstrong,is muchstronger
thanin the
lowerestuary
- theoverallentrainment
velocity
is 1.5x 10'5m s4. Thesalinity
of thesurface
waterincreases
from 0 to 1.5%oby station4 but only increases
to 2.0%o by the mouth.
1988). Mixing occurredat about 0830 on 6 February,4 hoursbefore slack water in the
mouth. This partially mixed characterremainsuntil new salinewaterof 34.7%0 and 17øC
intrudesbeneaththe 32.5%0 and22øCambientestuarinewaterat about1400 (Largier, 1986) -
this corresponds to a dilutionof seawater(35.5%0) of about25% througha combinationof
surf-zoneand intrusionprocesses.This 2%0 and 5øCstratificationremainsfor the 3 hour
flood tide and is erodedagainon the ebb tide after 1700 at station1. The basalintrusionis
seenat stations2, 3 and4 by about 1430, 1515 and 1630, respectively- corresponding to a
realistic
speedof propagation
of theorderof 0.1 m s'• andnegligible
furtherdilution.A
similarbasaldensitycurrentwas observedon the previousday (5 February).
'•.....
......
'"'"
ß' _-.".:X:• ........
'•5•
''•'"'.':5:'
.i/t•:'-•"'- ....
•-'::z..:'::•'
":':':":':"
4 .;t:' "- .":'• .:':':'-:':':':
_ ":'-"
'-'.."..."........'•...-........,`
'.-0._:-./..
:.'::.'i
½! .ii:.:.½:i:ii:i:'.½:::i:½.:.'
"':'"':'
':".
;-]
ß :"'. .'".'•'.'i'.".': X \
5:!:• )."'
t / ß"
"':'
'-'
':""
'"
..•.:.:....:....?•.......... ..:.....•.:.:........:........:.
•.•. - j ...... .
I I :.•.,..-z...:.-.:, -.•:oz-.:-:.:;:r.':.'..,z...[.
,-..;.,.
,,-.,
o,-.,::,,--•:•-.½:
•:.-.,'.-..:;..
>,.:-.: ...,..
,,.:.,,.;_
0700 0900 1100 1300 1500 1700 1900
Figure5: Laterallyaveraged
watervelocitiesat a channelcross-section
throughstation1 on 12February
1985. Velocities
arein unitsof cm s'• andnegative
valuescorrespond
to landward
flow.
velocities
required
for suspension
trodto scourthesandbedbetweentheplungelineandthe
hydraulicjump. A closedareaof over3 m depthis recorded
aroundstation
0 in summer.
As thewaterlevel rises,thenatureof the tidal intrusionfrontchanges.The incomingseawater
flowsoverthefloodtidalshoalsanda newcurvedplungelinedevelops
alongtheedgeof the
shoalasthedensewaterspillsdownintothedeeper channel
landward
of theshoal(Figure6) -
thisis a typeA tidalintrusion
front(Largier,1990).
Figure6: Thecurved
foam-line
andcolour
change
associated
withtheplunging
flowalongtheedgeof
thefloodtidaldeltaduringlatefloodon 12February1985.Station0 is to therightof thefoamlineand
the boatis at station1. The U- andV-shapedplungelines,whichoccurred on the earlierflood,were
locatedin theleft foreground of thepicture.The photographer
wasstanding at thewesternendof the
sandspit, lookingnorth.
AlthoughFigure5 clearlydemonstrates
thistwo-layervelocitystructure,
it doesnotdisplay
the presence
of a transient
thirdlayerwhichis observed as a smallbasaldensitycurrent
intrusionin summerconditions.It does,however,displaya velocitymaximumin the lower
halocline,as notedfor the winter intrusions.The third layer developsthroughthe tidal
intrusionprocesst3irlylateon thefloodtide. Initiallytheinflowingwaterplunges
belowthe
estuarinewaterand is substantiallydilutedby the strongshear,producinga salinelayer of
about 32.5%0 which fills most of the estuary. It is suggestedthat as the flood tide
strengthens,
sotheplungelineis pushed backintodeeperwaterand,eventually,
theseawater
flowingin alongthebottomis sufficiently
removedfromtheinterfacial
shearthatthebillows
do not dilute it to the sameextent as the rest of the intowing seawater. This appearsat
station1 (Figure7) asa basalintrusion
of 33.5%0and 13.5øC;it maycontinueto draininto
thedeeperchannelat station4. In additionto thetemperature-salinity
character
displayedin
Figure7, thisbasalintrusion
wasobserved to haveenhancednutrientlevelscharacteristic
of
the recentcoastalupwellingin the ocean.
SURFACE
lWARMING
25-
ß ', 1/,February_1985
12h50
river
water
•!• ,•AINMENTS•'.
I
1 2
3 4'
,
20'
TEMP.
(o[)
15'
',
sea •ater
100.... 10
SALINITY { %0 )
Figure 7: Temperature-salinity data collectedat about 1300 (high tide) on 14 February 1985 (data
plottedfor surface,depthincrements of 1 m andbotto•nat eachstation).River waterwas0%o and23øC
whereasseawaterwas35%0 and 13øC.The separation of waterintotwo typesis clear. A is thesurface
water, exhibitingsurfacewinning and increasingsalinityowing to upwardentrainmentof halocline
water.B is the old salinewater at middledepthand D is the old basalwater in the scourholesof the
upperestuary.C is the newly introducedsalinewaterwhich is noticeablycolderandin the processof
intrudingunderthe olderwaterresidentat stations 4 and5. The coldest,•nostsalinepointis the bottom
densitycurrentinitiatedlate on fl•e flood fide.
In comparison,
thefreshwater
fractionof theestuaryhasresidence
timesof lessthana tidal
cycle(asfor winter). Typicalebbtidesurface
velocities
average
0.1 in s4 or morefor6
hours,yieldinga tidal excursionof 2200 m whichis clearlylongerthanthe estuary(Figure
1). There is thus no residentfleshwaterlayer. This outflow of surfacewater appearsto
inducea sympathetic seawardflow in thesalinelowerlayer(Figure5). However,thisdenser
wateris blockedby the sill acrossthemouthandverticalvelocitiesare negligiblesincebulk
Richardsonnumbersare in excessof 10. It is suggested that the interfaceis in the process
of beingset up with an opposingbaroclinicpressuregradientgrowinguntil it matchesthe
imposedstressof the overflowinglayer.
The two-criteriaclassification
diagramsof HansenandRattray(1966) and Jay and Smith
(1988) are more illustrative. Hansen and Rattray (1966) use indices of tidal-average
stratification
AS/•qandcirculation
U,/U!(Figure
8),whereas
JayandSmith
(1988)useinternal
(FB) and barotropic(Fr) Froudenumbers. This latter schemeseeksto classifythe estuary
throughevaluationof barocliniccirculationandbarotropictidalforcing. Palmietvaluesof FB
10•01•4
I I Type4
.D400A4
10
-1
1•l Type
2
10-2
1 10 102 103
Us
In Figure 8, the winter and summerdata are plottedon Hansenand Rattray's scheme.
While mostof the pointslie in a smallregionin the highly-stratified upperleft corner,their
similarityis a bit deceptivein that this is alsoa regionof largegradientsin criterialike the
saltflux diffusion/advection ratio v. It doeshoweverprovidethe essenceof the classification:
the Palmiet is highly stratifiedmost of the time and the advectivesalt flux, due to the
gravitationalcirculation,is at least as importantas the diffusive fluxes (v -- 0.5) - and
generallymore importantin the upperestuary(station4). Even whenthe mouthis closed
(next section),the datafall into thistype4. The one exceptionis the datafrom station2 on
the springtide duringvery weakfiver flow on 6 February1985. In thiscasethe estuaryis
partiallymixed(type2) andit is predictedthatdiffusiondominates the saltflux in the lower
estuary.
top of the halocline. On the fourthday, the 'affectof a recentfreshetis seenin coolerand
moresaline(owingto increased entrainment of haloclinewater)surfacewater. At aboutthis
time the estuarybeginsto overflowthebar acrossits mouthandlargershear,associated with
this throughflowof surfacewater,accountsfor morerapiderosionof the haloclineby days
5 and6. However,themixingcontinues to exhibitsupercriticalRichardson numbersandonly
the top of the haloclineis removed. As the haloclineis erodeddeeperit retainsmuchof its
strength(Figure 9).
26
// ?,._/
24
22
2O
x5
6.• 6 ,Surface
/
+
ß
o
2m
3m
4m
i.6 X 5m
0 10 20
/ ,
Salinity (%o)
What is particularlyinteresting
is the apparentabsence of mixingbelowthis halocline.
Slingerand Largier(1989) calculateddiffusionratesbasedon an assumption of Fickian
diffusion.Bothsalinityandtemperature ratesindicatethatturbulenceandadvectionis absent
at depthandthatdiffusion occursvia molecular action(thermaldiffusivityKt '-' 10'3cm2 s'•,
salinediffusivity
Ks- 10'scm2s'•). Thesurface
layerturbulence,
dueto surface
windstress
and fiver inflow (and intrusion),
cannotpenetrate
the strongpycnocline.The deeper
Acknowledgements
We gratefullyacknowledge the supportof the NationalResearch
Institutefor Oceanology
(CSIR) andthe SouthAfricanCommitteefor Oceanographic Research(FRD, CSIR).
8.6 References
Armi, L., 1986: The hydraulicsof two flowinglayerswith differentdensities.J. Fluid Mech., 163,
27-58.
Largier,J.L., 1991:Estuarine
circulation.Proceedings
of theEstuaryandOceanProductivity
Workshop
- 1990NE PacificChinookandCohoSalmonSymposium(C. Tuss,ed.),HumboldtStateUniversity,
Sept. 1990, (in press).
Simpson,
J. E., 1982:Gravitycurrents
in thelaboratory,
atmosphere
andocean,Ann.Rev.FluidMech.,
14,213-234.
Simpson,J. E., 1987: Gravity Currentsin the Environmentand the Laboratory, Ellis Harwood,
Chichester,244 pp.
Taljaard,S., G. A. Eagle and H. F.-K. O. Hennig, 1986: The PalmietEstuary: A modelfor water
circulationusingsalinityandtemperaturemeasurements overa tidalcycle, Water S. A., 12 (3), 119-126.
Wilkinson, D. L., 1983: Studiesinto the structureand motion of densitycurrents, Water Research
LaboratoryResearchReport 160, Universityof New SouthWales,Australia,36 pp.
Abstract
9.1 Introduction
1Atributary
estuary
is defined
hereinasanestuary
located
withina largemainesmarine
system
where
the largerestuaryservesas the tributaryestuary's"coastalocean"salt source.
WeeksBay is locatedon the easternshoreof the Mobile Bay estuary(Fig. 1) in the northern
Gulf of Mexico. It is nearlydiamondshaped,with a 3.4 km north-south
longitudinalaxisand
a maximumeast-westextentof 3.1 km. A small, 5 to 7 m deepscourfeatureis locatedin
the narrowmouthof the bay and a similarscourfeature,3 to 4 m deep,occursin the Fish
River approximately200 m upstreamof whereit emptiesinto the bay. The bay hasa mean
depthof 1.4m anda surface
areaof 6.95x 10• m2 (Crance,
1971). Tidesareprincipally
diurnalwith a meanrangeof approximately0.4 m. Two riversdischargeinto the bay; the
Fish River at the northernend, accountingfor approximately73% of the total fresh water
input,and the MagnoliaRiver on the easternshore. Mean combineddischargeis estimated
at 9 m3s'1 , withfreshets
up to 4 timeslargeroccurring
throughout
theyear. Shortcurrent
meter recordsfrom the mouth of Weeks Bay are available(Schroederet at, 1990b). These
recordsare generallyfrom a singlemeterand,thus,do notresolvethe barocliniccomponent
of the flow. A singledeployment of two metersdoesindicatethatbaroclinicexchanges are
occasionallyimportam,but the barotropiccomponentgenerallyis dominant. Tidal currents
rangein amplitude
from0.15to0.3 ms'1andsubtidal
(40-hour
low-passed)
currents
from0.1
to 0.3 ms'1.
9 10
!•" .i,'
•, phi_?
I's ana Gulf of /•/exico I I
..
..
..
WEEKS BAY
USGS-WLil ...i::?ii??!
•....... •iiili
..
•:.::.
DEPTH IN METERS
--2
MOBILE
BAY i:•: 0I • 1
!
• Km
Daily averagefiver dischargemeasurements for the Fish River were obtainedfrom the
USGS (USGS-R)(Fig. 1) startingin November1986. For theperiodMay to November1986
Fish River dischargewasestimatedfrom local precipitationdata. Precipitationwassummed
overthe threereportingstationsin theFishRiver drainagebasin,Bay Minette,Fairhope,and
Robertsdale (U.S. Departmentof Commerce,1986,1987,1988). A linear,multipleregression
model,with independent variablesof zero-,one-,two-, andthree-daylaggedprecipitation,was
fit to 480 days(December1986throughMarch 1988)of knowngaugedFishRiverdischarge.
In effect,we considerthecatchment basinto be a linearsystemwith FishRiver discharge as
the output. The impulseresponse of the systemintegratesa weightedversionof the runoff
over a 4-dayperiod. The impulseresponse decaysrapidly,with time. This simplemodel,
withanR valueof 0.815,adequately explainedlow andmoderate dischargevalues,buttended
to underestimate highFishRiver discharge.Never-the-less, it wasusedto estimatetheearly
missingFishRiverdischarge data. Daily averageFishRiverdischarge is depictedin Figure
2c. To extrapolatefrom daily averagedFishRiver dischargeto the totalfreshwaterflow into
WeeksBay,twofactorswereused.ForFishRiverdischarge, <2.83m3 s4, a factorof 2.76
wasused. For higherdischarges,
>2.83m3 s4, a factorof 3.65 wasused(J.L.Pearman,
USGS, personalcommunications).
W W Schroeder,
S P DinnelandW J Wiseman 159
I 6 7A 7B
0,6--
0.3-
-0.3-
-0.6
24-
18-
I 6 7A 7B 12
12
Forfill timecalculations,
WeeksBaywaspartitionedinto16segmentswithonestationper
segment; segment borderswereapproximatelyequidistant
between
adjacentstations.Two
stations
in theFishandMagnoliariverswereexcluded.The surfaceandbottomsalinities
wereaveraged andthisaveragesalinitywasassumedcharacteristic
for eachsegment.The
freshwaterfraction,F, for eachsegmentwas determinedas
WhereSi is the averagesalinityof the i-th segmentandSois the salinityof the Mobile Bay
stationimmediatelyoutsidethemouthof WeeksBay. The freshwaterfractionmultipliedby
the stationdepthandthe segmentsurfaceareagivesthe freshwatervolumefor thatsegment,
andsummingovertheentirebayproduces theexcessfreshwatervolumeof WeeksBay. The
total freshwaterflow was summedbackwards in time, beginningwith the day priorto the
survey,until the excessfreshwatervolumein thebay presentat the time of the survey,was
accountedfor. This lengthof time is the fill time.
Figure3' Longitudinal
salinity(psu)cross-section
fromMobileBaystation,alongthechannel
on the
easternsideof WeeksBay to theFishRiver station,survey1, on 20 May 1986.
9.3 Horizontalandverticalsalinitypattems
In the followingsectionwe will hypothesize
thatspecifichorizontal
salinitypatterns
and
9.3.1 -SubtidalExchange-
As a resultof fallingwaterlevelsthehighsalinitybottomwatersmixed,spreadlaterally
to the westoverthe deeperportionsof the lowerbay andmovedseaward.At the sametime
lowersalinitywaterflowedseawardin thesurfacelayerandcoveredtheshallowbottomareas
adjacentto themouthsof bothrivers(Figs.4a and4b). Duringsurvey1, windswereweak
from the north(Fig. 2a) and fiver discharge wasjust comingunderthe influenceof recent
rainswitha moderate to highincreasein flow (Fig.2c) seenasverylow surfacesalinities
at
the headof the bay (Figs. 3 and4a).
Figure4: Surface(a) andbottom(b) salinity(psu)for Week Bay for survey1, on 20 May 1986.
L,'mdsat
satelliteimage(not shown)whichrevealsa ribbon-likeplumeof turbidwaterexiting
WeeksBay andorientedto the westandnorthwestalongthe shorelineof Mobile Bay.
9.3.3 -Bathymetry-
i •.'n
o.8'
0.9
.3.0 (
Acknowledgements
This work is the resultof researchsponsored
by the U.S. Departmentof Commerce,NOAA,
NOS, Officeof OceanandCoastalResource Management, MarineandEstuarine
Management
9.6 References
Amorocho, J. and J.J. DeVfies, 1980: A new evaluation of the wind stresscoefficient over water
surfaces.J. Geophys.Res., 85:433442.
Schroeder,
W.W., S.P.DinnelandWm.J.Wiseman,Jr., 1989: Stratification
variabilityin a tributary
estuary. Abstracts,Tenth BiennialInternationalEstuarineResearchFederationConference,Baltimore,
Maryland,Oct. 1989. p. 105.
Schroeder, W.W., S.P. Dinnel and Wm.J. Wiseman, Jr., 1990a: Salinity stratificationin a river-
dominatedestuary. Estuaries,13(2):145-154.
Schroeder,W.W., Wm.J. Wiseman, Jr. and S.P. Dinnel, 1990b: Wind and fiver inducedfluctuationsin
a small, shallow,tributaryestuary. ResidualCurrentsand LongtermTransport,R.T. Cheng,Ed.,
Spfinger-Vedag,pp. 481-493.
10
On meteorologicallyinduced subtidal
motions in Hangzhou Bay
J L S u and W Chen
Abstract
10.1 Introduction
HangzhouBay is a largefunnel-shaped
bay alongthe EastChinaSeacoastof China (Fig.l),
whose width is about 100 km at the mouth and narrows down to around 20 km some 100 km
to the west. Severallargecitiesare locatednearthe bay and a nuclearpowerplant is being
built on the north shore of the bay. Consequently,there have been interests in the
understanding of the environmentalqualityof the bay water.
120'E 122'E
32'N
oSS
• vsc
ZAP
•SNS
oo CT
30 'N
Figure1:Locationsof datastations:
tide(circle);meteorology
(solidcircle);wind(solidtriangle).Double
concentriccirclesdenotethe Shanghaicity.
The data stationsare shownin Fig. 1. The two wind stations,YSC and SNS, providethe
remote (over the shelf) wind data and the wind stationZAP gives the local (over the bay)
wind data. Throughout thisstudytheterm"windstress" refersto the quadraticwind velocity
partof the usualwind stressformula. For the wind stresscomponents the east(cross-shelf)
is takenas the positivedirectionof the x-axisand the north(alongshelf)the y-axis. The sea
level stationCT providesthe dataof the coastalsealevel whichalsoforcesoscillationsin the
bay. In thefollowingwe will usethetermscross-shelf andalongshelfwindswhendiscussing
the coastalsealevel oscillations.Methodsandformulasusedin dataanalysisare thestandard
ones(e.g., Mooresand Smith, 1968;Jenkinsand Watts, 1968).
Forsubsequent discussions
we lista fewformulae of multipleinputspectral
analysis.For
simplicity,
considera linearsystemwith oneoutput(u) and two inputs(xI and x2).
Assuming thatXl andx: arecorrelatedit canbe shown(ChenandSu, 1991)thattheoutput
spectrumof this systemcan be written as
andlikewiseforEznandE2•. "S"denotes
eithertheautospectrum
or thecross-spectrum,
"H"
the frequencyresponsefunction and "R" the coherencesquared. The slashsymbolin the
subscriptof eithera spectrumor a coherence squared
indicates
a "partial"quantity.El/2is the
contributionof the uncorrelatedpartof x• to S•uandE•: is the contribution
to Su•dueto the
partof x• thatis correlated
with x:. We takethemodulusof (El/2+ E•2)asa measure of the
contributionof x• to S•. The meanings of E• andE,.•canalsobe explained accordingly.
We notethat the sum(F_q: + E:•) is a real number.
5
• •o•- • 105-
E
4
7. 10.- 7' lo,-
GI:' >.
I-- 3.
•
z
lO ...... I--10-
Z
',," "•;'v
• >'- 10•-_
• •o•
z '• z
lO- ! i
•7 i i
I LU 10 I I I
Figure 2: Sea level spectrainsidethe bay (GP) and at the coast(CT) (solid lines), and sea level
difference
(GP-CT)spectrum (dashed line). Invertedtriangleontheabscissa
indicates
the15 dayperiod.
(The 95% confidenceinterval is •narked)
105.
104-
t
103-
102•.
10,
O. OI m O.Im • m
I 10- m
0.01 I J,
0 I I
FREQUENCY (cpd) FREQUENCY (cpd)
spectrum.
Spectral
peaks
atfrequencies
corresponding
tothepeaks
ofthesealevelspectra
are
alsofoundin oneor bothof thewind-stress
spectra,
including
thefortnight
frequency
(Fig.
2). At thehighfrequency
endthewind-stress
spectra
dropsoff moresharply
thanthesea
levelspectra.
Thisisbecause,
atthehighfrequency
end,thesealevelinside
thebayisalso
excitedby continental
shelfwaves(ChenandSu, 1991).
1.0
C? v$ %,
rD
w 0.5
0.0
0.01 0.1 1
FREQUENCY (cpd)
]-0-
CT vs %,
0.5,'
0.0-
0.01 o'.1
FREQUENCY (cpd)
Figure4: Squared
coherences
between
thecoastal
sealevelat CT andtheshelf(YSC)wind-stress
components.(The95% significance
levelis at thehorizontal
dashed
line)
The squared
coherences betweentheshelf(YSC) windstress componentsandsealevels
at thecoast(CT) andinsidethebay(GP) areshownin Figs.4 and5. Thefiguressuggest that
subtidalmotionsinsideandoutsidethebaywithperiodsup to 70 daysaredriven,directlyor
indirectly,
by meteorologicalforces,exceptfor oscillations
nearthefortnight
period.In the
1.0-
0.0'
0.01l 0.1 1
FREQUENCY (cpd)
rr 0.5'
0
0
I I
O.Ol o.1 1
FREQUENCY (cpd)
Figure 5: Squaredcoherences
betweenthe bay sea level at GP and the shelf (YSC) wind-stress
components.(The 95% significance
level is at the horizontaldashedline)
104
WINTER SUMMER
GI:
103
GP
JS
I0
JS
IO
(GP
HH
IO
CT
(JS)
CT
10-
( HH
I0 I ! I I ,
(CT)0 0. I 0.2 0.3 0.4 0 5 0 0. I 0.2 0.3 0.4 0.5
FREQUENCY (cpd)
Figure 6: Sea level spectrainside the bay (GP, JS, HH) and at the coast(CT) during winter and
summer.
To seebetterthe spectralcharacteristics
at the highfi'equency
endtheseasonal spectraare
computed.Againthesealevel spectrainsideandoutsidethebay havesimilarforms(Fig.6).
The winter sea-levelspectrahavepeaksnearperiodsof 7, 4.5, 3 and 2.5 days,whereasin
summerthe peaksare closeto periodsof 10, 4.5 and 3.3 days.Comparedwith the summer
casethewintersea-levelspectraarenoticeably higherat periodsshorterthan5 days.Thisis
probablydueto theeffectof theshelfwaves(ChenandSu, 1991)whicharemoreenergetic
in winter(ChenandSu,1987).Peaksat periodscloseto positions of sealevelspectrapeaks
are also found in the respectiveseasonalspectraof one or more of the four wind stress
components(Fig. 7). The local windsare more attenuatedthanthe shelfwinds. However,
at the coastthe east-westwindsare moreenergetic,whereasinsidethe bay the north-south
windsare stronger.
-
10
3
\ .._•/
10_ ., i-_ • • i I i I I -
0 0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 0.5
FREQUENCY (cpd)
acceleration,
H the averagewaterdepth,andx the distancemeasured from the bay mouth.
Therefore,as far as the effect of the local east-westwindsis concerned,the modelpredicts
that the amplitudeof the sealevel response at GP is abouttwiceas largeas thatat JS.
Table 1: Percentage
contributions
to the winterbay sea-levelspectra(0.1-0.5 cpd):three-input
model.
Station GP JS HH
Local winds
33 14 9
14 12 10
Coastalsealevel, qcr 78 93 96
Residual 11 3 2
amongthe inputs.
I'01 WINTER
,, ,.-:, :':
I'\, I•, A t,, • '
• '•'11•/ : ..,. • •
,,z,
o.s•- •:'--•- •-,-• •'-/• .... •1-•- • -•--
0.0
FREQUENCY (cpd)
Figure
8:Thewintercoherence
squared
between
'riotandXy(dash-dotted
line),'riotandx, (dashed
line)
andx, andxy(solid
line).(The95%signific,'mce
levelisatthehorizontal
dashed
line)
r'r' 0ß 5.
UJ
o
o
0.0'
FREQUENCY (cpd)
Figure9: Thewintercoherence
squared
between
thesealevelat GP and(a) •lc•-(dashed
line),(b) x,
(dash-dotted
line)and(c)Xy(solid
line).(The95%significea•ce
levelisatthehorizontal
dashed
line)
FREQUENCY (cpd)
Table2: Percentage
contributions
to theseasonal
bay sea-level
(0.1-0.5cpd):modifiedthree-input
system.
, ,
winter summer
Station GP JS HH GP JS HH
Local winds
•x 30 17 12 36 31 25
'cy 10 9 14 6 4 11
Coastalsealevel, qcr 64 85 79 65 75 72
Residual 18 7 7 15 11 8
regressionmethod.
31 WINTER
SUMMER
2
z z
<
1
1 ...
FREQUENCY (cpd)
FREQUENCY (cpd)
/ f I L/ f II I I
,,,,/ ! •
:'U"l f f f I; I
/, f
_u.:] Ill
I l, I I I I
-9oJ !' , ' t ,
I:' ,--'..
_180:!, "/ ,
0.0 0.1 0.2 0.3 0.4 0.5
Figure11' Thegainandphasefunctions
of thesealevelat GPxvithrespect
to qcr(solidline)andxx
(dashedline). (Two-inputmodelresults)
Fromtheabovediscussions we therefore
ignoretheeffectof thecross-bay windstress and
treatthebay oscillation problemas a two-inputsystems. The two inputs,qcr and%, are
correlatedbutnottoo highly.The computed frequencyresponse functions
at GP andJS are
givenin Figs.11 and 12. The gainfunctions at GP showreduced oscillations
thanthe
previousresults obtainedfromthethree-input system(Fig.10).Themeanvaluesof thegains
over0.1 to 0.5 cpd are listedin Table3. The one-dimensional modeldiscussed before
predictsa constant unityvaluefor thegain withrespect to qcr and CpJxl/PgH for thegain
withrespect to 'c,. Thecomputed gainwithrespect toqCThasaverage valuesranging from
1.03to 1.24. If we takeC=l.5x10'3, p•/p= 1.25x10 '3, H=10 m andIxl =100 kin, the
distanceof GP fromthebay mouth,thepredicted valuefor thegainwithrespectto 'c,at GP
is 1.9x10
'3m/(m/s) 2,whichis one-third
to one-halfthevaluesgivenin Table3. Thepredicted
31 WINTER SUMMER
O1 ' 0
0 0.1 0.2 0 3 0.4 0 5 0 0.2i t
0.3 l
0.4 0.5
Figure 12: The gain functionof the sealevel at JS with respectto qcw(solidline) andx: (dashedline).
(Two-input modelresults)
Table 3: Mean valuesof the gainsof the bay oscillationsover 0.1 to 0.5 cpd.
Winter Summer
Station GP JS • GP JS •
Forcing
coastalsealevel, qcr 1.08 1.03 1.24 1.22 1.15 1.21
east-westwind stress,% 6.2 3.4 2.2 3.5 2.9 2.7
(10-3m/(m/s)
•)
If we usea largerdegree-of-freedom
for the spectralcomputations
the oscillations
in the
cross-speclral
resultswill becomesmaller. The conclusionsremain,however,unchanged.
period
coastal
sealeveloscillations
thealongshelf
windstress
component
dominates,
whereas
forlongperiod
oscillations
thecross-shelf
windsu'ess
component
becomesequally
important.
Thetwo-input
systemusedabove does notworkwellherebecause
thetwoinputs,
thecoastal
sealevelandeast-west localwindstress component,
arenowhighlycorrelated.However,as
discussedbeforeit is sufficient
to usethecoastal
sealevelastheonlyinputto studythelong
period
oscillations
inside
thebay.Alternatively,
wecanalsousethetwocomponents
ofthe
shelfwindstressas theinputs,alsodenoted
respectively
as x• andXyin thefollowing
discussions.
In thiswaywecanalsogainsome inforlnation
ontheeffectofeast-west
winds
over the bay.
o l0a_
i I
-2
FREQUENCY (10 cpd)
periodsgreaterthanabout18 dayscontribution
from'cxto the sealevel oscillations
at GP is
actually
moreimportant thanthatfrom•y.Thissuggests
that,liketheshortperiodcasestudied
above,thecontribution
fromthenorth-south
component
of thelocalwindstressissignificantly
lowerthanthatfromtheeast-west
component.
Wenotethedominant
contribution
from
aroundthe15-dayperiod.Thisis anartifactsince,aswe commented onearlier,thepeakof
theGPsea-level spectrumatthe15-day
periodisduemainlytoaliasingfromtheincompletely
removed tides.In Table4 thepercentage
contributions
from'cx,'Cy
andtheresidual tothesea
level spectrum(0.01 to 0.06 cpd) at GP, JS, HH andCT are listed.
ß .
ß
ß
ß
I I I -1
3 6 9 12
FREQUENCY
(10-2cpd)
Figure
14:Contributions
tothebay(GP)sealevelspectrum
(heavy
solidline)fromxx(dash-dotted
line),
xy(lightsolidline)andfl•eresidual(dottedline).
Table4. Percentage
contributions
to thesealevelspectra
(0.01to0.06cpd): two-input
model.
Station GP JS HH CT
'c• 69 69 64 49
'Cy 28 25 33 45
Residual 12 15 13 16
,
2•
(a) (b)
o o
o o 3 6 9 12
FREQUENCY
(10-2cpd
) FREQUENCY
(10-2cpd)
Thegainfunctions
at CT andGPareshownin Fig. 15. Thetwogainswithrespect
to
are quite similar,indicatingthat the contributionfrom the cross-baywinds to the bay
oscillations
is of minorimportance.The gainwith respectto % at GP is higherthanthatat
CT by about10 m/(m/s)
2. Thisdifference
is abouthalf of the valuepredicted
by the
one-dimensionalmodelfor the gain at GP due to east-westwind stressover the bay. The
phasefunctionsat CT and GP were foundto vary closeto 180 or -180 degrees.This is
expectedsincebothnortherlyandeasterlyshelfwindswill lowerthe coastalsealevel andthe
easterlywindsover thebay will lowerthe sealevel at GP.
10.5 Summary
Applyingmultipleinputspectralanalysisto sealevel dataof HangzhouBay it is foundthat
a. Coastal sea level has the dominant influence on the subtidal sea level oscillations inside
thebay. The localwind stresscomponent alongthe axisof the bay is alsoimportantfor
the bay oscillationsat periodsshorterthan 10 days. The local cross-baywind stress
component is of minorimportancefor theseoscillations.
b. Linearinviscidone-dimensional modelcanbe usedto describethe sealevel response of
the bay to forcingby both the coastalsealevel and the alongbay winds.
10.6 References
Bendat,J.S.,andA.G. Piersol,1980:Engineering
Applications
of CorrelationandSpectralAnalysis,John
Wiley & Sons.
Chen, W., and J. L. Su, 1991' Low-frequencysea level fluctuationsin the HangzhouBay, Acta
OceanologicaSinica, 10, (to appear).
11
Abstract
The Louisiana coastal marsheswere built over the last 8,000 years by the successive
formation, abandonmentand subsequentsubsidenceand erosionof delta lobes from the
MississippiRiver. The AtchafalayaRiver Delta is one currentsite of depositionand land
buildingin thisprocess.Sedimentsof the AtchafalayaRiver havebeenfilling Atchafalaya
Bay since1950,and new islandsfirst emergedin 1973. This new land is beingformedin a
coastal
regionwhichis currently
experiencing
landlossat a rateof about155km"peryear.
The surrounding marshesare basicallyfreshdue to the presenceof the fiver but also havea
waterlevel signalthatis stronglyinfluencedby astronomical tides,atmospheric forcingand
the AtchafalayaRiver itself. The resultingwaterlevel regime(inundationand frequency)is
a key elementcontrollingthe marshvegetationdynamicswithin thisdeltaicsystem.
Data for the analysiscame from four recordingwater level stationslocatedwithin the
system.The generalwaterlevelpatternwassimilarfor all stations,showinga strongdiurnal
tidal signalsuperimposeduponotherlargerscaleevents. Theselarger scaleeventsincluded
bothfrontalpassage eventsaswell asthefloodingeventsof theAtchafalayaRiver. The fiver
flow explainsabout 86% of the variationin water levelson an annualtime scaleand about
45% of thevariationon monthlytimescales.On timescalesof severaldaystheatmospheric
forcingbecomes important,with thenorth-south
windcomponent explainingabout62% of the
variation in water levels.
Theseyearto yeardifferences
aredetermined
by theflowof theAtchafalaya
River.
11.1 Introduction
Oneof thekeyelements
in thevegetation
dynamics
of coastalmarshes
is theroleplayed
by thewaterlevelregime. Of particularconcernis thedurationandfrequency of inundation.
It hasbeen shown,that the periodicfloodingof a marshsystemmay serveto enhancethe
productivityof theemergentvascularplantsTeal (1962), HopkinsonandHoffman(1984),to
enhancethe flux of nutrientsbetweenthe marshproperandadjacentwaters,andincreasethe
availablefeedinghabitatfor smallestuarine-dependent fishspeciesBoeschandTurner(1984),
Zimmermanand Minello (1984), Minello and Zimmerman(1985). Thesestudieshavebeen
largelyconfinedto saltandbrackishmarshsystems.Fieldstudiesinvestigating thehydrologic
regimeas it relatesto the vegetation,indicatethat the degreeof inundationmay alsobe an
importantparameteraffectingfreshmarshvegetation.Sasser(1977), Mendelssohn andMcKee
(1987). Recentwork on vegetationwithin the Atchafalayaislandsby Johnsonet al. (1985)
indicatedthat the environmentalfactorswhichvary mostconsistently with the vegetationare
thosewhich dependdirectlyon hydrology. Theft work however,did not includea detailed
In connection with its mission of Flood Control and Channel Maintenance within the
Atchafalaya
Delta Complex,theU.S. Army Corpsof Engineers
(COE) routinelymonitors
waterlevelsat four locationswithinthissystem(Figure1). The stationsof interestincluded
DeerIsland(DI), in theLowerAtchafalaya River, the AmeradaHessPlatform(AH) in the
deltaislandsproper,and EugeneIsland(EI) an offshorestation. The fourthstation,Belle
Island(BI), is locatedin an oil fieldon a saltdome,andwasnotusedin thisstudy.
The recordsof water stageat thesestationswere obtainexlfrom the COE and were
machine digitizedat onehourintervals.TheycoveredthetimeperiodfromJanuary1979
throughDecember 1986. Thedataweredigitizedusinga "Numonics2400"digitizingtablet
interfaced
withan "IBM AT" computer.Duringdigitizing, notesweretakenregarding any
gaps,anomalousdata or time basedrift on the records. These noteswere usedlater for data
editing.
MISSISSIPPI
LOUISIANA
TEXAS
29 ø
N GULF OF MEXICO
94ø 92ø
/ 90o
I i I
•E•2
EVENT 1 EVENT 3
A1 A2
-10,
TIME IN HOURS
MAR 82
ß
8o
t
40-
:2 20
-2o ß i ß i ß i ß i ß i ß i ß I ß i ß i ß I ß i
ß
ß i
.
ß i
.
ß
80
MAY 1982
60
ß 40
:s 20
-20 ' i ß i ß i
ß
ß i ß i ß i ß i ß i
.
ß i ß i ß i
.
ß I ß i ß
80
JULY 83
ß 40
z
• 20
øt i
8
ß i
16
ß I
24
ß i
32
ß i
40
ß i
48
ß i
56
ß i
64
ß i
72
ß
80
i ß i
88
ß
96
i ß
104
i ß
112
TIME (HOURS/3)
In conducting
theanalysis,
it wasnecessm'y
to havethewaterleveldatafromthedelta
islandsproper(Amerada
Hess)asopposed to offshore
(EugeneIsland)or fromAtchafalaya
Figure
4 presents
threehourmeanwaterleveldatafromAmerada
HessforJanuary
thruApril
for 1983and 1984. Figures5 and6 presentexamples of the daily meanwaterlevelsat
Amerada HessandAtchafalaya Riverflowat Simmesport for 1983and1984. TheAmerada
Hesswaterleveldatashownin Figure3 is typicalof thedatafromall years,whichshowa
strongdiurnaltidalsignalsuperimposed uponotherlargerscaleevents.Theselargerscale
eventsincludeboth frontalpassages that influencesealevel and floodingeventsof the
AtchafalayaRiver. Thistypeof signalis quitecommon to waterlevelsmeasured in coastal
LouisianaByrneetal. (1976),Chuang andSwenson (1981),Swenson andTurner(1987).The
fiverflowdatashows a typicalpatternwitha springflood,followed by a fairlyrapiddropin
flow duringthe summer.
•D A
•o• B
600
n00
z00
zøø11
OtJAN83 01NAR83 OtNAY83 O1JUL83 01SœP83 01DœC83
•5•
•.
•oo
75
50
O.
0IJAN84
0•84 0m•¾84 0•J6•84 0•s•P84 01D•C84
20000 50
ß !i ,•I WATER
LEVEL
15000
;I :I •, /! 40
ß /•i , ,;:, :• ..,•
::":
i¾' t. '. 30 •
>,
\. : : •,, , 20 z
10000
", ; ; ":•
x '.. ; ow
10
5000
0
12000 ß 35
RIVER
FLOW • WATER
LEVEL 3O
YEAR
6O
5O ,A .o.,'ø•WATER
LEVEL 500
4O
..'.. A v\ ILl
Z
3O
2O
v: WIND STRESS -500
>-
10
0 10 20 30 4O
01 DEC82 00:00 HRS.
TIME IN DAYS
the river flow. Althoughthere are only 6 data points (1979 - 1984), regressionanalysis
betweenthe two indicatedthat river flow may explainabout86% of the waterlevel variation
at time scalesof a year. At shorterti•ne scales(daily) the fiver flow explainsonly 20% of
the variation. At these shorter time scales, both the diurnal tide and the effect of frontal
passagesbecomeimportant.Figure9 presentsan exampleof waterlevelandnorth-southwind
component for December,1982. Linearregression indicatesthatthewindeventsexplainabout
62% of the water level variation,in this particularinstance.
-- 1979
8000 -' 1980
-- 1981
; 1982
6000
4000
2000
o
-25 25 $0 75
10000
= 1983
8000 ß 1984
ß 1985
= 1986
6000
4000
2000
0
-25 0 25 50 75
ELEVATION(CM NGVD)
Figure10. Yearlyfloodingduration
curvesfor AmeradaHessfor 1979'through
1986. Indicatedarethe
numberof hoursa givenelevationwasfloodedduringthe year.
10000
-10 CM
8000
0CM
10 CM
6OOO
20 CM
4000
30 CM
2000 40CM
50 CM
0 ' I ' I ' I ' I '
1978 1980 1982 1984 1986 1988
2O3
E M Swenson and C E Sasser
u. •00
0
ß lOO
z
0
-25 0 25 50 75 100 125 150
-- 1984 EVENT8
300 ß 1985 EVENTS
200
100
0
-25 0 25 50 75 100 125
ELEVATION CM NGVD
Figure
12: Yearlyflooding
frequency
plots
forAmerada
Hessfor1979through
1986.Indicated
arethe
numberof timesa givenelevationwasfloodedduringtheyes.
Theannual
flooding
duration
curves
for theDeltaIslands
(AH) from1979through
1986
arepresented
inFigure10. These
plotspresent
thecumulative
distribution
of hours
flooded
asa functionof elevation.The elevations
arein centimeters
relativeto theNationalGeodetic
VerticalDatum(NGVD). In general,
it canbe seenthatelevations
lessthanabout-25 cm
NGVD areflooded100%of thetime(therem'e8,760hoursin a commonyear),thecurvethen
11.5 Conclusions
The emergence of the vegetationin the springis controlledby thedurationof the flood
peakof theAtchafalaya River. OncetheRiverstagehasdecreased to a levellow enoughto
allowthemudflatsto havesomesub-aerial exposure, thevegetation emerges.Duringthe
summermonthswhen the vegetationis activelygrowing,the water level patternsare
controlledby the astronomicaltides. Recentresearch SasserandFuller(1988) indicated
grazingstresswithinthe islandsto be an important bioticcontrolof the vegetation.At
elevations wherethe inundation is high,the resulting floodingfrequency and duration
becomes a majorabioticfactorinfluencing the vegetationgrowthanddistribution.In these
instances,thegrazingstress, coupledwith thephysiological (flooding)stresscanleadto a
reductionin thevegetation cover,particularly for theSagittariaspecies.
Table3: Listing
of themajorhydrologic
forces
controlling
thewaterlevelswithintheAtchafalaya
Delta
Islands.Indicated
arethemajorforcing
functions,
byseason,andthehypothesized vegetation
response
to this forcingfunction.
In thefall, southerly
windsassociated
witha majorfrontcanmovemoresalineGulf of
Mexicowaterintothedelta,flooding
theislands.Thus,thevegetation
mustalsocopewith
periodic
saltstress.
Wehavehypothesized thatthissalinity
sixess
mayberesponsible
forthe
Acknowledgements
The authorsthankthe U.S. Army Corpsof Engineers,particularlyMr. RobertCunningham,
for supplyingus with copiesof the raw datarecordsfrom the gagesusedin thisstudy. We
alsoacknowledge the assistanceof Ms JessicaKastlerwho spentcountlesshoursdigitizing
theraw recordswhilekeepingcarefulnoteson theprocess.Her diligencemadetransferring
and editingthe dataa muchsimplertask. We also acknowledge commentsmadeby two
anonymousreviewers.
This study was supportedthrough the National Marine Fisheries Service, Habitat
Conservation
Division,U.S. Departmentof Commerce Additionalsupportcamefrom the
LouisianaBoardof RegentsResearchandDevelopment ProgramandtheLouisianaSeaGrant
CollegeProgram,a part of the NationalSea Grant Collegeprogrammaintainedby the
NationalOceanicand AtmosphericAdministration,
U.S. Departmentof Commerce.
11.6 References
Byrne,P., M. Borengasser,
G. Drew, R. Muller, B. L. Smith,Jr., andC. Wax. 1976. BaratariaBasin:
Hydrologicand ClimatologicProcesses.Centerfor WetlandResources, LouisianaStateUniversity,
BatonRouge,Louisiana70803. SeaGrantPublicationLSU-T-76-012. 175 pp.
SAS Institute,Inc. 1985b. User's Guide: Statistics,Version5 Edition. Cary, NC: SasInstituteInc.,
956 pp.
Sasser,C.E. 1977..Distribution
of vegetationin Louisianacoastalmarshesasresponse
to tidalflooding.
M.S. Thesis. LouisianaStateUniversity,BatonRouge,Louisiana,70803. 40 pp.
Swenson,E. M. and R. E. Turner. 1987. Spoil Ba16:s: Effectson a CoastalMarsh Water Level
Regime. Estuarine,Coastaland Shelf Science. 24:599-609.
12
Modelling of low-frequencysalinity
variations in the St. Lawrence Estuary
K T Tee
Abstract
12.1 Introduction
The St. LawrenceEstuaryis a long and deepestuary(Fig. 1). Its lengthfrom QuebecCity,
wheresaltintrusionends,to Pointe-des-Monts,
wherethecoastlinesuddenlydiverges,is about
400 kin. An abruptrise in the LaurentianChannelnear Tadoussacin the middle estuary
separatestheupperestuary,with typicaldepthsof 20-50 m, andthelowerestuary,with typical
depthsof 200-300 m. The bottomtopography in the upperestuaryis fairly complexwith a
deepchannelin the northernsectionanda shallowchannelin the southernsection.
210 K T Tee
7?_
ø 71ø 70 ø 69 ø 68 ø 67 ø 66 ø 65 ø
i i i i i
50 ø
50 ø ST. LAWRENCE
ESTUARY
'. . .
0 km 50 MANICOUAG•N•.'."2
ß'
ßRIVEjR.:i•:'
BAIE-COMEAU
' I•TE'DE•.
49 ø 49 ø
MONT-JOLI
RliViOUSKI
ß
ß..
ß
.....
48 ø LAURENTIAN 48 ø
CHANNEL
ß '...". '
BAlE-SAINT-PAUL
ß ..
48*
4.• o
ß... .... ,,,,,,,,
ß i ß.:.'.5 " ..':'.:."
.... .. .-; •. ' 44 o
72 ø 71ø 70 ø 69 ø 72 ø 68 ø 64 ø 60 ø 56 ø 52*
In addition to these oscillations, the 1982 observationsalso show an oscillation with the
periodof 40-50 days. A simpleanalysisindicatedthatthe oscillationin the middleestuary
occurredearlier than that in the upperm•dlower estuaries.Becauseof shortcurrentmeter
records,an accuratedescription of the oscillationusingspectralanalysiscannotbe cardedout.
K T Tee 211
The data were collected at 30 minute intervals. The currents were resolved into
along-channel and cross-channel components.The along-channel axis is directedfrom the
head to the mouth of the estuary,and the cross-channel axis from northwestto southeast
shores. To studythe low-frequencyoscillations,the time seriesof salinity,temperatureand
currentcomponents were first low-passedwith a Cartwrightfilter (half power at 35 h) and
subsampled at 6 h intervals. The low-passed datawere thenfitted (leastsquaresmethod)to
10thorderpolynomials.The low-frequencyvariationof the temperature is closelycorrelated
with the salinityvariation,with high (low) temperatureassociated with low (high) salinity.
Becausethe low-frequencyvelocity variationhas strongsignalsof spring-neapvariation
generatedby tidal rectification,and/orseasonalvariationgeneratedprobablyby freshwater
ranoff(Tee, 1989),the40-50 dayssignalin thevelocityvariationcannotbe identifiedclearly.
In thefollowing,onlythe low-frequency oscillationin salinityis discussexlandcomparedwith
the model prediction.
212 K T Tee
a. FORCING
lO
Uw
1035
•-o
1015
r-
•995 •
27
C. MIDDLE
33
•'31
d. UPPER
o• 25
23
t (julian day)
K T Tee 213
50m
40m
30m
20m
T(DAYS)
Figure3: Theverticalstructure
(20, 30, 40, 50 m) of theobserved low-frequency salinityvariationat
stationC in themiddleestuary.The upwardpropagation of the salinityvariationis shown.
214 K T Tee
numericalmodel, the readeris referredto Tee and Lim (1987). Only thoseaspectsthat are
relateddirectlyto this studyare outlinedhere.
12.3.1 Governingequations
In the Cartesianco-ordinates,
the governingequations
of the model,averagedoverthe width
of the channel, are:
1 a(Bu) + aw = 0 (12.2)
B ax az
--.- (12.3)
(12.5)
p(z)
- p,+• gpdz
wherep, is the pressureat surface,and g is gravity.
K T Tee 215
(12.6)
and
where
- _ g(apl&)l(aul&) (12.8)
is the Richardson number, and a and b are constants. These forms of N and K were chosen
by BowdenandHamilton(1975) for theftstudiesof the estuarinecirculation.The constants
a andb arechosen
to be3x10'4m2s
4, thevaluesfoundto givethebestfit between
computed
andobservedsalinityin FebruaryandMay of 1963in the St. LawrenceEstuary(Tee andLim,
1987).
The bottomboundaryconditionsare:
where u, is the normal velocityat the bottom,ubis the horizontalvelocitynear the bottom
takento be the valueat the deepestlevel, and•, is the linearfrictioncoefficienttakento be
0.001ms4. At thesurface(z=0),theboundary
conditions
are:
216 K T Tee
Aprilcanbe affected
by icemelting
in March.Notethattheresponse
of theestuary,
estimated
byusing
theratioof length
scale(500km)tovelocity
scale(0.1ms'l),isabout
two
months.
30-
20-
LU10-
Z
,,,,,
I I I I I
6 26 46 66
T (DAYS)
Figure
4: Theicethickness
in theSt.Lawrence
Estuary
fortheyear1982estimated
fromicecharts
produced
byIceForecasting
Central,
Ottawa,
Canada.
Theestuary
issubdivided
intofivesections
of
equalhorizontal
spacing.
Thesection
number
increases
fromoneneartheheadto fivenearthemouth
oftheestuary.
Solid:
section
1; dotted:
section
2;dash-dotted:
section
3; dashed:
section
4;long
and
short
dashed:
section
5. Thelow-frequency
variation
of theicethickness
usedin themodel
isobtained
byaveraging
thethickness
over
allthesections,
andthen
fitting
(least
squares
method)
theaveraged
data
intoan1l thorderpolynomial.
Theresultisplotted
asa thicksolidcurve.
K T Tee 217
ALONG- CHANNELWIND
lO
1040
C) ATMOSPHEREPRESSURE
1020
lOOO
98O
o •o •o •o •o •o
t (DAYS)
218 K T Tee
12.3.2 Forcingfunctions
The ice thickness andcoveragefor the year 1982 wereestimatedfrom ice chartsprovidedby
Ice Forecasting Central,Ottawa,Canada. The chartwasproducedeverytwo to threedays.
Most areasof the estuaryexceptthosenear the head were ice free after March 12. The
modellingarea is subdividedinto five sectionsof equalhorizontalspacing. The average
thicknessis computedfor each sectionand shownin Fig.4. The sectionnumberincreases
from one near the head to five near the mouth of the estuary. The thicknessat the two
sectionsnearthe headof the estuary(solid-anddotted-curves) are basicallythe sameexcept
for theperiodbetweendays18 and25 andnearday 65 whenthetwo thicknesses are slightly
different. As expected,the ice near the headof the estuarywherethe salinityis low is
generallythickerthan downstream.The exceptionare thosebetweendays47 and 58 when
theice nearthemouthof theestuaryis significantly thicker thanthatinsidetheestuary.The
low-frequencyvariationof the ice thickness(thicksolidcurve)usedfor computingsaltflux
on the surface(F_xt.12.14) is obtainedby averagingthe thicknessover all the sections,and
thenfitting (leastsquaresmethod)the averageddatainto an 1lth orderpolynomial.
The hourly wind data were resolvedinto along-channel (uw) and cross-channel (vw)
components. The resolveddatawerelow-passed with a Cartwrightfilter (half powerat 31
hours)andsubsampled at 6 hourintervals.Figure5 showsthe low-passed (solid-curves)
and
low-frequency (dashedcurves)windvelocityandatmospheric pressure.The low frequency
data are obtainedby fitting (least squaresmethod)the low-passeddata to a 20th order
polynomial.As indicatedby thefigure,the low-frequency dataof u• andv• are inversely
correlatedwith thatof Po; maximain u• andv• correspond to minimain Po,or vice versa.
Withintheperiodof observation (betweendays105and250), thealong-channel component
of thewindvelocitywasmuchlargerthanthecross-channel component. In the2-D modelling
carriedout in this study,only the formercomponent is needed.
The freshwater
runoffinto the St. LawrenceEstuaryis contributed
mainlyfrom the St.
LawrenceRiver (78-80%, Tee and Lim, 1987; EI-Sabh, 1988). The contributionsfrom
Saguenay
Riverin themiddleestuary
andtheManicouagan
RiverSystem
in thelowerestuary
are respectivelyabout 10 to 13%, and 9 to 10%. The runoff from St. LawrenceRiver is
characterizedby seasonal peaksin springandfall, whilethatfromSaguenay hasa seriesof
shortpulses(10-20days)between May andJuly(Tee,1989). TherunofffromManicouagan
River system,becauseit is highlyregulatedfor hydroelectric
power,is moreor lessuniform
throughoutthe year. Figure6 showsthe 1982 freshwater ranoffthroughMontreal(solid
K T Tee 219
16
14
6 -
12.4Comparison
betweenobserved
andcomputed
salinityvariation
In the numerical
model,a steadystateestuarine
circulationwasfirstobtainedby usingthe
observedwind,ice andrunoffon January7, 1982. The computationwasthencontinuedfor
ninemonthsusingthe longperiodforcingdescribed in theprevioussection.
220 K T Tee
I I I i
35
b) MIDDLE
c) LOWER
MODEL '-•
3O
T(DAYS)
Figure
7: Thecomparison
between
computed
andobserved
low-frequency
salinity
variation
attheupper
current
meters
at(a)station
E in theupper
estuary,
(b)stations
C andD in themiddleestuary,
and(c)
stations
A andB inthelowerestuary.
Solid:
observed
salinity
atthesouthernstations
(A,C,E);dashed:
observed
salinity
atthenorthern
stations
(B,D);dash-dotted:
computed
salinity
withalltheforcings;
dotted:
computed
salinitywiththewindforcingonly.
E. At station
E in theupper
estuary,
wecanseefromFig.7athattheobserved
long-period
salinity
variation
canbereproducedapproximately
bythenumerical
model.Notethatbex:ause
of thesimplicities
of themodel(e.g.cross-sectional
averaged)
andthecontamination
of the
observed
40-50dayssalinity
variation
by spring-neap
signals
generated
by strong
tidal
K T Tee 221
30 • , , ,
35 •
I I
I •C,D
30 •I
,
I I I
I
I
25- ',
20-
L
t (DAYS)
222 K T Tee
K T Tee 223
well with the modeltime lag of 2.7 days. The sameresultcanbe foundfor the salinity
variationat stationsA and B in the lower estuary(Fig.9c),wherethe observedtime lag of
3.5 daysbetweenthe salinityat 20 and30 rn compares well with thecomputed timelag of
5.2 days. At stationE in theupperestuary,thesalinityoscillation
at thelowercurrentmeter
(at 19 m) lagsthoseat theuppercurrentmeter(at 14 m) by 0.8 days. This smalltimelag
disagreeswiththemodel'sresultswhichshowsthatthe salinityvariationin thelowercurrent
meterleadsthatof the uppermeterby 1.8 days(Fig.9d).The causeof the disagreement is
unclear. It may be becausethe observed time lags(0.8 days)is too smallto be estimated
accurately.
2O (a) (b) 0 5 10
E•. 20
LU 40
6o
i
T- To (days) T- To (days)
-10 0 lO 0 5 10
, ,
(d)
•
T
F-
2o
(c) •'1o
I
Q_
Q_
LU 40 c3 20
60 3O
224 K T Tee
K T Tee 225
34-
32-
;
; ' / • "-15
;
tt
30-
28-
26-
0 I I
100 I •
2 0 I I
300
T (DAYS)
226 K T Tee
35
30
25
20
15
10-
5-
\ /
\
0 I I \ I .,,• I I I
T (days)
K TTee 227
-4.0
-7.0
-8.0
-8.0
0.0
3O
i i i i I i
T (days)
12.5 Conclusion
BetweenAprilandSeptember1982,fourteen
currentmeters
weremooredat sixstations
along
theSt.LawrenceEstuary:
twostations
eachin thelower,middleandupperestuaries
(Fig.l).
228 K T Tee
Becauseof the shortcurrentmeter records(about 150 days), the time lags betweenthe
stationscannotbe estimatedobjectivelyusinga spectralmethod.To confirmtheexistence and
propagationof thisoscillationin theSt.LawrenceEstuary,a two-dimensional (cross-sectional
averaged)estuarinecirculationmodelwith the observedlow-frequency forcingof wind, ice
and freshwaterrunoff was developed.The observedsalinityoscillationin the upperestuary
is simulatedwell but that in the lower estuaryis underestimatedsignificantlyby the model.
In themiddleestuary,thesummer-time (days190to 250) salinityoscillationis simulated
well,
but the spring-time(days 100 to 190) oscillationis underestimated by the model. This
underestimationin the middle and lower estuariesis probably becausethe runoff from
SaguenayandManicouagan Rivers(Fig.l) are not includedin the model.
12.6 References
K T Tee 229
Mertz, G., M. I. El-Sabh and V. G. Koutitonsky,1989: Low frequencyvariabilityin the lower St.
LawrenceEstuary,J. Mar. Res.,47, 285-302.
Tee, K. T., 1989: Subtidalsalinity and velocity variationsin the St. LawrenceEstuary,J. Geophys.
Res.,94, 8075-8090.
Tee, K.T. andT. H. Lim, 1987: The freshwaterpulse- a numericalmodelwith applicationto the St.
LawrenceEstuary,J. Mar. Res.,45, 871-909.
Weisberg,
R. H. andW. Sturges,
1976:Velocityobservation
in theWestPassage
of Narragansett
Bay:
A partiallymixedestuary,J. Phys.Oceanogr.,6, 345-354.
230 K T Tee
13
Abstract
13.1 lntroduction
The North Sea and the Baltic Sea communicate via a transition-region with complex
topography (fig. 1). It consists
of theSkagerrak,theKattegat,theBelt SeaandtheSoundand
theWesternBaltic(Kiel Bigh0. In theKattegat,thecirculationandtransformation of thewater
massesis cruciallydeterminedby both,local processes (winds,stratification)and by the
influenceof the far-fieldNorth Seaand Baltic,respectively. Thereare weatherconditions,
which,dueto the combinedeffectof windsandair pressure anddueto theshallowness of the
sea,cancausestrongin- andoutflowevents.(Moreextensive descriptionsof thehydrography
and the dynamicsof the transitionregionare givenby Aure and Saetre(1981), Svansson
(1984),LarssonandRodhe(1979),Thorstensson (1978),Knudsen(1899),Dietrichand$chott
(1974).) Theseeventscausedby the meteorological and hydrodynamical far-field have
predominantly synoptic time-scalesof 1-5daysandtheycontributesignificantlyto theextreme
variability,whichis typicalfor theregion.As a resultthecurrentsmayvarybetweenzeroand
(locally) up to five knotsand the salinitycan vary within a rangeof fifteen psu (practical
salinityunits).In the transitionregionan increasingabundanceof oxygendepletionsdue to
anthropogenic eutrophication is observedfor thepast,saytin years.Theseeventshavecaused
severeharm to the local fisheriesand to the marineenvironment.They are causedby local
nutrientinputsvia the atmosphereand via run-off from the mainlandand from numerous
islands in the domain and, on the other hand, from remote sourceslocatid in the southern
North Sea and the eastirn Baltic, respectively.For the contributionfrom the west the
circulationand advectionin the North Sea, in particularalong the Danishwest coast(the
Jutlandcurrent),playsan importantrole for the nutrientbalancein theKattegat.The nutrients
carriedwith the Jutlandcurrententerthe Katiegatat depth,via a deepchannelalong the
Swedishwestcoast,due to the strongdensitycontrastto the muchlesssalineBaltic run-off.
If not disturbedby the atmosphere thisrun-offis a geostrophic
buoyantplumewhichfollows
the Swedishwestcoasttowardsnorth.The extinsionof thiscoastalplumewithinandbeyond
the Skagerrakis called the Norwegiancoastalcurrent.The salinityand also the nutrient
balancewithin the Kattegatis hencestronglydeterminedby advection(North Seainflow and
Balticran-off)andby entrainment betweenthein- andoutflowingwatir masses, respectively.
The lattir is dominatedby local atmospheric disturbances.
233
N Winkel-Steinberg,
J O Backhaus
andT Pohhnann
10'
$9'
N
0R
WlA
Y fiord
58'
Goto River
Fluden
NORTH SEA
57'
Laholm
KATTEGAT B,ght
DENHARK
GRENAA
ZONAL SECTION
•56'
SJAELLAND
GR. BELT
LT. BELT
BALTIC SEA
55'
KIEL BIGt•T
GERMANY
56'
The modelis definedon the Arakawa-Cgrid andit usesa two time-levelscheme.For the
approximationof the advectionof momentum,heat and salt a Lagrangian,"selective
vector-upstream
scheme"was used.This will be describedherein somemore detail,because
it is missingin theabovecitedpapers. The scheme is, dueto theupwindalgorithm,a priori
diffusive.However,comparedto a straightforward,classicalupwindschemeit has the
advantageof a small cross-flowdiffusivity,which is worsewheneverthe flow directionis
alignedwith or closeto thediagonalof thecoordinate system.Theimprovement wasobtained
by a Lagrangianextensionof the straightforwardupwindoperator.The ideaof the selective
vector-upstream schemeis readilyexplainedin a graphicalway by meansof thebelowsketch
(fig. 2). As an example,andalsoto comparethedifferentupwindalgorithms, we consider a
flow fromthesouth-west quadrant,which,in theLagrangian sense,advectsthepropertyP to
the centralgrid-pointC within one time-step(for the sakeof claritythe other,irrelevant
quadrants aroundC havenot beendrawn).The positionof P is uniquelyder'reed by the
advectionvelocity,i.e. the Lagrangianpathduringonetime-step,whichendsat C.
N Winkel-Steinberg,
J O BackhausandT Pohlmann 235
+ w c
+SW + S
0 - 10
,.•
ß
10 - 35
.i:i.., :-:-35 - 50
•- •i•:i:
85 - 150
' I :;.•.
• 250
1so- 250
- SO0
.• •i•soo
•ooo
:•• "/'::'"':'::•:•:!*::•
""":••
..:...!. !
.....
•!i::
•
' '::J:i-. [
I -'•• •
Figure3: Modeldomain(discretized
topography,
depthsin m)
18.0
0 - 20.O
•-•71•18.0
'
- 20.0 20.0 22.0 20.0
.....................................22.0-
- 22
2•.0
0
2•.0 28.0
26.0 28.0
• _-
.-.
-•J•' -'.::.-':!:.::!*•:.::!:i:•!
•;.-';::
28.0 30.0 •:•; 28 0 30 O
• ' 30.032.5
[] 32.5 35.0
.:... ,,-;:,-,:..;ill
. ::
..:.-zT?
-•-'
: .. :
=.. ,.1'.
,.
9 10 11 12 13 ,,,•--_
.• ,•,..
9 1o 11 12
.• • 13
N Wi]ficel-Steinberg,
J O BackhausandT Pohlmann 241
ß, I < 2 I
• % 2 - 5
% 5 - lO
% 10- 15
% 15 - 20
r-'-':5-,,•--) , •-
•, 20- 2s
,o -
% BO 70
N Winkel-Steinberg,
J O Backhaus
andT Pohlmann 243
b) Shorttermoutflowsituationwith strongnorth-easterly
wind (fig. 8 b) After threedayswith
a fairlystrong
north-easterly
windof 10ms'• andthesealevelattheBalticboundary
risenby
0.50 m only the local effect of the wind tendsto broadenthe coastaloutflow plume. As a
resultthe salinityincreasesat the Swedishcoastin conjunctionwith a local upwardmotion
(cf. the 28 psu contourin figure 8 a).
N Winkel-Steinberg,
J O Backhaus
andT Pohlmann 245
10 26
2o 29 >"
25
30
(m)
GRENAA BAY
OFLAHOLM
GRENAA BAY OF LAHOLM
10 26 26 2
15
15
27 20
20
25
25
30
30 •
(m
(m) Z
z
13.7 A tracerexperiment
The import of water massesfrom the North Sea into the Kattegatis demonstrated by a
(passiveand conservative)tracerexperiment,where at the westernboundaryof the model
Figure9: Saltentrainment
ratesat 15 m depth(ms-'x psu)
As known from water mass characteristics the waters of the Jutland current enter the
domainvia thedeepchannelalongtheSwedishwestcoast.Theychangethesignature of the
buoyantsurfaceplumeby entrainment.Thesefeatures
areshownin thefigures10 a-c,which
depicttracerconcentrations
(dissolved
arbitrary
matter)atthreedifferent
depthranges.
Clearly
N Wildeel-Steinberg,
J O Backhausand T Pohhnann 247
I !0 !! 12 13
Figure10:Concentrations
of passive
andconservative
tracers
in percent
a) surfacedistributionafter 106 days
b) distribution
in depthrange:10-15m after 1.06days
Figure10:Concentrations
of passiveandconservative
tracersin percent
c) distributionin depthrange:20-25 m after 106 days
N Winkel-Steinberg,
J O Backhausand T Pohlmann 249
Acknowledgement
The authorswish to thankJ. Svensson
of the SMHI at Gothenburg
for conducting
all model
experimentsdescribedin sectionV.
13.9 References
Backhaus,
J.O., 1985:A three-dimensional
modelfor thesimulation
of shelfseadynamics.
Dt. hydrogr.
Z., 38 H. 4, 165-187.
Backhaus,
J. O., J. Bartsch,D. Quadfasel,
J. Guddal,1985:Atlasof monthlysurfacefieldsof air
pressure,wind stressand wind stresscurl over the North EasternAtlanticOcean:1955-1982,Techn.
Rep. 3-1985,UniversitaetHamburg,Inst.f. Meereskunde.
250 N Winkel-Steinberg,
J O BackhausandT Pohlmann
Mueller-Navarra,S., 1983:Modellergebnisse
zurbaroklinenZirkulationim Kattegat,im Sundundin der
Beltsee,Dt. hydrogr.Z., 36 H. 6.
Stigebrandt,
A., 1983:A modelfor theexchange
of waterandsaltbetweentheBalticandtheSkagerrak,
J. Phys.Oceanogr.,13, 411-427.
Svansson, A., 1972: Canal modelsof the seaand salinityvariationsin the Baltic and adjacentwaters.
Fisheriesboardof Sweden,SeriesHydrographyrep. no. 26.
Thorstensson,
B., 1978:Decademeanvaluesof hydrographical
parameters
fromMarstrand
to Skagen
(theP-section),Medd. Havsfiskelab.,
Goeteborg,Sweden.
Winkel,N., 1989:Salz-Entrainment
im Kattegatuntersommerlichen
Bedingungen
- eineModellstudie,
DiplomaThesis,UniversitaetHamburg,Inst.f. Meereskunde.
Sedimentation
Dynamics and Exchanges in
Coastal and Estuarine Studies Estuaries and the Coastal Zone Vol. 40
14
Observations of fine-sediment
concentrationsand transport in the
turbidity maximum region of an estuary
R J Uncles,J A Stephensand M L Barton
Abstract
Verticalprofilingdatashowpronounced stratification
duringtheearly ebbfollowinghigh
water,with subsequent reductionof verticalmixing.It is notuntilthestratification hasbeen
erodedsignificantly in the faster currentsandwith the approach of the freshwater-saltwater
interfacethat SPM can be mixed into the water column.Vertical mixing is, as an average
overthe watercolumn,moreintenseduringtheflood. The mixingmaximizesastheinterface
passesthroughthe siteon the floodand SPM is mixedinto the near-surface waterswhere
considerableup-estuary transportcanoccur.Thesedata arecomplemented by a longitudinal
transectexperiment. In thestratifiedregiondown-estuary of thefreshwater-saltwaterinterface,
immediatelyfollowinghigh water,a considerable periodof time is availablefor settlingof
SPM. As the ebb currentsincrease,the stratificationerodesand vertical mixing increases,
leadingto high SPM loadsas the interfacepassesdown-estuary.Much of the SPM is
advecteddown-estuary,where it has little time to settleand depositat low water, and
subsequentlyadvectsbackup-estuary in closeassociation
with the interface.Thesefeatures
are also apparentin data obtainedfrom two longitudinallyseparated instrumentpackages
deployedat 0.25m abovethe bed in the turbiditymaximumregion. SPM concentrations,
longitudinal
velocityand salinitydatarecordedat 5 minuteintervalsover a 10 dayperiod
demonstratethegrowthof theturbiditymaximumovertheneapto spring-fide periodandits
dependenceon bothlocalresuspension, and thereforethelocationof thestockof bed-source
sediment,andthepresence of the freshwater-saltwater
interface.
14.1 Introduction
for sometime (Officer 1981 review paper). The followingtidal processes can be important
(Allen et al., 1980; Uncles et al, 1985 a,b,c; Dronkers,1986): (a) asymmetryin the tidal
currentsin which flood currentsexceedebb currentsand high-waterslack periodsexceed
low-waterslackperiods,(b) suspension of erodedbottomsediments,and (c) the existenceof
an up-estuarymaximumin the tidal currentsand thusin the erosionof sediments.In this
paperwe providesomeevidencefor the importanceof tidal processes and demonstratethe
modificationof these processes by intratidalvariationsin water-columnstabilityand the
presenceof the fleshwater-saltwater interface.
14.2.2Suspended
sediments
O-•.We
irHead
5
Tamar
Tavy
lkm
Lynher
3O
Plymout
Sound
Figure1: TheTamarEstuary,showingitssub-division
into51onsegments
and thelocationsof Stations
CK andHQ in the turbiditymaximumregionwhich havebeenusedas deployment sitesfor the
long-terminstrumentpackages.
R J Uncles,
J A Stephens
andM L Barton 259
sediment
andlocalresuspension
anddeposition.
Localtermsarebased
onformulations
given
by OddandOwen (1972). OddandOwen(1972)assumed thattherateof erosion
is
proportional
totheexcess
shear
stress
(x-xc),
wherexcis thecritical
erosion
stress
andx the
bed shearstress.Our modelusesa quadratic
relationship
betweenbed shearstressand
cross-sectionally
averaged
velocity.Based
onlimitedprofiling
dataandmeasurementsof the
shearstrength
of intertidal
mudflats,whichindicate increasingstrengthfromheadto mouth,
we takethecriticalstress
for erosion
to increasefrom0.14N m'"neartheheadto 1.6N m'"
nearthemouth.An erosion rate"constant"of 3x10'• kg m'"s4 is used.
Themodelsimulates
a high-waterturbidity
maximum in theupperreacheswhichisroughly
of thecorrectmagnitude
andin thecorrectposition.A comparison of dataon 11/03and
09/04clearlyshowstheeffectof highm-off for a similartidalrange(5.0 and4.9m);the
maximum
is displaced
down-estuary
at highran-off(60asopposed
to 9 m• s4) andhasa
muchsmallermagnitude. Thisbehaviouris reproduced
by themodel.A comparison of data
on 11/03and27/06clearlyshowstheeffectof tidalrangeon themagnitude of theturbidity
maximumfor a similarfresh-water ran-off(9 and8 m• s4);the maximumis locatedin
roughlythesameplacebutis muchreduced forthesmallertidalrange(3.2asopposed to 5.0
m). This behaviouris also reproduced
by the model.
A featureof thehigh-water
datawhichis notreproduced
by themodelis thenarrowness
of the turbiditymaximumpeak,the tidal modelsmearsthe SPM maximumover a tidal
excursion. It is shown later that this narrownessmay be associatedwith the
freshwater-saltwater
interfaceand the effect of water-columnstability down-estuaryof the
interface.
(A) (B)
SPRING TIDE SPM DISTRIBUTION (1985)
22/02
11,4 5
I f• 11/03 O9/04
60,4 9
r• NEAP
TIDE
SPM
\
DISTRIBUTI
(1985)
30/01 28/02 15/03
\\73,20 8,21 9,19
/-\\
/ \,,,
29/04 13/05 09/09
27/O6
Figure2: Measuredandcomputed(dashedlines)distributions
of SPM at highwaterof (a): springtides
and(b): neaptidesalongtheestuaryduring1985. Eachverticalaxisrepresents
0-250ppmin (a) and
0-50ppm
in (b). Eachhorizontal
axisrepresents
0-30km.Thedate,run-off(m3s-•) andtidalrange(m)
for eachperiodof observation
are shownin thetoprightof eachdiagram.Reproduced
fromUnclesand
Stephens(1989),with permissionof the AmericanGeophysical Union.
14.2.3 Fine-sedimenttransport
The tidal model and observations establishthat the turbiditymaximumwithin the Tamar
migrateslongitudinallyin responseto changingfiver flows. The sameoccursin the Gironde
Estuary(Allen et al., 1977, 1980);the maximummigratesup-estuaryduringlow fresh-water
discharge,andis locateddown-estuary duringhighdischarge.Thereis anaccompanying shift
in thesedimentshoalingpattern,whichconsists of sediment accumulation
in theupperestuary
in summerand autumn,and in the lowerestuaryin winterand spring. A similarseasonal
patternof sedimentmovementhasbeenobservedin theTamarEstuary(Baleet al., 1985)and
hasbeenmodelledusingpurely tidal processes (Uncleset al., 1988).
The sedimentsmodel was run for 12 years of simulatedtime. During the simulation,
environmentalforcingfunctionsof tideandfresh-waterrun-offrepeatedthemselvesthreetimes
with a four-yearperiodicity. This was done to ensurethat quasi-periodic
solutionsfor the
sediment-mass distributionscouldbe achieved. ObservedTamar forcing functionsfor the
period1982-1985wereused.Thedistribution of sediment massin theupper20 km of estuary
was essentiallythe samefor the secondand third four-yearperiods.
(B)
(c)
2 kin, 8 km (- -) (D) 2 kin,18km(- -)
2
o
82 83 84 85
Figure3: Esti]nated
fluvialsedimentinputsandsimulated
massacculnulationonthebedof theestuary
asfunctionsof timeduring1982-1985.(a): Fluvialsedi,nent
illpUtS(107kg month-i);
(b):Cumulative
fluvialsedimentinputs
(10skg);(c):Massof mobilebedsediment perunitlength
of estuary(103kgm-1)
at 2 kin (full line) mid8 kan(dashedline - plottedat twiceactualmagnitude
for clarity)fromhead;(d):
Massof mobilebedsediment
per unitlengthof estuary
(10• kg In-1)at 2 km (full line)and18 km
(dashedline - plottedat twiceactual•nagnitude)
fromhead.Reproduced
frmnUncleset a1.(1988),with
permissionof the EsmarineResearchFederation.
is tidally-pumped
up-estuary
earlyin theyear,whentherun-offis sufficiently
low thattidal
pumpingcandominate.Whenthe run-offis very low duringsummer,sedimentis tidally
pumpedup-estuaryfasterthan it is being suppliedfrom down-estuary, and sedimentis
depleted.Duringthe onsetof highrun-off,sedimentis transporteddown-estuaryfrom the
upperreaches.Someof thissedimentaccumulates at 8 km untiltherun-offachievesvery
highvaluesandthe sedimentis thentransported
furtherdown-estuary.
0
o i i !
7.5 9.5 11.5 131.5 15.5 17.5 19.5 7.5 9.5 11.5 13.5 15.5 17.5 19.5
TIME (hours) TIME (hours)
• 4
0 i ! i
7.• 915 11.5 13.5 15.5 17!5 19.5 7.5 9.5 11.5 13.5 15.5 17.5 19.5
TIME (hours) TIME (hours)
1500 ,
(A):
!
SPM(-), !
U(-*-), S(- -)
! ! !
•ooo-
500 -
0
6 20
60 !
(B):
!
0.1*SPM(-),
!
Kz(-*-)
! ! i
40-
20-
o
6 2o
Figure5: (a):Depth-averaged
SPM (-), speed(U, -*-) andsalinity(S, - -) and(b):Depth-averaged
SPM
(-) andthe estimatedverticaldiffusivity(K,, -*-) at 0.25m abovethe bed as functionsof time for the
contoureddata shownin Figure4.
14.3.2 Longitudinaltransectdata
LW*2 +0 4m/s
!
lO'O
,, ,
7o /
250
HW-1.5 +015m/s
/
6 7 8 9 lO 11 12 13
300/ 2•0•
9 10 11 12 13
mud at the interfacein the strongcurrents. Someof this SPM moveswith the interfacebut
much of it settlesand depositsas high-waterslack approaches.Some SPM remainsand
reachesthe limit of intrusionof the interface,where it forms a slowly erodingturbidity
maximumas particlessetfieand deposit. At the beginningof the ebb in the freshwater the
lower SPM concentrations advect down-estuary.In the stratifiedregionthe SPM ebbsin the
surfacelayers. Depositionoccursin the stagnant,near-bedwater. As the near-bedcurrents
increaseon the ebb and the interfacemoves down-estuary,strong vertical mixing and
resuspension occurin the unstratifiedwaterbehindthe interfaceandthe associated turbidity
maximumadvectsdown-estuary.
14.4.1 Instrumentation
14.4.2 Deployment
Numerical solutions have been derived for turbidity maximum formation due to
density-driven circulation(Festaand Hansen,1978), and this circulationpatternmay be
expectedto contributeto SPM transportin the presenceof fidally-inducedresuspension.
Figures7(c,d) showtheresidual,longitudinal currentat 0.25mabovethebedfor StationsCK
and HQ, respectively. An up-estuary flow occursat HQ duringthe neapfidesand smaller
springtidesbutreverses duringthelargerspringfides(Figure7(d)) whenSPM concentrations
arelargest(Figure7(a)). At CK thenear-bed residualflow is directeddown-estuary
(Figure
7(c)), so that density-drivencirculationsalonecannotbe invokedas a causativemechanism
for turbiditymaximumformationin the upper estuary.
We will focuson the singletidal cycle during9.0-9.5 daysat the end of the record,when
SPM levelsandfluxesarehighest(Figures7(a,b)). Thebehaviour of theturbiditymaximum
is similarduringthepreceding
springtides. FloodandebbSPM peaksalsooccurat neap
tides,butthesehavemuchlowerSPM concentrations dueto thegreatlyreducedinfluenceof
resuspensionprocesses.
R J Uncles,J A Stephens
andM L Barton 269
4OOO
(A): SPM, (CK) 4000
(B): F(-), <F>, (CK)
•3000
2000-
•ooo
0
• 1000
0 -2000
0 5 10 0 5 10
....
-0.2 -0.2
0 5 10 0 5 10
Figure7: Time seriesdataat 0.25m abovethebedfor StationsCK andHQ locatedin Figure1; times
0.0 to 1.0 daysrepresent
November14, 1988. (a): SPM concentrations at StationCK; (b): SPM fluxes
at StationCK (F, flood positive)wifl• a 25 hourrunningmean,<F>, drawn as the bold line; (c,d):
25-houraveragedlongitudinalcurrents(<U>, flood positive)at StationCK (c) andHQ (d).
thenear-bexl,ebb-directedtidalflow immediatelyfollowinghighwater.Therapiddecelemtion
following maximumflood currentsat CK and the long periodof near-zerocurrentson the
mid-flood appearsto be triggeredby resuspension of high SPM concentrations.This
behaviouris currentlynot understood, althoughit is consistently
observedon severalearlier
tidal cyclesand is possiblyrelatedto the dampingof turbulenceandreductionof bed shear
stressin the presenceof high SPM loads,as modelledby ShengandVillaret (1989).
8 (A): HQ(-),CK(--)
! ,
(B): HQ(-), CK(- -)
15 -
10 ' -
0 • '
215 220 225 230 215 220 225 230
(C)' !
HQ(-), CK(- -)
, 5000
(D)' HQ(-), CK(- -)
0.5
0 .....
-0.5
-1 0
215 220 225 230 215 220 225 230
Figure8(d) showsthe SPM concentrations at HQ (-) andCK (- -) asfunctionsof time for the
final tidal cyclecovering9.0-9.5 daysof the record. Flood $PM levelsare smallerthanebb
levelsat HQ, thereversebeingtrueat CK, despitesimilarnear-bed, peakcurrentvelocities
at bothsites (Figure8(c)). Thisimpliesa scarcity
of sourcemudon thebed,down-estuary
of HQ. Eachmainfloodandebb$PM peakcomprises several
sub-peaks,
someof whichmay
be theresultof patchiness
in thebed-sourcemud,whichin turnmayleadto patchiness
in the
resuspendedSPM.
Both sites exhibit low SPM levels around low-water slack (roughly 217 hours) and
extendedperiodsof very low $PM levelsaroundhigh-waterslack(roughly221 hours). The
mainfloodSPM peakat HQ leadsthatat CK andthemainebbpeakat CK leadsthatat HQ.
However,thisis not unequivocalevidencethatadvectionof $PM dominates localprocesses
becauseFigure8(c) showsthatfloodcurrents(andpossiblelocalresuspension) at HQ lead
floodcurrentsat CK, whereasebbcurrentsat CK (andpossiblelocalresuspension)
lead those
at HQ.
The flood SPM levels (times greaterthan 217 hours in Figure 9(a)) show a dramatic
increaseas the criticalshearstressis exceeded.Salinityis zeroat thistime andthe enhanced
mixing on the flood tide associatedwith strongercurrentspeeds(implied from profiling
measurements but only marginallyreflectedin the near-beddatapresented)lead to elevated
$PM levelsat 0.25m abovethe bed. The near-bexl SPM-induceddensitygradientsappearto
'stall' the near-be• current and leads to a decline in SPM levels as SPM either settles or is
replacedby particlesmixed down from flood-directedcurrentshigherin the water column
where $PM levels will be smaller. The increasingsalinityimpliesthat a similar vertical
mixing is occurringfor salt. The passageof the freshwater-saltwaterinterfaceleadsto a
subsequent peakin $PM levelsas $PM associated with the interfaceadvectsthroughCK.
U(- - ), S, (CK)
6000 (A)' SPM(-),
,
3000 (B)' SPM(EBB,CK)
4000 o o
2000-
2000 %%0
lOOO- øo o
0 :_•-
o%•
a-,-•000 o
215 220 225 230 -1 -0.5 0 0.5
o _
2000- •3000- o
•zooo- o•ø
cPc• -
.o%
0 • 1000-
oOgo o
o
14.5 Conclusions
Thesemodelled
tidalprocesses
aremodifiedby buoyancy
effects.Datafromthreetypes
of experiments
in the turbiditymaximumareaof the Tamarhavebeenusedto illustratetidal
processes and their modifications
by buoyancy.Verticalprofilingdatashowpronounced
stratification
duringtheearlyebbfollowinghighwater,with subsequent reductionof vertical
mixing. Therefore,althoughbedshearstresses maybe high,it is not until the stratification
hasbeenerodedsignificantly withtheapproachof thefreshwater-saltwater
interface thatlarge
amounts of SPM canbe mixedintothewatercolumn.Verticalmixingis moreintense,asan
averageover the watercolumn,duringthe flood. The mixingmaximizesas the interface
passesthroughthe site and SPM is mixed into the near-surfacewaterswhere considerable
up-estuary transport can occur.
14.6 References
Dronkers,J., 1986:Tide-inducedresidualtransport
of finesediment,
In: Physicsof shallowestuaries
and
bays, (ed. J. van de Kreeke),Lecturenoteson coastaland estuarinestudies,16, Springer-Verlag,
228-244.
Hamblin,P.F., 1989:Observations
andmodelof sedimenttransport neartheturbiditymaximumof the
Upper SaintLawrenceEstuary,J. GeophysicalResearch,94, No. C10, 14419-14428.
Harris,J.R.W., AJ. Bale, B.L. Bayne,R.F.C. Mantoura,A.W. Morris, L.A. Nelson,PJ. Radford,R.J.
Uncles,S.A. WestonandJ. Widdows,1984:A preliminarymodelof the dispersalandbiologicaleffect
of toxinsin the Tamar Estuary,England,EcologicalModelling,22, 253-284.
Odd, N.V.M. and M.W. Owen, 1972: A two-layermodelof mud transportin the ThamesEstuary,
Proceedings
of the Institutionof Civil Engineers,Paper 7517S, 175-205, 1972.
Officer, C.B. and M.N. Nichols, 1980: Box model applicationto a studyof suspendedsediment
distributions
andfluxesin partiallymixedestuaries.In: EstuarinePerspectives,
AcademicPress,London,
329-340.
Uncles, RJ., R.C.A. Elliott and S.A. Weston, 1985c: Lateral distributionsof water, salt and sediment
transportin a partlymixedestuary,In: Proc.of the 19thInt. Conf. on CoastalEngineering,
ASCE,
Chapter204, 3067-3077.
West, J.R., K.O.K. Oduyemi, A.J.'Bale and A.W. Morris, 1990: The field measurement
of sediment
transportparameters in estuaries,
Estuarine,CoastalandShelf Science,30, 167-183.
Wolanski, E., J. ChappeR,P. Ridd and R. Vertessy, 1988: Fluidizationof mud in estuaries,J.
GeophysicalResearch,93, No. C3, 2351-2361.
15
Velocity asymmetriesin
frictionally-dominated tidal embayments:
longitudinal and lateral variability
C T Friedrichs,D R Lynch and D G Aubrey
Abstract
15.1 Introduction
To investigate spatialvariations
in velocityasymmetry
in a varietyof shallowembayments,
a zero-inertiadiffusivemodelof tidaldynamicsis developedfor well-mixedtidal flow. We
showthatunderconditions appropriate to manysmall,well-mixedtidalembayments, friction
is one-to-twoordersof magnitudelargerthanthe acceleration term in the cross-sectionally
integratedmomentumequation.Linearanalyticsolutions to thezero-inertiaequationareused
Table1' Magnitudes
of local(/.,4) andadvectiveacceleration
(AA)relativeto thefrictionterm(F), along
with relevant characteristic scales.
*Thesetwoexamples
areincluded
to illustrate
limitsof thefrictional
dominance
assumption.
Seetextfor
discussion.
c- g• 1+-- _ +O ,
2 h
15.2.2 Scalaranalysis
I I
ax
+ at
- 0, contiauity (15.2)
282 C T Friedrichs,
D R LynchandD G Aubrey
mommmm (15.3)
(15.4)
(]5.5)
Weassume
O(•SQ)
.• O(Q)andsolve
forfixin(14.4).Thensubstituting
fixinto(15.5)gives:
If weassume O(•SA/A½)
< O(1)(which
isthecaseforallobservedand/or
modeledembayments
considered inthispaper),
thenthesumofthethreeinertia
terms isO(f)Q/•St).
Combiningthe
firstthreetermsof (15.6)anddividing
(15.6)bythescaleof thesurface
gradient
gives:
___F.
•t%e• (]5.8)
where
Uois thecharacteristic
scaleof velocity,
averaged
overthemomentum-transporting
portionof the cross-section,
and R½ = Ac/P½is the hydraulicradiusof the
momentum-transporting
channel.In shallow
tidalembayments
of interest,
Uoisof theorder
0.5ms
'•, c,••- 10':,thesemi-diurnal
period
& = 4.5x 104 sec,and1 m < Re< 10m.
Therefore
F will typicallybe one-to-two
ordersof magnitude
largerthanI in these
[,,) (15.9)
We also solved (15.2) and (15.3) numerically,with and without the inertial terms, to
demonstrate
furtherthe validity of the zero-inertiaapproximation
in shallowregimes. The
modelembaymentconsistsof a sevenkilometerlong, wide rectangularchannel(i.e., Rc =
channel depth,Pc= bc= constant),
witha linearlysloping bottom.We let ca= (0.013mz/a)
R•'z/a,usingtheManningformulationfor bottomfriction. ForRc= 2 m, ca= 0.01. The
channel bottom rises from two meters below mean sea level at the ocean end to one meter
abovesealevel inland, and is forcedby a one meteramplitudeoceantide. Eqs. (15.2) and
(15.3) with the inertial termswere solvedexplicitlywith a 1-d finite differencemodel (At =
15 s, Ax = 250 m, from Speerand Aubrey, 1985). Omittingthe inertial terms,(15.2) and
(15.3) were solved implicitly using Crank-Nicolsonin time (At = 450 s) and 1-d finite
elementsin space(Ax = 250 m). Withoutthe inertialterms,the solutionis free of advective
time-stepping constraints,allowing stablesolutionat much larger At. This may be an
importantconsideration in the solutionof highlycomplexgeometries in two dimensions.
284
C T Friedrichs,
D R LynchandD G Aubrey
0.5
a)
0
-1
0 5 10 15 20 25 30 35 40
0 7
0.35
o
•ß -0.35
I 0.3
:•
5//,•x 0
//
-0.35
-0 7
0 5 10 15 20 25 30 35
1. temporal
acceleration
2. advectiveaccelera•on
3. pressure
gradient
4. bottom friction
time (hours)
15.3.1 Formulation
•Ox +nZRc
-4aIulu- O -toO( (15.10a,b)
Ox
whereu is velocityaveragedoverthemomentum-transporting
portionof thecross-section
(Fig.
1),n = RcU•(c,,,/g)
mistheManning
frictioncoefficient,
andtheothersymbols
areaspreviously
defined. Cross-s•tionally averaged continuity (15.2) may be re-expressed,with
momentum-transport
confinedto the channel,as
b c9•
c3t
+•Ox bchu
I 01 (15.11)
a• 1 a (Da• I 0 (15.12)
at b ax!• -•xx} '
whereD is a non-lineardiffusioncoefficientgivenby
(15.13)
Eq. (15.12) may be solvedto first order analyticallyif we assumethat D is finite and
slowlyvaryingin timeandspaceduringmostof thetidalcycle. Thisanalysisis inappropriate
for tidalflow at slackwaterbecause D is undefinedfor zerosea-surface gradient.However
this linearized analysis still provides valuable insight into the behavior of
frictionally-dominatedtidal propagation
duringthe majorityof the tidal cycle. Underthese
assumptions, (15.12) may be restatedas
a• I [ a2• (15.14)
Ot CgX
2'
with a "constant"
diffusioncoefficient,,c,defined(with Rc-- h) as
bc h• c3C[
-m (15.15)
b n axl
cx•(•Lj,-
•t- 9),-•ttt+
oos(•L•+
•0, •ttr*+cos(x4L•-
•0e-•t¾*
+cx•(•Lj,+
•t- 9),•ttt
2 (eo• 9 + cosh9)
(15.16)
where
t0isangular
tidalfrequency,
L/= (2r,/t0)
mand• = 2L/Le
½(x,t)- CoC•(x/L•-
•Oe-•tty. (15.17)
Weestimate
L/asa function
ofknown
parameters
byfirstevaluating
la•fdxlatx = 0. From
(15.16), it may be shownthat
a½
0x
(x-0)- (sin•eos(•t-
n/4)+sinhq•sin(•t-
n/4)) . (15.18)
L(cos,
C T Friedrichs,
D R LynchandD G Aubrey 287
(z=O) - 2meø
sin(•/4-
eO= (15.19)
0.5
o 1 2 3 4 I 2 3
t/me (a•n) t/me (a•n)
(d) I///=4
.
1
1
0.5
-1
1 2 3 4 0 1 2 3
time(an/n) time
Figure3: Timeseries
of Eq.(15.16)during
twotidalcycles
recorded
atx/L= 0, 0.2,0.4,0.6,0.8and
1.0:(a)L/L/=1/2;(b)L/L/= 1;(c)L/L/=2; (d)L/L/=4.
Wesolve
directly
forL/bysubstituting
thisrelation
into(15.15)
along
withthedefinition
L/
= (2r,/co)a:
(15.20)
According
to(15.20),
propagation
speeddecreases
andamplitude
decay
rateincreases
asdepth
is reduced,
frictionis increased,
or totalembayment
widthis increased
relativeto thewidth
of thechannel.Eq. (15.20)is alsoconsistent
withobservations
thatindicateshallowtidal
embayments actaslow-pass filters(AubreyandSpeer,1985).
(a) SwashBay andChannel,VA, USA L/Lf= 0.38 (b) StonyBrook,NY, USA L/L/= 0.57
.... 1 ß
0.5
-0.5
0 5 10 15 20 25 0 5 10 15 20
time (hours) time (hours)
0.4
0-5
0,' /,
-0.2
-0.4
_
-0.6
0 5 10 1'5 2'0 25 0 5 10 15 20 25
time (hours) ume (hours)
The zeroth-orderdecayand delay of the M2 tide in four tidal embayments (Fig. 4) are
consistent
withbehaviorpredictedby simple,linearsolutionsto thezero-inertiaapproximation
(Fig.3). Table2 summarizes
theparameters
usedin calculating
L/for fourrealsystems.
We
choosebc/bequalto embaymentareaat meanlow waterdividedby embaymentareaat mean
tide level (exceptfor The Fleet, for which Robinsonet al., 1983, providebcand b). The
valuesgivenfor Manning'sn (Table 2) are basedon the n or cavaluesusedby the authors
in 1-d numericalmodelsof the embayments.Robinsonet al. (1983) assumed a typicalshelf
valuefor the dragcoefficient(ca= 0.002), possiblyunderestimating the frictionaleffectsin
the tidal channel.They thenmatchedobservations of tidal elevationby tuningthe valuesof
bc/bin theftnumericalmodel,a modelwhichlimitsmomentumtransportto thecentralwidth,
bc. Eq. (15.20) demonstrates that decreasingbc/b is analogousto increasingn in a
frictionally-dominated
system.SinceRobinson
etal.tunedbc/b,theL/L/valuecalculated
for
The Fleet (Table 2) should still be accurate,even if Robinsonet al. underestimatedn.
Robinsonet al. notedthatlow valuesof bc/bstronglyincreaseamplitudedecayandphaselag,
but theymistakenlyassumedthatthetidein The Fleetpropagates asan Airy-typefrictionless
long-wave.With ca,bc, Q and h as givenby Robinsonet al., F/I = 40 for a semi-diurnal
tide
in The Fleet. By assuming thedominance of accelerationoverfriction,theyderiveda phase
speed
of (ghbJb)
m,whichbecomes
smaller
atlowbdb.Robinson
etal.alsoneglected
friction
in their discussion
of tidal non-linearityand attributedobservedflood-dominance
only to
Table2: Geometricandfrictionallengthscales(andtheparameters
usedfor calculation)for four shallow
tidal embayments.
Location
L h M2o bc/b Manning's
n L/Lf Sources
(km) (m) (m) (m-1/3s)
15.3.3.Non-linearities
predictedby the linearizedequation
Simple
analytic
solutions
suchasthose
displayed
in Fig.3 assume
L/L•,to beconstant
throughout
the tidal cycle. However,non-dimensional
parameters
importantto non-linear
behavior
canstillbeidentified
byconsidering
variations
inthemagnitude
ofLj•during
thetidal
cycle.Sincethetidalwavepropagation speedis proportional
to thefrictionallengthscale(c
= LI c0),thetidalwavewill propagate
mostquicklyduringtheportionof thetidalcyclewhere
Lj•is greatest.
If Lj•is significantly
greater
around
highwaterthanlowwater(L•i•n>>L.
now),
thenthe propagation
speedwill be greaternearhigh water,and the crestof the tide will
propagatemorequicklythanthetrough.Theresultwill be a shorter
flood,a longerebb,and
occurrence
of highest
velocity
currents
during
theflood.Analogously,
if L•i•n<<L.
now,
the
troughwill propagate morequickly,andebb-dominant currents
will result.From(15.20)the
ratio of high- to low-waterfrictionallengthscalesis dependenton two non-dimensional
parameters,
bnign/btow
andhhign/htow.'
290 C T Friedrichs,
D R LynchandD (3 Aubrey
Thisconceptual
extensionof lineartheorymaybetestedby measuring tidallagsrelativeto
boundaryforcingin real systems.Tidal lag, •, is inverselyproportional
to the tidal
propagation
speed
and,therefore,
inversely
proportional
toL•, Thustheratioof tidallagsat
highandlow wateris givenby
(15.22)
Theratioof Oh•Sh
to Olo,•provides
a measure of thedegreeof tidalasymmetry and,by our
conceptual
extension
of lineartheory,a roughmeasureof theratioof frictional
lengthscales
nearhighandlowwater.Table3 compares
ratios
of Oh•s•
toOlow
observed
in fourshallow
embayments
withvalues
predicted
by (15.22).In all fourcases
thegeneral
trend(>1 or <1)
predictedby (15.22) is in agreement
with observations.
predicteA observed
hhighbhigh6}high• 6}high
-- -- 6} AD Sources
Location hlowblow6}low
(hours)
(hours)
6}low
1. SwashBay and Channel, 1.6 4.0 1.49 0.65 0.28 1.55 Byrne et al. (1975);
Wachapreague,VA BoonandByrne(1981)
4. The Fleet,England 2.1 1.0 0.44 4.9 -1.6 0.72 Robinsonet al. (1983)
15.4Numerical
treatment
of thenon-linear
zero-inertia
equation
Linearsolutions
to thezero-inertia
equation
identify
parameters
(L•s•L•o,•,
b•s•b•o,,,,
hnisn/hlo•,)
whichlikelygovern
non-linear
behavior. We nowutilizefullynon-linear
solutions
toconfn'm
theimportance
ofthesenon-dimensional
parametersandtoinvestigate
fivespecific
patterns
of velocity
asymmetryobserved
in thefield:(a)Embayments characterized
bylarge
changes
inmeanchannel depth
during
thetidalcyclehaveflood-dominantmainchannels.
(b)
Embaymentscharacterizedby large changesin total width during the tidal cycle have
ebb-dominantmain channels. (c) Systemswith flood-dominantmain channelstend to have
ebb-dominanttidal inlets. (d) Systemswith ebb-dominantmain channelstend to have
flood-dominantshallowmarginsand intertidalareas. (e) Tidal inlet cross-sections
are often
composedof separatedeep or ebb-dominantand shallowor flood-dominantsections. The
followingsub-section addresses the first two observedpatternsof velocitydistortionand
discusses empiricalrelationshipsbetweentidal prismandchannelgeometry.Sub-sections 2
- 4 individuallydiscusspatterns(c) - (e).
Z In a ]•
15.4.1 Velocity-dominance
of the main tidal channel
104
103
104
102
10• +
+•'•
ß
103 lOO
104
107 108 109 103 104 105 106 107 108 'i'09
spnng
prism
(m3) •ng •sm(m•)
Thames x Delaware
ß InletalongU.S.AtlanticCoast
with oneor nojetties Forth + Unnamedestuary
Savannah ß Wrecked Recorder Creek
Figure5' Cross-sectional
areaas a functionof springtidalprism:(a) Measuredat 37 tidalinletsalong
the Atlantic Coastof the U.S. with one or no jetties (data from Jarrett,1976); (b) Measuredat stations
alongthe lengthof six tidal channels(datafxomChantlet, 1974).
In solving(15.12)we chosemeantideconditions
of • = 1 m, L = 7000m andn = 0.036.
Manning's
n = RcU•(c,t/g)m
= 0.036corresponds
to caTM
0.01forRc= 2 m or ca= 0.006for
R, = 10 m, a roughapproximation of the rangeseenin Table 1. The distribution
of intertidal
storagefor the modelis shownin Fig. 6. We assumed maximumspringdischarge to be
one-and-a-half timesthatof a meantideandthendetermined channelgeometries on thebasis
of the empiricalrelationssummarizedin Table 4. Jarrett'sdata were used to determine
channel cross-sections at the inlet, whereas Chantler's data were used to determine
cross-sections
within the innerportionof the basin.A linearcombinationof the two relations
wasusedin-between.Basinwidthsrangingfrom 100m to 10 km resulted
in 10• m3 < fl <
108m3andthedistribution
ofembaymentcross-sections
shownschematically
inFig.7. Small
tidalprisms
areassociated
withshallow
channels
andnarrowintertidal
areas,
i.e.,hhign[hto
w>>
1. Large
tidalprisms
areassociated
withdeep
channels
andwideintertidal
areas,
i.e.,bh•sh/bto•
>> 1. For the rangeof tidalprismsexamined,
Fig. 8 displays
Lfhis•/L•o,•
(averaged
along-channel)as predictedby lineartheory. This extensionof lineartheoryto highly
non-linearsolutionsis accurateto lowestorderonly. Nonetheless, resultsare generally
C T Friedrichs,
D R LynchandD G Aubrey 293
consistent
withthefullynon-linear
solutions
tobediscussed.
Aslineartheorysuggests,
small
prismswithL•s, > Lj•o•areassociated
withflood-dominant
mainchannels,
andlargeprisms
withL•s, < Lj•o•areassociated
withebb-dominant
mainchannels.
maximumwidth, bmax,adjustable
intertidal
ai'½a
L=7km
constant
localwidthof embayment
= bmaxsin (n:x/L )
channelgeometry,
functionof prism
•'o= 1 m
constant
Figure
6:Schematicdiagram ofmodeltidalembayment. Tidalbasin
areaisspecified
bya sinefunction.
Withtheforcing
tidalamplitudeandembayment length
fixed,thechoice
ofmaximum embaymentwidth
determines
thespringtidalprism,
which,in turn,empirically
determines
thechannel
geometry.
Spring
prism
(m3) Cross-section Geomewi½
parameters
bhigh
= 150m
106_ 107
n,••
....'• T• •' = 3m
hhig
hbhig
h
hlow= blow
bhigh
=3000
m "-I
hhigh bhigh
107-1 htow
<<brow
Figure7: Schematic
diagramof the relationships
(with •o andL fixed) betweenspringtidal prism,
embaymentcross-sectional
geometryandthe non-dimensional
geometricparameterswhichgoverntidal
asymmetry.
(Lf•,i½,
> L3o,,)
in flood-dominant
channels
(Fig.10b).
1.2
0.8
0.6
(b)
bhigh1213
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
•
hhighllOD
th'-'•o•t
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
, .........
====================================================================
.•...,,. • • • 10/9
h• b high]-2131h
Lfhig._._.• high
I
107 108Lftow b'•o•
I •hlow
!
spring
tidal
prism
(m3)
IL -40.95.......
:,xlo, ebb-dominant
__
__•0.9L' '•
.:::::::iiiii::•!::::::i•ii::::::i•i::::i!•::iiii::ii•::i::•ii!!::i•::!::iii::i::•ii?•:i:•:::iiii!i!i:..::•.::•:!iiii!
...........................
9
0 1 2 3 4 5 6
distancealong channel(km)
Figure9: Results
of 30 numerical
solutions
to(15.11).Theratioof cross-sectionally
averagedmaximum
floodto maximum ebbspeedduringa meantidalcycleiscontouredasa function
of localposition
along
thechannelandtotalembayment springprism.If maximum floodspeed/maximum ebbspeed is greater
thanone, the channelis flood-dominant;if the ratio is lessthanone, the channelis ebb-dominant The
dashedlines indicatetwo modelsexaminedin more detail, resultsfrom which were usedto force other
models(seeFigs. 10, 12, 13).
0 0
,• -0.4
0 10 20
L• 0 l0 20
0.3
-0.3 , ,
0 !0 20 o lO •o
time (hours) time (hours)
Figure 10: Time seriesof tidal elevationsat the forcingend and 5500 m up the tidal channel:(a)
ebb-dominant mainchannel;(b) flood-dominant mainchannel(seeFig. 9). Tingeseriesof tidalvelocities
producedby thezero-inertiaequationanda geometrically identicalmodelwhichincludestheacceleration
terms,5500 m up thetidalchannel:(c) ebb-dominant mainchannel;(d) flood-dominant mainchannel.
(15.23
we seethata muchsteepersurfacegradientnearlow waterleadsto highestvelocitiesat the
inlet duringtheebb(assumingtheeffectof theincreasein surfacegradientis largerthanthe
effect of the accompanying
decreasein hydraulicradius).
15.4.3Flood-dominantintertidalareasadjacentto ebb-dominantmainchannels
O.(DO½}-
0½
c9t
. 0' (15.24)
L!,,,H• {••ts
(nto)-m, (15.26)
T-16 • ,
T-14A
C-UNH
T-UNH
South Ehot
K•ttery
T-13
T-12
T. 11
Seavey
T-19
Stratham
km
around
lowwater(L•ish
> L•ow),
andhighwater
propagates
morequickly
thanlowwater.The
fasterpropagation of highwaterfurtherenhances flood-dominance whenthetideat theedge
of theflat is forcedby a shorterdurationflood(Fig. 13a,c). If theboundaryis forcedby a
shorterdurationebb,however,the fasterpropagation of highwaterfirst mustovercome the
pre-existing shorterebbbeforeeventuallyproducing a shorterfloodandflood-dominance
fartheroutacrossthe flat (Fig. 13b,d).The sharppeakin floodvelocitywhichdevelops with
distanceoverthe intertidalflat is consistent with observationsof sedimentresuspension
at the
GreatBayEstuary,NH. Anderson (1973)indicated thatfloodtidesediment resuspensionis
at at leasttwice that of the ebb over flats at the Great Bay.
Gauge
ormeter distance M2 M2phase M4/M2 relative dominance
(km) (morms-l) (dog) phase(dog)
C T Friedrichs,
D R LynchandD G Aubrey 301
theshallower
channeltypically
contained
lowervelocity,
weaklyflood-dominant flow. These
observations
areconsistent
withqualitative
descriptions
of tidalinletflood-andebb-channels
by Hayes(1975;1980). KjerfveandProehlspeculated
aboutthe observed
velocity
distribution,
including
theinfluence
ofchannelcurvature,
theshapeoftheflood-tide
delta,and
thenearby junction
of twotidalcreeks.
Fromtheconsideration
of thezero-inertia
equation,
however,it appears
thatthevelocity
pattern
observed
atNorthInletmaybea directresultof
thedistribution
of cross-sectional
bathymetry.
b-do nant
0 i I i I l
distanceacrossintertidalfiats (m)
Figure
12: Theratioof depth-averaged
maximumfloodtomaximum ebbspeedduring
a mean tidal
cycle,
plotted
asa function
ofdistance
across
anintertidal
flat. Theforcing
elevations
wereproduced
bythetwomodels highlighted
inFig.9,recorded
6500mintothetidalchannels. Thesolid(dashed)
lineis forced
bya timeseries
witha shorter
duration
flood(ebb).
If flowthrough
a channel
cross-section
is approximately
straight
andparallel,
it is
reasonable
to assume
sea-surface
elevation
andgradient
arerelatively
constant
across
that
givenchannel
cross-section
at anyonetime(e.g.,Henderson,
1966).Therefore,
by the
two-dimensional
equivalent
to (15.23)
depth-averaged
velocities
willconsistently
besmallerovertheshallower
portions
of the
channel
cross-section.
Furthermore,
thedifference
invelocity
between
thedeepandshallow
!
(a) 0.2
(c)
........
o 0
-0.2
• 0 1o 20 • 0 !o 20
.....
' 0.15 (d)
o 0
-0.15
-1
0 10 10 0 10 20
time(hours) time(hours)
Figure13: Timeseries
of anundistorted
sinecurve,tidalelevations
usedforboundary
forcing,andtidal
elevationsrecordedat 500 m intervalsacrossthe intertidalflat: (a) forcedby a time serieswith a shorter
duration
ebb;(b) forcedby a timeserieswitha shorterduration
flood(seeFig.9). Timeseriesof tidal
velocitiesrecordedat 500 m intervalsacrossthe intertidalflat: (c) forcedby a time serieswith a shorter
durationebb;(d) forcedby a time serieswith a shorterdurationflood.
depth(m) depth(m)
2.5 4
3 5
6
3.5
3.5
6
3 5
2.5
--1.5 -- 2 3 4
, ! 2 •
0 2• 4• 600 8• I000 20O 4OO 600 8OO lOOO
• ' ß •o•v-:-o'•-•
....:.:.:.:::::::!:i:i:i:i:i:i:!:i::.:.:..i!.i•:i:.::::::i:i:i:i:i:i:•:•:i:•:$•i!ii•i•iiiiiiiii•
20 ..........
:::::::::::::::::::::::::::::::::::
'"'"'"'"' ":':':':':':'""i
........ ...::!'•iii
0 •:i-i-:-:•i-_:-i:i:i:i:i:i::_:•:•:..':?:v:'•!•_._..:
.......................
;.•.;
...............
.•.•.•.•
.........
.?:
......_._
...... •
0 200 400 600 800 ß 1000 2OO 460 6OO 800 1000
distancealong channel(m)
Figure14: Two-dimensional
finiteelementmodelsof tidalchannels
with distinctdeepandshallow
sections.The modelsareforcedby a sinecurveat oneendandelevationsrecordedfrom the two models
highlighted
inFig.9 attheotherend,a distance
of onekilometer
upstream: (a)Thefiniteelement
grid
usedforbothmodels;(b),(c) thebathymetry
of thetwochannels;(d), (e) theratioof depth-averaged
maximumfloodto maximumebbspeedin thetwo channels.
15.5Implicationsfor sedimenttransportandmorphologicevolution
Severalauthorshaveseparated
tidalembayment
morphologies
intotwocategories.
Dronkers
(1986b)discussed two fundamental typesof basins:
Type 1 basinshavedeepchannels (• /
h << 1) andhigh ti_dalflat areas,on the averageabovemeansealevel.Type 2 basinshave
shallowchannels(h at mosta few timeslargerthan•) andlow tidalflat areas,whichat high
waterbecomepartof the streamcrosssection.Dronkers'Type 1 and2 systemsareequivalent
to typicalebb- and flood-dominant geometries discussedby others(e.g., Boon and Byrne,
1981; Speerand Aubrey, 1985; Friedrichsand Aubrey, 1988). Boon and Byrne, (1981)
suggested thatin a systemwithflood-dominantmainchannels andlackingextensivetidalflats,
high velocityfloodscouldfill an initially deepbasinto a leveljust floodedat high water,
increasethe areaof flats,and eventuallyproducean ebb-dominantmainchannel. But results
from SpeerandAubrey(1985) andFriedrichsandAubrey(1988) indicatedflood-dominance
of the mainchannelsdoesnot leadto eventualebb-dominance. Net bed-loadtransportmore
likely will causeshallow,flood-dominantchannelsto shoal,whereasunder ebb-dominant
conditions,deeperchannelsmay be maintained.
This paper utilizes empirical relationsbetween maximum spring tidal prism (or,
interchangeably,
maximumspringtidal discharge) and channelcross-sectional
geometryand
discussesthefeedbackbetweenflow andmorphology.For example,hydraulicengineers have
longrecognized the rapidfilling of tidal inletswhichoftenfollowsenlargementof inletsby
dredgingor reductionof prismby engineering works. Channelmorphodynamics respond on
much shortertime scalesthan changesin overallbasingeometrydue to sedimentation or
changesin relativesealevel. Gardnerand Bohn(1980) presented geologicevidencefrom
Holocenetidalchannels whichsupports thiscausalinterpretation.
Thuspre-existing
geologic
controlmay dictatedominancetype: Smallerinlet/baysystemsthat form with limitedtidal
prismswill tendto haveshallower channels,
whereas largersystems withrelictmorphology
that dictatesa largertidal prismwill have deeperchannels.In basinsforcedby similar
offshoretidalamplitudes,initiallylargeor smalltidalprismswill favorebb-or flood-dominant
mainchannels, respectively.Alonga barrier-beach coast-line
backedby a low-lyingcoastal
plain, the occurrence
of flood- or ebb-dominantembayments speculatively
canbe relatedto
tidalrangealone.Hayes(1979)andFitzGerald(1988)relatedthespacing of tidalinletsalong
barrierbeachcoastlinesinverselyto thetidalrange.Thereforelowertidalamplitude maylead
to widerspacing of inlets(bhigh
>> bto,,,),
largerprisms,
deeperchannels
(h•,ign
• hto,,,),
and
ebb-dominantmainchannels. Highertidalamplitudemayleadtocloser
inletspacing
(b•i•=
bt,,,,),
smaller
prisms, shallowerchannels
(h• >> hto,,,),
andflood-dominant
mainchannels.
Theserelationships
are beingexploredas ongoingresearch.
15.6 Summary
1. Underconditions appropriate to manysmall(L <-20 km),well-mixed(•o / h > 4).1) tidal
embayments, friction is one-to-twoordersof magnitudelarger than accelerationin the
cross-sectionally
integrated momentum equation.Thustidalflow in thesesystems maybe
described
by a zero-inertia diffusivemodel.Thisrepresentationcapturesthemostimportant
non-linearities
which contributeto asymmetricpeak tidal velocities(and, therefore,to net
near-bedsedimenttransport), while greatlysimplifyingconceptual
interpretation.
2. Thelinearanalytic
solution
tothezero-inertia
equation
describes
anexponentially
decaying,
partially
progressive,
partially
standing
wavegoverned
bythefrictional
length
scale
L•, L/
represents
boththepropagation
speed(c = L/•0) andtherateof amplitude
decay(e-folding
length= L). Propagation
speed
decreases
andamplitude decayincreases
asdepthisreduced,
frictionis increased,or total embaymentwidth is increasedrelativeto the width of the
channel.Thezeroth-order decayanddelayof theM2 tidein fourexistingtidalembayments
areconsistent with behaviorpredicted
by thelinearsolutions.
3. Non-dimensional
parameters
important
to non-linear
behavior
areidentified
byconsidering
variations
in the magnitudeof Lj,duringthe tidalcycle. Sincepropagation speedis
proportional
toL/,thetidalwavepropagatesmostquicklyduring
thatportion
of thetidalcycle
havingthegreatestLe An extensionof linearresults
indicates
thattheratioof frictional
lengths
nearhighandlowwateris givenbyL•sdL.
ao,,--(h•,•sdhto,,)•ø'S(b•sdbto,,)
'2a. In four
existing
tidalembayments,
non-linear
behavior
asindicated
by M2andM4tidalcomponents
is consistent
with conceptual
predictions.
4. Non-linear
numerical solutionsto thezero-inertia
equation
recreate
patterns
of velocity
distortion
observed in thefield: (a) Largechangesin channel
depthduringthetidalcycle
(hn•sn/h•o•,
>> 1) areassociatedwithflood-dominantmainchannels.(b)Largechangesin total
widthduring
thetidalcycle(bn•db•o•,
>> 1)result
inebb-dominant
mainchannels.
(c)Systems
withflood-dominant mainchannels tendto haveebb-dominanttidalinlets.(d) Systems
with
ebb-dominant mainchannelstendtohaveflood-dominantshallow marginsandintertidal
areas.
(e) Tidalinletcross-sections
areoftencomposed of separate
deepebb-dominant andshallow
flood-dominant sections.
5. Bothanalytic
andnumerical
solutions
indicate
thatvelocityasymmetry
in thesesystems
reinforcesmorphologyand likely plays a centralrole in its creationand maintenance.
Shallowerareastendtobeflood-dominantandenhance landward
near-bed
transport;
deeper
areastend to be ebb-dominant and enhanceseawardnear-bedtransport.
Changesin
morphology are associated
with reversals
in velocity-dominance.
The velocitypatterns
reinforce
themorphologies:
flood-dominanceenhanceslandward
near-bed
transport,
channel
308 C T Friedrichs,
D R LynchandD G Aubrey
shoaling,
andfurtherreductionsin depth;ebb-dominance
enhances
seaward
near-bed
transport,
channelerosion,and furtherincreasesin depth.
Acknowledgments
Specialthanksis extended
to Dr. Ole S. Madsenwhoprovidedmanyhelpfulcomments onthe
manuscript.Thisworkwassupported by theOfficeof NavalResearch throughtheAmerican
Societyfor EngineeringEducation;by theNationalScienceFoundation, undergrantnumber
CEE-8352226; by theWoodsHoleOceanographic Institution's
CoastalResearch Center;by
theNOAANationalSeaGrantCollege ProgramOffice,Department of Commerce, undergrant
numberNA86-AA-D-SG090,WHOI Sea GrantProjectR/B-82;and by the A.W. Mellon
Foundation.The U.S. Government is authorized to produceand distributereprintsfor
governmental purposesnotwithstanding anycopyright notation
thatmayappearhereon.
15.7 References
Anderson,
F.E., 1973:Observations
of somesedimentary
processes
actingon a tidal flat. Marine
Geology,14, 101-116.
Aubrey,D.G., 1986:Hydrodynamic
controlson sediment
transport
in well-mixed
baysandestuaries.
Physics
of ShallowEstuaries
andBays,J. vandeKreeke,Ed.,Springer-Verlag,
245-258.
Hayes,M.O., 1979:Barrierislandmorphology
asa functionof tidalandwaveregime.BarrierIslands:
From the Gulf of St. Lawrenceto the Gulf of Mexico, S.P. Leatherman,Ed., AcademicPress,1-28.
Hayes,M.O., 1980:Generalmorphology
andsediment
patterns
in tidalinlets.Sedimentary
Geology,
26,
139-156.
Henderson,
F.M., 1966:OpenChannelFlow, Macmillan,522 pp.
Lincoln, J.M., and D.M. FitzGerald, 1988: Tidal distortionsand flood dominance at five small tidal inlets
in southemMaine. Marine Geology,82, 133-148.
Nummedal,
D., andS.M. Humphries,
1978:Hydraulics
anddynamics
of NorthInlet,1975-76.US Army
CoastalEngineeringResearchCenter,G.I.T.I. Report16, 214 pp.
Swift, M.R., and W.S. Brown, 1983: Distributionof bottomstressand tidal energyin a well-mixed
estuary. Estuarine,Coastaland Shelf Science,17, 297-317.
van de Kreeke, J., 1967: Water level fluctuations and flow in tidal inlets. ASCE Journal of the
WaterwaysandHarborsDivision, 43, 97-106.
Westerink,J.J.,K.D. Stolzenbach
andJ.J.Connor,1989:Generalspectral
computationsof thenonlinear
shallowwatertidal interactions
withinthe Bightof Abaco. Journalof PhysicalOceanography,19,
1348-1371.
advectiveaccelerationterm of momentumequation
b totalwidth of cross-section,
includingintertidalflats
width of momentum-transporting portionof tidal channel
widthsaveragedover embaymentlength
bdb embayment
surface
area(a•) at meanlowwaterdivided
by a• at meanfidelevel
c phasespeed
c,• bottomdragcoefficient
D non-linear diffusion coefficient
Mean valueover the ebb for quantity•
•bb
F fiction termof momentumequation
__flooaMean valueover the flood for quantity•
g accelerationof gravity
_H localtotaldepth
h depthat meantide level, averagedover areaof embayment
h depth,averagedacrossmomentum-transporting portionof channelcross-section
__high high tide valueof quantity_
I inertiaterm of momentumequation
__inlet valueof quantity__ at inlet
L systemlength
localaccelerationtermof momentumequation
frictionallengthscale
low tide valueof quantity•
Manning'scoefficient
perimeterof entireembaymentcross-section, includingtidal flats
perimeterof momentum-transporting tidal channel
volume transport
maximumspringdischarge
hydraulicradiusof momentum-transporting tidal channel
time
depth-averaged velocity
characteristic
scaleof depth-averaged
velocity
voiume stored in intertidal areas/volume in channels
horizontal coordinate
constants
determinedby least-squares linearregression
characteristic
variationin quantity•
numericaltime step
surface elevation relative to a vertical datum
amplitudeof sea-surface
forcing
tidal lag
"constant" diffusion coefficient
dependentvariablein least-squares
linearregression
2 L/Lf
angulartidal frequency
16
Abstract
The cross-sectional
shape,width and heightof muddyshoresis relevantto a wide rangeof
engineeringissues,but little is knownof how thesevary in nature. It hasbeenappreciated
for manyyearsthatbothpredominantly accretionary andpredominantly erosionalcoastsoccur
in natureand more recentlyit was shown,for a large numberof mud fiats, that shorecross-
sectionalshapevariessystematically, at leastover a narrowbandof tidal range.
314 R Kirby
16.1 Introduction
Of thosefew investigations
of muddycoastswhichareavailable,the majorityinvolvereal-
time monitoringof short-termchanges,over periodsrangingfrom hoursto a year or more.
Mathematical models which consider one or a few ebb and flood fides are available. In
contrast,limited consideration hasbeengivento longerterm evolution,eitherin the pastor
asit is likely to occurin thefuture. Mathematicalmodelsareespeciallyproblematical
in areas
wherelongtermpredictionis required,whereunder-,normallyor over-consolidated substrates
are presentand where mixed or layered substratesoccur. It is these long term, real
complexitieswhich need to be assessed so that consequences
of changesin waterlevel on
varioustimescalescan be predicted.
The relative elevationof the land and sea can changesuddenly,for exampledue to
engineering workssuchas the construction of largedamsfor tidal powergeneration,or due
to rapid earthmovements.Relativechangescan alsooccuron muchlongertimescales due
to sea level rise or earthmovements.Each of thesewill have major effectson a muddy
coastline,which may take many yearsto adjust,even if the changeitself occursrapidly.
Changes arelikely to affecttheheightandwidthof thecoastalzone,its cross-sectional
shape
and whetherit is predominantlyerosionalor alepositional.
R Kirby 315
Sincetheearly1950s
and1960s
scientific
investigations
ofmuddy
shores
experiencing
long
termaccretionhavebeenavailable(van Straaten,1950,Evans,1965etc.). In suchregimes
a coast-aligned sequence of mud and sandfiats occursacrossthe shore,whichis closeto a
long-termequilibriumwith the hydrodynamicconditionsshapingthe shore. Physical
propertiesof thesediment, suchassediment grainsizeandgeotechnical propertiesareadjusted
to the hydrodynamics of the overlyingwatermass. Similarly,biologicalcharacteristics,
both
floralandfaunal,arein someformof dynamicequilibrium.Saltmarshes riseby deposition
andaregenerallytransitional withthetidalfiat witha seaward
limit closeto MeanHighWater
Neaps(MHWN). In thesameway, hard-shelled, burrowinganimalsremainin positionat the
baseof their burrowson deathand are steadilyentombedby the risingtidal fiat surface.
Finally,tidalfiat sediments containa widerangeof anthropogenic chemicals,reflectingboth
deposition of sediment containing adsorbed contaminantsandincorporation by animals.The
sediments consequently reflect,andarein balancewith,thechemistry of theoverlyingwaters.
316 R Kirby
indicators, suchastheover-consolidated
natureof thedeposits,
mudcracks,saltmarshor tidal
fiat cliffs andotherindicators
of erosion,peatbeds,fossilforestsandby a widerangeof
archaeological evidence,(Kirby,1990,Kirby,in press,andSELRC,1990). In additionto
thesenaturalindicators,measurements againstman-madestructures, especiallyof linear
features crossing
theshoresuchaspipelines andjetties,maybe important,(Kirby,1988).
In manytidalmudflatareas,biologicalindicators,
in theformof scoured-out
examples of
burrowing orboringbivalves,suchasMyaarenaria,Scrobicularia
plana,andPholasdactylus,
indicatethat an areais experiencing
long-termerosion. Largescale,short-termepisodic
changes needtobedistinguishedfromlongerterm,slower, evolution,
butfrequently
thereare
a rangeof indicators
of thetimescale
of change.Exposure of suchbivalveshasbeenusedin
a numberof areasto indicatethe sensebut nottherateof long-termevolution,(Kirby, 1990,
Kirbyet al.,in press,
andBradley,1957).Frequentlytheprogressive
lowering
of themudflat
surfacewhichexposes thebivalvespermitsepifaunalspecies
suchasbarnacles
andalgaeto
colonisethe bivalve surfacesin bands. Band age decreases towardsthe sedimentsurface.
Measurements of bandthickness, reflectedin yearlyspaffall,providesa directmeasurement
of erosionrate, (Kirby, 1990).
Biological
indicators
maybe complemented
by chemical
evidence,
suchastheabsence
of
radionuclides
from surficialsediments
underlyingcontaminated waters,(Kirby and Parker,
1980)or theabsenceof otheranthropogenic
chemicals,(Little, 1989).
R Kirby 317
specified
in the tidalrange,andcoastal
morphology
is so general
in naturegivessome
confidencethat thereis somelink betweenthe changingdominanceof wave and current
processes
andthecoastalmorphology theygiveriseto.
Table1' Meantidalrangedivisions,
followingDavies(1964)andothers
, :
Figure1:Variation
in morphology
of depositional
shorelines
compared
to tidalrange(Hayes1975).
318 R Kirby
+2,0
.ø
, ..9MLW•/
, , , , , , ,
0.1 0,2 0.3 0/, 0.5 0.6 0.7 0.8 0.9 1.0 1.1
relafiveconsfanf-[eve[
area(A/AMHW )
Figure2: Relationship
between
shapeandtidalrangeestablished
for barrier-island
tidalfiatsby
Dieckmannet al (1987).
Microtidal
regimes
arelargelydominated
by waveswhereas,
astidalrangeincreases,
the
relativeimportanceof tidal currentsmustincrease.
R Kirby 319
320 R Kirby
Almostincidentally,but consciously,
the siteschosenin the four localitiesaroundthe UK
coast,whenaddedto theGermanwork,permittedalmosta completespectrum of hypsogmphic
curvesfrom microtidalthroughto hypertidalto be assembled
for the first time.
Figure 3: Locality map showingthe four coastal embaymentsin which measurements were
undertaken.Boundaries
of numbered
andnamedsub-areas measured
areindicated.Notescales
vary.
R Kirby 321
16.4 Methods
Area km 2
Wash 162.1
The curvesmeasured
from the eightrepresentative
sub-areas
presented
herediffer in a
numberof respects.
16.5 Results
A totalof 43 sub-areas
werecalculated andbothnon-dimensionalanddimensionalplotswere
prepared.Sub-areas wereamalgamated suchthataslargean areaascouldbe justdiedhas
beencalculated in orderthatresultsareasrepresentative
aspossible.The43 sub-areaswere
reduced to 15. The 10 Humbersiteswereamalgamated into2 hypsographiccurvesfrom
322 R Kirby
continuous
areas,14 sitesin the Medway werecompiledinto 3 curvesfrom continuous
areas
and 10 sites in the Wash were consolidated into 1 curve. The 9 sites measured in the Severn
couldnot be consolidated
furtherbecauseall are from non-adjacent areasand all aredifferent
onefromtheother. Only 8 of thecurvesarereproduced here. In additionto individualplots,
all Severn Estuary curves are individually compiled into a single plot. In addition,
representative
curvesfrom this studyare compiledtogetherwith the earlierDieckmannet al
datato producea compositespanningtheentiretidalrangespectrum from 1.8 to 9.4m MTR.
The arealcoverageof the curvesvariesconsiderably from site to site,Table 3.
MHW *6 55m
MTL -"4 1 m
OD(N)
.39m •
MLW ß1 75 m •
0m CD
-0.1 m LAT
Figure4: Hypsographic
curvefor areas1-3 SpurnBight,HumberEstuary.
R Kirby 323
16.5.1 Humber:SpurnBight
MHW .6.5m
MTL +3.8m
OD(N)
MLW +1.7m
Figure5: Hypsographic
curvefor areas1-10 The Wash.
16.5.2 Wash
324 R Kirby
MHW ,-5.25m
OD(N) -'3.75 m
MTL*3lm j
MLW .1 05m
f 0m 'CO
+0.1rn LAT
• _
-8-
_
-12-
_
-14-
Figure6: Hypsographic
curvefor areas1-3 SledeOoze,MedwayEstuary.
This tidal flat is entirely muddyother than for a narrow zone with shellbanks. The
hypsogmphic curve(Figure6) hasa pronounced sinuositywith a verybroadflat zonejust
below MTL. The tidal flat is at a low elevation,onlyjust reachingaboveMTL closeto the
coast. Thereis no uppershore,the tidal flat only risingto +3.80m (closeto the national
datum,Ordnance Datum(Newlyn),(OD{N})), andthemid-shore sectionis narrowandsteep.
Thecurveis mostcomparable withthatfromDeadman's Islandin theMedwayandthecurves
for sections1,2,3, 5 and 6 in the Severn.
325
R Kirby
2-
MTL +6 45m
OD(N] •
•MLW
+2
25m
6.3m below OD(N) LAT (CD)
-10 • , , '
o o'.• o'.• o'.•
Relative Constant Level Area
Figure7: Hypsographic
curvefor CardiffBay,SevernEstuary.
,• .2
•-4
-8
.10
-].2
-].4
0 ' 0:2 ' 0:4 ' 0:6 ' 0•,8 ' 1.0
Relahve Constant Level Area
Figure8: Hypsographic
curveforStertandBerrowFlats,Bridgwater
Bay,BristolChannel.
326 R Kirby
Area 3
6-
MHW ,10.8m
MTL ,6.45m
OD(N)
MLW ,•2.25 m
-10 • , ,
0 0'.2 014 0'.6 ' 0. '8 ' 1.0
Relative Constant Level Area
Figure9: Hypsographic
curvefor Gold Cliff, SevernEstuary.
R Kirby 327
The hypsographic
curve(Figure8) showsa pronounced
andcontinuous
concavityextending
to the subtidal limit. The curve extends above both MTL and MHW but the inner section is
narrowand in relativeterms,steep. Only 20% of theprofile is aboveMTL. There is no salt
marshat the top of the tidal flat for muchof the coastalsectionand the salt marshbetween
Steart village and Hinkley occupiesonly 2% of the total intertidal zone. The section
consequently providesa markedcontrastwith thehypsographic curvefor CardiffBay, despite
the latterbeinga similardistancedown-estuary on the oppositebank.
Area 6
6-
MHW ,11.6m
2-
MTL ,6.92m
ODCN)
MLW +2.2m
-8-
-10-
-14-
0 ! 12
0. [ 0.4 ! ! 0.6 o18
Relative Constant Level Area
Figure10: Hypsograpic
curvefor Clevedon,SevernEstuary.
328 R Kirby
- 4........
Stert
&Berrow
Flats 4.•."/'"
5
7
.... Gold
Rumney
6......
Clevedon
Cliff •
•
. •' .? •
2y.,•3,5,7-9
9 ..... Avonmouth •. • "•' 7•'• ./
-
,. •'..•" .
........
? Referenced
toMTL
/" .
. ... /
.'
0 • 0. •2 • 0,4' ' '
0.6 • 0•.8 • '0
1. 1.2
Figure11' Superimposed
Hypsographic
curvesfor all nineSevernEstuaryandBristolChannellocalities.
R Kirby 329
1 •;F Severn
2 CB Severn
--Cardiff
Stert
Flats
Bayt
Hypert•dal MTR 8.55m
II 3 The
4
Wash
t
5......
Humber - Spurn Bight
Medway- SledeOoze)
Low macrotidal MTR
t4'8m
•4.8m
4.2m 1
1
/
II MTL
6.........
7.....
8..... Meldorfer
Bucht
West
Meep •,•',•
mesohdal
MTR
Otzumer
M,c'rohdal
MTR
1.8m3.3m
BelleLow
• •
mesotldal
.•5
2.5m J 4
,.•"....'•''"•'
.........
1 ß• •.-• • *• • ....
/ x 1
!
/,/
{,• , , , ,
.e,eence,o.,
, I
330 R Kirby
complicated
by thefact thatwhilsttheUK datais referenced
to MeanTide Level (MTL) the
Germandatais referencedto MeanSeaLevel(MSL). A strictintercomparison
is notpossible
because MSL is not known for the UK curves.
16.6 Interpretation
It hasbeenpointedoutthatmuddyshorelines havebeenpoorlysurveyed andonlyveryfew
determinations
of shoreshapeareavailable.Arisingfromthisanyfindings fromthisstudy
mustbetreatedwithcaution.Nevertheless themeasurements of shoreshapegainin reliability
fromthefactthatthe long-termsedimentary regimein the areasunderscrutinyis sowell
established.
TheHumber,especially in theSpurnBightregion,is knownto be experiencing
infillordepositionoverthelongerterm,Wilkinson et al (1973),asistheWash,Evans(1965).
Boththesesitesapparently havean adequate sediment inputandbotharegainingsediment
fromseaward aswell aslandward.In contrast, theMedway,at leastin respect of its outer,
main channelcoastline, has beenshownfrom a varietyof evidence,includingstake
measurements,archaeologicalandmapevidence, aswellasbiological evidencefromMyaand
otherbivalves,to be erosional.Thisswingto anerosional regimehasoccurred sometime in
thelast300 years,priorto whichtheMedwaytidalflatswereaccretionary, Kirby(1990).
R Kirby 331
332 R Kirby
16.7 Conclusions
R Kirby 333
Acknowledgements
The authorwishesto thankthe SevernTidal PowerGroupfor permissionto publishtheresults
of contractwork. Surveydatawassuppliedby BritishPortsAssociation(HumberandCardiff
Bay), Binnieand Partners(Wash),MeAlwayPortsAuthority(MeAlway).Surveysfor Severn
Estuarymain channelcoastswere undertakenby contractaerial surveys. The surveyof
BridgwaterBay wasavailablein scientificliterature.The earlystagesof the studybenefitted
from guidanceprovidedby Dr-Ing ManuelaOsterthunof the FranziusInstitutein Hanover,
Germany.
16.8 References
Goss-Custard J.D., McGrorty S and Kirby R., 1990 Inshorebirdsof the soft coastsand sea-levelrise.
In: Expectedeffectsof climaticchangeon marinecoastalecosystems,
Editors:BeukemaJ.J.,Wolff, W.J.
and Brouns,J.J.W.M. Kluwer AcademicPub., p.189-193.
Kirby R., (in press) The evolutionof the fine sedimentregimeof the SevernEstuaryand Bristol
Channel.BiologicalJournalof theLinneanSociety.Proc.Symp.onevolutionandchangein theBristol
Channeland SevernEstuary.
334 R Kirby
Straaten L.M.J.U., van 1950 Environment of formation and facies of the Wadden Sea sediments.
Tijdschr.Kon. Neded. Aardr. Gen., LXVII, No. 3, p.354-368.
17
Abstract
Bedformsarecommonfeaturesin ourestuarineandcoastalwatersandthedynamicinteraction
betweenthebedformsandtheflow is of considerable interest.In strongturbulenttidalcurrents
thesuspended loadis thepredominate formof transportandit is essential
thatthesedimentary
processes of thiscomponent be understood. The work reportedhereis partof a studyaimed
at clardyingthe role of suspended transportuponsandwavedevelopment and migration,and
the feedbackof thesemodificationson the suspended load. To discernthe mechanismsof
transport,measurements of the turbulentflow were takenat four heightsabovea sandwave
andrecordsof the response of the suspended sedimentconcentration wereobtainedusingan
AcousticBackscatter Probe,ABP. In conjunction with theABP themeansuspended sediment
concentration was measuredusingconventional pumpsampling.The ABP is increasingly
becoming a prime techniquefor measuringsuspendedconcentrationfluctuationsand
consideration is given here to the assessment of the acousticmethodfor making such
measurements.
17.1 Introduction
17.2 Acousticanalysis
17.2.1 Theory
/•s' as/•oro
D2
2r
2if. l (17.1)
axis
•nearfleld I'arfleld
(17.3)
if.. ½
, --I' dr (17.4)
• -(lJpa,){15(kaj)nJ[l+(kaj)
2+(-•)[3(kaj)n]} (17.5)
2
</>2>
r292
ee(%.,.)
16a,
p, ".2Jl(karqinO)
sinOdO
(17.6)
t•r• • 3xc kat•inO
fm= 1
100
I0ø' pOs•:
075 b
10-1
10-2
10-3
10-4 I I I I I I I I
17.2.2 Laboratorymeasurements
Vibrator.
Sediment
container
Sieve•
Impeller
mixer Hopper
Mixer
Hydrophone
beam
width
Pump
sampling• •.. '¾ Homogeneous
3m
suspension
Electronics
8roadband
Tronsducer
IH,ydrophone
Figure3: Experimentalarrangement
for laboratorymeasurements
of backscattering.
To establish
experimentally
the valuesfor fmand • a numberof measurements
at a
selectionof concentrations
were carriedout. Measurements
of the attenuationfor a rangeof
concentrations
were obtainedby recordingthe signallevel measuredusinga hydrophone
placedatthetopof thetower.Themeasurements gave•--0.8+0.15,
thiscompares
witha value
obtainedfromequation(17.5)of • = 0.35. Theformfunctionwasestimated
fromreadings
takenat low concentration
whenthe effectof or,wasnot significant.This wascarriedout by
rearrangingequation(17.2) and makingf•, the dependent variableon the backscatered
pressure.
(17.7)
<P,.,>,r
3xcM
J [•.[ •a-•;• SinOdO
This resultedin a valueof fro=0.7+0.1,equation(17.3) givesfro=0.5.The form function
valuesarecomparable whilefor attenuationthedifferenceis significant.At presentwe are
not in a positionto resolvethe latter discrepancy definitivelyalthoughif a, is being
underestimated
by20%thereadjusted
calculated
values
wouldgivefro----0.58
and•=0.53which
wouldbe roughlywithinthe expecteduncertainties.
However,repeatedmeasurements of a,
showa variabilityof theorderof 10%.Alsotheanalytical
expression
for ( is a meanfit to
the data of Flammer's which does include a scatter about a mean of 10%. However, these
variationsare not sufficientto accountfor the discrepancy
in the attenuationconstantand no
convincingexplanation
is availableat presentto resolvethe inconsistency.
For the current
laboratory
analysis
themeasured valuefor • hasbeen.•mplemented.
100[ • 0171
kgm
3
1011 11 b
10ø 0.1
589kgm3
101
10 2 mp
10 3
01
100
[• 3 10
1 217 kgm3
101[ 1
0.1
10 2 10 1 10o 05 10
R(m) R(m)
• Ol ot i.o to.o
M,(kcjrn
-•)
Figure
5: Comparison
of acoustic
measurement
of suspended
loadM•,,withthepumped
sample
values
M..
The variation
aboutMA=M p is of the orderof 20%, someof whichis assigned
to
uncertainties
in Mpof 10%,andtheresttotheacousticmeasurements.
Theresults,
however,
do show that the theoreticaland measuredconcentrationsare in excellentagreementand
confinn the applicability of the acoustictechniquefor measuringsuspendedsediment
concentration.
31dHz
transducer
•-•
'•F Electl-omagnet•C
current
meter I • (tOm)
Estimatesof concentration
profiles,computedusingequation(17.6), obtainedfrom one
minuteaveragenns acoustic
pressure dataareshownin Fig 7a. The measurementsnearpeak
flow,around1700hours,showsignificant
concentration
levelsof greater
thana Kgm'3.The
suspension
is well mixedthroughthe watercolumnwith valuescloseto 1 Kgm'3 being
estimated at l m above the bed.
Towards the end of the flood, 1900 hours, when concentrationsare seen to be about two
ordersof magnitudedownon the peak thereis an almostuniformconcentration profile. At
thehigherconcentrations thereis thepossibility
thatuncertainties
in thevalue of ( canlead
to significanterrors in concentrationestimatesat the longer ranges,ie near the bed. As
mentioned previouslythestep-wisesequentialcomputation of M(r) is proneto accumulating
errors,as M(r) is usedto calculateor,and is then usedto continueevaluatingM(r). An
exampleof thevariationwhichcanoccuris shownin Fig 7b. Thereis a steadydivergence in
predicted
concentration asthebedisapproached
andtherefore theconfidence
in concentration
estimates
is relatively
lownearthebedforhighconcentrations whenthevaluefor • is not
precisely
known.However, havingtakencognisance
of this,theresults
showninFig8, where
a comparisonof theconcentration
estimates
obtained
usingtheABPwith theesmarine pump
sampledatacollected
overtwo consecutive
floodsshowmoderately
goodagreement.
0 102
100
1735
R(m) R(m)
101 101
•s = 0.6
Figure7: a) Acoustic
measurements
of suspension
profilesat differenttimesovera floodperiod.b)
Effect of varyingthe attenuationconstant.
,Olin
+02m
•04m
aOSm•+• oIm
o2m
o4m
08m
.•
M.(kg•s) MA(kgn•
3)
Figure8: Comparison
of thefieldmeasurements
of thesuspended
concentration
usingacoustic,
M•,, and
pumpsample,
Mp,dataontwoseparate
days.
346
P D Theme,R L Soulsby
andP J Hardcastle
T(s)• '."--',.......
--'-,,•;?__
• ...g•:-=
R(m)
Figure
9:High
temporal
resolution
acoustic
measurements
ofsuspended
lead.
Values
below
0.2kg
m-3
havebeensetto zeroforclarity.
The
results
are
sufficiently
similar
toprovide
encouragement
forprogressing
and
examining
the
data
topresent
itshigh
temporal
resolution
characteristics.
Anexample
ofthe
preliminary
results
presentlybeing
calculated
arethose
shown inFig9.Thiscovers
oneminute
ofdata,
equivalent
tooneprofile
inFig7 andshowsthetemporal
andspatial
variability
ofthe
concentration
field.
Surges
inconcentration
arefollowed
seconds
later
byrelatively
quiescent
periods
ofsimilar
duration,
withtheprocess
being
extremely
variable.
•
0.0 oo
o.8rt•-
> 0.02
0.02
o.
ø'8
I
0.8
0.0
•-0.
• o0o•
• o oo•
-5. •
0.0o6
0.00•
SCALE (m)
0 • a 3 4 5 6 7 8 9 •0
BAND
• • t t J t I I I I I •
BAND •
BAND 3
• BAND 4
Figure10: a) Longitudinal
diffusive
flux. b) Verticaldiffusive
flux. Scalekg m-:s-'.
17.4 Conclusions
17.5 References
18
F Jiangand A J Mehta
Abstract
The behaviour
of softmudunderprogressive
waveshasbeenbrieflyexaminedin the field.
Themainobjectivewasto demonstrate
thatunderwaveconditions
thataretoomildto cause
significant
particulate
erosion,
deeplayersof fluidmudcanundergo
measurable
oscillations
dueto waveloading.Thefieldsitewasin theshallowlittoralmarginof LakeOkeechobee,
Florida. In a test duringwhichwind wave frequencywas on the orderof 0.4Hz and
significant
waveheight
around
10cm,wave-coherent
mudmotion
wasmeasured
20cmbelow
the mud-water interface.
A shallowwaterwavemodel,whichconsiders
thewatercolumnto be inviscidandthemud
layertobea highviscosity
fluid,hasbeenusedtoaidin datainterpretation.
Giventhewave
energyspeca'um,themodelapproximately
simulates
thewatervelocityandmudacceleration
spectra;
deviations
between
simulationandmeasurementarepronouncedin thehighfrequency
rangewhereintheshallowwaterassumptionbreaksdown.A low frequency wavesignalat
about0.04Hz,believedto be forcedby thewindwave,wascharacteristic
of measured
spectra.
Thissignalin themudwasenhanced relativeto theforcingsignaldueto thedependence of
waveattenuation onfrequency, andledto horizontal muddisplacementsontheorderof 2mm.
It is possible
thatthissignalis akinto surfbeatwhichcauses themudto oscillateslowly,
therebycontributingto its mobility.Sucha motionmayhaveimplications for theratesof
nutrientexchange
betweenthebottomandthewatercolumn.
18.1 Introduction
It is generally
wellrecognized thatin shallow,
episodic
coastalor lacustrine
environments
with muddybeds,reworkingof mudby wavescauses thebottomto becomefluidized,with
the fluid statepersistingas longas wavescontinueandthereafter,untilthebottommaterial
dewaters sufficiently
to leadto hardening undercalmconditions.Laboratoryevidencepoints
to fiuidizationundercyclicloadingtobedueto thebreakup of thestructural
matrixof thebed
18.2 A physicalperspective
A simplephysicalperspective canbechosen to dealwitha rathercomplexproblemwhichis
characterizedby time-dependent changesin mud properties with continuedwave action.
Althoughsuchchanges havebeentrackedto someextentin laboratoryexperiments, field
evidenceis scarcedue to evidentproblemsin deployingrequisitetransducers.Furthermore,
the basisfor any theoreticalexamination
of the time-variability
of suchpropertiesas mud
shearstrengthis presentlyrudimentary.Thereforeit mustbe bornein mindthatin treating
theproblemtheseconsiderations imposecertainoperationallimitations
in datagathering
and
analyticconstraints
in analysis.
Below the fluidized mud is the cohesivebed having yet higher densities(greaterthan
-1.2g/cm3).
Laboratory
observations
byMaaandMehta(1987),andtheoretical
workbyFoda
(1989) for instanceshow that the wave orbits can penetratethe bed, therebyleadingto
deformations of theupperpartof theotherwisestationary bed. Undercontinued waveloading
suchdeformations cancausefluidizationby eliminationof theeffectivestressassociatedwith
the bed matrix. Due to the generallylow ratesof upwardmassdiffusionabovethe wave
boundarylayerhowever,theproblemof mudmotionby wavescanbeconveniently considered
to be uncomplicated by the effectsof particulateerosiondue to stress-induced
breakupand
entrainment of the cohesivesedimentaggregates at the mud-waterinterface(vanRijn, 1985).
In fact,laboratoryaswell asfield observationsshowthatwaveconditions requiredto generate
measurable bottommotioncanbe quitemoderatecomparedwith conditionsrequirexlto cause
significantparticulateerosion(RossandMehta, 1990;MehtaandJiang,1990). Accordingly,
MWL
UB
Mobile
Suspenion
• Lutocline
..... • I•md'•'u c•
Deforming Bed
Figure1' Schematic
of mudbottomresponse
to wavesin termsof verticalprofilesof sedimentdensity
and velocity.
_ (•(x,t)
1P• (712(x,t)
. i mm Water m m m mm mm
x ....•;:...•...:.?:.:•.....;.::.•..............:•:.....•?.•?•:..•...•?..•?:•:..
..'.."..':.:'.:.'
Bed
Upper layer:
C•U
1 i•q1
•t
+g ax
- 0 (18.1)
i5(q•_qz)+H••u 1 -0
c•t
(18.2)
Lower layer:
&h
dz+c3qz
.0 (18.4)
tll (eø'0'112(øø'g't)'
Y[1(eø'O
• Y]2
(oo,
t) --,0 (18.5b)
u•(x,O,0- 0 (18.5c)
Upper layer:
+ 0 (18.6)
iSt' F•
at'
, - v12
/5(vl• ') + c3u•'
ax'
=0 (18.7)
Lower layer:
%'(x',0,t')- 0 (18.]0c)
au:'(x,H:',t')/az'
- o (18.10d)
[1 - cosh(mz')+ u•h(m•')
(18.14)
sinh(mz')]exp [i(k'x' - t')]
where
(18.17)
[R[1+R- cos-0
2
(p2+q2)¬]
+
2r(R: +I:)
(18.18a)
I[I-sin•(p
z+q2)•
]}
Yi '
2r(Ra+Ia) [R[l-sin-02
(p2+q2)•
]- (18.18b)
I[ 1*R-c,
os-0
2
(p2.q2)•i]
]
where
0=tan-I
(•) (18.19c)
exp(4X)
- 1+2sin(2x)
ßexp(2x)
1 (18.19d)
I=H'•[1
-2:1exp(4x)
-+11;-2•(2X)
[exp(nz) 2•-• 'e•2X)
I (18.19e
(18.190
1.0
H:•=0.1
0.2
0.3
0.4
0.5
0.6
0.7
r = 0.15
0.8
0.9
1.0
0.7
0 I 2 3 4
7,=H:•(Re/2)I/2[=H2
((•/2v)
1/2]
Figure3: Dispersion
relationship
basedon theinviscid-viscid
model
In Fig.4 thenormalizedattenuation
coefficient
k;/F,which
alsodependsonH•,r andZ,
isplotted
againstZ forvalues
ofH•ranging
from0.1to1. Wenotethatsince
k•/F•= k•Co/o,
theplotcanbe considered to represent themannerin whichk• changes with increasing Z,
starting
withZ TM 0 atwhichk• -- 0 (rigidbottomcase).An interesting featureof theobserved
variationin k• is theoccurrence of a peakin k• asZ approaches unity. In otherwords,wave
damping is greatest whenthemudlayerthickness (H2)is of theorderof theboundary layer
thickness.As Z increases beyondthisvalue,k• decreases with decreasing thickness of the
boundary layerandapproaches zeroasZ -• o•,i.e.thelowerlayeralsobecomes inviscid.As
Z decreases belowone,although theentiremudthickness constitutestheboundary layer,the
rateof energydissipation reduces relativeto thatat Z = 1, hencek• reducesaswe!!.
0.16
] I f I • I i I
H•,=1.0
0.12
r= 0.15
0.08
0.04
0.1
0.00
0 I 2 3 4 5
Z=H:•(Re/2)l/2[=H2
(c•/2v)
1/2]
1.0 I i I • I
•- 1.5=7.
O•'
0.8 ao=8cm
=0.15
• 0.6 WATER
mO
0.4
• _ 0.5
• 0 1.5
=7. FLUID
MUD
• 0.2
o[,• • I • I • I
0 0.1 0.2 0.3 0.4
VELOCITYAMPLITUDE,urn(m/s)
27ø1
80ø40'
Test
Site
26ø45._.•
81ø00'
I 80ø40'[
0 5Kin
Depths in Meters
Below Datum
Mudviscosity
wasmeasured
in a Brookfield
viscometer
(modelLVT) withminiature
vanes,
at different
mudbulkdensities.
Fig.8ashows
thestress-shear
raterelationship
obtained
ata
density
of 1.05g/cm
3. Thebehaviour
of themudisobserved
tobepseudoplastic
atrates
less
thanabout3s'•,andNewtonian
athigher
rates.Sincethewave-induced
shear
ratesin themud
duringthe studyperiodwere on the orderof 0.1-0.2s
'•, it was decidedto calculatea
representative
viscosity,
p, astheslope
of thetangent
tothecurve
passing
through
theorigin
asshownby thedashed line. Thestress-shear
raterelationships
at all otherdensities
were
foundto be qualitatively
similar,andwereusedto obtainthecorresponding valuesof
viscosity.
Theresultingrelationship
between theviscosity
(relative
towater,Pw)anddensity
isshowninFig.8b. Duetolimitations intheapparatus,
muddenser than1.12g/cm 3couldnot
betested.It wasthereforeassumed thatathigher(upto 1.18g/cm
3)densities,theviscosity
couldbeobtainedbyextrapolationof thelog-linear
curveat higherdensities
in Fig.8b,i.e.
assuminga power law.
BULKDENSITY
(g/cm
3)
1.1 1.2 1.3 1.4
01.0 I ii I •
I
-10 - I
\,
ß I
• -20
Mud
_
-30 • Ler_
-40
-50
.....
-6O "Hard"
I ! I
Any interpretation
of viscometerdataobtainedusingvanes,in termsof a relationship
between viscosity and the structure of the sediment matrix, must be treated with
circumspection sincevanes,by their very presence,breakup the matrixat the cylindrical
surface definedby thevanedimensions.Furthermore, viscosity
isnotcharacterized
bydensity
in a uniquesense,insofaras viscosityand densityare independent physicalpropertiesof
fluids. Neverthelesstheobserved trendin Fig. 8b doesseemto suggesta rathersignificant
influence of sedimentpacking,asreflectedby thebulkdensity,on viscosity.Apparently,at
densities
lessthanabout1.02g/cm
3,theaggregated
matrixrapidly
became
tightlypacked
with
increasing
density. However,furtherincreasein densitydid not seemto havedrastically
alteredthecompact
arrangement attainedat 1.02g/cm3. Sincethe"granular"density
of the
organics-rich
sediment
was2.14g/cm 3, the sediment volumefractioncorresponding to
1.02g/cm3wouldbe0.018.It isconceivablethatthisvolume fraction
isequaltotheso-called
criticalvolumefractionabovewhichthe rigidityof the matrix(as reflectedby the shear
modulusof elasticity),increases
rapidly(WilliamsandWilliams, 1989).
Duringtheexperiment,time-series
of waterlevel,horizontal
watervelocityandmudmotion
were obtained. Water level was measured with a subsurfacemountedpressuregage
4-
P2
=1.04
g/cm
3
!• =1.90Paos
3--
I
/
I
I
I
/
0 2 4 6 8 10 12 14
SHEAR
RATE
(•1)
Figure8a:Stress-shear
raterelationship
for LakeOkeechobee
mudat a bulkdensity
of 1.04g/cm3
The transducers
were connectedto a dataacquisitionsystem(Tattletale,Model 6) mountedon
a woodenbaseat the top of the tower. The testwas conductedfor a total of sevenhours,
duringwhichthe wavedirectionwasconsistently westerly,corresponding to a fetchof about
50kmo
I I I
103
uJ 102
•W = 1.0087x 10'3Pa ßs -
22oC -
o
O3 10
1
1.00 1.02 1.04 1.06 1.08 1.10 1.12
BULK
DENSITY,
p2(g/cm
3)
Figure8b: Selectedrelationship
betweendynamicviscosityanddensityfor Okeechobee
mud
The time-series
wereanalyzedin termsof theirspectralproperties
andcentraltendencies
via standardprocedures.
Fig. 9 showsthevariationof thesignificant
waveheight,I-I•, over
testduration.Notethateachhourlydatapointis a 10minaverage overtherecordlength.H•
is observedto havebeenrathersmall,peakingto 10cmat 3hr.
spectrum
itselfis shownin Fig. 11 The modalfrequency
variationis compared
in Fig. 10
with thesamedetermined from thewatervelocityspectra,an exampleof which(corresponding
to Fig. 11) is shownin Fig. 12.
o•,10
I I
0 I 2 5 6
TIME (hr)
(mean 0.51Hz) in the caseof accelerationin comparisonwith thosefor the waves and the
current are believed to be attributable to anomalies associated with the in situ mount used for
the accelerometer.The accelerationspectrumshowsa markedpeak at a very low frequency
corresponding to a long periodoscillation.Commensurate but comparativelysmallerpeaks
also appearin the wave and watervelocityspectraof Figs. 11 and 12. This low frequency
is distinctfrom the wind wave-inducedfrequency,and is discussedfurther later.
1.0
-1- 0.8
E
Mud Acceleration
0 0.6
0.2
0.0
0 I 2 3 4 5 6 7
TIME (hr)
5O
[ I [ I [ I [ I
• 40
30
o
o 0.2 0.4 0.6 0.8 1.0
FREQUENCY (Hz)
LU 240
i-
180-
•Itl•
Calculated
FREQUENCY (Hz)
0.30 j i
•-,
••. 0.24
•E I
o3z o.18
z•o - Calculated•,,,,•
!•
Measured•
I" i
Q. ,• 0.06
/'
0 0.2 0.4 0.6 0.8 1 .c
FREQUENCY (Hz)
Figure13:Measured
andcalculated
mudacceleration
spectra
at 1 hr
18.5 Response
of LinearizedFluid Mud-WaterSystem
Giventhewaveenergyspectrum, whichessentially
characterizes
themeasuredwaveamplitude
(A) variationwith wavefrequency, Eq. 18.13canbe usedvia the correspondingspectral
transferfunctions(MehtaandJiang,1990)to calculatethewatervelocityspectrum,
andthe
time-derivative
of Eq. 18.14to calculatethe mudacceleration
spectrum.
model results.
'•' 0.4
• 0.3
0
•: 0.2
o
Mud Acceleration
•,
0
0.1
- Surface
Wave
/ '--
Water Velocity -
0.0 I I I I I I
I 2 3 4 5 6 7
TIME (hr)
Figure14:Dominantlow frequency
spectral
signature
duringthetest
Components of
Forcing Wave
Forced
Long
Wave
FREQUENCY
fœ fl f2
---• •'••Af •'[ •Af
At/--Forcing
Wave
•' I/. , I• t-orcec•
wavem . .t !• I
-5
v v
= azcot - (18.20b)
Here, el and ee are the phaselags. Assumingfurtherthat the amplitudea = a• = a,., the
resultant(forcing)wave, •lr = •l• + •l,., is
n,=2acos(ot
-e)-cos(,aø•:
ae) (18.21)
- =-
n a2k [1 * cos(ao-t- e)] (18.22)
Assuming
thetwo peakfrequencies
to be theprimarycontributors
to thecorresponding
long
wave peak, we have the following parameters(obtainedfrom the Fourier seriesof the
corresponding surfacewave record):a• = 2.4cm,a: = 2.6cm,long wave amplitudea• =
0.30cm,f• = 0.38Hz, f: = 0.42Hz, f• = 0.049Hz, e• = 191ø,e: = 171ø, and e• = -65ø. Note
thataf -- 0.040Hz, whichis reasonably closeto fe The forcingwave (q•) and the forced
wave (q•) are plottedagainst• = ao.t-ae in Fig. 16. It is evidentthat the forcedwave is
almost;• radiansoutof phasewith the shortperiodwaveenvelopeas theorywouldrequire.
If we assume theshallowwatercondition, theforcedwaveamplitude(fromEq. 18.22)would
bea:/H.Givena -- 0.5(a•+ a•)= 2.50cm
andH -- 143cm,
a:/H= 0.044cm.
Thusthetheory
under-predicts
the amplitudesignificantly,which might be due to inherenttheoretical
limitations
aswell as therathergrossassumptions
madein applyingthetheoryto data.
18.7 ConcludingRemarks
Notwithstanding
thelimitednatureof thedataobtained, thecomplexities
in modelling
mud
motion,and the ratherobviousconstraintsin simulationarisingfrom the shallowwater
assumption
aswell asothers(e.g.inviscidwaterlayer,linearized
response,
homogeneous mud
layer), it is demonstratedthat measurablemud accelerationscan occur tens of centimetre
belowthemud-water interface,
underwaveactionthatis mildenough toprecludemeasurable
particulate
erosion of theinterface.Although bioturbationdoesnotseemto be a significant
factorin LakeOkeechobee (Kirbyet al., 1989),theeffectof persistent
mudoscillation,
even
thoughverysmall,canbe germane to likelychanges in theratesof exchangeof phosphorus
andotherconstituents responsible
for thehighlytrophicstateof thislake. Similarly,there
may be an effect on the formationand upwardtransportof gas bubbleswhich occur
abundantly in the muddyareaof the lake (Kirby et al, 1989).
In thisstudy,weconsidered
muduptoa density
of 1.18g/cm
3tobea highlyviscous
fluid.
In general,however,carefulmeasurements
(e.g. SillsandElder, 1981) indicatethat,at least
underquiescent
conditions,at densities
exceeding
about1.1g/cm 3,clayeymudtypically exists
asa poroussolidratherthana fluid. Vaneshearmeasurements (Hwang,1989)suggest that
theorganics-rich
Lake Okeechobee mudis proneto occurin a fluidizedstateat densities
at
leastuptoabout1.065g/cm
3in quiescent
conditions,
although
viscosity
datapresented
here
seemto suggest
theoccurrence
of a structured
phasefordensities
greater
than1.02g/cm
3.
It is unclearif at 1.18g/cm
3 densitythismudis normally (i.e. in theabsence of wave
action)fluidized,
butit is worthexaminingthisissuebycalculating values of •. Thusgiven
H,.= 0.283m,x>= 1.76x10'" m'/s(selectedvalueformodelcalculations) ando = 5.03rad/s
(correspondingto thedominant forcingfrequency of 0.4Hz) yields• = 2.4, whichwould
mean,asnotedpreviously,thatthemudwouldrespond likea "watery"
material.On theother
hand,at the low frequencyof 0.04Hz, • = 0.76, whichwouldmeanthat the mud would
respond
asa "harder"
material
atthisfrequency.
Hencethechosen
modeldescription
maybe
approximate
in thislatterrespect.Nonetheless,
sincein generalthemodelcalculated
mud
Acknowledgement
Supportprovidedby the SouthFloridaWater ManagementDistrict,West Palm Beach(Lake
Okeechobe• PhosphorusDynamics Study), and the U.S. Army Engineer Waterways
ExperimentStation,Vicksburg,MS (ContractDACW39-89-M-4639), is acknowledged.
18.8 References
Kirby R.R., HobbsC.H. andMehtaA.J. (1989). Fine sedimentregimeof Lake Okeechobee, Florida.
Rept.UFL/COEL-89/009,CoastalandOceanographic Engineering Dept.,Univ. of Florida,Gainesville,
FL, 77 p.
Longuet-Higgins
M.S. andStewartR.W. (1962). Radiationstressandmasstransport
in gravitywaves,
with applicationto 'surf beat'. Journalof Fluid Mechanics,13, 481-504.
Maa P.-Y. and Mehta A.J. (1987). Mud erosionby waves:a laboratorystudy. ContinentalShelf
Research,7(11/12), 1269-1284.
Sharma J.N.andDeanR.G.(1979).Developmentandevaluation
of a procedure
forsimulating
arandom
directional
second
orderseasurfaceandassociated
waveforces.OceanEngrg.Rept.No. 20, Dept.of
Civil Engrg.,Univ. of Delaware,Newark,DE, 139 p.
Shibayama
T., Aoki T. andSatoS. (1989). Mud masstransport
dueto waves:a visco-elastic
model.
Proceedings
of the 23rd Congressof I.A.H.,R., Ottawa,Canada,B567-B574.
Wells J.T. and Kemp G.P. (1986). Interactionof surfacewaves and cohesivesediments:field
observations
andgeologicsignificance.In: Estuarine
CohesiveSediment
Dynamics,a.J.Mehtaed.,
Speinger-Verlag,
Berlin, 43-65.
19
Abstract
The waveswere registeredby threerecordersplacedat low tide level and situatedat the
comersof a fight-angledtriangle,suchthatthe secondrecorderwasin mostcases10 m away
from the first,measuredalonga line parallelto thebeachandthe otherrecorderwasalso 10m
from the firstbut alonga line normalto thebeach. FromFourierandcorrelationanalysisof
threewaverecordstakenat or nearhightide, it waspossibleto calculateandplot the phase
differencesagainstthe frequencyfor all threecombinations.It wasalsopossibleto calculate
thewavedirection.If thewaveswereprogressive,
at thisdepth(wherec2 = gh,independent
of frequency),the variationof phasedifferencewith frequencyshouldbe linear. In actual
fact, however,the trendline hasa "wavy"characterdue to the effect of reflection.
378 J Darbyshire
19.1 Introduction
(blC
1 - b2c2)l(blC
1 + b2c2
) (19.1)
2blCl/(blC
1 . b2c2
) (19.2)
where
61,2
= (•,hl,2)l/2 (19.3)
Lamb had somereservations aboutthe accuracyof the formulabecauseof the rapid
variationsin depthand width at the pointof discontinuity.
(C1 - C2)](C
1 + C:•) (19.4)
Thisformula
was derived
forlongwaves(wherec = d'gh),by equating
thefluxonboth
sidesof the boundaryand equatingthe displacement.By this meansthe sameformulacan
be extendedto any gravitywave. However,in the longwave case,the rateof passageof
energyacrossthe boundaryis alsoconserved whereasthisis not so in the generalcaseas the
groupvelocityvarieswith depthin a differentmannerfrom thephasevelocity. Even so,as
J Darbyshire 379
If theoriginisreferred
tothemeasuringposition,
xøreferstothedistance of thewater-line
fromtheorigin,andx• refers toanypointbetween them,thenthereflected waveat x• will
becontributedto byall positions
fromit to thewater-line
xø. Thephase of thecontribution
will dependon thetimetakenby thewaveto travelfromx• to x andback. Expressed
mathematically andusingtheformula,
thenthereflected waveat x• will begivenby:
(19.5)
Theintegrationcanbecarriedoutnumericallyforeachvalueof x• between
0 andxøand
thusgivesthereflection
coefficient
atallpoints
between0 andxø. A computerprogramwas
preparedto do this.
(ah/Ox)
/ kh<<1 (19.6)
380 J Darbyshire
In derivingtheequations,
conditions
(19.6)and(19.7)havebeenassumed. Equation(19.5)
is derivedfrom moreelementaryconsiderations and no conditionis immediatelyapparent
besidesthe shallowdepth one. However, theseconditionswill be borne in mind when
discussing theresults.As equation(19.8)is correctto a firstorderin g, a value=of g less
than 0.33 shouldgive an accuracyof 10% or more.
381
J Darbyshire
0.8
0.7
0.6
0.5
0.4
0.2
O. 1
b/d
Figure1a: Comparison
of observed
andcalculated
reflection
coefficients
fora sinusoidal
bottom
profile.
382 J Darbyshire
0.8_
8.7 _
8.5_0 ß
d=2.5 c m•.
0.4_ '
ßß
0.2
0.1
0.8_
0.7
1 2 5
Iß ' ß 4
0.5 •
0.4
0.2
•
mmmm
ß ß
'I d=5.0
cms.
0 1
1 2 • 4 5 -Fr'ect. Hz
0.6
0.4 - ß ß mm
_ •
0.2 _
0.1
I I ! • I I I I
1 2 5 4 5 •req. Hz
dep!:h=d+sin<&.28518x/12)+sin(&.28•18x/6)• dmx cms.
Figurelb: Comparison
of observed
andcalculated
reflection
coefficients
forthebifrequency
bottom
profile shown.
J Darbyshire 383
0.9
0.8
0.?
0.6
imm
m
0.5
0.4
0.2
0.1
' .
d=2.5cms.
0.8
8.7
0.6
/ ,
0.5
0.4
• • ,•] d=5.0
cms.
0.2
0.1
3 II 4 5 •.e• Hz
0.6
0.5
0.4
0.5
-
0.2
0.1
i 2 5 4 • •re•.
H••
384 J Darbyshire
0.5
0.2
t4' mm
m ßßß '
0.1 _
I I I I I
0.8
5 -Freer.•-74
0.7
0.•
i
I
ß
0.4 _
0.5 _ •.'
0.2
0.1
J Darbyshixe 385
Table 1: Gradientandgradienff(kh)
for tankexperiments.
Taken as a whole, the resultsare satisfactoryand one can use the programwith some
confidence with real conditions as found on a beach.
386 J Darbyshire
a- (19.10)
(19.11)
f:•?
(1/2c)
(Xo
-x)"d•/(Xo
-x•,)"
(19.12)
. "=
•ldx - - O.•(gdDua
(Xo-X)"n-•l
(xo-x•,)"n (19.13)
(19.15)
-n/4
f,:?
(Xo-X)"-•dxl(Xo-Xt,)
"- 1/4
The value1/4 is thusindependent
of n i.e. of the shapeof the profilein thebreakingzone.
J Darbyshire 387
LLYN
PENINSULA
........ IOfIr
........... 2Of•
i I
388 J Darbyshire
AUGUST 11 1977
ß 180
"" 160
• 140
• 1:20
• 4•
• 2•
'• -28
0.08
_ 0.0•
_ 0.04
0.02
100
•e.•s
e0.4
o
4 6 8 10 12 14 l& 18 20 22 24
! I I I I I I I I I I I ! I I I I ! I !
1.5
1.8
5 10 15 20 25 :10 55 40
389
J Darbyshire
JULY 9 1979
ß, 8o
•lTI9*
OF
• ½1)
2--3
•&• r18
.0.20
0.12
o.
04ß re<t. x 100
•
o
-:0.2
•2 • •4
4.0
5.0
39O J Darbyshire
$EPTE•E• 1• 1•?•
2-1
'"' 1,•0
-,= 140
o
• 120
•- 40
ut
0.10
0.06
0.04
0.0:2
ß •-e<•. x 100
0 '
.,-:0.2
4.0
ii1/111
10 20 :50 40 50 60 70 80 90 lee 110 120
J Darbyshire 391
OCTO•œ• •0
ß-. 80
0.10
I
•0.08
_ 0.0•
.0.04
, '-'"'"
14 mi• 0m 72i ?4
•e,a.. x 100
•e.6
e04
0 "
Figure3d: Variation
of observed
andcalculatedphasedifferences
between
waves
on(2) and(1), and
(2) and(3) andthecalculated
reflection
coefficients
againstthefrequency
forOctober
30 1978. The
wave powerspectrumand the beachprofile are shown.
392 J Darbyshire
'"" 1 •0 -
-,=, 140 _
o
a, 120 -
e- 40
.,,..e
0.15
•0.4
o
ø0ø2
J Darbyshire 393
OCTOBER 20 1980
140
120
•- 40
_0.12
_ 0.88
_ 0.04
e 32
•r-e<t. x 100
,,: e.6
o • 0
2.0
394 J Darbyshire
NOV•M• • 1•80
'"' 1•,0
• 140
o
• 100
q- 40
ß 20
0.10
0.08
0. O&
0.04
0.02
I i4._ , E• I ?2 , i•4
•r-e<l. x 100
'0ø2
4.e .
Figure3g: Variation
of observed
andcalculated
phase
differences
between
waves
on(2) and(1),and
(2) and(3) andthecalculated
reflection
coefficients
against
thefrequency
forNovember
5 1980.The
wavepowerspectrum
andthebeachprofileareshown.
J Darbyshire 395
FEBRUARY ? 1981
•2-t
"" 240
'• 218_
o
u
• 188 -
• 158 -
• 128
.,, :/;"'",,•,,,-,,,/
/-. Zlq•Zll•qC:8 -t t •.
r1.0
, 92
,/•, •;20 , ;2:2,
2-3
;24
0.88
8.84
, • , ,• , ? , ,•e, , , , ?e, •2 i •4
-Freq.. x 180
•84
0 '
ø8ø2
t4
•'•- 6I I I I I I i I I I 8,
4.0 -
396 J Darbyshire
FEbRUaRY ? 1781
140
• 120
,• 100
ß". 80
!
'" 60
,q- 40
ß 20
19.25
0.20
0.15
_0.10
_ 0.05
a•0.4
o
Figure3i: Variation
of observed
andcalculated
phasedifferences
between
waves on(2) and(1), and
(2) and(3) andthecalculated
reflection
coefficients
against
thefrequency
for February
9 1981.The
wave powerspectrumand the beachprofile are shown.
J Darbyshire 397
MA•CH 5 1981
240
210
180
150
120
80
&O
_ 0.08
_ 0.06
0.82 x
I I I I6 ! 8
I I I10 I I12 I I14 I I16 I I18 I •0 I •2• •4
4.8
5.0
2.8
1.8
m/m
18 2.8 •0 48 58 68 70 80 •0 100 110 128
398 J Darbyshire
_?
(•)
_6
;24
_7'
<:b) 9.?.79
_6
_5
_4
_2
_4
4 '""*'"
& 14 ' 16 18 20 22 24
I I I I I I I I I I I I I I I I I I I I I
J Darbyshire 399
3601:•3sia0/I, (19.17)
In shallow
waterthevelocity
becomes
4'ghwhereh isthedepth.
If f isthefrequency
then
400 J Darbyshire
(19.18)
and the formulaefor the phasedifferencesbecome:
(19.19)
•2' 360112fcøs01
g• aud •23
Thusthephasedifferencebecomes
directlyproportional
to the frequency.The direction
can be found from the two phasedifferences
•231
•)12- (123/112
)t/triO (19.20)
Thusknowing
theIsandthe•)s,canbefoundandthen4'ghobtained
fromboththephase
difference formulae.
J Darbyshire 401
Figure 3a is the most strikingexample. This examplewas differentfrom the othersin that
the waverecorderswere installedspeciallyonly 25-30 metresfrom a sharprise in the beach.
This was the closestapproachto a laboratorytankexperiment.The resultsshowa classical
caseof a wavyphasedifference frequency variation.In thiscase1•2was5 metresand123 was
10 metres. The reflection coefficientvaries from 25% to 60%. In all the other cases,both
1•2and123are 10 metres. In thesecasesconditions werenot so idealisticas in Figure3a as
the recorders had in manycasesbeeninstalledsometime before. The beachsurveyswere
cardedoutat low springtidesothewaveobservations couldnotbe cardedoutat exactlythe
sametime and therewas usuallya differenceof one or two tidal cycles,so that therecould
be somechangesin the beachprofile. This couldexplainthe constantphaseshift foundin
Figures3d and 3g. Nevertheless in mostof the ninecasesthe resultsare satisfactory.In
Figures3c, 3f, 3g, the wavesare movingnormalto the beach. The agreementis goodfor
Figure3c and Figure3f. In Figure3g, theactualvariationsin thepredictedvaluesagreewith
thoseof the observedvaluesapartfrom the constantphaseshift. Figures3b and 3d showan
angleof approach of nearly45øto theshoreline. The variationsof thepredictedvalues for
Figure3d, like thoseof Figure3g agreewith thoseof the observedonesbut againthereis a
constant phaseshift. The agreement for Figure3d is not unsatisfactory. Figures3i and 3j
showexampleswherethe wavedirectionis about20øoff the normal. Agreementwith both
is satisfactory.A mostremarkablecaseis shownin Figure3e. Here the wave directionis at
70øto theshoreline. In thiscaseno observed profilewasavailablebuta constant depthwas
assumed in thedirectionof 123andthena suddenriseabout40 metresbeyondposition3, (on
the other side to position2). On this basis,someagreementcan be obtainedwith the
observed results.
In mostof thecases,thereflection
coefficients
varyappreciablywith thefrequency,
from
10% to 60%. The 60% valueis reachedin Figures3a, 3e, 3f and3g.
402 J Darbyshire
h G
Anotherpossible
methodof mmckis to extendthereflection
program to coverthewidth
b aswellasthevelocitysothatbcratherthanc becomesthevariable
parameter. A separate
waverefraction
programis constructedwhichstoresthe trackof an initiallyparalleland
J Darbyshire 403
There is an interestingconsequence
of thismethod. The reflectedwaveswouldretracethe
pathof the incidentwavesand so in theorya straightincidentwavefront,althoughgreatly
modifiedby refractionin shallowwater,would still ultimatelygive rise to a reflectedwave
in the openseatravellingin an oppositedirectionto the incidentwave,althoughthis would
be relativelysmall.
404 J Darbyshire
19.9 Conclusions
A method of evaluating the reflection coefficient for any wave frequency for a
one-dimensional profile is developed.It hasbeenappliedsuccessfully with bottomprofiles
in tanks. The methodcan be appliedto real beachesby usinga triangulararrangement of
wave recorders. In the casesdealt with in this paper,somevery interestingresultswere
obtained,indicatingthat for somefrequencies,thereflectioncoefficientin somecasescanbe
as high as 60%.
The one-dimensional
approachcan be extendedto two dimensions.
Acknowledgments
The author wishes to thank Mrs. Eileen Pritchard and the late Mr. W.R. Davis for theft
invaluableassistance
throughoutthe yearsin obtainingthebeachprofilesandmaintainingthe
pressurerecorders. My thanksare also given to Drs. Belzonsand Guazzelli for the useof
their data.
19.10 References
Longuet-Higgins,
M.S., 1950: A theoryof the originof microseisms.
Phil. Trans.Roy. Soc.Lond.,
A243, 1-35.
J Darbyshire 4O5
Modelling (Sediment)
Dynamics and Exchanges in
Coastal and Estuarine Studies Estuaries and the Coastal Zone Vol. 40
2O
Abstract
20.1 Introduction
In realistictidalembayments,
thebasingeometry
andwaterdepthdistribution
(bathymetry)
are usuallyquitecomplex. The distributions
of tidal currentspeedanddirectionare rectified
by thebasingeometryandbasinbathymetry resultingin a complexstructureof transient
tidal
flow. Thenonlinearities in theflowareembedd•in thecomplex structureof velocityshears
whichgiveriseto macro-scale mixing(tidalpumpingandtidaltrapping).The variations in
velocityshearcanbe largeandvariable,theweaklynonlinearapproximation maynotbevalid
for partor all of theembayment.As theflowschangefromweaklynonlinearto nonlinearor
stronglynonlinear,the practicaluseof the resultsderivedfrom weaklynonlinearanalysis
becomes increasingly unsatisfactory.An extensionof theproperties of Lagrangianresidual
currentsandlong-termtransport to generalnonlineartidal flowsis highlydesirable.
20.2 Tidally-averagedtransportequation
O<C>
+ •,,'V<C> + <g,"VC'> = 0 (20.2)
Following
thedynamics
ofalabelled
wate_>parcel
inLagrangian
mechanics,
anaccepted
definitionof the Lagrangianresidualcurrent,u t. is given as
Z(T+to)
- Z(to) (20.3)
where
•(T+to)and•(to)aretheending
andbeginning
positions
ofa labelled
water
parcel
releasedat the time to (referringto a tidal phase),and T is the tidal period,(Zimmerman,
1979;ChengandCasulli,1982). ChengandCasulli(1982) pointedout thattheLagrangian
residualcurrentis a functionof the tidal phaseto. Undertheweaklynonlinearapproximation,
Feng et al. (1986a) have shownthat the Lagrangianresidualcurrentcan be givenin a small
parameterexpansionas
u-;,(t
o) - a,, + + ß au(to) + o(,:) (20.4)
in which u •ais the Stokes drift, e is a smallparameteron the orderof the ratio of residual
currentover tidal current. The time dependency of the Lagrangianresidualcurrentappears
in theLagrangian
drift,•d(to),[seeCheng
etal (1986)].TheLagrangian
driftinscribes
an
ellipseon the hodograph planein a tidal cycle. Within the weaklynonlinearanalysis,Feng
et al. (1986b)continuedto showthatthe first ordertidally-averaged transportequationfor a
conservative soluteis a pure convectionequation,
c•<C>
(20.5)
Ag-
g•(t,
to)
- g,(t)
-[f•,dt'].V•,+
Higher
Order
Terms, (20.7)
where
•(t, to)andu•(t)aretheLagrangian
andEulerian
tidalvelocities,
respectively.
Specifically,the Stokes'drift is definedas
(20.8)
parcelat time t releasedfrom to. If at any time in a tidal cyclethe valueof the [ ] is of the
sameorderof magnitudeasthecharacteristic lengthof thevelocitygradient,thenthetruncated
Taylor seriesin Eq.(20.7) is insufficientto approximatethe differencebetweenthe Eulerian
and Lagrangian tidal velocities (Zimmerman, 1979). When the weakly nonlinear
approximation becomesinvalid,it furtherimpliesthatthephysicalsignificances of theStokes'
drift (as definedabove)and the masstransportvelocityare alsoquestionable, (Zimmerman,
1979).
20.2.2 GeneralizedLagrangiandynamics
ensemble-averaged soluteconcentration
andthetime-averaged concentration at (Xo,Yo)canbe
established.To helpvisualizethiscomplexsituation,consideragaina weaklynonlineartidal
system. The dispersionpatch in this case can be shownto be a mirror image of the
Lagrangianresidual ellipse (Cheng et al., 1986) becausethe governingequationsare
Hamiltonian.In the generalnonlinearcase,a dispersion
patchexistsbut it is not necessarily
a mirror imageof the regionoccupiedby the Lagrangianresidualcurrents.
Dispersion
Patch
A:[x(t-T), y(t-T)]
O: [Xo, Yo]
CO: liean Lagrangian
residual current
)>= •1 •feo*l•12
(20.9)
<C(xø
'yø
'tø t•-ra C(xø
'yø
' t)dt
On theotherhand,whentheturbulentdispersion is neglected,
Eq.(20.1)impliesthatthesolute
concentrationis constant alongeachstreakline.In particular,C(xo,Yo,t
) = C[x(t-T),y(t-T),t-T],
whereIx(t-T), y(t-T)] is a generalpointon the edgeof the dispersionpatch. Thus,Eq.(20.9)
can be written as
<C(xo,y•to)> - •1 •,o-r/:
•'to.
rt2C[x(t-T),y(t-T),t-T]dt (20.10)
' •1 •'t,.rl2
<C(xø'Yø'tø)> <C[x(t),y(t),to_T]>dt(20.11)
1 fyrt2x(t)dt --=-•
, and Y 1 at•-r12
y(t)dt. (20.12)
By subtracting
<C(•,•,to-T)>fromeithersideof Eq.(20.11)anddividingEq.(20.11)by T,
Eq.(20.11) becomes
1 ft•+TI2
<c(xo,•o,to)>
- <c(•,•,t o- • ato-r/2<C[x(t),y(t),to-T)l>dt- <C[E,j',to-TI>
T T
(20.13)
Thetruncated
Taylorseriesexpansion
of <C(xo,Y0,to)>
and<C[x(t),y(0,to-T]>
about(•,•,to-T)
can be written as
c9<C> c9<C>
- <C(•,Y,to-r)> + Tog<C>
<C(xo,Yo,to)> + HOT
+(xo-•)• +(yo-•)oy
(20.14)
417
R T ChengandV Casulli
<C[x(t),y(t),to-T]>
- <C(j,•,to-T)>
+[x(t)-x-]
Ox
+ [y(t)-•]
+_21 [x(t)-œ]
• cgZ<C>
i}x2
+[x(t)-œ]
[y(t)-y-']
O•-<C> + _1[y(t)_y--]
ilxc3y 2
2O •Oy
<C>+ HOT
2
(20.•5)
whereall thederivativesareevaluated
at (•,•,to-T)andHOT standsfor higherorderterms.
Substitution
of theseTaylorseries
expansionsintoEq.(20.13),
afterneglecting
thehigherorder
terms,yieldsthefollowingfidally-averaged
transportequation,
c•<C> 02< C> 02< C> 02< C>
+ <•t,> 'V<C> ' +2 + .
i•t v• Ox
2 v• Oxi•y v•y •y2
(20.16)
In Eq.(20.16),
is themeanLagrangian
residual
currentfollowingthedefinition
givenin Eq.(20.3). The
dispersion
coefficients
are givenby
1 fto+?12 dt (20.17a)
2T2•o-rt2
[x(t)-•]2
1 (20.17b)
2T2•,o-r/2
[x(t)-J]
[y(t)-•]
dt
(20.17c)
¾yym
2T• •eo-rlz
[y(t)-•]=
dt
Thetidally-averagedtransport equation,
Eq.(20.16),is validupto thelimitof theTaylor
seriesexpansions
usedin Eqs.(20.14) and(20.15). Theexactcondition for theconvergence
of theTaylorseriesis difficultto establish,
however,thecharacteristiclengthscalefor the
dispersion
patchisexpected tobeanorderof magnitude smallerthanthelengthscales of tidal
velocitygradients
usedin thedefinition of theStokes' drift,Eq.(20.7).Thus,thevalidityof
thisequationhasbeenextended to generalnonlinear
flowswithoutthelimitation of weakly
nonlinearapproximation. In strongly nonlinear
flows,thedispersion patches canbe highly
a<c>
+ <tZt,>.V<C>-v,Va<C> , (20.•8)
at
20.5 Examples
Resultsfrom two examplesare givento illustratetheproperties andthespatialdistribution
of
tidally-averaged dispersioncoefficient.In bothcases,the numericalsolutionsof the shallow
waterequationsare solvedby a semi-implicit,Eulerian-Lagrangian, finite-difference
method
(Casulli, 1990). The detailsof the Tidal,Residual,andInter-tidalMud-flat model(TRIM) are
reportedin Chengand Casulli (1991), and they are not repeatedhere.
The TRIM modelis usedin the computations of tidal and residualcurrentswith 0.5 km x
0.5 km finite-differencesquaremeshes. The numericalmethodusedin the TRIM model,
underlinearizedanalysis,hasbeenshownto be unconditionally stable(Casulli,1990). The
numerical integrationis carriedoutwith At = 15min.withoutanysignof instability(thegrid
Courantnumberis on the orderof 20). Thirteendroguesare followedfor five days,the
startingpositions
of thedrogues areshownin Figure2a. The Lagrangian
trajectories
of these
drogues,Figure 2b, show a complicated patternin the headlandregions. The drogue
trajectories
exhibitregular
patternsawayfromthecenterof thebasin.Thecomplex trajectory
patternis alsoanindicationthatthespreading
of theLagrangianresidual
currentcanbequite
large. This patternshowsa high degreeof nonlinearinteractions
betweenthe tide and the
currentin regions of highvelocityshear.In thisexample,thespatialvariations
of thevelocity
field (tidalcurrentshears)are inducedby theheadland constriction.A typicaltidalcurrent
distributionis depictedin Figure2c showinga pair of tidal vorticeslocatednearthe inlet
constriction (headland) in theinnerbasinduringflood.A pairof vorticesaredeveloped near
theheadland in theouterbasinduringebb,Figure2d. A slightantisymmetry canbeobserved
in theflow field,whichis caused by theCoriolisacceleration.
In regionswherethevortex
quadripoles developed, thetidalvelocityshearsarehigherthantherestof theregion. The
sizesof thevelocityeddiesareof thesameorderof magnitude asthecharacteristiclengthof
thebasin.Therefore, theweaklynonlinear approximation
is notvalidfortheheadland region
coveredby the vortexquadripoles.Figure 2e and 2f are the computedEulerianand
Lagrangianresidualcurrentdistributions.
Lagrangian Trajectories
Model Tidal Inlet (b) in A Model Tidal Inlet
Open Boundary
Followingtheprocedures
described
above,thedistribution
of thetidally-averaged
dispersion
coefficient,
% iscomputedanddepictedin Figure3. Thecomputed x)tranges
between
5x10
•
and5 x 10• cruZ'/sec.
Theregionin whichthevalues of x)tis greater
thanlx105cm•'/s
nearly
coincideswith the residualquadripoles.The meanLagrangianresidualcurrentpatternis
similarto theLagrangianresidualcurrentdistribution at a giventime. The sizesandshapes
of the dispersion patchesare measures of the varianceof the Lagrangianresidualcurrents.
In regionsnearthe constriction headland,the dispersionpatchesare largeandirregularand
the nonlinearinteractionsare significant.In thisregion,the weaklynonlinearapproximation
is violated,andthephysicalmeaningof theEulerianresidualcurrentis questionable, Imasato
(1983) . Likewise,the usageof the Stokes'drift and the masstransportvelocitybecome
questionablein regions of strong nonlinearity(Zimmerman,1979). Away from the
constriction,the nonlinearityis expectedto be weak. The computeddispersion patchesshow
neatlyclosedloci. In somecasesthe dispersion patchescan be approximated by ellipses
consistentwith the resultsof weakly nonlinearanalysis.
Tidally-Averaged
Dispersion
SxlO 6 cm2/sec
5xlO • cm2/sec
5xlO 4 cm2/sec
HONKER BAY
GRIZZLY
P E TA L UMA
RIVER
SAN PABLO BA
•UINEZ BAY
38 ø00' STRAIT SAN JOA Q UIN
RIVER
GOLDEN
) 10 20KILOMETERS
37ø30
CREEK
Figure4: SanFrancisco
Bay, a bathymetrically
complextidalestuary.
The inter-tidaldispersion
is resultedfrom the time dependency of theLagrangianresidual
currents,whosedistributionis principallya functionof thetidalcurrentshears.While thetidal
velocity shearsare stronglydependentupon the spatial variationsof the basin depth
distribution,the valuesof inter-tidaldispersion
coefficientsare alsocloselycorrelatedto the
variabilityof thebasinbathymelry.High valuesof inter-tidaldispersion coefficientarefound
in regionsof hightidalvelocityshears.Or highvaluesof theinter-tidaldispersion coefficients
canbeanticipated in regionswithcomplexgeometryandlargewaterdepthvariations.Central
Bay is just sucha region,thedispersion
coefficientin CentralBay is on the orderof -106
cm"/sec.
Thevelocityshears
in themiddleof deepchannels
in bothSouthandSanPablo
Bays,andin CarquinezStraitarenotnecessarily
high,thecomputed
valuesof thedispersion
coefficient
aresmallto moderate
(- 5x10
• cm'/sec).Similarly,
overtheregions
of broad
shoalswherethe weaklynonlinearapproximation
may be valid, the computeddispersion
coefficients
arelessthan10• cm"/sec.However,
at theedgeof thedeepchannels,
the
dispersioncoefficients
becomelocallylargebecausethe tidalvelocityshearis highin the
transition
zones,Figure5. The computed dispersion
coefficients
in thewesternSuisunBay
aregreater
than106cm2/sec.
Thecomplex
basingeometry
induces
highvelocity
shear,
and
thushightidal dispersion
for thatpartof SuisunBay
Tidally-Averaged
Dispersion
5X10 6 cm2/sec
5x10 s cm2/sec
5x10 4 cm•/sec
Figure5: Thecomputed
tidally-averaged
dispersion
coefficient
distribution
in SanFrancisco
Bay.
20.7 References
Cheng,R. T., and Casulli, V., 1991, A Tidal, Residual,Inter-tidalMud-flat (TRIM) model using
semi-implicit,Eulerian-Lagrangian
Method,to appear.
Fischer,H.B., E.J. List, R.C.Y. Koh, and N.H. Brooks,1979, Mixing in inlandand coastalwaters,
Academic Press.
Hamrick,J.M., 1986,Long-termdispersion
in unsteady
skewedfreesurfaceflow, Estuarine,Coastaland
Shelf Science,Vol. 23, p. 807-845.
Middleton,J. F., and J. W. Loder, 1989, Skew fluxesin polarizeAwave fields,J. of PhysicalOceano.,
Vol. 19, No.1, p. 68-76.
Ridderinkhof,
H. andZimmerman,J.T.F., 1990,Mixing processes
in a numericalmodelof the western
DutchWaddenSea,in ResidualCurrents
andLong-TermTransport, Ed.,RalphT. Cheng,Spring-Verlag,
p. 194-209.
21
Abstract
21.1 Introduction
The work describedin this paper was motivatedby my being invited to watch some
spectacular largescaleexperiments conductedby Guymer,Brockie& Allen (1990) in the
SERC flood channelfacility at HydraulicsResearchWallingford. The main concernof the
experimentswas to investigatethe lateral spreadingof neutrallybuoyantparticlesin a
meanderingchannel. However, some flow visualisationexperimentsindicatedthat the
directionof travelandrate of dilutioncouldbe markedlydifferentfor particleswhichare not
neutrallybuoyant. Indeed,it is intendedthat over the next 3 yearsa numberof research
groupswill utilize the SERC flood channelfacility to investigatea varietyof questions
involvingsediments and buoyantparticles. This paperis a theoretician's
attemptto predict
quantitative differences
in the transportanddilutionof bothrisingandsinkingparticles.
430 R Smith
(21.1)
with
with
(21.4)
wherethereference
levelZois chosen
sothatthedepthaveraged
valueI • I = 1 is correctly
reproduced.
Forairbubbles
of radius5 x 10'• m (.05mm)witha dirtysurface
therisevelocity
is 5 x
10'3ms'• (5 mms'•).Forspherical
particlesof sand
of thesamesizeandwitha density
twice
thatof water,thefall velocitywouldhavethesamevalue. If we specifythewaterdepthand
verticaldiffusivity
then the P6clet numbersfor the bubblesis P = 1, and that for the sandis P = -1. For bubbles
or particlesof othersizestheP6cletnumberscalesas the squareof theradius.
R Smith 431
• ' ß
u/ u
Figure 1: Velocity profilesof the longitudinaland cross-stream flow. The dottedcurvesshow the
verticalconcentrationprofiles•(z) for rising(P = 1) andfor sinking(P = -1) particles.
R+y
uv
R+y R+y
r 0•I• I * 0•(,0•u), (21.7)
R uO,,v
R+Y
+vOffR+yu2- -0•IIp
l +a,(va,
v) (21.8)
with
432 R Smith
and
_Rh• (21.12)
lul-
3(R +
with
R Smith 433
21.4 Weighted-averagevelocities
The verticalsamplingof the flow by the particlesleadsto a weightedaveragehorizontal
velocity
The explicit formulae (21.5, 21.11, 21.13) for 'y, u, v, enable us to evaluate the
dimensionless
proportionalityfactorsU,V:
5O4O
• {5 66840
v'=50-exp(-/9) pe+
+exp(-P)
[ 32144
1680 p /,2
+
(21.17)
At P = 0 thereare no singularities:
U- 1, V-0 (21.18)
21.5 Sheardispersion
Fischer(1969)showedexperimentally andtheoretically thatfor neutrallybuoyanttracersthe
effectivecross-stream
mixingratein a curvedchannelcouldbeupto 10 timeswhatonewould
expectin a straight
channel.Thesecondary flowcauses theaugmented cross-streammixing
by thesheardispersion
mechanism firstexplained by G.I. Taylor(1953).Recently, theauthor
(Smith 1991)hasderivedexpressions for the effectsof rise (or sinking)velocitieson the
horizontalsheardispersiontensor
434 R Smith
• 6-
• 4-
-• 2--
• -6
(21.19)
Daa=•_
•-• ¾(u-
•rllul)dz
t dz,
•'-•' •1f-,
ø •1 [f_•¾(v-Vv3dz.
tr& (21.20)
•(v- vo)dz
•]dz
(21.21)
It is comparatively
easyto evaluatethe innerintegrals:
(21.22)
•(u- Vlul)dz•-HI Vl
R Smith 435
(21.23)
where
I (o -
2( 1 1•
2+2p p2 • p2
+
2(• -cxp(-•) • •
(21.24)
{exp(P•)
I5
=33•
2+35•4-
7•
6+ 66• = 140•s+42•s
P
5(1-exp(-•) a •
(21.25)
In terms
ofthetwofunctions
I, • wedefine
thebulkflow,curvature
andmixed
coefficients
436 R Smith
as(p
) . 1 - exp(-P)
,p f-ioexp(-p•)12d• (21.26)
au(p
) . I - exp(-P)
P f_• oexp(-P•)/•
d• (21.28)
In principle
theseintegrals
canbeevaluated explicitly.
However,
forourpresempurposes
numericalvaluessuffice
(seefigure3). In theneutrally
buoyant
case(P = 0) theexplicit
values of the coefficients are
Withthedependence uponvertical
driftaccounted
forin theaB,aR,amcoefficients,
the
formulae
(21.19,20,21)fortheshear
dispersion
tensor
takethesimpleform
/•aah2 IIu12'as,D2
• h2•
aa
' o•2 /• h2
I uI 0 (21.30)
Fromfigure3 it isnoteworthy
thatsinking
particles
havegreater
longitudinal
butsmaller
transverse
shear
dispersion
thanrising
particles.
Thiscanbeexplained
intermsofthevelocity
profiles.
Thelongitudinal
velocityhaszeroshear
atthefreesurface
butthetransverse
velocity
also has zero shear near the bed.
21.6Gradual
separation
between
plumes
of neutral
andof rising
particles
In practice
theeffective
longitudinal
velocity
U I u I greatly
exceeds
theeffective
transverse
velocity
V• ofparticles.
Thisleads
togreatly
elongated
particle
distributions.
Consequently,
thestrongest
gradients
andstrongest
sheardispersion
is in thecross-stream
direction.The
corresponding
simplified
version
of thetwo-dimensional
sheardispersion
equation
is
R Smith 437
(21.31)
Here
cisthevertically
averaged
concentration
andI•c:]istheappropriate
weighted
averageof the transverse
turbulenteddydiffusivity.
0.00 0.01 0.02 0.03
I I I
10
4-
0-
-2-
-4-
-6
t
-8
-10
0.00 0.01
I
0.02
I
0.03
I
with
438 R Smith
x-x o
az' 211•. II + D22] I uI (21.33)
u'
To derivethissolution
we regardall theflowproperties
h, U, I u I V 0, I •nc2,
D22as
beingconstant
andwe neglecty/R.
If a neutrally
buoyant
(V = 0) andrising(orsinking)particles
arereleased
together
then
eventuallytheplumeswill become distinct.If we denotethe velocities
anddispersion
coefficients
for thetwospeciesby superscripts{o>,
{•>thenwecanestimate thenecessary
downstream
distancefor separation
otO)
' + oO)
•.
IA1)0
(x-Xo)
o•) l u I I (21.34
i.e. x-xo-2
(21.35)
Thisformulaequates
thesquare
of thecentroid
separation
to thesumof thevariances
of
theindividual
plumes.So,thespreadingandseparation
areof comparablesize.
Theformula
(21.30)for D22allowsusto assess
howthedistance
(21.35)for plume
separation
varieswiththestrength
• of thesecondary
flow. Thefunctional
formis
U•ø[37888
where X.(P) - 2
iAo•[
1126125 +aa(P)] (21.37)
Thus,
thedistance
forplumeseparation
decreasesrapidlyasthestrengthofthesecondary
flowincreases.
TheworkofFischer
(1969),asupdated byGuymet, Brockie & Allen(1990),
reveals
thatforneutrally
buoyantparticles
D22canbeupto 10times I •c2I in meandering
channels.Forneutral
andrising
particles
releasedtogether,
thedistance (21.36)forplume
separation
canbewithin10%of theasymptote (i.e.retaining
justtheXo.term).
R Smith 439
8-
6-
4-
2-
o-
--2 --
--4 --
--6 --
--8 --
Distance X. for separation
-10
0.0 0.5 1.0 1.5 2.0 2.5
Acknowledgements
I am mostgratefulto Ian Guymer,Nick Brockie& CathAllen for lettingme seewhatreally
goeson in turbulentflow roundbends.The typescript
waskindlypreparedby theDepartment
of Electronic& ElectricalEngineering.When thiswork was commenced the authorheld a
post-doctoralfellowshipfundedby the Royal Society.
21.7 References
440 R Smith
22
Abstract
22.1 Introduction
There have been several recent studies of the combined effects of waves and currents in the
bottomboundarylayerof shelfseas.Thesestudieshavebeenconcerned with the influence
of wave-current
interaction
on therateof energydissipation
at theseabed,on verticalmixing
processes
and,hence,on suchpractically importantmattersasstorm-surgeprediction,scour
aroundoffshorestructures,
and sedimentresuspension.
Oscillatory
boundary
layersarequalitatively
similarwhethertheyareinduced
by wavesor
tides. However,whereasturbulentwaveboundarylayersare thin [O(10-20cm)]andconfined
to thenear-bedlayer,tidalboundarylayersarethick[O(20-80m)]andmayoccupythe entire
442 A O Davies
22.2 Formulation
A G Davies 443
0u
•t _ 1 ap
0 •x + •0 (K•}
•u p
0== pressure
fluid density
boundary
conditions
bed
[ : u = 0 on z = z K = K(z,t) = Eddy Viscosity
0u
surface : K • = 0 on z = h K = K(turbulent energy
I distribution)
Turbulent Energy Equation
[ Generation
of• turbulent
• Diffusion • •k(z,t)Dissipation
energy (c)]
Scaling Laws
Mixing Length l(z,t) < >
K,c : k,1
boundary conditions
444 A G Davies
The vertical distributionof suspended sedimenthas been modelledon the basisof the
sedimentcontinuityequation(seeFigure 1). Solutionof thisequationrequiresa knowledge
of the sedimentsettlingvelocity (W.) and physicallyplausibleassumptions for both the
sedimentdiffusivity(e.) and the bottomboundarycondition. The resultsin SectionIll are
basedon threeassumptions.Firsfly, it has beenassumedthat the presenceof suspended
sedimentdoesnot aff•t the distributionof t.k.e in the flow; while thismightbe reasonable
for low sedimentconcentrations, it is not a generallyvalid assumption (Soulsbyand
Wainwright(1987)).Secondly,thesedimentdiffusivity(e.) hasbeenassumed to be equalto
A G Davies 445
the eddy viscosity (K). Again, this is probably an oversimplification(see Dyer and
Soulsby(1988)).Thirdly,andmost importantly,a knowledgeof the "reference concentration"
at the bed hasbeenassumed.This form for the bottomboundaryconditionhasbeenadopted
mainlyfor simplicity. In particular,it hasbeenassumedthat the referenceconcentration at
level Z=Zocan be prescribexl as a functionof the instantaneous bed shear stressin the
wave-current cycle. At the freewatersurface,the conditionof zero sedimentflux is applied.
The systemof equationsin Figure 1 has been solved in two stages,by means of
numericallyconservative,finite differencemethodson a log-linearverticalgrid. Initially, the
required hydrodynamicalsolution has been obtained for wave, current and combined
wave-currentflows. The respectivevelocityand eddyviscositydistributions, determinedas
functionsof heightabovethe bed andphaseanglein the wave cycle,havethenbeenusedin
the solutionof the sedimentcontinuityequation.The resultspresentedbelowrelateto a case
studycarriedout with the followingparametersettings.The bed roughness lengthscale has
beentakenas Zo= 0.5cm,corresponding to sandtippleshavingthe approximatedimensions:
height--5cm, wavelength=50cm. The water depthhasbeentakenas 10m.
(i) Steadycurrent
As expected,the verticalprofileof horizontal(mean)velocity(U(z)) is logarithmicin the
lowerpartof the flow (Fig. 2i), andthe (mean)shearstressincreaseslinearlyfrom zeroat
thefree-surface
toitsmaximum
valueof•(Zo)=8.72dyn/cm:
(u.=2.95cm/s)
atthebed(Fig.2ii).
446 A G Davies
L• •vels
1000.0 -
500.0
.•, 100.0
• 50.0
N
• 10.0
•- 5.0
1.0
0.5 ß
Figure 2i: Vertical profiles of horizontalvelocityU(z) for the steadycurrentand <u(z,t)> for the
combinedwave-current(W+C) flow. (Angle brackets<> denotetime-averaging.)The modelgrid is
shown on the inner vertical axis.
Levels
ooo
8oo
600
400
20O
0
0 2 4 6 8 10 12 14
(Me(:]n)SheerStress (dyne/cm2)
Figure 2ii: Vertical profilesof shearstressx(z) for the currentand <x(z,t)> for the combinedwave-
current flow.
A G Davies 447
Levels
1000.0 -
500.0
• 100.0-
E
• 50.0
N c
._o, 1o.o
T 5.0 ß
Current
1.0
0.5
ß' ' 5'0 •60 •50
(Meen)T.K.E.
Levels
1 ooo
8OO
W+C
Current
400
200 Waves
448 A G Davies
The associateddistributionsof t.k.e. (k) (Fig. 2iii) and mixing length(not shown)give rise to
an essentially
parabolicdistributionof eddyviscosity(K) (Fig. 2iv).
(ii) Waves
The verticalprofile of the amplitudeof wave velocity(Fig. 2v) hasthe familiarstructureof
a free-streamregion(z > 20cm), an "overshooting" region(5 < z < 20cm),anda logarithmic
near-bedregion. The associated cycle-averageddistributionsof t.k.e. andeddyviscosityfor
thewavesare shownin Figures2iii and2iv. The solutionconvergedafter60 cycles(10 min
real time).
Levels
1000.0 -
500.0
W(]ves
• 100.0 -
E
o 50.0
o, 10.0
-r 5.0 ,.
1.0
0.5
0 ' 1'0 ' 2'0 ' 3'0 ' 4'0 ' 5'0
VelocityAmplitude (cm/s)
A G Davies 449
Extrapolation
of thestraight-line,
logarithmic segmentof thewave-current profile(15< z
< 200cm)in Figure2i indicates
an"apparent roughness"
forthewave-current boundary layer
of 2.5cm.Thisis 5 timesthephysical roughness lengthscale
(z0). In addition,
theslopeof
thestraight-line
segment suggests
an"apparent meanu."valueof 3.97cm/s. Thisissomewhat
largerthanthetruevalueof 3.74cm/s(i.e.thevaluecalculated
fromthemeanstress atthebed
level).Thisdiscrepancyisexplainedbythenon-logarithmicformof themeanvelocity profile
in the upperpart of the flow.
(i) Steadycurrent
Verticalprofries
of suspendedsedimentconcentration
(c) areshownin Figure3i for three
grainsizeshavingthesettling
velocities
W, = 0.1,0.25and1.0cm/s.Eachprofilehasbeen
normalizedby thereference
concentration
c---Co•
atlevelz=z0.No attempt
ismadetoevaluate
Co•in whatfollows;Co•is simplytakenasthevalueof reference
concentration
appropriate
to u. = 2.95cm/sfor therespective
grainsizes.Thustheproblemis reduced
to theprediction
of the verticaldistributionof concentration
relativeto Co•.
Levels
1 000 -
800
600
400
s=O.1 cm/s
200
O I ß I ß I ß I ß I '
0.0 0.2 0.4 0.6 0.8
Concentrotion C/Coo
450 A G Davies
Levels
000 ,
800
600
4-00
Ws=O.
1cm/s
200 Ws=0.25
0
o .... .... .... 15' .....
Flux UC/Co (cm/s)
Assuming thatthehorizontal
velocitycomponentof thesediment
grains
isequaltothatof
thefluid,thesedimentfluxateachlevel(normalized
bythereferenceconcentration
Co,)is
U(z)C(z)/Co,.Vertical
profiles
of thisquantity
areshown inFigure
3iiforthethreesettling
velocities
W, = 0.1,0.25and1.0cm/s.Thelevelsof maximum
sediment
fluxare445,41 and
1.8cmin therespective
cases.Valuesfor thedepth-averaged
concentration
andsuspended
sedimenttransport
ratefor steadyflow arediscussed
in Section22.4.
A G Davies 451
whereXo•= 8.72dyn/cm
•'is thevalueof bedshearstress
for thesteady
current.It shouldbc
noted,firsfly, that C(Zo,Ohas been taken as a power of the magnitudeof the normalized
instantaneous bed shearstressand, secondly,that no sedimentthresholdmotioncriterionhas
beenincludedin the formulation.Variousexpressions for c(zo,0havebeen suggested by
previousworkers,incorporating differentpowerlaw behaviours,for exampleN = 1/2 (Shi,
1985) andN = 1 (HagatariandEidsvik,1986). Sincehighervaluesthanthesein the range
N = 3/2 to 3 may be inferredfrom resultsfor the cycle-averagedreferenceconcentration(see
Dyer andSoulsby(1988) andNielsen(1988)), thisremainsan areaof considerable uncertainty.
The sensitivityof thepresentmodelresultsto changesin N is discussed in Section22.5. The
omissionof a thresholdshearstressfrom the expression for C(Zo,0takessomeaccountboth
of the persistentnatureof near-beAsedimentmotionin oscillatoryflow and of the effectof
settlinggrains.
Levels
1000-
800
600
400
L.x•Ws=1.0
2OO J Ws=0.25 .
452 A G Davies
(iii) CombinedWave-CurrentFlow
As for the caseof wavesalone,theeddyviscositygenerated by thehydrodynamical
modelhas
beenusedin the solutionof the sedimentcontinuityequation.The referenceconcentration
has
beencalculatedin the way described earlierandan initial valueproblemhasbeensolvedfor
c(z,t). Convergedsolutionswere obtainedafter 1400, 800 and 200 cycles(233, 133 and 33
minutesreal time) in the casesWs = 0.1, 0.25 and 1.0 cm/s, respectively. Each of the
cycle-averaged profilesin Fig 5i takesthe value <c>/C0,= 5.02 at the bed level z=z0(as
comparedwith the value4.83 for wavesalone). The concentration distributionsfor the case
of combinedwave-current flow maybe compareddirectlywith theresultsfor thecurrentalone
in Figure 3i. An enhancementin the suspended sedimentconcentration is evidentfor each
value of settlingvelocityconsidered.This arisesboth on accountof the increasein mean
bottomconcentration, and also on accountof the enhanceddiffusivity.
u = tJ+u,,,+u
t andc = C+c,,,+ct (22.2)
<uc>= UC+<u,,,c,,,>
+<u,c? (22.3)
A G Davies 453
800
0.1 cm/s
600
Ws=0.25
400
200
800
600
400
200
454 A G Davies
Levels
6O
5O
'•' 40 Ws=1.0
N ,30
• 20
lO
o ß
o io'4'o''o do
Mean Flux: <u> <c>/Co•; ...... <uc>/Co• (cm/s)
Figure5iii: Vertical profilesof cycle-averagedsedimentflux for the lower part of the wave-current
flow, for threevaluesof settlingvelocityW,. Profilesbasedupon<u><c> (full line) and<uc> (dotted
line) are shown in each case.
Levels
35-
3O
._.25
E
'•20
N
m15
•10
0
o 2 4 6 8 10 12
Mean Flux'
<u><c>/Co• ...... <uc>/Co• (cm/s)
Figure5iv: Verticalprofiles
of cycle-averaged
sediment
fluxasin Fig.5iii, butfora largervalueof
A G Davies 455
22.4 Depth-averagedconcentration
and sedimentflux
22.4.1 Notation
AverageconcentrationC =
1 f••C(z)dz,
h-zo
(22.5)
$uspendeA
scdim•t flux q, = C(z) lJ(z)dz.
q, = Co•(h-z
o)I, (22.6)
whereU is thedepth-averagedvelocity.Heretheflux is dividedintotwoparts:(i) a "stirring
term"givenby the referenceconcentration
Co,,whichis assumed to dependonly on the bed
shearstress
andproperties
of thegrains;
and(fi_.)
a "transporting
term"comprising
theproduct
of the volumetricflow rate per unit width U(h-Zo)and the shapefunctionI½. The later
calculationsconcernonly the transporting
term.
456 A G Davies
For wave-current
flow,depth-averages
aretakenaftercycle-averaging
asfollows:
Average
concentration
• - 1 f•'<c(z,t)>dz(22.7)
Thesuspended
sediment
fluxis calculated
onthebasisof thetwoapproaches
discussed
abovefrom the expressions:
(22.8)
q,•=f•'<u><c>dz
=f•:U(z)C(z)dz
(22.9)
whereCorn
= <C(Zo,t)>is themeanoveran integral
numberof waveperiods of thereference
concentration
at Z=Zo.[It should
benotedthatCorn
andCo,will, in general,
bedifferent.]The
quantity_V
is thecycle-averaged,
depth-averaged
velocity
in wave-current
flow,suchthatthe
product
V(h-zo)is thevolumetric
flowrateperwidth. Sincethewavesaresuperimposexl
on
thecurrent
co-linearly,
•/' hasthesamedirection
asO.
22.4.2 Results
Calculationsof theabovedepth-averagedquantities
havebeenmadefor a rangeof valuesof
settlingvelocityW,. The resultstabulated
belowfor the current,waveandwave-current
casesare baseduponthe flow solutions
discussed earlier.
(i) Steadycurrent
The resultsfor depth-averaged
concentration
C and transportrateper unit widthq, (both
normalizedby thereferenceconcentration
Co,),together
withtheshapefunctionI,, arelisted
in Table 1. As expected,
eachquantitydecreasesrapidlywith increasing
W•.
Further
modelrunshavebeenperformed
withtheroughness
todepth
ratioZo•=10
-4. In
Figure6 theresults
obtained
forI, areplottedagainst
theparameter
S, = Ci)U/3W,,wherethe
dragcoefficient
Ci)= X(Zo)/pU:,
onthegraphof DyerandSoulsby(1988).
Theresults
relate
to a singlesettlingvelocityW,, anda singleroughness lengthscale
Zo(i.e. thesimultaneous
presence of differentbedformtypesis not permitted).Thevarious
curvesplottedin theinset
are for differentfunctionalformsfor theeddyviscosity,andhencesedimentdiffusivity(with
A G Davies 457
w I
q½/c Oc
(cm/s) (cm2/s)
-! -!
0.1 5'22x10 2 ß54x 104 5'09x10
-!
0.25 2.06x10 -• 9 ß55x 103 l'91xlO
-2 -2
0-5 5'12x10 2' 14xlO 3 4'29x10
-2 -2
0.75 1'54x10 5' 57x102 l'12xlO
-3 -3
1.0 5'73x10 1' 69x102 3'38x10
-3
1.5 1'47x10 2 ß47x101 4.95x10 -•
2.0 6.77x10 -4 6' 39x10 ø 1' 28x10 -•
-5
3.0 3 ß03xlO -4 1' 26x10 ø 2-52x10
-1 -5
4.0 1.91xlO -4 4.46x10 8'94x10
-2 -6
6.0 1' 07x 10 -4 7' 39x10 1'48x10
458 A G Davies
Sc: CoO/3Ws
001 0.1
1.0
XPresent model
10-I _ P
FC
10-• _
Izo/h
--10
-•'I L
Ic
0'•
$m
F
z/h
ø'6 FHFC L
10-•_ 0•
02
•o-6 _
0. 0 O0
[ I
0.1 02 K/•cu.h
03
I
04 05 0.6
Figure6: Shapefunction
I• versus
Scforsteady
Howwithz0/h=10
-4. Thelabelled
curves
correspond
to the vertical profilesof eddy viscosityshownin the inset, as explainedin detail by Dyer and
Soulsby(1988) andSoulsby(1988) (•: = Von Karman'sconstant).Thepresentmodelresultsareshown
by the crosses.
A G Davies 459
Table2: Cycle-averaged,depth-averaged
concentration
results
obtained
withI• = 1 andN = 1 forboth
wavesonlyandcombined wave-current
(W+C)flow. Flowparameters:z0= 0.Scm,h = 1000cm;mean
velocityV = •'• = 50 cm/s;wavevelocityamplitude
• = 50 cm/s,period= 10s.
<C>/ <C>ratios
Ws Co½
(sin/s)
Waves W + C W + C W + C
Waves Current
460 A G Davies
Ws q+•/Co
c q+x//qo I +1 q+
2/coc q+
2/ /qo I+2 I +2I+1
(cra/s) (cra
2/ s) (cra
2/ s)
O'1 1' 65x105 6 ß5 6' 59x10-1 1' 65x105 6 ß5 6' 59x10-1 1' O0
0'25 8.15x104 8'5 3'26x10-1 8.16x104 8'5 3'26x10-1 1'00
0'5 2-70'104 12'6 l'08xlO -1 2.70x104 12'6 l'08xlO -1 1'00
0'75 9.91x103 17-8 3'96x10-2 9-94x103 17'8 3'97x10-2 1-00
1' 0 4 ßOlxlO3 23 ß8 1' 60xlO-2 4 ß04xlO3 23 ß9 1' 61xlO-2 1' O1
1'5 9-13x102 36'9 3'65x10-3 9-42x102 38'1 3'76x10-3 1'03
2'0 2'45x102 38'3 9'77x10-4 2.73x102 42'7 1'09'10 -3 1'12
3'0 4.72x10 ! 37'5 1'88x10-4 7.44x10 ! 59'1 2'97x10-4 1'58
4 ß0 1' 74x101 39 ß1 6' 96x10-5 4 ß04xlO1 90' 5 1' 61xlO-4 2 ß32
6'0 4'87x10ø 65'9 1'94x10-5 1.74x101 235 6'94x10-5 3'57
8' 0 2 ßOlxlOø - 8' 03xlO-6 8' 35x10ø - 3' 33x10-5 4' 15
12ß0 5 ß23x10-1 - 2' 09xlO-6 2 ß51xlOø - 1' OOxlO-5 4 ß81
16'0 1.71xlO -! - 6.83x10-7 8.90x10-! - 3-55x10-6 5'20
A G Davies 461
zo/h
=10
-4,
(z+=
5,/•+=1001
10-z x Presentmodet
TS2
1-0
Eddy viscosity
10-4
C J2
0'8
I+ I+2
10-6
0-6 /•"GM
z/h
I+1 0-4
10-8
0.0
0.0 0-2 0.4 0.6 0-8 1'0 1-2
K/•<u,>h
Figure7: ShapefunctionI+versusS+forwave-current
flowwithz0•=10-4,ct+=5
andI]+=100.The
labelled
curves
correspondtothe(time-invariant)
vertical
profiles
ofeddyviscosity
shown intheinset,
asexplained
in detailbySoulsby(1988)
(K= VonKarman'sconstant).
Thepresentmodel results
for
I+l and1+2areshownby thecrosses.
22.5 Discussion
Theresults
inSections
22.3and22.4wereobtained
withthesediment
diffusivity
setequal
to
theeddyviscosity
(B=1) andwithN=1 in theformulafor thereference
concentration.
These
assumptions
are now discussed.
22.5.1Effectof changing15
Forsteady
flowthesediment
continuity
equation
(seeFigure1) reduces
to
462 A G Davies
de
% dz + WsC' 0 (22.11)
(22.12)
dz + - o
A G Davies 463
Table 4: Mean sedimenttransportrate (q+per unit width) and shapefactor(I+) for combinedwave-
currentflow. Values for power N in the referenceconcentrationformula are as indicated. Flow
parametersare as for Table 2.
Ws q+ (cm
2/ s) q+2 (cm2
/ s)
(ore/s)
•/co• /c o•
N=l/2 N=i N=3/2 N=l/2 N=i N=3/2
-1 -1 - 1 -1 - 1 -1
0' 25 3 ß 13x10 3 ß 26x10 3 ß 37x10 3 ß 13x10 3-26x10 3 ß 37x10
464 A G Davies
C(•o,t) Co•]X(•o,t-
At)lxo• (22.13)
A numberof runs have been conductedto test the sensitivityof the model resultsto
variationsin At. For a lag At corresponding to 36ø of phase(such that the reference
concentration is roughlyin phasewith the outerfree-streamvelocity),the depth-averaged
concentration for wavesalone increasesby 12% for W,=lcm/s and by 3% for W,--4cm/s.
Theseresultswereobtainedwith N--l; percentage changeswereslightlylesswith N=l/2, and
slightlygreaterwith N=3/2. For the caseof wave-current flow with W•=4cm/sandN=I, the
samephaselag givesrise to a 5% increasein <c>/Co,,a 12% increasein I+• and an 11%
decreasein I+•_.A morecompleteaccountof thesecomparisons will be givenelsewhere.This
is a topicwhichwarrantscarefulinvestigation,not leastin connectionwith vortexformation
andsheddingaboverippledbeds,whichis knownto be importantin the processof sediment
entrainment(Nielsen, 1988).
22.6 Conclusions
A G Davies 465
The present results are dependent upon the assumed relationship between the
(instantaneous)bed shearstressand the (instantaneous) bottomreferenceconcentration.The
powerlaw relationship betweenthesequantifies exertsa significant
influenceon the transport
rate. Thisarisesprimarilythroughchanges in thereferenceconcentration,andonly to a small
extentthroughthe shapefunction. The phaserelationship betweenthe two quantifiesalso
affectsthe results,not leastin respectof the "wave-related"contributionto the flux. This is
a matterwhich warrantsfurtherinvestigation.
Acknowledgements
RichardSoulsby(HydraulicsResearch Ltd., Wallingford)is thankedfor hishelpfulcomments,
and also for allowinguseto be madeof Figures6 and 7. Bob Everitt and AngelaWilliams
are thankedfor their help with the figuresand the text. Financialsupportfor the work was
providedby the U.K. Ministryof Agriculture,FisheriesandFood.
22.7 References
Ballard, J.A., 1981: An energeticstotal load sedimenttransportmodel for a plane sloping beach.
Journalof GeophysicalResearch,86, Cll, 10938-10954.
466 A G Davies
Grass,AJ., 1981:Sedimenttransport
by wavesandcurrents.
ReportNo. FL29, SERCLondon Centre
for Marine Technology,26pp.
Nezu,I., andW.Rodi,1986:Open-channel
flowmeasurements
witha laserDoppleranemometer.
Journal
of HydraulicEngineering,112, 5, 335-355.
Nielsen,P., 1988:Threesimplemodels
of wavesediment
transport.
Coastal
Engineering.
12,43-62.
Soulsby,
R.L.,1987:Therelative
contributions
of waves
andtidalcurrents
tomarine
sediment
transport.
HydraulicsResearchLtd., Wallingford,ReportSR 125.
Soulsby,
R.L., andB.L.S.A.Wainwright,1987:A criterionfor theeffectof suspended
sediment
on
near-bottom
velocityprofiles.Journalof HydraulicResearch,25, 3, 341-356.
23
Abstract
Theconstructionandapplication
of a mathematical
modelformorphologicaldevelopment
in
tidalregions
isconsidered.
Themodel consists
ofa flowmodule,a sediment
transport
module
anda bedlevelchangemodule.Thestructure
of themodelandthechoiceof theflowmodule
andsediment transport
moduleare discussed. The application
aspects
of the modelare
consideredwith a casestudyfor the Yantze estuary.
23.1 Introduction
Themorphologicaldevelopmentof estuarine
andcoastalregions isa longtermprocess.The
timescaleof morphological
processes mayrangefromweeksto decades or evencenturies.
Morphologicaldevelopment
changes thegeometryandbathymetry of a regionandthereby
it
influences
theflowpattern
andallotherphysicalprocessesin theregion.Thereforeknowledge
aboutthemorphological
development is important
for hydraulic engineering aswellasfor
studying
longtermvariability
of otherphysicalprocesses in esmarine andcoastalregions.
Morphologicaldevelopment ofesmarine
andcoastalregions isa verycomplicatedprocess.
Manyfactors
areinvolved in thisprocess:
fiverflow,fide,density flow,windandwaves,
sediment
properties,etc.Although considerable
knowledge existson all theseindividual
elements,
very little is knownhow to includetheminto a mathematical modelfor
morphological
development. Mathematical
modellingof thisprocessisstillinanearlystage.
At DelftHydraulics
themorphological
modelDELMORis underdevelopment.
At this
momentit is not wiseto try to developa generalapplicable
modelby includingall the
significant
factors
onecanimagine.
Thiswill leadto a verycomplicated
model.Littleis
knownaboutthebehaviour
of sucha complicated
model.Thereforethestepby stepstrategy
468 Z B Wang
is followedfor the developmentof DELMOR. In the first stepthe modelwill only include
fiver flow and tidal motionas drivingforcesfor sedimenttransport.Densityflow, wind and
wavesare not takeninto accountfor the time being.Furtherit is restrictedto non-cohesive
sediment.The fundamentalaspectsof sucha simplifiedmodelwerestudiedby Wang (1989),
who developedthe researchmodelESMOR. Someconsiderations on the consauctionand
applicationof sucha relativesimplemorphological modelaremadein thispaper.In thefuture
when the behaviourof such a simplermodel is sufficientlyunderstoodthe model will be
extendedstepby stepby taking more factorsinto account.
The morphological
developmentinvolvesthevariations
of flow, sediment
transport
andbed
level.Thesethreesubprocesses
all interactwith eachother.This meansthat the equations
describing
theflowfieldvariables
(h,u,v....), thesediment
transport
variables
(c,s•,Sy...)
and
thebexllevel(zO arecoupledwitheachother.ff thesecoupledequationsaresolvedtogether
thetimestepusedshouldbe muchsmallerthanthesmallestsignificanttimescale,i.e. thetidal
period. Obviouslysucha computationfor a long period (decadesor even centuries)is
impossible.
In otherwords,simplifications
in morphological
modelsarevirtuallyinevitable.
Z B Wang 469
LINITIAL
STATE
CURRENT
FIELD
ONE
INTIDAL PERIOD
VELOCITY FIELD :•
CONCENTRATION
FIELD •
TRANSPORT
FIELD nuj
o•
UPDATE
TOTAL
TRANSPORT •
I BED
LEVEL
470 Z B Wang
The Flow Model: In the caseof esmarineand coastalproblemsone can choosefrom two
possibilitiesfor the flow model, viz. the two-dimensional(horizontal) and the three-
dimensional model.At present,computercapacityrestrictsaffordablelongtermmorphological
computations basedon a fully 3D flow model.Thereforethe quasi-3Dapproachis chosenin
ESMOR aswell asin DELMOR. The depthaveragedflow is computedwith a 2DH tidalflow
model.A profile modelis appliedto describethe verticaldistributionof the velocityin the
longitudinaldirectionanda secondary flow modelis appliedin orderto takeinto accountthe
velocity componentin the normal direction.In DELMOR, the depth averagedflow is
computedwith TRISULA. The secondary flow modeltakesinto accountthe influenceof the
curvatureof the streamlines of the depthaveragedflow andthe geostrophic acceleration(de
Vriend, 1981, Kalkwijk and Booij, 1986). It shouldbe mentionedthat TRISULA is a multi-
dimensional flow modeldevelopedat Delft Hydraulics.which is able to performfully 3D
computations. This makesit easierto extendDELMOR to a full 3D modelin the futureif
necessary.
471
Z B Wang
in which
c = depthaveraged
sediment
concentration;
u, v = depthaveraged
velocitycomponents;
T. = adaptationtime;
¾ = coefficients;
c. = depthaveraged
equilibrium
sediment
concentration.
Aftersolvingthismeanconcentration
equation
the sediment
transport
ratescanbe
calculated.
Detailedinfomarionaboutthismodelis givenin Galappatti(1983) andWang
(1989).
Remark:Equation(23.1)applies
forthecase
of sand.
In caseofcohesive
sediment
thefight
handsideof theequationshouldbereplaced
bythesedimentation
anderosion
rate(Krone,
1962,Mehtaand Partheniades,
1982).
Extensiveanalyses,
theoretically
aswellasexperimentally,
onthevalidityof thismodel
wascarriedout(WangandRibberink, 1986,Wang,1989,1990).Fortunately,
theanalyses
showthatthisquasi-3Dmodel is sufficiently
accurate
toreplace
thefull3D modelin most
practical
problemsin estuarine
andcoastalregions.
TheBedLevelModel:The bedlevelmodelis themostsimpleoneof the threesub-models.
Thebedlevelchange
is described
by themass-balance
equation
for sediment.
(23.2)
Thisequation
isexactandhastobesolved
withthelargetimescale
soit should
be
integrated
overthe tidalperiod.
+ + Y -o (23.3)
In this equation
472 Z B Wang
Tx,Ty=totalsediment
transport
perunitofwidthduring
a tidalperiod.
TheYantze
riverwitha length
of6380kmanda catchment
areaof 1.8million
km2isthe
largest
fiverin China.It flowsintotheEastChinaSeanorthof Shanghai. Its estuary
covers
anareaof about100x100km2.Unfortunately duringthecasestudyit wasimpossibleto take
thewholeestuary in themodel,becauseof therestricted
computer
capacityaswellaslackof
fielddata.However,themostinteresting
partof theestuary,i.e. thesouthbranch,wherealso
themostfieldmeasurements havetakenplace,is modelled.
Themodelcoversanareaof about
80x20km2 (seefig.2).With a spacestepof 1 km in bothhorizontal directions
themodel
containsabout1000 activecomputational
grid points.
473
ZB Wang
The morphological
time stepAttoandthe numberof continuitycorrections M, are model
parameters
only.Theydonothaveanyphysical meaning andtheyhavetobechosen suchthat
the computational
effort requiredis as smallas possiblewhile sufficientaccuracyof the
computations
isensured.
Forthispurpose a sensitivity
analysis
wascarriedout.In thisanalysis
a numberof runshavebeencarriedout with differentvaluesof Attoand M•. The computed
474 Z B Wang
TRNT?E
ESTUR•T
Figure
3: Netsediment
transport
during
spring
tide(top),averaged
tide(middle)
andneaptide(bottom)
475
Z B Wang
ii
I (unit = m)
;• • above2
':'" r'---1 i - 2
i -1 - 0
• I -2 - -1
: :• -2
Figure
4: Bedlevelchange
during
spring
tide(top),averaged
tide(middle)
andneaptide(bottom)
An interesting
resultwasthatwithM•=3thecomputational
effortusedfor the flow
computation
andthatfor the sediment
transport
computation
are aboutequal.Thisis
remarkable
asin thesediment
transport
modelonedifferential
equation
(i.e.theconcentration
equation)
hasto besolvedwhilein theflowcomputation 3 differential
equationshavetobe
solved.
Fora totallydifferent
case,(steady flowin a meandering
fiversection
(Waalin the
Netherlands)
withonlybedloadtransport),it alsoappeared
thatthemostefficient
computation
is obtainedwhen the effort for flow computation
and that for the sedimenttransport
computation
areabout equal. It seems
tobethatin a wellconstructed
morphological
model
thecomputational
effortforflowandforthesediment transport
shouldbeof thesameorder
of magnitude.
Afterthedetermination
,ofthemodelparameters
somesensitivity
analyses
onthephysical
parameters,
e.g.thefallvelocity
andthebedboundary
condition
etc.,werecarried
out.Next,
themodelwascalibratedanda 5 yearsimulation
wascarriedout.Detailedinformation
about
thissimulation
is givenin Wang(1989).Hereonlysomeresultsof the5 yearsimulationare
illustratedin fig.6 andfig.7.
476
Z B Wang
Cross section S
10
I J..
-10
I 2 3 4 5 6 7 I 2 3 4 5 6 7
day In measurlno period day In measuring period
JI calc.
ebb
-•
transp
• calc.
flood
transp
meas ebb transp • meas flood transp
I 2 3 4 5 b 7
day In measuring period
Figure5: Totalebb-andfloodtransport
through
thethreecross
sections
(unit = m)
above 0
-4 - 0
-8 - -4
L_ -12
-16
-20
-
-
-8
-12
-16
-24 - -20
-28 - -24
-28
Figure
6: Theinitialbedlevel(top)andthecalculated
bedlevelafter5 year
Z B Wang 477
478 Z B Wang
•}C C•-C
(23.4)
23.6 Conclusions
Z B Wang 479
to savecomputational
costs.
The choiceof the morphological time step and the numberof times of continuity
corrections were investigated with a sensitivityanalysis.The conclusionis that a large
morphological time stepis allowedbecauseof usingthe quasi-3Dmodelfor the sediment
transport.
It is alsoconcluded thatthecontinuitycorrection
methodis a usefultool for saving
computationalcost.
Further, the numerical schematizationof the source term in the sediment concentration
equationand the upstreamboundaryconditionhavebeendiscussed. It is recommended to
discretisethe sourceterm implicitly.For the caseof lackinginformationat the upstream
boundarya solutionfor the boundaryconditionis presented.
23.7 Acknowledgement
The mainpartof the workdescribed in thispaperis carriedoutduringthepreparation of the
doctoralthesisat the Delft Universityof Technology.The writer wishesto expresshis
appreciation
of the adviceandencouragement givenby the supervisorsprof. dr. M. de Vries
andprof.dr.C.B. Vreugdenhil.Thanksaredueto Mr. J.C.Winterwerp for hisencouragement
andconstructivecomments duringthe preparationof thispaper.
23.8 References
480 Z B Wang
Galappatti,
R.andVreugdenhil,
C.B.(1985),
A depth-integrated
modelforsuspended
sediment
transport,
J. of Hydr. Res., Vol. 23, no. 4.
Hauguel,
A. (1978),Utilization
desmodeles
mathematiques
pourL' etude
dutransport
solide
sous
1'
actiondestourantsde Maree,ReportFA2/'78.41,
EDF, DirectiondesEtudeset Recherches.
Kalkwijk,
J.P.Th.
andBooij,R. (1986),Adaptation
of secondary
flowin nearlyhorizontal
flow,J. of
Hydr. Res., Vol. 24, no. 1.
Krone,
R.B.(1962),
Flume studies
of transport
of sediment
in esmarial
environments,
Finalreport,
University
ofCalifornia,
Hydraulic
Engineering
andSanitary
Engineering
Laboratory,
Berkeley.
Mehta,A.J.and
Partheniades,
E.(1982),
Resuspension
ofdeposited
cohesive
sediment
beds,Proceedings
oftheXVIII Coastal
Engineering
Conference,
ASCE,
Vol.2,Cape
Town,
SouthAfrica,
pp1569_1588.
Rijn,L.C.van(1987),Mathematical
modelling
of morphological
processes
in thecaseof suspended
sediment
transport,
Doctoral
Thesis,
DelftUniversity
of Technology
Vriend,
H.J.de(1981),
Steady
flowin shallow
channel
bends,
Doctoral
thesis,
DelftUniversity
of
Technology,Delft, The Netherlands.
Vriend,
H.J.de(1985),
Flowformulation
in mathematical
modelsfor2Dhmorphological
Changes,
Reportno.R1747-5,Delft Hydr. Lab.,Delft, The Netherlands.
Vries,
M.de(1959),
Transients
inbed-load
transport
(Basic
considerations),
Report
no.R3,DelftHydr.
Lab., Delft, The Netherlands.
Vries,
M. de(1965),
Considerations
about
non-steady
bedloadtransport
inopen
channels,
Publ.no.36,
Delft Hydr. Lab., Delft, The Netherlands.
Wang,Z.B.(1984),
Thevalidity ofadepth-integrated
modelforsuspended
sediment
transport
andthe
extension
ofthismodel
totidalrivers,
Report
no.10-84,
DelftUniversity
ofTechnology,
Dept.
ofCivil
Eng.
Wang,
Z.B.andRibberink,
J.S.(1986),
Thevalidity
ofadepth
integrated
model
forsuspended
sediment
transport,J. of Hydr. Res.,Vol. 25, no.1.
Wang,
Z.B.(1989),
Mathematical
modelling
ofmorphological
processes
in estuaries,
Doctoral
thesis,
Delft Universityof Technology.
Wang,
Z.B.(1990),
Theoretical
analysis
ondepth-integrated
modelling
ofsuspended
sediment
transport,
papersubmitted
to J. of Hydr. Res.
24
Abstract
24.1 Introduction
In thepresentthree-dimensionaltransportmodel,thenecessary 3D deterministic
flow field
is derivedby extendinga computedverticallyintegratedvelocityfield into threedimensions
by meansof simpleanalyticalfunctions.The analyticalvelocityprofriesare steadyandapart
from additionaldistortions
by wind or densitygradientstheyare unidirectional,sotheydo not
accountfor phasevariationsand skewedsurfaceandbottomEkmanlayers(Hamrick1986).
Thisprovisionalsimplificationseemsto bejustifiedby certainresultssuchasthoseof section
23.3.1 and it alsomakespossibleto investigatethe sensitivityof the dispersion
processes to
variousparameters. Neverthelessit is intendedto evaluatethe improvements to be obtained
by quasi-3Dor "2•/D" approaches (Prandle1982, Van der Giessenen Jansen1988, Davies
1990)for obtainingtherequired3D velocityfields.Finally,theidealway to derivethesefields
would be by completethree-dimensional hydrodynamicnumericalmodelling which has
becomefeasiblein recentyears.Thesevariousalternativesfor obtainingbetter 3D velocity
structureswould leave the structureof the transportmodule unchanged:discreteparticle
approach,displacementin three directionsat each time step including supplementary
displacements which accountfor vertical exchangeand for spectrallystructuredhorizontal
"sub-grid"velocities.
24.2.2. Supplementalhorizontalvelocities
The analyticalexpressions
for the verticalvelocitydistributions
werechosenas simpleas
possiblei.e. Van Veen distributions(Van Veen 1936, 1938) for undistortedverticals
(neglectingphasedifferences):
(24.1)
(V} = givenvertically
averaged
velocity;
H = waterdepth;
ot= shape
parameter)
andthesum
of two expressions
of the sameform to accountfor possibledistortions
vd (z) by wind or
densitycurrents:
(24.2)
(•do= value
atthesurface;
[•and7shape
parameters).
Bothdistributions do not effectthe given2D horizontalvelocityfield. The verticalaverage
of (24.1) is equal to the given velocity v; the verticalaverageof (24.2) is zero. Both
distributionslie in verticalplanes,but (24.2) canmakeany anglewith (24.1) and a sumof
(24.1)
and(24.2)
isgenerally
notina plane.
Thedistortion
vd (z)istaken
neutral
(vertical
averageis zero);the effectof the causeof the distortionuponthe verticallyaveragedflow
field is takeninto accountin the numericalflow computation.
v (z) -v (a + 1) g
Figure 1: Construction
of "distortion"
vd(z) of verticalvelocitydistributionby wind or densitycurrent.
or)is extremely
small.Theeffectof thisdifference
upondispersion
patterns
wouldnotbe
noticeable.
Z0
0,9-
0.8-
0,7-
0.6-
0,5-
0,4-
0,3-
0.2-
0, 1-
verticalvelocitydistribution.
A possiblerefinementshouldbe preceded by sensitivitytestsas
to theeffectsupondispersion. The matteris probablyratherinsignificant
for dissolvedmatter
dispersionbutit couldbe of interestin thecaseof sediments.Anotherref'mement couldbe to
accountfor the time variationof I• in a tidal flow.
In (24.2), oneof the exponents,at leastin practice,hasto be > 1, say {3> 1, while the
otheris < 1. The termwith ]3> 1 canbe considered as theprimaryinfluenceof the wind,
downfrom the surface(fig. lb). A highervalueof ]3meansa smallerrelativepenetration
depth,so that• couldbe higherfor deeperwater.The termwith ), (normally),< 1) canbe
seenas the compensation currentthatneutralizesthe effectof the otherterm uponthe depth
averagedvelocity.Thereforeit seemsappropriate to take), roughlyequalto ix. The heightz
= zowhere
vol(z) = 0 liescloser
tothebottom
as• issmaller
(relative
penetration
depth
greater).Figrife2 giveszo/Has a functionof {3for fixed valuesof ¾.
Thetwoscalerangesdonotactentirelyindependent. Thelargescalestructure
of a velocity
field alonedoesnot inducethe samepatternsas its combinationwith smallerscalestructures.
Smallerscalevelocitystructures havea smoothing effectupondispersionpatternsas would
be causedby the large scalestructurealone.
E• = cnk-•' (24.3)
with F_•= energyper unit of massand unit of wave numberk; ckand m positiveconstants.
On thebasisof comparisonbetweennumericalsimulationandexperimentsthe valuesusedfor
theNorthSeaat present
areck= 2.87x 10'4m2 S'2,m = 1 (VanDam 1987).A spectrum
of
the simpleform (24.3) in two dimensionsand in absenceof othereffects(in particular3D
mechanisms) corresponds with a behaviourof the maximumconcentrationof a singlepatch
(instantaneous
point source)
If thereis a structuredgiven(computed)velocityfieldpresent,
therandomstephasto be
levelledoff (gradually)at thescalewherethegivenvelocityfieldbeginstoprovidethecorrect
structures.
or T - (24.5)
K'' 2r= = 2Kz
r = }/2o•on (24.6)
if • and q are the directions
of the principalaxesof the patch.The useof r insteadof t
impliesthat• and{Jnhaveto be computed fromtheparticledistribution
quitefrequently.
10-2
io--4 io-4
TZ = 0
/
Z = 3600
= 7200
io-6 lO-6
Z = 14400
io-8 i0-8
Tz = 28800j
T = 8640
Z
It is interesting
to compare figure3 withthecompiled datafromtracerexperiments in the
NorthSea(fig.4). Thecomparison supportsthechosen approximationof theenergyspectrum
andit strongly suggests thatthenaturalvariations
of T,_(thehighervaluesarerealisticin stra-
titled areas)are the main causeof the spreadamongthe variousexperiments, ratherthan
regionaldifferences in kineticenergycontentor its spectralstructure.
The correspondence
betweenthe bandof observations andthe bandof simulations is quitesatisfactory.
Note that
the bandof simulationshasa lowerbounddue to the fact that from a certainvalueof T,_the
increase
of dispersion
rateswith increasing
T,_comesto an endandevenchanges
into a
decreaseat still largerT,_values.
10-6
zo-8
10-10
].o hours
Figure4: Compiled
NorthSeadataof instantaneous
pointreleases
of tracer(seerefs.).
Thelattereffecthasbeenstudied separately
in a vertically
two-dimensional (2DV) model
(essentiallythesamemodelwaspresented in 1982by Allen).A resultof thesestudies is
presented in figure5. The dispersion coefficientsplottedin the verticaldirectionare the
constantvalues (nofide,netcurrent only)or theeffectivetidalaverages,bothreached after
a certaininitialperiodrelatedto theverticalmixingtime.
Deff
300 m2s'1
200 6
81% osc.
•n.
100
3O 98
•0
Tz: T/6 T 6T
!!
1/3
1/2
2/3
516
....•,':'
.. • ..•...• . :)
ß:%•2.',½.: x T
Tz = T Tz >> T.
Figure6. Illustration
of (vertical)sheardispersion
of aninstantaneous
verticallinesource
for 100%tidal
current(comparefig. 5), showinghowthelongitudinal dispersion
decreases
comparedwith thecaseTz
= T, whenTz becomes(much)smalleror (much)larger.
during
onetidalperiodof a contour
which
isinitially
a circleof 1500m radius,using the
computedflowvelocities
only.Onthisscale,
therelative partof thehorizontal velocity
structure
in thedispersion
is dominant
andthemainpartof itsenergy is reproduced in the
flowcomputation.
Similar
results
havebeenpublished
byRidderinkhof (1990).In figure 7c
thedistribution
of a dense
setof particles,
fillingthecircle,is givenat theendof thesame
tidalperiod,
including
theeffects
of "sub-grid"
scale
velocity
variations
andvertical
shear.
0 • 20000
s
_ 5000
s 2•0•00•
•50•
0•a
45000
-period)
• • • 15000s
b
4500
C (eM2
peri
d) _
Figure
7. Deformation
ofa circular
contourduring
onetidalperiodinEmsEstuarymodel (mesh= 300
m)bycomputed velocity
fieldonly.Difference
in initialphase
betweena andb is onehalfof tidal
period
(22500s).Extradispersion
by"sub-grid"effects
(vertical
shear
withvertical
mixing;"sub-grid"
horizontal
velocity
variations)
forcasea shown
in inset(c) (situation
atendof period
only).
a
b
5oo j
t = 4000
Figure
8: Effects
similar
tofigure
7,obtained
bysimulation
with2Dsynthetic
eddyfields
(8pairs
of
harmonic
components)
in anunbounded
region.
a. Stationaryfield
b. Eddy life time 1.5 revolutions.
In case
b,onlythecontour
ofthefluidblobhasbeendrawn;
graphs
(candd) show
evolution
ofoverall
sizeandcontourlengthin time.
Similar
patterns
wereobtained
intheseparateinvestigations
of synthetic
velocity
fields.
Figures
8aand8bwereobtained
with8 "eddycomponents"(8pairsofharmonic
functions),
figure
8ainastationary
mode,fig.8bwithfinite
lifetimes
(1.5revolutions
pereddy).
Units
Thenumerical
experiments
withsynthetic
fieldsarea quickmeans
to interpret
dispersion
patternsin termsof energyspectra(section24.2.4).
24.4 Practicalapplications
ß .
..
Figure9: Bird's-eye
viewof particlesetsin a tidalcurrent
(Northgoingphase)
withmoderate
wind
fromtheWest,computed for threedifferent
verticalmixingtimesTz.
Pointreleaseat thesurfaceat t = 0; distributions
drawnafter40, 80, 120,200 and360 minutes.
Patchwithoutwinddrawnfor t = 360 minutesonly.
The simulation
of periodsup to a few dayshasbeenoperational with pr•eding 2D
versions
of themodelfor accidental
spillsat sea(DelftHydraulics
Laboratory
1989)or in the
Western
Scheldt
(VanDam1985%
1986b;
Lefevre
andVanderMale1988).Forthesame
type
of applications
a 3D versionof the modelis now beingimplemented
for the North Sea
(Diependaal
1990;Hendriks
1990
a'b)
andwillbeadopted
forotherregions
in thenearfuture.
An illustration
of an applicationon a shorttimescaleis givenin figure9, presentinga
bird's-eyeviewonly(particles at all depthlevels).Thechosennumberof particles(300) is
smallerthanin cases whereiso-concentration finesare required.The picturegivesan
impression of windinfluence(moderate crosswind)anditsdependence uponT•.
I I
•ookm
Figure10. Simulation
with 3D modeof transport
modelof a continuous release(nearthemouthof
riversRhineand Meuse)in the North Sea. Periodicflow; (periodic)steadystateconcentration
distribution.
a. Computed
withnon-uniform
account
of density
current
perpendicular
to thecoastandsuppressed
verticalmixing,decreasing
with distancefrom river mouth.
b. No densitycurrentandmixinguniformover the wholearea.
White= land(figures
10,11,12).Shades
in figures
10through
15represent
concentration
ranges
of a
factor 4.
Figures10,11and12givesometypicallongtermapplications, whichwereonlyfeasible
bytheuseofa strictlyperiodic velocity
field.Of course,theperiodicity
oversucha longtime
(orderof oneyear)limitstherealisticmeaning of theresult.Theperiodicitydoesnotonly
enable theseparation
of the(relatively
short) flowcomputation andthetransportcomputation
butit further
enablestheuseof superposition (VanDam1985a),whichsaves a greatamount
of computer time.Theresults havebeenobtained by following only300particles fromthe
source, amounting
toa totalof some100000particles in thefinaldistribution
bysuperposition
.............................................................
............................................ 2..•_;,.-.•:.•
•"•' :::::::::::::::::::::::::::::
::::::::::::::::::::::::
,•;•.z•.•:••5=• .•:=; ß,
================================ .• ....:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
......
=:==.==.= :,2•.-
7•. ._'•-
Figure l lb and 12b as comparedto 11a and 12a are to illustratethe effects due to
differencesin advection(Van Dam 1984,1986•, 19889thatariseif Eulerianaverages of tidal
velocitiesare used in order to save computertime. It is seen that the effects are more
pronouncedin one area than in another.However, thereis little reasonto use the fasterbut
lessaccurateapproach(figures1lb and 12b)nowthecompletetimedependent simulationhas
become feasible.
'--.•.E•-.- ,•..--:•-'
7•2=g•.• •i-':•.•. •.,•;-nñ .............................................
.
;.•?.:..•.:/•=g•:=----====:_z.:_z.=E-!-!'-"!-•..•--=_=
•'....,,--l•- r-"":-7.:-
::.•.--.7.:•.:.7.:•:--
•
...................................................
...............................................
_ Iw•l
z (24.7)
c (z)- c
where
_ttl,•el• (24.8)
(l%l= absolute
value
of fallvelocity;
withz positive
upwards,
we< 0) or,with(24.5)
formixingtimeTzandTv= H/l% I forfallingtimeandz* = z• forrelative
vertical
position:
(24.9)
c (z,) = c (o) e
iii!i!ii!!i:i::::::::::!i::!::!::::::::ii:ii!:i:iiiii!!!!iiii
!!!!i i i i•.iiii:::•::i::!:.:•:.•:•....`.•.:•;i:•i:..!i•:•.:•i:.::•i!ii•ii•.:i!!..?.`•::
•i!;•;:'.':,'•i:..'.:::!i•iiiii::":::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
.;.;::.,,..:,•.:.•..•
.',.
•••'-'•,•.
;.:.;•,:
;:...,.::•:::::
•.:(...!.
:.:...•:.:'::..'.:
................
!i::i•::::::i::!:iii::::i!iii::iii::ii::!::!!::::iiiii:::i:!•i•ii::i:::.:.•i
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::::::
v..•
::::::::::::::::::::::::::::::::::::::::::::::::::::::
.--.-'.'
::::::::::::::::::::::::::::::
:,_:..,-.-.
::::::::::::::::::::::::::::::::
•...
......................................
........................................
..............................
..............................
::::::::::::::::::::::::::::....
........................
::::::::::::::::::::::::::::::::::
!•iiiii•;•iiiii
..............
ii::.... IOO m ........................................
i!•i::::::::iii::"i
........ iii:.":
........
..............................................................
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::::::::::
=================================
....................................... ::::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::
............................
:::::::::::::::::::::::: ....
........................................ ....................
::::::::::::::.. ..::::::..::
..................................................................................................
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
..::::::::::::::"
..................................................................
..................................................................
..................................................................
..................................................................
..................................................................
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::
•:' ;.•:.::.::•
:::::::::::::::::::::::::::::::::::::::: ::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::: ............................
:::::::::::::::::::::::: ....
::::::::::::::::::::::::::::
....
........................
........................................
......................
24.5 Conclusions
24.6 References
Barrett,N.J., D. Munro and A.R. Agg, 1969. Radiotracerdispersionstudiesin the vicinity of a sea
outfall.Proc.4th Int. Conf. on Water PollutionResearch,PergamonPress,New York, 863-882.
Delft HydraulicsLaboratory,1989.Gebmikershandleiding
MARS (MarineSpills).Rept.T 470, 47p.
Hendriks,
A., 1990
b.MARS2.0,Deconstructie
vaneenhydrodynamische
database.
DelftHydraul.
Lab.,
T 607.
Leendertse,
J.J.,1967.Aspectsof a computational modelfor long-period
waterwave
propagation. The
RandCorporation,SantaMonica,Calif., rpt. RM-5294-PR/Ph.D. Thesis,TechnicalUniversityDelft
1967, 165p.
Lef•vre, F.O.B. and C. van der Male, 1988. Gebmikersdokumentatie programmapakket CALM,
Rijkswaterstaat
(Neth.),Tidal WatersDivision,rpt. GWWS-88.407,62p.
Prandle,D. 1982. The vertical structureof tidal currentsand otheroscillatoryflows. Cont. Shelf Res.
1, No. 2, 191-207.
Suijlen,
J.M.,1990•. Measurement
of turbulent
diffusion
andresidual
displacements
bydyeexperiments
in the southernNorth Sea in 1982. Data Report.Rijkswaterstaat
(Neth.), Tidal WatersDiv. GWAO-
90.023.
Talbot, J.W. and G.H. Talbot, 1974. Diffusion in shallowseasand in Englishcoastaland esmarine
waters.Rapp.,P.V. Cons.Int. Explor.Met., 167, 93-110.
Van Dam, G.C., 1982. Models of Dispersion.In: Pollutanttransferand transportin the Sea, G.
Kullenberged. CRC PressInc., BocaRaton,Florida,91-160.
VanDam,G.C.,1986
• Particle
modelling
of2DHtransport
phemomena
intheWestern
Scheldt
Estuary.
Abstracts
International
Symposium
onPhysical
Processes
in Estuaries,
Noordwijkerhout,
TheNeth.:1.1.2.
VanDam,G.C.,1988
b.Themathematical
modelling
of transport
phenomena
intheNorthSeainrelation
to experiments.
Rijkswaterstaat
(Neth.),TidalWatersDivision,rept.GWAO-88.043,27p.
VanDam,G.C.,1989'.Eddysimulationof dispersion.
Workshop on applying
non-linear
dynamics and
chaosin fluidmechanics,
June1989,Woudschoten,
Zeist,TheNetherlands/Rijkswaterstaat
(Neth.)Tidal
Waters Division, GWAO-89.313.
VanDam,G.C.,1989
b.Dispersion
ofdissolved
matter
intheNorthSea.ICES,Hydrogr.
Committee,
C.M.
1989/C:13,21p.
VanDam,G.C.,1990
•. Two-andthree-dimensional
models
of dispersion
based
uponmeasurements
in
theNorthSea.Proc.IAWPRC/EWPCAInt. Conf.on NorthSeaPollution,Amsterdam,
14p.
VanDam,G.C.,1990'.Numerical
simulation
of dispersion
by eddyvelocityfieldsin twodimensions.
Submittedfor publication.
VanderGiessen,
A. andR.R.Jansen,
1988.Het2,5D-model.
Eennieuwe
methode
terbepaling
vande
vertikalestromingssmfictuur
in deNederlandse
kustzgne.
Rijkswaterstaat
(Neth.)TidalWatersDivision,
GWAO-88.013.
VanVeen,J,, 1936.Onderzoekingen
in deHoofdenin verband
metdegesteldheid
derNederlandse
kust.
Alg. Lands•erij, Den Haag,Neth.,258p.
25
Abstract
25.1 Introduction
25.2.1 General
which, unlike the bed component,is unlikely to attain equilibriumwith the unsteadyand
non-uniform ambient flow conditions. For this reason, therefore, the model treats the
suspended andbed components of the sedimentload separately.In addition,a cross-shore
transportmoduleis includedin orderto reproducethe development of beachprofiles.
The model computesthe bed level changesin two stages. Firsfly, the depth-averaged
versionof theunsteady-state
diffusion-advection
equationis solvedfor thesuspended
sediment
concentrations.This equationtakesthe form:
DISPERSION
(25.1)
+--••&(ht,
UC)
+•(hrVC)
+Ro
- R•. 0
ADYECTIOIg SOURCE
/SINK
where:
(2 = depth-averaged
suspended
sedimentconcentration
(kgm'3);
Dx = longitudinal
dispersion
coefficient
(m
Dy = lateral
dispersion
coefficient
(m2
hv = flow depth(m);
Rv = sediment
deposition
rate(kgm'2s4);
R•. = sediment
entrainment
rate(kgm'2s4);
t = time (s);
0 = depth-averaged
longitudinal
velocity(m s4);
•/ = depth-averaged
lateralvelocity(m s4);
x = longitudinalco-ordinate(m);
y = lateralco-ordinate(m).
minimal.In addition,
a checkonthequantity
of sediment
present
duringtheprogress
of the
slug showedthat masswas conserved.
z
_o 6O
< 40
z
,,, ø
z
o
o 1.2m
• 1.0 1.0
1 '2m
1.o
0.8
0.•6
0.8
0.4 0'4
0'8
0.2
(o) Advection
I
•" 100
'E I
• 80 I
z
o_ 6O I
4O
z
z
o
.2m
lo2m'•1.0 ' 1'0
1.00.'••6 0••.6
0-8 ,", 0.
0-8
0.2
(25.2)
(1-p)p,at
where:
The only unknownin Eq. 25.2 is theelevationof thebed (%), detailsof the computation
of the independent
variables(Q•x, Q•v, Qcx, Qc•, Rx•,R•) being set out below. As the
deposition
rate(Rx))is a function
of thedepth-averaged
concentration
((2),thisvariable
is
dependent
on the solutionof Eq. 25.1. Eq. 25.2 is solvedby meansof an explicitfinite
difference scheme.
The structureof the model setup to solveEqs. 25.1 and 25.2 is containedin Fig. 2. A
description
of the constituent
partsof the modelfollows.
I INITIAL
BATHYMETRY
I
i HYDRODYNAMIC
DATA
I
CROSS-SHORE
TRANSPORT
i ENTRAINMENT
I
I
I BATHYMETRIC
CHANGES
[ =,NAL
BATHYMETRY
I
Figure2: Model structure
25.2.2 Bedforms
The heightof the bedformsat any pointin plan is neededto fix the thickness of the mobile
bedlayer. Bothcurrentandwaveactionmaybe presentat anygivenlocationbuttheauthors
are unawareof any quantitative informationin theliteratureaboutbedformsgenerated under
suchconditions.This problemwasovercome,therefore,by determiningthe bedformheight
producedby currentsand wavesseparately and thentakingthe largerof the two to be the
representative
value. The determination of the characteristics
of the current-generated tipples
is basedon the work of Yalin (1977) and thatof wave-generated tipplesis basedon the work
of Nielsen(1985) or Swart(1976). The presentversionof the modelmakesno provisionfor
the occurrence of marine dunes.
The results of a comparisonof the Nielsen and Swart methodsof deriving the
wave-generatedtipple heightsis shownin Fig. 3. Clearly,thesetwo approaches
yield values
whichdiffer markedlyin trend. However,recentwork by RibberinkandA1-Salem(1990) has
confumedthat the Nielsen methodgivesthe more qualitativelycorrectresults,a sandbed
tendingto becomeplaneunderintensewave activity.
0'012
q(m)
0.010 _
0 038 -
NIELSEN (1985)
o-00z. _
--0 SWART (1976)
0-002 --
H(m)
0
0 0-2 0-4 0-6 0 8 1-0 1-2 1.4 1-6
25.2.3 BedloadTransport
Secondly,the ratio of the steady-state suspended load transportrate to the bed load
transportrateis determinedusingthe Bijker (1968) versionof the methodoriginallyproposed
by Einstein(1950). The Bijker-Einsteinmilo, however,hasbeenslightlyalteredfor usein
thepresentmodelin thatthe thickness of themobilebedlayeris setequalto half thebedform
heightratherthanthe totalbedformheightadoptedby Bijker. This resultsin an expression
of the form:
where:
I• = first Einsteinintegral;
I: = secondEinsteinintegral;
QB= bedload
transport
rate(kg m4
Qs= steady-state
suspendedloadtransport
rate(kg m
rB= bed roughness
height(m).
(25.4)
where:
D9o= ninetiethpercentile-by-weight
groinsize (m);
q = bedformheight(m);
X = bedformlength(m).
50 I i I 1 I I I I O-5
Qs/% Oe(kgm-1s-1)
40 O-4
GRAIN SIZE = 250 10'6 m
30
-- FLOW
VELOCITY
=1.0
m$-1 • 0.5
WAVE
PERIOD
10-0s •sI••
2O 0.2.
0.1
H(ml
0 I I I I I I I 1 o
0 0 2 0 z. 0-6 0.8 1 0 1 2 1./.. 1-6 18
25.2.4 Cross-ShoreTransport
Bailard (1982)
Nishimuraand Sunamura(1986)
Watanabe(1985)
Shibayamaand Horikawa(1985).
producederodingbeacheswhentheyshouldhavebeenstableandstablebeacheswhenthey
shouldhavebeenaccreting.Clearly,this is a subjectrequiringfutureimprovement.
Ho/L
ø I •
FIELD SUNAMURA
&HORIKAWA
(197/.) • J
10'2
10'3 ,J I
10-•
{iø)-ø
:•?(Dsø/Lø)ø'•?
10-•;
- % c; (25.5)
where
C•ø = steady-state
near-bed
suspended
sediment
concentration
(kg m'3);
Ws = sediment
fall velocity(m s'•).
Direct approach:
Indirectapproach:
30
i I I ! I ! i I
RE (kgm-2s-•) VAN RIJN AND TAN [1985)
............... O'CONNOR
..J
.J
Figure6: Entrainment
rate(R•.)versuswaveheight(H)
25.2.6 SedimentDeposition
- (25.6)
where:
Cs = instantaneous
near-bedsuspended
sedimentconcentration
(kg m'3). However,the
solutionof thedepth-averaged
diffusion-advection
equation(F-xl.25.1) yieldstheinstantaneous
depth-averaged
concentration
(2. Hence,if theratiobetween
thesetwoquantities
is written
as:
(25.7)
- ca
I• "0.22 (25.9)
where:
12 I I
r•(lO3)
I
WAVE GENERATED RIPPLES:
NIELSEN (1985}
1.0
SWART (1976}
GRAINSIZE = 25010'Gm
FLOW DEPTH = 2-5m
0.8
WAVE HEIGHT = 0-Sm
0.6
0.2
I I I ! I I • (ms-l)
0 0.2 0-4 0-6 0-8 1.0 1-2 1-4 1-6
Figure7: [•-function
versus
flowvelocity(0)
approximately
40m wide by 6m deepwhichhadbeendredgedacrossthe inshorezoneof the
DanishNorth SeaCoastin orderto bringashoreoil andgaspipelines.The trench,the layout
of which is given in Fig. 8, had been surveyedat regular intervals and continuous
measurementsmade of the wave climate. Details were known, therefore, of the siltation in
the trench and the associated wave conditions.
ß ß
....
ß ß ßßII 'ßOm /
ß Ii
ß ß ß
ß
ß e
e e
ßß,
ß ß ß ß ß ß
COASTLINE
DEPTH [m)
6
5
5
6
TRENCH
1600m
beachconfiguration.
Tidaleffectswereconsidered
to benegligible,
themeantidalrangebeing
less than one metre.
10s
• 10
,
103
103 10•, 10s 10•
DURATION is)
105
/.."'
•z 10•
/
•/B E___DF
ORM$ .ENTRA_I__
NIELSEN VANRIJN& TAN
-- SWART DE RUITER
SWART VAN RIJN & TAN
SWART F. LUOUE
I
103
103 10s 10•
DURATION
1000
NEASURED
I
800 ------- CONPUTED
• 600
• 400
200
0
5OO 700 900 1100 1300 1500
CHAINAGE {m)
Figure11: Comparison
betweenmeasured
andcomputedlongitudinalinfilling
25.4 Conclusions
!CHA,NAGE
700i
• ORIGINAL
..... MEASURED -
COHPUTED
T
ee e * ß ß ß ß ß ßßß ß ß
W ß
C• ß
ß
ß
ß
12
20 0 20 40
OFFSET Iml
[CHAINAGE
1350J
•
ORIGINAL i
..... MEASURED
COMPUTED
• ".2
12
40 20 0 20 40
OFFSET Im}
Figure12: Comparison
betweenmeasured
andcomputed
lateralinfilling
Application
of themodeltoa simplified
fieldsituation
involving
thesiltation
of a dredged
trenchin a coastalenvironment
yieldedrealisticanswers, whichwerewell withinengineering
accuracy. Thefieldapplication
alsoshowed theimportance of including "feedback"between
morphological changes andthehydrodynamic flow field,theneedto choose a suitable
model
timestepandtheneedto operate themodellongenough to ensurethatany"scaling-up" of
short-termruns,whichreduces modeloperation costs,is representative
of thefinalconditions.
Finally,a detailedexaminationof the longitudinalsiltationpatternin the dredgedtrench
showed that the model underestimatedthe siltation in the inshore sectionof the trench, which
suggests
thatadditionalfield testingis requiredin orderto providefurthercheckson the
accuracyof the scheme.
Acknowledgement
This work was undertakenas part of the MAST G6 CoastalMorphodynamics research
programme.It wasfundedby theCommission of theEuropean
Communities, Directorate
Generalfor Science,Researchand Development,
underMAST contractno. 0035.
Symbols
(2 = depth-averaged
suspended
sediment
concentration
(kgm'3);
Ca = instantaneous
near-bexl
suspended
sediment
concentration
(kgm'3);
C•ø= steady-state
near-bed
suspended
sexliment
concentration
(kgm'3);
D•o = mediangroinsize (m);
D•o = ninetyper centfiner-by-weight
grainsize(m);
Ds = longitudinal
dispersion
coefficient
(m"
Dy = lateral
dispersion
coefficient
(m"s'•);
ha = mobilebed layer thickness(m);
hv = flow depth(m);
H = wave height(m);
Ho = deepwater wave height(m);
io = initial beachslope;
I• = first Einsteinintegral;
I,. = secondEinsteinintegral;
Lo = deepwater wave length(m);
p = bed layer porosity;
Qax= longitudinalbed load transportrate due to tidal and wave-inducedlittoral currents
(kg m'• s'l);
QB¾
= lateralequivalent
of QBx(kgm'1
Qcx= longitudinal
cross-shore
transport
ratedueto undertow,
waveasymmetry
andlocalbed
slope(kg m4 s4);
Qc•= lateralequivalent
of Qcx(kgm'l s'l);
Qs = suspended loadtransport
rate(kgm'l s'l);
rB = bedroughness
height(m);
RD = sediment
deposition
rate(kgm'2s'l);
RE = sediment
entrainment
rate(kgm'2s4);
t = time(s);
T = wave period (s);
0 = depth-averaged
longitudinal
velocity
(m s'l);
9 = depth-averaged
lateralvelocity(m
x = longitudinalco-ordinate(m);
y = lateralco-ordinate(m);
Ws = sediment
fall velocity(m s'1);
za = elevationof the bedrelativeto an arbitrarydatum(m);
= cde;
,,t = modeltime step(s);
q = tipple height(m);
•, = tipple length(m);
p, = sediment
density
(kgm'•).
25.5 References
Deigaard,R., Fredsoe,J. and Hedegaard,I.B., 1986: Mathematicalmodel for littoral drift. Jour.
Waterway,Port, CoastalandOcean Engineering,ASCE, Vol. 112, No. 3, 351-369.
Fernandez
Luque,R., 1974: Erosionandtransport
of bed-loadsediment.
Dissertation,
KRIPS ReproBV,
Meppel,Netherlands.
O'Connor, B.A. and Yoo, D.H., 1988: Mean bed friction of combined wave-current flow. Coastal
Engineering,Vol. 12, 1-21.
Shibayama,
T. andHorikawa,K., 1985: A numericalmodelfor two-dimensional
beachtransformation.
Proc. JSCE, No. 357/II-3, 167-176.
Sunamura, T. and Horikawa, K., 1974: Two-dimensional beach transformation due to waves. Proc.
FourteenthCoastal EngineeringConf., ASCE, 920-938.
Applied Studies
Dynamics and Exchanges in
Coastal and Estuarine Studies Estuaries and the Coastal Zone Vol. 40
26
Abstract
Resultsare describedof a measurementprogrammein the outflow zoneof the Rhine river that
hasbeengoingon over the pastyearswith varyingintensity. Its aim is to describethe three-
dimensionalhydrographicstructurein the area at differenttime and spacescales.
Underextremedischarge conditions
combinedwithstrongwindson-shore nearbedcurrents
of ca. 10cm/swereobserved.Duringthiseventstratification
wasveryslxong throughout
the
coastalzone. In the upperlayerlongshore
currentspeedsappearedof up to 50cm/s,whereas
in the lower layer thesewere only a few cm/s.
26.1 Introduction
[ I I
0 50 100 km
SOUTHERN 8/GHT
0•• 5•70
ß
' 5850
Callan
5•20
IJ•iden
•oo
o
y78o
y/7o
Tert•lde
o
5750
Y7•O
6000.
'1500.
E
_c•ooo.
o 1500.
--
I C •
o
•,
o oo
Figure3: Discharge
andwindconditions
duringtheobservation
periods.A: Rhinedischarge.
B: wind
vector. C: periodnumbersindicatingindividualcruisesin the coastalzone
Thebulkof theRhinewatercombines
withtheMeusefiverdischarge
(averaging
300m3/s)
anddebouches
into theNorth Seathroughboththe RotterdamWaterwayandthe Haringvliet
sluices
(Fig.2). Thelatteraremanipulated
withtheaimto guarantee
a discharge
of 1500m3/s
throughthe Waterwayas long as possibleto preventsalt wedgeintrusion. Above a fresh
waterdischarge
of 1500m3/stheremainder
entersthecoastal
zoneviatheHaringvliet.Only
at discharges
over6000m3/s
anextratransport
throughtheRotterdamWaterwaytakesplace.
In front of the Dutch coastthe fiver water mixes with open sea water that has enteredthe
southern NorthSeathroughtheDoverStrait. Thisinfluxof Channelwatervariesconsiderably
with wind speedand direction(Prandle,1978).
NoorthftJk 890926
•um
Temper•t,Umre
, I Tempermture
-B
TerHeyde
-10
-20
•5 20 t0
-25
t)Jstance to coast
11•
-5 _
-10
.4
• -•0
0 • • 15 10 5
-•5
t)Jstance to coast
Noord•jk
• • • 115 IlO I7 I4
33 5 5 .5
33. ß
5 2. t.5 t. 30.5
• 33.
-5
-t0
-•5 '6
6
-2{)
20 t5 t0 5
-25
O•stance to coast
Egmond
-iO . .5 . i.5 3t .5
ß
-20 33.5 ß
-25F 25 20 15 t0 5
• E•5 E•O •l• •lO EN07 ,Era ,ENI
-20
5 20 t5 t0 5
-25
Distance to coast
Callantsoog
-5
-t0
32.5
-20
20 t5 t0 5
-25
-5
-t0
-is
-20
-25
538
W P M deRuijter,A vanderGiessen
andF C Groenendijk
It is interesting
to notetheincreasedgradient of salinityvariabilitybetween10 and15km
offshore fromCallantsoog (fig.5d). Thisis alsotheareaof relatively maximum stratification
and crossshoredensitygradient,suggesting that herethe positionof the frontalzone
separating openseawaters fromtheregionof majorfiverinfluence is relatively
stable.This
maybe a consequence of therathersteepbottomslopeinhibiting (byrotational constraints)
cross-shore displacementof watercolumns andsteering theflow alongit. However,sofar
the numberof cross-shoresectionstakenat this locationis too small to draw firm conclusions.
In thisareaalsoa minimum
of suspended
matterhasoftenbeenobserved,
mostlyduring
wintercruises(Visseret al, 1991).
..
Figure
6: Average
salinity
fields(over8 cruises)
andtheirstandard
deviation
(ino/oo)at lm and10m
belowtheseasurface.
Clearly,variability
intheupperlayerismuchlarger
thanin thelowerparts
of
the water column.
The above three-dimensional picture of the salinity field is further confuted by its
horizontaldistributionat a few selecteddepths(fig. 6). However,thesemeanfields don't
appearexcitingas manyof the interestingfeatureshavebeenaveragedout (thoughthe fields
of standarddeviationindicatethat there'smore goingon).
After its debouchmentinto the North Sea some fiver water is transportedsouthward,
particularlyin caseof northerlies.Consequently, in the Goeree-section
some20km southof
the RotterdamWaterway(Fig. 2) the influenceof the outflowingfiver is still significantup
to 15kmoffshore.Variabilityis of thesameorderof magnitude asin theNoordwijk-transect.
Verticaldensityinhomogeneities are smallerdueto the shallowness
of thecoastalzonenear
Goeree.Underpersistent windsfrom thenorthRhinewatermayevenreachthemouthof the
ecologicallyvulnerableEasternScheldt.In suchan extremesituationcombinedwith a high
fiver dischargea fiver water fractionof about25% wasobservedat this location,which is
situatednearly40km southof the Haringvliet(De Ruijteret al., 1991).
-5 _
.5
-10 _
33.5
-15 _
33.5
-20 _
TerHeyde 870213
0 15 lO 5
-25
Distance to coast
-5
2.5
-t0
-15
32.5
-20
Noordwi jk 870213
5 20 t5 1o 5
-25
Distance to coast
Figure 8: Salinitydistributions
in the Ter Heijde (a) and NoordwijkCo)transectsat averageriver
discharge
(2300m3/s)andwindswithaneasterly
component.
Dateof survey:
February
13,1987
541
W P M deRuijter,A vanderGiessen
andF C Groenendijk
In the situation
with dischargecloseto averagein theTer Heijde-transect
up to 20km
offshorestratification
isstrong.Mixingprocesseshavesubstantially
reducedthestratification
in theNoordwijk-sextion,whereit appearsmainlyin a zonebetween 5 and15kmoffshore.
Fig. 8 alsoshowssomeearliermentioned phenomena observedin thesetmnsects.For
instance,nearTerHeijdethepattern of isohalines
indicatesanestuarine-like
circulation,
with
lesssalinewaterdrivenoff-shoreandopenseawatertransported on-shore.Ca. 4km from
shorea frontnearthebottom(TerHeijde)canbeobserved. Northerlies
andoff-shorewinds
maystimulate
the occurrence
of thesephenomena.
ß . . ß
• TerHeyde 860i30
5 20 ',5 f.O 5
-25
OJstance to coast
-2O
Noordwijk 860130
20 t5 •0 5
-25
Figure
9:AsinFig.8,nowatabove
normal
discharge
conditions
(6000m3/s).
Dateofsurvey
January
30, 1986
Extreme
fiverdischarge
conditions
leadto substantial
deviations
fromtheaverage
salinity
pattern.A salinity
distribution
during
a peakdischarge
(:!:6(X}0m3/s)
is depicted
in Fig.9.
Stratification
is extremenow, throughout
the coastalzone. Comparedto the Ter Heijde
transect
onemaynoticethedeepening of theupperwindmixedlayerin thestratified
system
nearNoordwijk,30km furthernorthward.It showsthe impactof strongwinds(5 Bf)
occurring
duringthisspecific
period.However,thereareindications
thatalsoataverage wind
conditions
thebuoyantspreadingandmixingof thefiverwaterplumeis affectedstronglyby
542
W P M deRuijter,
A vanderGiessen
andF C Groenendijk
wind-induced
stirringof theupperlayer.
lb20 lh15 fido lh07 Th04 lhl.5
-5
•.5 •. 3t. 29.
•
-tO
-t5
.
•.5i• 3t
•.5 5t5 t5 .5 29 . . 3. .5 ß
-20
ß
TerHeyde 871110
20 t5 tO 5
-25
Distance to coast
-5 __
-10 __
-t5 __
-20 __
Noordwiik 871110
.'5 20 t5 tO 5
-25 ,
Distance to coast
Figure
10:AsinFig.8,nowatrelatively
lowRhine
river
discharge
(1400m3/s).
Date ofsurvey
November 10, 1987
During
periods
oflowfresh
water
discharge,
stratification
isalmost
absent
(Fig.10)except
in a small
areaof theTerHeijde-section,
closetothemajor fiveroutlet.Nevertheless,
significant
horizontal
densitygradients
stillexist.
Inthelong-term
average
situation
these
lateral
density
gradients
decrease
withincreasing
distance
fromshore.
However,
ina specific
casetheactual
meteorological
andhydraulic
conditions
may
disturb
this
pattern
considerably.
Therefore
anyindividual
density
pattern
may
deviatesignificantly
fromtheaverage
situation.
Currents
weremeasured
continuously
in thetransect
nearNoordwijk
fromAugust
1985
through
September
1986.(see
Fig.11forlocation
ofmoorings
and'positioning
ofcurrent
543
W P M deRuijter,A vanderGiessen
andF C Groenendijk
meters
in thewatercolumn.)Someremarks
aboutthetidalcurrents
in thisregionweregiven
in section 26.2. Here we will focus on the residual currents.
NOORDWlJ K-transect
30km 20 10 5 3
I
(
5m-
10-
15-
20-
¸ Current
meter
::: if::
• i I--IPressure
recorder
Figure11:Location
ofcurrent
metersduring
theone-yearobservational
period
inthecross-shore
transect
nearNoordwijk.
Currentmeterswererobust
self-recording
Flachsees.
Every10minutes
theyregistered
10 minute-averaged
velocities
andinstantaneous
currentdirections
Residualcurrentsareinducedby several
physical mechanisms.
A basiccontribution
is due
to nonlinearfide-topographyinteractions
(e.g. Zimmerman,1978). Anotherdominant
componentis wind-induced.In caseof windswitha predominant
coast-parallel
component,
within severaltensof kilometresfrom the shore,the flow mainlyfollowsthe coastlineas a
geostrophic
jet in downwind
direction
ateverydepth(decreasing
in magnitude
withdepthdue
to friction).Besides
alongshore
flow,windsalsoinducea cross-shore
circulation
component,
eitherbya combination
of fictionandCoriolisforce(asanEkmanspiral)orasa driftrelated
towindsetup.In coastal
waters density
drivenresidualcross-shore
currentscanbeimportant.
For instance,
for the Irish SeaHeaps(1972) hasshownthatthe observed averageon-shore
flow neartheseabed mayresultfromthebuoyancy effectdueto thelateralvariation
in the
densityfieldwith relativelyfreshwaterinshore
andhighersalinities
offshore.The same
applies
to theDutchcoastal zonewherethehorizontal
densitycontrastis largerdueto the
highermeanfiver discharge comingfromMeuseandRhine.
Suchan estuarine-typecirculation
component is constant
in directionandthereforemay
contribute
significantly
to thelong-termdistribution
andtransportof waterproperties.
Thelongtermaverage currents
nearthebottomvaryfrom2.5 to 3.5cm/s(Fig. 12). The
maximumof i}p/3x(=2.10'4kg/m
4,Fig.4b)between5 and10kinoff-shorefromNoordwijk
corresponds
with a maximumin theon-shore
nearbexlflow anda relativelylargelongshore
30km
•'•
•, /....
......>.-
Mid-column
.................• Bottom
.....
ili!
•:
Figu;'•:12: Mean currents(closeto the surface,mid-columnand near the bottom)in the Noordwijk
transect. The averagewas taken over the full periodof observation,
August1985-September 1986.
Relativelylargelongshoremeancurrentsat 5 and3km offshoreare indicativeof frequencyoccurrence
of a narrowcoastaljet.
spiralsdrivenby cross-shore
windsis alsoclear from the observations
(Fig. 13).
• DWIJK
..'......DWIJK
DWIJK
5cm/s
:,, Surface
........................
-• Bottom
Figure13:Residual
currents
20kmoffshore
fromNoordwijkunderpersistent
windsblowingfromfour
distinct sectors
• I
o 2Okm
3o
E-bottom
26
27 28 29 30
27 D-surface
26
JANUARY 1986
winds
Figure14: Progressive
velocityvectordiagrams,nearthe surfaceat 10km(D) andnearthe bottom(E)
at 20km offshorefrom Noordwijk. Periodof observation:January25 - January30, whenthe Rhine
discharge
waslarge(+6000m3/s)
andwindsstrong
andvariable
(shown
inthelowerpartoftheFigure).
•o
CM/S'
3 KM
- 10 near bottomcross-shore
currentcomponents
-
t0
l0
10
•0
1o
'0 30
- t0
B000
6.000
2 000
t t ' APRIL
JANUARY FEBRUARY MARCH
1.o,86
Figure15:Low-pass
filteredcross-shore
currentcomponents atfivenearbottomcurrentmetersoff-shore
fromNoordwijk.Persistenton-shore
currents
existaftera periodof highRhinedischarge.Fluctuations
on time scalesof severaldaysare mainly wind-induced.
The horizontaldistribution
of meanstratification (Fig. 7) suggests
two favouritemodesof
fiverplumespreading.In bothcasestheoutflowjet separates fromthefivermouthandturns
to thefight influencedby earthrotation.In thisfirst modetheplumereattaches to the coast
(probablybetweenTer HeijdeandNoordwijk),to forma narrow(ñ10km)coastal jet. In the
othermodereattachment doesn'ttakeplace,or whilemovingnorthward, separationfromthe
coastis inducedby windsfavouringoffshoreupperlayertransport (particularly
windsbetween
North and Eas0. In the mean field a bandof relativelystrongstratificationis observed
between10 and 15kmoffshore.Thisbandmay alsobe relatedto the frequentexistence of
an offshore front.
Thelargevariabilitynecessitates
anelaborate
analysis
of individual
density
patterns
(fronts,
stratification),
characterizing
a widevarietyof windandfiverdischarges
conditions.
Extensive
fieldstudies,
including
repeated
hydrographic
sections
andlongtermobservations
from mooredarrays,are indispensable for the development and validationof 3-D
hydrodynamic anddispersion
models.Theresultsreported
hereindicate
thatemphasisshould
beona 25kmwidezonebetweenGoereeandNoordwijk(i.e. some40kmnorthof theoutflow
port). Longshoreresolutionof the observational
networkhasto be muchfiner thanwas used
forthisstudy.(Recently,
aspartof thePROFILE-project
(Processesin Regionsof Freshwater
Influence)
withintheEuropean
Community MAST-program a collaborafive
efforthasstarted
to fill in this gap).
All of the above showsthat for mostapplicationsa treatmentof the Dutch coastalzone
usinglong-termmeanhydrographic parameterscanleadto totallyerroneous
results.The same
holdsif verticallyintegratedinsteadof three-dimensional
currents(and models)are applied.
Acknowledgement
We gratefullyacknowledge themanypeoplewhocontributed to thecompletionof thisstudy,
in particularthe crew from the RV "Holland",the personnelfrom the North SeaDirectorate,
The Tidal Watersdivisionandfrom the formerdivisionHellevoetsluisof Rijkswaterstaatwho
organisedand assistedin the data collectionand analysis.
26.6 References
Prandle, D. 1978: Monthly-mean residual flows through the Dover Strait, 1949-1972.
J.Mar.Biol.Ass.UK, 58, 956-973.
27
Abstract
27.1 Introduction
The SouthernBight of the North Seais exposedto high inputsof contaminants, especially
through rivers and dumping activities. Part of those contaminantsare dissolvedand
transportedwith the (residual)circulation,othersinteractwith suspended
particles(silt,clay).
Thereforethe distributionand transportmechanisms of suspendedmattermay give more
insightinto the fate of pollutants.
552 M Visser
Folkard(1969)andPostma
(1981).Postma
(1981)assumed
itsposition
to coincide
withthe
seawardendof a densitydrivencirculation.
1
0 tOO &, m
SEA
,!!iiii!iii!!!i
B E LG I U
Figure
1:SouthernBight
of theNorth
Seawithisobaths
andinvestigated
area.The76measuring
locations
aredistributed
overtencross-transects
along
theDutch
coast.
Thelocations
areindicated
by
thefirstletter(s)
of thetransect
andtheirdistance
fromtheshore(km). Fromsouth
to norththe
transects
are:Appelzak
(A),Walcheren
(W),Schouwen
(S),Ooeree
(O),TerHeijde
CTH),
Noordwijk
(N),Egmond
(E),Callantsoog
(C),Terschelling
(Ts)andRotmmerplaat
(R).
In thispaper
westudy
thepossibleinfluence
ofthe(cross-shore)
density
drivencirculation
on thesuspendedmatterdistribution
usinga simpleanalytical
model.In section
27.2a
description
of theDutch
coastal
zoneisgiven.Section
27.3shows
observed
salinity
and
M Visser 553
29
Figure2: Averagesalinitydistribution
(O/DO)
overthefull period(1975- 1983).The freshfiverwater
is spreadnorth-eastward
alongthe Dutchcoast.
554 M Visser
The Dutch coastalzone is locatedin a shallowpart of the southernNorth Sea with water
depthslessthan30 m (Fig. 1). The greaterpartof thewatermassoriginatesfromtheEnglish
Channel.A smallpartentersthecoastalzonethroughthecontinental rivers,in particularthe
Rhine.
27.3 Observations
M Visser 555
salinity,
temperature,
suspended
matter
concentration
andarange
ofwater
quality
parameters
weremeasured.
Thesuspended
matter
concentration
wasdetermined
froma water
sample
of
1 lilxethatwasf'fitered
witha 0.45pmcellulose-acetate
filter. Theresidue
waswashed
with
demineralized
watertoremove
thesalt.Afterit wasdriedata temperature
of 105øC,
the
residue
wascooledandweighed.
Nodistinction wasmade betweenorganic
andinorganic
suspendedmatter.
lO %
10
0 •0gm
Figure
3: Average
suspended
matter
distribution
(mgl
-•)over
thefullperiod
(1975
- 1983).High
concentrations
aremeasured
infront
ofthecoast.
Furtheroff-shore
theturbidity
is muchsmaller
and
morehorizontally
homogeneous.
Howevera weaklocalminimumispresent
atsome 25kmoff-shore
northof Noordwijk.
556 M Visser
5 ß
..4
3 ,50
,
,/50
10'
6 *
20;/
5O
IO0
0 50 km
Figure
4: Average
suspended
matter
distribution
(mgl
4) over12January
cruises
(1976- 1983).A clear
minimumis presentat some25 km off-shore.
M Visser 557
Especially
in wintera pronounced
suspended
matterminimumis present.The average
Januarypattern (Fig. 4) shows a clear minimum at some 25 km off-shore. Individual
measurements oftenshowa localminimumin whichabruptconcentration
changes
occur.
Thesefrontalchangesoftencoincidewith a largedensitygradient.
/
/
/
/
,/
/
-".31
/
/
34 /
,/
25
35
25
33
o 50kin
Figure
5a: Salinity
in o/oo(a)andsuspended
matter
in mg1-1
Co)measured
during
thesurvey
of 5 - 8
March1979. A salinityfrontis coinciding
witha suspended
matterminimum.
558 M Visser
2x, , ./ .--" ,5 7
' •" ..................
•.10<?
1
34' '
/
2
10
ß
lOO
o 5okm
M Visser 559
(0, y, •(0))
(0, y, 0)
/I
/
! I
! I
! I
! I
! I
!
(0, y,-h)
! !
! !
! !
! !
! !
!
! ........:.........
! !
! !
! !
! !
! .... ::::::::::::::::::::::::::
! .................
! !
! !
! ! .... :.:.:.:.:..........................
! ..:.:.:.::.:.:.:.:.:.:.:.:.:.:.:.:
ß
!
/ !
z = ((x)
• • z y
I !
I !
I !
I !
z=-h I• (o, o,-h)
(x0,0,-h) x
560 M Visser
1028.0 i I I I I I I 3.5e-04
-3.0e-04
1027.0
- 2.5e-04
- 2.0e-04
1026.0
- 1.5e-04
1025.0
1024.0
1023.0
0
1028.0 3.5e-04
-
- 3.0e-04
.
1027.0
- 2.5e-04
.
- 2.0e-04
1026.0
- 1.5e-04
1025.0 - 1.Oe-04
p-gradient
- 5.0e-05
1024.0
- O.Oe+00
.
1023.0 I I I I I I I , -5.0e-05
0 5 10 15 20 25 30 35 40
M Visser 561
27.4 Modelling
To illustratethe possibleeffect of a densityfront we usea linear, stationarymodel,with a
coordinatesystemas shownin Fig. 6. In the modelwe describethe advectivetransportdue
to the densitydrivencirculation. In the northernpart of the Dutchcoastalzone (northof
Noordwijk) long-shoreand vertical density changesare inferior to the still very large
cross-shore gradients. In this part of the area a local suspended matter minimum often
occurs. Thereforelong-shoredensitychangesare not takeninto accountin the model.
in whichPo= 1028kgm'3thestandard
density
of seawater;Pdthedensity
difference
between
coastaland sea water; and Xoa cross-shore lengthscale,approximatelythe positionof the
largestdensitygradient. This densityfield is in reasonable
agreement with measurements of
the densityprofile in the CallantsoogWansectduringJanuary(Fig. 7).
Oz
g (27.2)
f u=kaZv (27.4)
Oz2
• + ' 0 (27.5)
ax az
562 M Visser
u a(
• I
w (27.8)
•x
w(x,-h) - 0 (27.9)
f_:
su(x,z)dz' 0 (27.10)
Properlywe haveto requirethatthevelocitycomponent normalto thecoastvanishes.But
in realitythebottomz = - h(x) depends
on thecross-shorecoordinate.Approaching thecoast
the depthdiminishes and is zeroat the coast. In thatcasewe haveto usethe boundary
conditionthatthevelocitycomponent normalto thebottomvanishes andthatu(0) = 0. In this
simplemodel,in whichwe usean uniformdepthh, we haveto usea transport conditionon
the coastlinex = 0. This impliesthat we allow that sUeamlines intersectthe coast. We
assumethatthereexistsa 'small'boundary layerin whichthedepthdiminishes to zero. The
coastalboundary conditionprovidesthatthereis nonettransport
intoor outof thisboundary
layer.
M Visser 563
z- ½ z- - k
[/1_••]
•--•,. Po
x•
k ' [0•]
•--t,. Po
x•
k (27.12
with thebottomstressproportional
to thebottomvelocity(Csanady,1982):
(27.14)
ß •-½ Po
k' •.½ Po
k
x W• :• . Wy (27.15)
For W we choose:
•(x,-h) - 0 (27.16)
Representative
parameter
values
arek = 0.01m:s'• (Bowden,
1983),Xo= 2 10• m,h = 20 m,
L = 10'3ms'• (Csanady,
1982)andp(x)prescribed
by(1) withPoTM
4.0kgm
'3. Withthese
boundaryconditionswe candeterminethehorizontalvelocitiesu andv, the 'stream'function
• andthesurface
slope• (theformulae
aregivenin Appendix
A).
564 M Visser
I I , I
-2
-/4_
-6_
-8_
-1/4_
-16_
-18
-2O
-35 -30 -25 -20 -15 -10
X (' 1000 m) •
u(x,z)
OCr
ax +v(x,z)
i]C+[w(x,z)
3-Y'y +W,]i]C
az . 0 (27.17)
describingadvectivetransportof suspended matterwith constantsettlingrateW,. Becauseof
the fact that we only look at advectivetransportwe can determinethe characteristic curves
alongwhichthe suspended matteris transported.The setof characteristiccurvesis givenby
the intersectionof two familiesof surfaces(seee.g. Kevorkian,1990):
M Visser 565
(27.18b)
h v(x,•)d• - y [w(x,z)+ W,]" cxmstaat
In absenceof wind, the projectionof the trajectorieson the x-z plane is given in Fig. 9.
For differentsettlingvelocitiesthe patternsshowconsiderable differences.Obviously,if the
verticalvelocitycomponent w is muchsmallerthanthe settlingvelocityall theparticleswill
immediatelysink to the bottom(Fig. 9a). If the verticalvelocitycomponentis muchlarger
thanthe settlingrate the particleswill follow the residualcirculationand hardlysettledown
(Fig. 9c).
566 M Visser
Fig.9a
X (' 1000 m)
0 I .... I , , • • I • • • • I , • • • I , , , • I • • I • , • , I , , ,
Fig.9b
_
-i
-1
X (* 1000 m) ;
Fig.9c
-•6•
-20
-35 -30 -25 -20 -15 -10 -5 0
X(' 1000 m) •.
Figure9: Characteristic
curves
of suspendedmatter(projection
onthex-zplane)withdifferent
settling
velocities
(a) W, =- 10.4ms-';(b)W, =- 104 ms-';(c)W, =- 10-• ms-'.
M Visser 567
Fig. 10 a
X (' 1000m)
X (' 1000 m) •.
Figure10: Characteristic
curvesof suspended
matterwithdifferentwinddirections
anda windspeed
of 5 ms4 (projection
onthex-z plane).a:northerly
(long-shore)
wind,b: easterly
(off-shore)
wind,
c: southerly(long-shore)wind, d: westerly(onshore)wind.
568 M Visser
-4_-
Fig.10c
•2
14
•6
•8
20
-35 -30 -25 -20 -IS -•0 -5
X (* 1000 m) --•
Figure10: Characteristic
curves
of suspended
matterwithdifferent
winddirections
anda windspeed
of 5 ms-1(projection
onthex-z plane).a:northerly
(long-shore)
wind,b:easterly
(off-shore)
wind,
c: southerly(long-shore)
wind, d: westerly(onshore)wind.
M Visser 569
Not only the wind directionis importantbut alsoits magnitude.Fig. 11 showsthe pattern
witha southerly
windwitha magnitude
of 10 ms']. In thecaseof a windvelocityof 5 ms']
(Fig. 10c) the suspended
matterminimumis locatedat some28 km off-shore,while in the
caseof a magnitude
of 10 ms'] theminimum
is located
closerby thecoast(at some24 km
off-shore). This is due to the fact that the circulationin the surfacelayer is predominantly
forcedby the wind. The influenceof the densitydrivencirculationis smallerin the caseof
a strongerwind. Thereforethe particlespresentfar off-shorein the surfacelayer can move
furtherto the coastbeforethey sink out of the surfacelayer.
-2
-4
-8_
-12_
-16_
-18_
-2O
-25 -•0 -15 -10 -5 0
X (* 1000 m) g
Figure11: Characteristic
curvesof suspended
matterwith a southerly
(long-shore)
wind of 10 ms-*
(projection
on the x-z plane). Comparing
to Fig. 10c(southerly
wind5 ms-*)thepositionof the
suspended
matterminimumis switched
to thecoastat some24 km off-shore(against28 km in Fig. 10c).
570 M Visser
M Visser 571
particlesremainin suspension
overthewholearea. The rangesizedistribution of suspended
mattervariesasthedynamicsof the systemchange.Storereel(1949) describes a comparable
sortingprocess.Flocculation
is notan important
process in thisarea:salinitydifferences
are
"small"andsuspended matterconcentrations
are relativelylow.
Acknowledgements
I would like to thank W.P.M. de Ruijter, H.E. de Swart and D. Visser for discussionsand
information.FurtherI acknowledge the North Sea Directorateof Rijkswaterstaat (Rijswijk,
NL) for puttingthe dataat my disposal.The investigations were supported by the Working
Group on Meteorology and Physical Oceanography(MFO) with financial aid of the
NetherlandsOrganizationfor the Advancement of Research(NWO).
27.6 References
572 M Visser
Murray,S.P.,1970, Settling
velocities
andverticaldiffusion
of particles
in turbulent
water,J.Geophys.
Res., 75, 1647-1654
M Visser 573
Appendix A
In this appendixthe formulaefor the pressurefield, the horizontalvelocities,the stream
functionand the surfaceelevationare given.
Substitutionof (27.3) and (27.A0) in (27.4) and applying(27.12) - (27.14) leads to the
horizontalvelocitycomponents u and v:
U •
2pofC c•x
2Poke
+
2PokC
•b
• (•-2a')
sin (•cosh
(q)
- co•
(•-2d)
anh
(• ,]
cosh(•2a'). ,x..(q sinh (•2a')
y sin(• +d)cosh
(•-d) - cos
(• +d)si• (•-a')-]
2pokC - an (•-a') cobh(• +d) + • (•-a') sinh(• +a')]
(27.A1)
574 M Visser
2p•fc ax
- sin
(n+eO
co•
(n-eO
+co•n
+a')
• (•-a')
.]
z
+
2PokC
•b
- sia
(n-2a')
• cosh
(n)+• (n-2a')
siah
(n)+]
(n) ½,mh(n-2a') - ex• (n) •.h (n-2a')
2PokC ,[-sia
(n+a')
cosn
sin (n-a')
- ex,s
(n+a')
siah
(n-a)
+]
2PokC sin
(q-2d)
cosh
(q)
- cos
(q-2d)
siah
(q)
+]
sin (n) ½osh(q-Za) + cos(q) •.h (n- •d)
+ [o(d- n)]
Po/ Ox
(27.A2)
where:
M Visser 575
z+h •+h
q- ; d-
(27.A4)
P a-• =- ax
,{, f [sin(d)
cos(d)
-sinh(d)
sic [- sia(d)•
(d) c,
osh(d)
+ c• (a') ]}
si• (a')J
f
% [sin (d) cosh(d) + cos (d) sinh(d)]
8kC
f
[sin(a')cos(d) + cosh(• sinh(a')]
8kC
(27.A6)
+ L-sinh(d)cosh
(d)- cos(d)sin(d)
(27.A7)
where:
576 M Visser
õL [sinh
(2d)- sin(2d)] (27.A8)
Nowoursystem
is completely
determined
if weprescribe
windanddensity
field.
28
Abstract
28.1 Introduction
The CentralZone of the Great Barrier Reef Marine Park (CGBR) encompasses
mostof the
coastal and continental shelf waters off Townsville in North-east Australia. The shelf in this
regionis about120 km wide and its bathymetrygentlyslopesfrom the coastto a depthof
about 100m at the shelf-break(figure 1) with the isobathsapproximatelyrunningparallel to
the shoreline.Thusthe along-shore directionof thispartof the shelfis definedto be roughly
alignedwith the north-westward, south-eastward axis.
95 147
19S
- g
North
• •' • Site
4
• Site 3
• x
Townsville x
Site 2
Site I
Bowling
Green
Bay
Upstart Bay
Site 5 •
0 20 ß
I x 65
Figure 1' Grid boundariesof the studyareawithin the cenlxalGreat BarrierReef. The currentdata
collectedfrom within the regionare from mooringslocatedat sites1-6.
Analyticalmodelsof the currentsin the CGBR predicttidal ellipseswith the major axis
predominantlyin the across-shelfdirection (Church et al. 1985). Further, numerical
simulationsby Andrewsand Bode (1988) showthat the tidally-inducedcomponentof the
currentsin this regionis generallyweak (peaktidal currents< 0.2 ms'l) and field
measurements indicatethatthetidalcurrentsarehighlymodulated anddominatedby thesub-
tidal flows(WolanskiandPickard,1985). Work undertaken by AndrewsandFurnas(1986)
suggestthat the tidal flows would not contributesignificantlyto the large-scaleadvective
processes of this region.
Duringtheaustralwinter,thewindfield is dominated
by thesouth-east
tradesandis highly
coherent
overdistancesgreaterthan1000km (Wolanski,1982). The windspeedgenerally
ranges
from0 to 15 ms4 andblowspredominantly
north-westward
(i.e.along-shelf)
in the
CGBR with typicalwindeventslastingfor a periodof 1-2 weeks. Non-tidal,along-shore
currentshavebeenfoundto be highlycoherentwith thelocalwindforcingduringthistime
(Wolanski andBennett,1983;Bunageet al. 1991a).
0.5
0.0
--0.5
0,5
0.0
• o.o
-0.,5
0.5-]
0.0
-0.5
12.0
4.0 e
-
-4.0 • • •
120 150 180 210
Time (d•ys)
Figure2: Time-series
plotsof thehydrological
andmeteorologicaldatausedfor modelcalibration
and
forcing.The sub-tidalalong-shelfcurrents
(V), fxomsites1 - 4, aredepicted
in graphs(a) - (d)
respectively.
Thelow-pass
filtered,
along-shelf
windforcing
(Wy),asmeasured
in Mackay,
is shown
in graph(e). The timeis givenas thedaynumberin 1985andcoverstheaustralwinterseason.
Site Regression
Eqm. R" Lag(HR)
4 V4= 4.92x 10'•'Wy- 0.312 0.463 24
3 V• = 4.00x 10'•'Wy- 0.182 0.737* 15
To understand
and studythe sub-tidalhydrodynamiccharacteristicsof the coastalflows in the
CGBR, a detailednumericalmodelof the entireshelfwas set-upin orderto encompass the
small-scale
andshelf-scale
influencesin thisregion. The numericalmodelusedto achievethis
is basedon the work of Leendertse(1970) andFalconeret al. (1986) and solves,usingan
implicit scheme,the followingdepth-averaged
two-dimensional hydrodynamic equations:
c• & C2 - W
-% p
(28.2)
8 LrH il VH
/}q * * - 0 (28.3)
The depth-averaged
equationsare valid for this applicationdue to the predominantly
horizontal flow which occurs in shallow shelf waters. Three dimensional currents have been
observedaroundreefsand in wakesbehindislandsand headlandsbut thesepresumablywill
contributelittle to the large-scaleflows and are mostlikely to be sub-gridscalephenomena
in thismodel(WolanskiandHamnet,1989;Deleersnijder, et al. 1989). Barocliniceffectsand
atmospheric
pressure
differences
duringtheaustralwinterare negligible.
model'sresponse
to thisforcingacrossthe shelf. Thesecomparisons
are listedin table2.
Table 2: Comparison
of the numericallypredictedalong-sheff
component
of the currentvelocity(Vm;
ms'l),withtheobserved
(V•,;ms-l)values
extrapolated
fromthecorresponding
regression
equation,
listed
in table 1, for the 6 givenwind conditions.
Run # 1 2 3 4 5 6 7
sit#
-0.31 -0.31 -0.21 -0.21 -0.12 -0.12 -0.02 -0.02 0.08 0.08 0.18 0.20 -0.31 -0.25
-0.18 -0.18 -0.10 -0.09 -0.02 -0.01 0.06 0.09 0.14 0.12 0.22 0.17 -0.18 -0.18
-0.10 -0.10 -0.02 -0.03 0.05 0.06 0.13 0.11 0.19 0.20 0.27 0.26 -0.10 -0.08
-0.06 -0.06 0.00 -0.01 0.06 0.05 0.12 0.08 0.17 0.20 0.23 0.22 -0.06 -0.03
gH8q
Ox
=fVH (28.4)
Simulations
wereperformedfor sixdifferentwindconditions
whichdominatethewindfield
duringtheaustralwinter(Wolanski,1982).Theseconditions
assume south-easterly tradewind
withvelocities
rangingfrom0 to 10ms'1at 2 ms'1increments,andareidentified asrun# 1 to
#6 respectively.A simulationwasalsoperformedidenticallyto run # 1, but withoutthereef
matrix (run #7) in order to assessthe influenceof the reef on the near inshorewaters.
The model was started from rest and allowed to run for 120 hours in order to reach a
steadystate. Testswere thenperformed,utilizingthe along-shelf
slopealongthe outer
boundaryunderno wind conditions,usingvarioustimesteps,to examinethe effectsof
numericaldiffusionon the simulation. The predictedflows for each test revealedno
significant
dependency
on timestep.The timestepwassetat 600 seconds.
Finally,thenumerically-predicted
currentsof thefirst6 runswereusedto drivea simple
advection
model.Theadvectionmodelutilised
thesheif-model velocities
todetermine
drogue
28.4 Results
IIIIIi111
IIIIIIii1111
•111111111
slllllillll
II•111111111
s%11•11111111
• • i I I I I I I I I I I I
] .... •11111111111111
/ --''•s•11111111111
Length
scale
,,,,
z/ /I//
/ ...... .......
, • •;;txilXl //
VelocityScale • 0.25
Figure
3b:Thepredicted
current
profiles
ofthemodeldomain,
forced
byanalong-shelf
slopeonly
(ran
#1). Thelengthof thearrowsrepresents
therelative
current
velocity
strength
anditsorientation
indicating
thecorresponding
direction
offlow.Part(a)shows
thepredicted
shelf-scale
flowswhile
(b)
givesa subset
ofthedomainfocusing
onthepredictedflows
within
thecoastal
zoneoftheCGBR
lagoonandits embayments.Notethevelocity
shearacross
thelagoon
andthevariation
incurrent
strengthalongthe lagoon.
Figure4: Snapshots
of drogue distribution
drivenby anadvectionmodelusingtheoutput of run#1.
Plot(a) shows
theinitialdistribution
of 26 drogues
alongthetransectY=55 for 10 < x _<35 at time=
0 days.Theremakning plotsgivethedistribution
after(b)2 days,
(c)4 days,
(d)6 days, (e)8 daysand
(f) 10 days.
Figure
5:Theflowpatterns
whichexistinthecoastal
region
fromsimulation
#2. Thedecoupling
of the
currents
across
theshelfthatresults
is highlighted
by a separation
zoneandstrong
shearpatterns.The
wind forcingis directedup the page.
a a b a
c d
e f a
: '•.
Asthewindvelocitystrengthensfromthesouth-east,
thelocalwindforcingdominates and
flowreversaloccursacross
theentireshelf.Undertheseconditions,
velocity
shearstilloccurs
nearthecoast andisparticularly
evidentin theembayments.Seefigure7 forexample.The
existence
of aneddyin BowlingGreenBay,in theleeof thecape,canalsobeobserved. This
will resultin sometrappingof the waterin thisarea.
•, / \.xx•tttttttt
, I • ltltttttlttl
• • • f ffttt tttt tt
..... ///flftttttltt lilt
..... ,•ftttttttttt 1111
..... •tttttttttttt
..... ,•T t tit t ttt t 1
...... ,\•111t tittit
....... \•111 tt ttttt
.. .ITtttt t flirt
- •z//lfff f ill I!
•,//II ftt f I It I f
_....• • / / I I f t f f t t I t 1
/,/zz//tttfftlttttt
/ •ttttttf fttttttttt
/..,•,,,,,x\\•11
\ ::
\ '"'-'/.//,///,;III
! / / / • /II
.. • /////,'/II
" ///IZll, II!
• /• //////,'//!
• ... /,/////////
/,////II/I////
/•////////////
• •/////////Z//
/,///////////ZZ
•////////////Z•
-_)s////////////Z•
_ ¾, /////////////Z•
• !, ////III//////Z•
• i %//////I/I/I//ZZœ•
/ .... ,/••////II/ZZœ•
,,,,,•, •//III///////
Velocity Scale • 0.25 m/s
Figure7: Increasing
windstrength
(run#3)causesflowreversal
across
theentirelagoon
andshelf.This
leadsto strongshearzonesacross
themouthsof theembaymentsandaneddydevelops in theleeof
CapeBowlingGreen. The wind forcingis directedup thepage.
a a b a
Length Scale
5000 m
28.5 Discussion
Thisstudyhasimportantimplications
for thesoundecological
management of thecoastal
andreefsystems.Thatis,anyland-derived
matter(suchasdissolved
nutrients,
pesticides
and
Finally,whiletheacross-shelf
tidalexcursiondoesnotsignificantly
contributeto theacross-
shelf exchangeof water, the role of tidal diffusionin mixing the watersis yet-to-be
determined.Although,from an ecologicalperspective, this processwouldhelpdiluteany
injectedcontaminants,furtherresearchinto its effectsis warranted,in orderto give some
measureof the mixing ratesin the coastalzone.
28.6 References
Andrews, J.C. and Bode, L., 1988, "The tides of the central Great Barrier Reef'. Continental Shelf
Research.,Vol. 8, 9, 1057-1085.
Andrews, J.C. and Furnas, M.J., 1986, "Subsurfaceintrusionsof Coral Sea water into the central Great
BarrierReef - I. Structuresand shelf-scaledynamics".ContinentalShelf Research.,Vol. 6, 4, 491-514.
Church, J.A., Andrews, J.C. and Boland, F.M., 1985, "Tidal currentsin the central Great Barrier Reef'.
Continental Shelf Research,5, 515-531.
Deleersnijder,
E., Wolanski,E. andNorro,A., 1989,"Numericalsimulation
of thethree-dimensional
tidal
circulation
in an island'swake".Proceedings
of the 4th International
Conference
on Computational
MethodsandExperimentalMeasurements,
Capri,Italy, 23-26 May, 1989.
Godfrey,J.S., 1973b,"Comparison
of theEastAustralian
Currentwith the westernboudaryflow in
BryanandCox's (1968) numericalmodelocean".Deep-seaResearch,20, 1059-1076.
Olafson,
R.W., 1978,"Effectof agricultural
activityonlevelsof organochlorine
pesticides
in hardcorals,
fish and molluscsfrom the Great BarrierReef". Marine EnvironmentalResearch,Vol. 1, 2, 87-107.
Wolanski, E. and Bennett, A.F., 1983, "Continentalshelf waves and their influence on the circulation
aroundthe Great BarrierReef'. AustralianJournalof Marine andFreshwaterResearch,34, 23-47.
Wolanski,
E., Burrage,
D.M. andKing,B., 1989,"Trapping
anddispersion
of coraleggsaroundBowden
Reel GreatBarrierReel followingmasscoralspawning".
Continental
ShelfResearch, Vol. 9, 5, 479-
496.
Wolanski,
E., Jones,
M. andWilliams,
W.T., 1981,"Physical
properties
of GreatBarrierReeflagoon
watersnearTownsville.2. Seasonal
variation".Australian
Journalof MarineandFreshwater
Research,
32, 321-334.
Wolanski,E. and King, B., 1990, "Flushingof BowdenReef lagoon,Great BarrierReef'. Esmarine,
Coastal and Shelf Science, 31, 789-804.
Wolanski, E. and Ridd, P., 1990, "Mixing and trapping in Australiantropical coastalwaters".In
"ResidualCurrentsand Long-termTransport",R.T. Cheng(Ed). Springer-Verlag,New York.
29
Abstract
29.1 Introduction
bottom waters since the 1930s (Newcombe and Home, 1938), and has become more
widespreadand of longerdurationduringrecenttimes(Flemeret al., 1983).
•7o
00'
) KILOMETERS
Figure1'. LowerChesapeake
Baywithitsfourmajorsubestuaries
(Theshaded
areasaredeeperthan20
m, the depthcontoursare 13 m; X marksthe stationwith time seriesmeasurements
and O marksthe
slackwatersurveystationin the bay).
29.2.1 Slackwatersurveys
DISTANCE UPSTREAM FROM MOUTH (Kilometers) DISTANCE UPSTREAM FROM MOUTH (Kilometers)
Figure2: Distribution
of dissolved
oxygen
(mg14)andsalinity(psu)in thelowerRappahannock
River.
a) August10, 1987 andb)August 17, 1987 (-marks the measuringlocation).
--1 1
--4
--7
--10
--1.3
/---
-• -
.
t
--19
/
--22
/
--28
¸
z
,
z
-34- -F
/
o) DO
-57-
-4.0 -
,,,/
o
0
z
z
--
--430 •0
I I I
20
I
50
I ,
40
I ,
50
Figure3: Dissolved
oxygen(mg1-1)andsalinity(psu)in themainstemof Chesapeake
BayonAugust
34, 1987 (The largertick markson abscissa
indicatemeasuringstations).
-3
-5
/
-6
-7
-8
(M)
-9
-lO
-11
-12
-13
91 121 51 181 211
-1
-2
-3
.'I••I r,
(I I /1I I11
Ii )ib)
iSalinity
ii i 1
-4
-5
D
E
-6
P
T -7
H
-8
(M)
-9
-lO
-11
-12
-13 I I
12_1 151 181 2_11 2_41
dulian Days
Figure4: Dissolvedoxygenandsalinityisopleths
in depth-timeplane(The tick marksat thetopof the
figure indicatesamplingdates).
Currentvelocitiesweremeasuredwith in-situ,selfrecordingmetersthatweredeployedwith
taut wire mooringsat the river mouthstationand the station16.6 km upriver. Two typesof
currentmeterswere used;Inter-OceanS4 metersand modifiedBrainconHistogrammeters.
The S4 meteris an electromagnetic typecurrentmeterwith solidstatememory,someof which
are also equippedwith temperature,conductivityand pressuresensors.The meterswere set
to recordthe averagevaluesof all variablesevery 30 min. The Brainconmeter measures
currentmagnitudeby a savoniusrotor, with a vane attachedfor directionmeasurement.All
threeBrainconmetersdeployedfunctionedproperlyfor a periodof only a few daysand then
succumbed to electronicproblemsor fouling. This leavesavailabledatafor two depthsat the
fiver mouth station,and for threedepthsat the upriver station.
06
oo
-06
06
O0
-06
22
"• 20
._
Julian Days
where
m(t) = - u(t)c(t)
0.02 0.72
DO (mg1-•) 5.63 5.61
- u x DO 37.9 41.1 -3.19
The net massfluxesof salt and DO by the tidal componentamountto only 1% and 8%
respectivelyof thoseby the subtidalcomponents (Table 1). Therefore,we needto consider
only the subtidaltransportin studyingthecharacteristics
of the estuary-subestuary
exchanges
of salt and DO in this case.
ug/u• = 0.32
whereugandu• are,respectively,
themaximum
density
drivenandwind-driven
velocities
in
the lower layer of the water column. Applicationof the formulaproposedby Hansenand
Rattray(1966) gives
u• = -3.3cms'•
Uw= - 10.2cm s'•
incoming
watermasses
werecharacterized
by lowDO andhighsalinity,
typically
3 to4 mg
1'• lowerDOand2 to 3 psuhigher
salinity
thanthose
of otherperiods.
Theimport
of low
DO, highsalinitywatermasses
wasalwaysaccompaniedby increased landwardcurrentand
southwest
wind. ComparingFigs.5d and5e with Figs.3 and4, it is evidentthatthe low
DO, highsalinitywatermasses
werederivedfromthelowerportionof thewatercolumnin
the bay. In the absenceof strongsouthwest
wind, the shallowsill off the fiver mouthblocks
the bottomwaterin the bay from enteringthe fiver, andtherefore,
the importedwateris
derivedfromthesurface layerin thebay,withhigherDO andlowersalinity.
29.5 Mechanisms
of windinducedmassexchange
Asindicated
inFig.5, theintrusion
of thelowDO,highsalinity
waterintotheRappahannock
Riverhasa timescaleof 2 to 3 days. Thisagreeswiththetimescaleof theeffectof local
windforcingin ChesapeakeBay. WangandElliott(1978)observedthatthesubtidalsealevel
fluctuations
in Chesapeake Bayhadspectralpeaksat periods of 20, 5 and2.5 days.Their
analysesassociatedthe 2.5 day fluctuations
with the localwindforcing,whilethe lower
frequencyfluctuations
wererelatedto non-local windforcingthrough theinteraction
with
coastalsea. With localwindas forcingfunction,therearetwo possible mechanisms that
contribute
tothetransportof thesubpycnocline
waterof thebayintotheRappahannock River.
For simplicity,we may look at the wind-drivencirculationaroundthe fiver mouthin two-
dimensional
space,
onein thehorizontal
planeandtheotherin thevertical
plane.
29.5.1 Horizontalplane
Fig.6 depicts
theforcing
of bottom
current
around thefivermouth.Theleftpanelis for
thewindfromthesouthwest. Surface
set-up
in twotransectsacross
thebayis considered
(Fig. 6b). The surfacecurrentin bothtransects
is towardtheeastbecauseof directwind stress
and Corioliseffect. The boundaryon the eastcausesthe watersurfaceto rise and a return
westward
flowalongthebottom.Thereis a difference
in thedegree
of surfaceset-downon
thewestern
boundary
betweensections
AA andBB. Owingto thesurface out-flowfromthe
fiver, the set-downat sectionAA will be lessthan that at sectionBB. This resultsin a
barotropic
flowawayfromthefivermouth
alongthewestern
boundary
of thebay(Fig.6c).
Therefore, thebottomwatertransported
intothefiveris mainlyderivedfromthemainstem
of thebay wherehypoxicconditions existin thesummer.On theotherhand,if windis
blowing fromthenortheast
quadrantthesurfaceset-upatthefivermouthwill belessthanthat
at sectionBB. Thisresults
in analongshorebarotropicflowtowardthefivermouth.If the
density-driven
gravitational
circulation
at thebottom of thefivermouth is strongerthanthe
wind-drivencurrent,
thebaybottomwaterwill stillbetransportedintothefiver. However,
thisbottomwaterisderived
fromnearshore regionstothenorthandsouth of thefivermouth.
Because oftheshallowdepths
inthese
regions,theimported watermass hashighDOandlow
salinity.
BAY
BAY
A•.•RIVER
1
b). SIDE VIEW
SECTION
AA ß ,
T •
SECTION BB
T
Ug
• • Uw
•
proposedtoexplain
theobserved intrusion
of hypoxicwaterontotheshallowflanksofthebay
(Tyler,1984;Maloneet al., 1986). On theotherhand,if strongwindis blowingfromthe
northeast,
thepycnocline
will bedepressed belowthesillonthewesternsideof thebay(Fig.
7c). Thewatertransported intothefiverwill be thatabovethepycnocline,thuswith high
DO, low salinity.
rivermouth
t v no
wind /
a) I g pycnocline
/•
,• '
29.6 Discussion
Acknowledgements
This studywas fundedby the Commonwealth of Virginia as part of Chesapeake Bay
Initiatives.The authorswishto expresstheirappreciation
to Mr. K. Curlingfor providingthe
dataof the main stemof the bay, collectedunderthe US EPA Chesapeake Bay Program.
29.7 References
Kuo, A. Y., andB. J. Neilson, 1987: Hypoxia andsalinityin Virginia estuaries.Estuaries,10(4), 277-
283.
Wang, D. P., A. J. Elliott, 1978: Non-tidal variabilityin the ChesapeakeBay and PotomacRiver:
evidencefor non-localforcing. J. Phys.Oceanogr.,8, 225-232.
30
Abstract
An interdisciplinary
studywascarriedoutaroundthetidalfromin theBungoChannel,Japan.
Temperatureand salinitydifferencesacrossthe tidal front are 3'C and 0.3, respectively.
Nutrientsare richer in the well-mixedregionthanin the surfacewater of stratifiedregion
wherethey havebeenconsumed by phytoplankton.Man-madeorganicchemicalssuchas
PCBs are concentrated in the surfacewaterjust on the front andthey are accumulated
in the
planktonwhich are gatheredthere. Neustonsuchas marine skatersand Vellela lata are
accumulatedon the front. The tidal front plays as a barrier for some oceanic and neritic
pontellidcopepods.
30.1 Introduction
30.2 Observations
bottomtopography
andthemapof theobservation
siteareshownin Fig.1.
131ø20'E 132ø00'E
33ø40'N
THE SEA OF IYO
KUNISAKI
60
PENINSULA
SHIKOKU
..
60
:!: HAYASUI
BEPPU STRAIT
¾• AJIMA
KYUSHU •v
)1) o+5
JAPAN• c;Pti +6
•
•oo• 1
BUNGO CHA'NNEL
32ø50'N
(Km)
Figure1: Observation
stations
in theBungoChannel.Numbered
contours
showthedepthin metres.
Verticalprofilesof temperature
and salinityweremeasured at 11 stations
alongthreelines
(W1-W3, C1-C5 andEl-E3) from 16 to 19 June1986,at 4 stationsalongone line (P1-P4)
from22 to 24 May 1987andat 6 stationsalonganotherline (1-6) on 21 May 1988by using
a CTD (ALEC Model AS-100). Water andplanktonsampleswere collectedat thesestations
with use of a pair of Van Dom water bottlesand planktonnet, respectively. Water
temperature,salinityandchlorophylla weremeasuredcontinuously at a depthof ca. 3 m
alongtheaboveobservation lines(waterwaspumpedup from 3 m depth)usinga multiple
environmentalsenser(Martec Mark VI) anda fluorometer(TurnerDesigns)in orderto detect
the exactpositionand sharpness
of the front.
30.3 Physicalaspects
Water temperatureand salinityabruptlychangedfrom 18C and 33.8 to 22'C and 34.3,
respectively,
acrossthe frontalregionwith a widthof ca. 500 rn on 18 June1986. Density
('C)
22.0
c-1 c-2
....
c-3 c-4 c-5
T('C)
t34.4
20.
o[
18.0[ t34.2
34.0
33.8
ira)
ø1 ' X-•' '24.o
18.0
II •..• :____'----.'--
:Z-_._2.4.2
. 24.4
....:•' 2 .0
:(•,'/. '....
lOO
The positionof the front was noticedeasily from the researchvessel,as the front was
accompanied by a narrowband of foam and variouskinds of drifting materialon the sea
surface.Suchaccumulation of driftingmaterialssuggests theexistence
of surfaceconvergence
at the tidal front. This basicphysicalstructurewas maintainedconsistently throughoutour
observation periodswith only minor spatialchangesaccordingto tidal phase. The detailed
physicalcharacteristics of this tidal front are describedby Yanagi and Tamaru(1990).
30.4 Chemicalaspects
Representativevertical distributionsof DIP (DissolvedInorganicPhosphorus)and PP
(ParticulatePhosphorus)on 24 May 1987 are shownin Fig.3. The front existedexactlyat
Sta.P3, while StasP1 and P2 were situatedin the well-mixedregionand Sta. P4 in the
stratified
region(Matsuda,1990).Theconcentration
of DIP ismoreimportant forthegrowth
of phytoplanktonthanthatof DIN because
phosphate
is a limitingfactorin the$etoInland
Sea(Joh,1986). DIP wasabout0.3 pg-at/1in the well-mixed
regionandit abruptly
decreased acrossthe tidal front to about0.15 pg-at/lin the surfacelayerof the stratified
region.On theotherhand,PP waslow (about0.1 pg-at/1)in bothsidesof thetidalfrontbut
high(about1.15pg-at/1)at the tidalfront. This is associated with higherphytoplankton
biomass asevidenced by higherconcentrationsof chlorophyll
a andpheo-pigments justat the
tidalfront(Fig.4). Thedissolved oxygenconcentration washighin thesurface layerat the
tidalfront(Fig.5). Thesefactsindicatesthattheactivityof phytoplanktonis highat thetidal
front.
P2 P3 P4
W ,
L
2O
. .X
0.2
4O
• 0.25'""
6O
8O
100
2• •ay • 987
(Ill)
Stn. P1 P2 P3 P4
2O
ß 0.10/•1 15 1.I0
0.10
4O
ß .
lOO
0
""";
'•t/• 5
) 10k•
(m)
Figure3: Vertical
distributions
ofDIP andPPfromSta.P1
toSta.P4
on24May1987.Tidalfrontexists
on Sta. P3.
Stn, P1 P2 P3 P4
20
40
80
lOO
20
40
6O
2O
lOO
(m)
Figure4: Verticaldistribution
of chlorophyll
a andpheo-pigments
fromSta.P1to Sta.P4on24 May
1987. Tidal front exists on Sta.P3.
Thereasonsfor thehighconcentration
of chlorophyll
a at thetidalfrontmaybe explained
asfollows.Phytoplanktoncannotgrowsuccessfully
in bothsides of thefront.Because
they
can not stayin the shalloweuphoticzonedueto strongverticalmixingin the well mixed
region,andnutrients arelackingin thedeepeuphotic
zoneof stratified
region. Ontheother
hand,phytoplankton
cangrowsuccessfully
nearthefrontwheretheextentof verticalmixing
Table 1' Concentrationof DDE and PCBs in planktoncollectedfrom the surfacesea water in the
northernpart of the BungoChannel(Tanabeet al., 1990).
Stn, P1 P2 P3 P4
or ß .-- • •___---. II ' i
6o
8O
DO (%)
.1.oo ß
(m)
Figure5: Verticaldistribution
of dissolved
oxygen
fromSta.P1 to Sta.P4on24 May 1987.Tidalfront
exists on Sta.P3.
623
T Yanagi,0 Matsuda,S TanabeandS Uye
Stn.1 2 3 4 5 6
I I I I i I
19.0
T e m p . (øC)
18 0
16 0
t -
Ch1. a(l•g/t2)
i ---- ___2_-
5.0- t --
---- i__'•-
4.0
3.0 ---i•
2_
z.oT
1.0 I___
0.0 . ._-----:
.....
60
70 - _
80
100
90 !_4_•
0 5 10km
Figure6: Horizontalvariations
of watertemperature,
salinity,chlorophyll
a andturbidity(reciprocal
of
beam transmittance)ca. 3 rn below the seasurfacefrom Sta.1 to $ta.6 on 21 May 1988.
0 m
Stn.
C-1 C-2 C-3 C-4 C-5
! I I ! I
50
lOO
(ng/[)
T Yanagi,O Matsuda,
S TanabeandS Uye 625
.• b•nd
/ 1000
m
2
Marine skaters 0
o ind./1000
m2
i:!iii!i!iiiiii!ii!i:i:i:i:!:i:i:i:!:i:i:i:i:!:!:i:!:i:
......... oC oEO///
'"'"'•'"•:•.•••"..••"...••ii:•i',
'""•:•'."•
...... •c •t••••_•
W30 •
0C4
0C5
33.8 •o
870o
• • 2 18øC
0C5
Figure8' Horizontaldistribution
of marineskatersandVelellalata in the surfacelayeron 18 June1986.
626
T Yanagi,O Matsuda,
S TanabeandS Uye
o Labidocera rotunda
E1 o o tnd./100
m3
-- t-10
c• ©"•/ -•-•oo
• C2 18øC
• C3 o
C4
-- O Labidoceradel:runcata _
OCEANIC
PONTELLIDS o L. acu•rrons
,• Pontella securtfer
[] Pontellopsts vttlosa
....:...• '•'
• Pontetllna
P mortl
olumata
E1
'""
• 3.8 V•,•E/•
•.•Xo
i•"'"""'"'•:':':•""•••
•o
• 2 __ '•"'•øø•
Jl
--11--100
C1 V -- 1-10
I!-''"'"'"--'•""""•"•:':'.'•:•••U:i::.,•
'18 øC _ • " •,.x,,,
•
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
,•. /•
"h •
E318•C
/
=======================
.......
18
•c •""
::i:i:i:i:i:!:!:i:i:!:i:!:i:i:i:i:!:i:!:!!!:i:!:!:!::"'• c3
:!:::':'"'
"':':':':'•:•i:i::•/ v o4
?:?!:i:i:i:i:i:i..........
:i:[:[:?i:!:i:?i:[:i:i:i:i:i:i:55i:[:i:!:55i:i:55i::':':"' v
Figure
9: Horizontal
distribution
ofneritic
andoceanic
Pontellid
copepods
inthesurface
layer
on18
June 1986.
627
T Yanagi,0 Matsuda,S TanabeandS Uye
30.5 Biologicalaspects
Epineuston
(whichliveson theseasurface)
suchas marineskaters
whosespecies
is not
identifiedandpleuston
(whosebodyexistsacross
theseasurface)suchas Velellalata are
accumulated
aroundthe tidal front as shownin Fig.8 (Uye et al., 1990).
0 ! I I 0 . . .I,
20 •s
3
2C ß
_
2O
40 2o 40 0
60
i Zooplankton
i. •0
( Number/I )
. .
8 .. ß :: : .....
CopepodNauplii ( Number/I )
pl P2 P3 p•,
I I • I
60
$0
,..........
.,.6:
. . .....;::...
Copepod Adult ( Number/I )
ß .
Figure10: Verticaldistribution
of zooplankton,
copepod
naupliiandcopepod
adultsfrom Sta.P1to
$ta.P4 on 24 May 1987. Tidal fxontexistson $ta.P3 whichis shownby an arrow.
Figure10 showstheverticaldistributions
of totalzooplankton,
copepodnaupliiandadult
copepodsacrossthetidalfronton 24 May 1987. Theirpopulation
densities
werehighin the
surfacelayer of the stratifiedregion.
50
•-•
O•
'-•::n
30
o
oo
o• 20
o ---
3 6X-16 9
,
o o
LU 10
i i i
Chlorophylla Concentration(pg/i)
Figure11. Relationship
betweeneggproduction
rateof thecopepod
Paracalanus
sp.andchlorophyll
a concentration.
30.6 Discussion
The tidal front is generatedin a transitionzone from the well-mixedregionto the stratified
one.Nutrientsare rich in the well-mixedregionandpoorin the surfacewaterof the stratified
region.On the otherhand,the biomassof phytoplankton is high at the edgeof the stratified
regionbecause the verticalmixingis too strongin the well-mixedregionfor phytoplankton
growth. The oleophilicchemicalsare concentrated alongthe tidal front and bioaccumulated
in theplanktonwhicharealsogathered alongthetidalfrontdueto thestufface convergence.
Coastal
area/T•dalfront
low
water temperature
/
_ /-'
/
r•
Offshore
area
low
salinity
// / hwater
iwgah
richnutrient.1•/h?_h /..
temperature
•/ PCBs.///' highsailnit
lowChl.a ,"'/high ,'/"/- '"-'7' Y
,,/' '•.•'h'l
• ,'/•'/ poornutrient
neriticzooplankton
//'
_h•gh neuston,/
Chl.a..// low
I --Chl.a-'-'
//abundances/'
•'• /_l,o_w
n,
aupli!/
oceanic
zooplankton
well-_.m
..... / ", ..
euphotic layer
Figure12. Schematic
representation
of thephysical,
chemical
andbiologicalstructures
of thetidalfront.
Acknowledgement
The authorsexpresstheir sincerethanksto Prof.T.Onbe and Mr.S.Ohtsukaof Hiroshima
University for their commentson zooplanktondistributionand the crew of T & R/V
Toyoshio-Marufor theirkind cooperation
duringthe field observations.
30.7 References
Uye, S., T. OhtsukaandT. Onbe (1990): Biologicalprocessaroundthe tidal front. In "The Scienceof
Siome"ed. by T. Yanagi,Koseisha-Koseikaku, Tokyo, 78-100.
31
Abstract
31.1 Introduction
632 T L Shaw
depth and width required to accommodatethe number and size of turbinesand other
equipmentrequiredto harnesstheavailableenergy.Eitherexcessive or insufficient
depth
and/orwidthcanintroduceseriouscapitalcostpenalties
to thistypeof project.Fig. 1 shows
the locationproposed
for the SevernBarrage(D.En. et al, 1989); thisis towardstheseaward
endof whatis knownas the SevernEstuarybut which,as we shallsee,hasfeatureswhich
makemuchof it moreakinto an opencoastline thanto a fiverestuary.
•nmOD
Severn Bridge
NEWPORT
;Avonmouth
i __
L_ CARDIFF
Laverhock
Point
///
if Flat
[---:::-;i
• ,,' •' Holm
, /
/
to r.• Brean
Down
Bridgwater
_
ß_ -10
0 5 1Okra
i I I
Figure1' Barragelocation
A barrage
in thislocation
wouldproduce some7-8%of thepresentelectricity
demandof
England
andWales,equivalent totheoutputof 2-3 majorthermal
powerstations.
Thiswould
makeit sometentimeslargerin sizethantheproposed
Mersey
Barrage,
thirtytimeslarger
thanthe RanceBarragein Brittanycommissioned
in 1966,andmorethana hundredtimes
largerthanmostpossible
UK tidalpowerschemes
(BinnieandPmrs,1989).Allhough
tide
rangeis a factorin thesecomparisons,
thedifferences
from siteto sitebetweenthesearesmall
comparedwith the basinareaswhicheachschemewouldcreate,ie., with the live water
volumes available
to eachsetof turbines
(volumes
whichlargely
determine
theappropriate
installed
generating
capacities).
Thebasin formed
bytheSevernBarrage
wouldhavea high
water surface area of over 500km 2.
T L Shaw 633
Thispaperoutlines
theworkdoneto determine
thehydrodynamic conditions
in whichthis
barragescheme wouldhaveto be constructed
andin whichit wouldoperate.Thelargesize
of theestuary
meansthatwaveclimatebecomesanimportantconsideration,
in additiontotide
rangeand associated
currents.
I i i • i, i I . i
Figure2: Representative
fide andestuarywaterlevel curves- RanceEstuary
634 T L Shaw
• l/-• 14
Scl :5963,000
6 7 8 $ I0 I 1213141516I118192021222324252•2?2823303132333435• 3?
32
31
30
29
28
Stackpol 27
'l II
25
24
23
II 22
II 21
II 20
I I
I I 8
II 7
II
II 5
4
3
2
1
0
8
7
5
4
3
2
1
I 2 3 4 5 t• 7 8 $ I0 11 12 13 14 15 It• 17 18 I$ 20 21 22 23 24 25 2S 27 28 2g 30 31 32 33 34 5 3• 3?
T L Shaw 635
',',I I I II .......
..........................
636 T L Shaw
./
AT PORT ISAAC
MODEL + ---.
OBSERVED E)
NEWPORT PORTBURY/AVON
eeeee
coooeec e
;HARPNESS
EPNEY
T L Shaw 637
HRS AFTER H W. o
AT PORT ISAAC
-2
-a
-,
OLDBURY EPNEY
638 T L Shaw
For thisinitial studyof the effectsof the barrageon conditionsin thefiver estuaries(D.En.
et al, 1989),it wasthereforeappropriate to focusdetailedworkon oneestuaryandto seekto
applythe lessonsso learnedto the others. In this way it provedpossibleto makegeneral
progress andto identifythe moredetailedeffortrequiredin duecourse.The geometryof the
fiver estuariesalloweda 1-D modelto be used,and this was appliedto the simulationof
turbidityand erosion/deposition (Wimpol, 1989). However,as previouslyexplained,the
consequences of the rise in meanwaterlevel in the basinareawouldencourage permanent
accretion on a muchlongertimescale thancouldsensibly be simulatedby a modelof fide-by-
tide processes.The techniques considered for this purposeand that appliedin detailare
described by theUniversityof Liverpool(1989). A timescale of about100yearsto reach95%
of theaccretionneededto give theultimatelystablebedformaccording to thechanged tide
regime and mean water level was deduced. The tide model was then run with this revised
bathymetry to showhowhydrodynamic conditions in thefiver estuaries
wouldthencompare
with thosewhichnowexist. Thosewhichwouldoccurin theshorttermfollowingbarrage
construction were also assessed.
The suitabilityof a I-D modelto simulateflows in the fiver estuaries,and of a 2-D model
for the main SevernEstuary,respectivelybecomeincreasingly questionable as the fiver
estuariesbroadentowardstheirmouthsandas the mainestuarydeepens furtherseaward.In
theseouterregions,bothgeometric anddensityeffectsprogressively influencethe simpler
regimeassumed with justification
to applyin eachcasefurtherlandward.Entending this
simplicityto areasin whichit is knownnot to be as strictlyapplicable givesrise to errors.
Thesemustbe quantifiedin orderto establish whethermorecomplexmodelswill be needed
in futureand,if so,whattheirspecificationsarelikely to be, thefield dataneededto validate
themandthecostsof collecting thosedataandof developing andrunningthemodels,etc.
T L Shaw 639
In the recent study of the Severn Barrage, an initial assessmentwas made of the
specifications for andfindingsof 3-D modelsappliedto thetideregimeof the SevernEstuary
and BristolChannel(McLarenand Collins, 1989). Their resultswere comparedwith those
of severalother2-D plan models.The resultssuggestthatthe overallstructureof the currents
in someshallowareasaway from the main flows and in deeperareascouldincludevertical
components.However, theseappearto be generallyweak comparedwith the main flows in
plan, and it remainsto be shownwhetheror not theyjustify furtherseriousconsiderationin
the contextof the barrage. Presentindicationsare that any effectswill only be of secondary
importanceand are more likely to be significantin some specific situationsrather than
generally.If this is confirmedit may still meanthatmuchof the SevernEstuarywill at some
stagehave to be modelledin threedimensions.
64O T L Shaw
Salinity (g/kg)
2..9
25
I0
1.5cms = 20km
T L Shaw 641
The reasonswhy the 1-D and 2-D modelsdifferedin their forecastsof the directionof
salinitychangein the basindeserveto be resolved,thoughthe absolutemagnitudeof the
differencebetweenthe forecastsmay not be too significant. The fundamentaldifference
betweenthe modelsis, of course,the way in whicheachrepresents lateralvariationsin water
levels,currents,etc.,acrosstheestuary.Early modelsof tidepropagation throughtheBristol
Channel and Severn Estuary (Fig 1) assumedlateral homogeneityand were able to
demonstrate the principaldetailsof the fide regime. However,lateral gradientsof levels,
currents,salinity,etc.,are knownto exist,hencethereis a casefor using2-D representation
for the presentregime.
RESIDUAL
VELOCITIES
-SPRING.
BARRAGE WITH PUMPS.
Figure7: a) Residualvelocities
- Spring.No barrage.
b) Residualvelocities- Spring.Barragewithpumps.
642 T L Shaw
Verticalgradients
of salinityin theSevernEstuaryhavebeenshownto be generallysmall,
lessthanlppt. Only in the deeperwatersof the BristolChanneldo greaterdifferencesoccur
but eventheretheyarebelievedto be of little consequence
to thewaterqualityandecological
regimes. Moreover,it is unlikelythattheywouldbe muchalteredby the barrage,thoughthe
possibleoccurrenceof locallysignificantchangesfor someplantandanimalcommunities, and
the needto designthe barrageto influencethis, shouldnot be ruledout.
T L Shaw 6•3
E.coli/100ml
2OOO
50O
I00
2.6cms = 20km
644 T L Shaw
The wave climatein the Bristol Channeldoesnot penetratefar into the SevernEstuary.
The coastfinesseawardof Clevedonand from Cardiff to Newport are exposedto conditions
whicharein partof thisoriginbutarealsodueto morelocallyraisedseas.Furtherup-estuary
and in the fiver estuariesit hasbeenconcludedthat apartfrom the influenceswhichreduced
currentsmay haveon refractionandits effects,the waveclimatewouldbe little affectedby
thebarragebecausethesewaveregimesare generated by localwindswhosefetchwouldnot
be affectedsignificantly,if at all (HRL, 1984).
The most uncertain influence of wave climate is its effects in shallow water and on
coastlinesas far as ecosystemproductivityis concerned.Referencewas madeearlierto the
informationnow availableon the historicevolutionof the SevernEstuary.Wavesarebelieved
to haveplayedan importantpart in this, in conjunctionwith currents. Over the pastfew
hundredyears,the processhas been alteredby the construction of sea walls, and further
changesmay be expectedto occurdue to the presenceof the barrageand how it wouldraise
mean water level in the basin and reducethe currents(more so on the ebb phasethan the
flood).
T L Shaw 645
31.6 Conclusions
1 The significancefor the effectsof the barrageof thr•-dimensionality in the flow structure
of the BristolChannel,thisbeingdeterminedby topographyand salinity.
2 The extentto whichthe flow structuresetup by the barragemay needto be and may be
modifiedby the designof the barrage,includingthe distributionand alignmentof the
turbinesandsluices,the sequenceof useof thisequipment,the extentandregularityof its
use includingthat for pumping,and the conditionswhich would occurduringbarrage
construction
accordingto the constructionmethodsandprogrammeadopted.
646 T L Shaw
31.7 References
Department
of Energy,CEGB and STPG, 1989, "TheSevernBarrageProject",EnergyPaperNo. 57,
HMSO.
Instituteof Oceanographical
Sciences,1980/81,"Tidal BarrageCalculationsfor the BristolChannel",
Rpts.to the SevemBarrageCommittee,Nos. STP 15-17.
Universityof Liverpool,1989,"Assessment
of RiverEstuaryGeometry",Rpt. to theSevernTidal Power
Group,N. SBP 10.
T L Shaw 647
WaterResearch
Centre,1989e,"TheEffectsof a Barrageon theDispersion
of Pollutants
in theBristol
Channel
andSevernEstuary",
Rpt.to theSevern TidalPowerGroup,No. SBP34.
Wimpol
Ltd.,1989,"I-D Mathematical
Modelling
Study
of theTidalReaches
of theRiverUsk",Rpt.
to theSevernTidal PowerGroup,No. SBP5/5A.
Vol. 14: A.J. Mehta (Ed.), EstuarineCohesiveSediment Vol. 37: C.M. Lalli(Ed.), EnclosedExperimental
Marine
Dynamics.473 pages.1986. Ecosystems:A Review and Recommendations.X, 218
pages. 1990.
Vol. 15: R.W. Eppley(Ed.), PlanktonDynamicsof the
SouthernCalifornia Bight.373 pages.1986. Vol.38: R.T. Cheng(Ed.),ResidualCurrentsand Long-
Vol. 16: J. van de Kreeke (Ed.), Physicsof Shallow termTransport.XI, 544 pages.1990.
Estuariesand Bays.280 pages.1986. Vol.39: M.I. EI-Sabh,N. Silverberg
(Eds.),Oceanography
Vol. 17: M.J. Bowman, C.M. Yentsch, W.T. Peterson of a Large-Scale Estuarine System.X, 434 pages.1990.
(Eds.),TidalMixingand PlanktonDynamics.502 pages. Vol. 40: D. Prandle(Ed.), Dynamicsand Exchangesin
1986. Copyright American Geophysical Union
Estuariesand the CoastalZone.648 pages.1992.