23
TIDE AND
CURRENTS
ALBERTO TOMASINDepartment of Applied Mathematics University of Venice,Italy
I.
Tide GenerationA.
The Tide Generating Forces B. The Development of Tidal Potential C.The Equilibrium TideD. The Development by Doodson E. Shallow Water Tides, Types ofTides 11.Hannonic and Non-hannonic Analysis
A.
Non-hannonic TermsB. Principles of Hannonic Analysis C. Least Square Methods
D.Tide Tables E. Cotidal Charts
lIl. Tidal Currents
A.
Considerations on Fluid Flow B. Current ObservationsC.Tidal Currents in Seas and Estuaries IV. Storm Surges and Extreme High Tides
A.
Atmospheric Effects, Wind and Pressure Actions B. Predictability of Storm SurgesC. Extreme High Tides
V. Models and Computing Methods A. Basic Equations
B. Friction and Stresses C. Finite Difference Methods D. Miscellaneous Methods
24
ALBERTO TOMASINI. Tide Generation
A.
THE TIDE GENERATING FORCES
The earth, the moon and the sun are the celestial bodies to
he
considered in studying the tides, since the other ones have a negligible influence. Indeed, the tide generating forces exerted on the earth by a celestial body are (obviously) proportional to its mass and (reasonably) to the cubic inverse of its distance from OUTplanet (the derivative of the Newtonian square inverse). This could also easily show why the moon is more important than the sun in this field.The mechanism to be considered in detail starts from the question of how much a point-Iike body on the earth is affected by the moon (or the sun). Consider the place where the moon is seen on the zenith (the closest one), then the one seen on the nadir (the farthest), then any intermediate one.
Due to the different distance from the moon, .there will be an obvious increase of gravitational attraction when moving from the farthest to the intermediate and then the closest one. Now, one should consider the force exerted by the moon on the center of the earth and remember that it has the special meaning of having a reference value for the terrestrial bodies. Indeed, we would not care of any possible force acting precisely the same way on all of us and on the center of the earth: by the same token, if the cable of an elevator is cut and it falls a1most freely, the objects in it behave as if they were
.weightless, since they have the same acceleration as the cabin. Likewise, what is relevant for the dynamics of the bodies on the earth is the deviation of the forces with respect to the reference one. One sees, by properly drawing the force vectors, that the farthest and the closest point will experience upward forces (thus they feel lighter), while the intermediate one will be almost unaffected. In a first approximation, the two upward forces will be identical for the two extremes (which corresponds to a linear change of the attractive force in passing through the earth).
We can immediately deduce that in two opposite points on the earth everything, and in particular sea water, is slightly lighter, giving origin to two bulges in the ocean, at the expense of the intermediate seas.
But things move: the moon turns around the earth in about a month, and much more quickly the earth has a spin motion. An observer, fixed on the earth close to the sea, will see the swell rising twice a day, when he passes close to the two particular points (since we know very weil that
they
are not the poles of the rotation: this would dismay OUTargurnents).
The day to be considered is clearly a lunar day (1.035 solar days), as long as we are considering the moon effects, but similar considerations hold for the sun.
TIDE AND CURRENTS 25
B.THE DEVELOPMENT OF THE TIDAL POTENTlAL
The above description is correct
,
but to go into practical evaluation one needs more
detail. So far, it was only stated that due to the moon the observer sees two high- and
two
low-tideconditions every lunar
day,and there are also two smaller bulges due to the
sun
.
This is due to earth rotation
,
but the picture is not statie, due to the slightly varying
distances of the two celestial bodies from the earth (elliptical, no circular orbits are
involved) and due to the varying trajectoryof the moon and the sun in the sky on a yearly
time scale (what familiarly is the sun's seasons).
We can consider (as such was indeed the developmentoftidal prediction) adding step by
step the various details of the ce1estialmotions in order to fit better and better the
theoretical description
to the observed world
.
This is closely connected to the
mathematical trick of a series expansion for the tidal forcing or for a potential function
from which the forces are derived.
Out of an infinite set, only a few terms turn out to be
relevant: better details will be given below
.
C.THE EQUILIBRIUM TIDE
Now
,
another important step is required. It was always assumed that an observer
passing by the "c1osest"point (with respect to the moon) sees aflood tide condition in his
area
:
this
-
is true only for an ideal sea which is able to respond instantly to the
gravitational change. If one prefers, the equipotential surfaces can accomplish it, but
whatever is real and material win suffer some delay. Under these conditions, one works
easily with an ideal "equilibrium tide"
.
The separate problem is then considered of the
real world ocean, where a huge wave runs east to west after this attracting potential.
There are many obstacles for this wave
,
i.e. islands, shallow areas, interposedcontinents,
so that the precise time and level of the high tide in a given place becomes something
frightening to be computed, from a theoretical basis
.
The help of observations will be
vital. It is vital also to rernark that the periodicity of the celestial motion is maintained
also in real life
:
for example, the seasonalvariation of the hight of the sun at noon win
certainly be found some way in the tidal records of any place in the world. This win
turnout to be the main help availablefor practical analysis.
26 ALBERTO TOMASIN
D. THE DEVELOPMENT BY DOODSON
The above idea of a series expansion of the tidal potential, giving origin to terms of decreasing importance, can now be connected to the periodicities in the tides. The work done mainly by Lord Kelvin, Darwin and Doodson shows tbat one can describe things
by
considering the six periodicities of interest:- 1 day is the period ofthe earth's rotation (relative to the Sun)
- 1 month is the period of the moon's orbital motion - 1 year is tbe period of the sun's orbital motion - about 8.85 years is the period of the lunar perigee
- about 18.61 years is the period ofregression of lunar nodes
-about 20,000 years is the period of solar perigee.
Consider now the frequencies fl ' f2 ,...,f6 related to the six periods above, sinee they are more useful in calculations; for example, one can subtly discover tbat other periodicities (Iike tbe lunar day) do not appear in the above list, as tbey may be deduced from tbe reported ones. This is tbe case of the lunar day, whose related fi:equency fmoon can be expressed as fl -f2
+
f3 (indeed, real expansions are performed in terms of lunar frequencies).Since the whole range of time scales is covered, one can see that the tidal behaviour is
.fully described by a relatively simple expansion.
Take the six frequencies defined above in a vector, f, and think of a similar vector k built by six integer numbers (ki = 0,
±l,
±2, ... and i=I,2, ...,6). Tben we express the tide in a certain place byz (t) = l:k Ck cos(21t k·f t - ek) or, if one pref ers,
In other words, we Iinearly mix the six fundarnental frequencies In all possible combinations, and from each of these we have a harmonie term witb tbe resulting frequency: its amplitude C and phase e (obviously dependent on the location coordinates, but not on time) can he theoretically determined to get the equilibrium tide. This way, a reasonable number of components can he selected, with an amplitude large enough to
he
reasonably observed.
TIDE AND CURRENTS
27
It
has already been remarked that this correspondsto taking into account step by step all
the eelestial mot
i
ons
:
i
f we dis
c
o
v
er that a h
arm
on
i
e function
,
a sine
, h
as a slight and
slow perturbation on its amplitude
,
we multiply it by a long period sinusoid
.
But then a
simple transformat
i
on g
iv
e
s, i
n
s
tead of the product
,
the sum of two harmoni
e
terms
,
where the frequenciesare in linear relat
i
onship with the originalones.
This gives origin t
o
the weil known tidal terms like M2 .Kj
,
S2
etc
,
where each one
corresponds to a definite choice of kl ,k2
,
.
.
.,k6
.
These are the basic ideas of the present
handling of the tides
.
E
.
SHALLOWWATERTIDES
,
TYPES OF TIDES
We have seen now a good
t
ooi to investigate the tides, since we have split the
astronomical input into a series of harmonie components
.
Do we really expect
tofind
these components in the real world
,
in spite of the disturbing effects that were
mentioned?
I
ndeed, it is essentially so, each component is found also in the ocean
,
usually lagged with respect to the theoretical one (but one could argue that this statement
is
,
strictly speaking
,
nonsense
)
and with an amplitude sometime reduced, sometimes
enlarged. What will be considerd now is the generationof spurious constituents, a fact to
be expected from the dissipative factors in the tide propagationaround the world
.
The equations that describe this propagation (LTE
,
the Laplace tidal equations are the
starting point
)
will necessarily include frictional terms
,
essentially nonIinear, mainly in
shallow water and in all coastal areas
.
Again, one can say that a monochromatic wave (i.e
.
, a perfect sinusoid) generates other
harmonies whose frequencies are twiee
,
three times, etc
.,
the fundamental one (they are
called overtides in this case). If, as it usually happens
,
a couple of extra terms are
suffieient, then the situation is easily manageable. A similar development is done for
compoundtides, i.e
.
to nonlinear interactions between the importantharmonie terms
.
Many studies have been developed in this field without using the harmonie technique,
from a purely hydraulic, so to say
,
point of view
:
one should remember that the estuaries
,
where the most important harbours in the wo
r
ld are situated, are typical plaees for these
phenomena and for centuries people have
beenconcerned with them
.
Animportant
remark is
,
for example
,
a very simple one about the wave d
i
storsion
,
sinee, in shallow
water
,
the high tide will progress sliding over a thicker water layer than the low tide, so
that it goes faster and we can guess the kind of distorsion
.
In harmonie development
,
simple considerationscan be made on the phase of a shallow
water term with respe
c
t to the parent component or components: this can be of
28
ALBERTO TOMASINsubstantialhelp if the number of tenns to be considered
is cumbersome.
As a conclusion for tidal considerations,one
is
left with a definite idea of a large variety
of observedtides
in differentparts of the world.
The latitude is certainlyrelevant, but the
morphology of the oceans and the
continents
count more.
We expect a large range for
possible amplitudes,
the possible presence of shallow water terms and even the overall
distinction of places
with predominant diumal or semidiumal tide
.
There are formal
methodsto c1assifythe tide of a certain place according to the latest distinctions.
IL Harmonie and Non-harmonie Analysis
A
.
NON HARMONICTERMS
Beforethe present developmentof the theory of tides,
or more exactly of the techniques
to handle it,
other methods were developed,
now called non-harmonie.
They follow
essentially the same step-by-step refinement in adequating observations to celestial
mechanics.
Much more interesting than describing a duplication of the present methods,we should
only remember the glossary of tbe non-harmonie approach, connected to empirical
aspectsand still used to give a picture of local tides.
One can start by from HWI,
the time interval hetween tbe moon's transit and the next
high water; next will be MHWI, the mean value of all HWI during 29 days. In the same
field is HWF&C,
high water full and change,
also called Establishmentof the Port,
i.e.
the HWI on the days of full and new moon.
Aremark can he made on the spring tide,
i.e.
the particularly strong
tidal effect related,
but usually not coinciding,
with full and new moon,
due to the relative alignment of the
sun,
the earth and the moon.
The "phase age"
is tbe
interval between the time of full or
new moon and the time of spring tide, which turns out to he different for variousplaces.
Other symbols like MHWS
(mean bigh water in spring tide)
and similar ones are even
more obvious.
.
B.
PRINCIPLESOF HARMONICANALYSIS
The developmentof the tidal potential (or of
the equilibrium tide)
was
introducedabove
to give a useful tooI to investigate something theoretical: it would be quite Iimited if it
had no return
in the practical use.
lndeed, if one thinks of the amplitude and phase of
each component as something to
hededuced from observation,
he concludes that from
TIDE AND CURRENTS
29
t
heory onl
y
fe
w
bu
t i
mp
o
rtant hints ha
v
e been
k
ep
t,
namely the frequencie
s
and the
crite
ri
a to
c
hoose areas
ona
bie numbe
r of co
mpo
n
ent
s
t
o
be in
v
est
ig
ated
.
The pra
c
t
ic
al
user will t
hen ha
v
e th
e
problem to anal
y
se a cert
ai
n amoun
t
of tida
l
o
bse
rv
at
io
n
s
(
u
sua
ll
y
h
ou
r
ly v
a
l
u
es of sea l
e
v
el
i
n a c
ertai
n p
i
a
c
e
,
taken b
y
a record
ing
tide
g
auge
fo
r a
r
e
a
s
o
nabie peri
o
d
)
i
n
or
d
er
to obta
i
n the harm
o
n
ie
constants
(
amp
li
tude
and phase
)
for a
few
c
o
mp
one
nt
s.
How m
an
y?
Say
,
tw
en
t
y i
n the del
i
cate c
as
e
s of
shall
o
w water
in
fluenee
,
but more usually only seven
;
e
v
en four would he better than
noth
i
ng. An add
i
ti
o
nal d
etai
l
:
onl
y t
he fir
s
t th
r
ee frequencies are actually used to d
e
fine
what is called a const
i
tuent
(
M2 for e
x
ample
,
or Kj
)
. In this structure
,
the three long
term period
i
ci
ti
e
s
d
i
sappear and are kept
i
nto account in another way. The previous
decompositionformula becomes,
i
n p
r
actice
:
z(t
)
=j C cos(wt
-
9 + v
)
where the summat
i
on is extended to the frequencies that are really taken into account:
the new factor j and the phase correct
i
on v are slowly varying (as an effect of the
igno
r
ed long term period
i
cities
)
and are tabulated for practical purposes
.
There are many practical recipes to est
i
mate the harmonie constants
,
depending on the
length of the observat
i
onperiod and on the availability of computingtools.
Generally speaking, one thinks of the classical methods (e.g. the Admiralty scheme or
the Tidal Institute Method
)
as numerical filters that acting on the available time series
make evident, in the different steps, one or another component.
Frequently a smart
principle is used
, i.
e
.
the fact that neighboring frequenc
i
es cannot he distorted quite
differently in passing from grav
i
tationalpotential to real tide. The world ocean and its
obstacles will presumably have a different action, since they are a selective filter
,
on a
diurnal or a semidiumal oscillation
;
but if two components have very similar frequencies
and theoretical amplitudes, say, in the rat
i
o 1
:
2 one expects that the practical realizations
of the two sinusoids
i
n the ocean will have an almost identical relation. This can he a
powerful help in calculations.
C. LEAST SQUARE METHODS
An approach to estimate the harmonie constants of a given place from the recorded
values of the sea le
v
el can be the least squares method
:
the best fit is sought between an
ideal summation ofharmonic terms and the observed values
(
at an hourly rate)
.
Clearly this method should be preferred in the case of irregular data
,
where a frequent
lack of information makes it impossible to rely on predetermined schemes. Should the
recording he uninterrupted
,
then the least squares could he equally satisfactory
,
but many
30
ALBERTO TOMASINalternative methods could have particular advantages, for example for short records. lt is just the case to mention that an expression like
C cos(wt- 9 ) can be wri tten as
c
cose
cos wt+
C sin 9 sin wt orx
cos wt+
Y sin wt.This way X and Y are Iinear unknowns, rnuch better to deal with in the least squares method. At the end, C and 9 are easily deduced.
The next refinement will allow the long term variation of the harmonie constants to be taken into account (the
j
and v quantities of the preceding section). In tbe Jatter expression we make evident the unknown part (X and Y), whilst cos wt and sin wt are something known. The algorithm is only slightly heavier if one considers, instead of cos wt,j
cos(wt-v ), withj
and v known from tables (or put once forever in the computer). The result, thus, is a1ready corrected for the long term variations and hence directly comparabie with other harmonie constants and ready for forecasting purposes.To a certain extent, the present large availability of good computing tools (which was implicit in the last considerations) reduces the importanee of many approximations that were developed in the past to run the least square algorithms more quickly, so tbat this well-known and multi-purpose method can be used rather straightforward.
D.TIDE TABLES
Tide tables, with the predicted values and the hours of the level extremes, have a1ways been precious for shipping, fishing and other coastal activities.
Usually the annual books give extensive tables for one or few important harbors, with some possibility to interpolate the values for intermediate places. The present availability of computing tools makes things easier, it becomes only a matter of good will to produce the tables for any place where the constants are known.
A severe limitation for the tidal predictions is to consider only the astronomical effects. One can never feel sure, in planning his trip at sea, that a change in weather might cause the sea level to deviate from the astronomical prediction: we will analyze how tbis can happen, but we know very weil from the beginning that a forecast given years in advance (as it is the case for something related to earth and moon movements) will never be given for weather conditions.
TIDE AND CURRENTS 31
One can conclude that tidal tables are valid only for normal atmospheric conditions,
otherwise they show at least the essential trend of the sea level. Surges and all meteorological efTects will develop around it as a perturbation.
E.COTIDAL CHARTS
Another use ofthe tidal constants (usually in the harmonie form, as we know) is related to tidal charts.
The
availability of tidal records is in most cases limited to coastal areas,but from them (and with the help of more or less complex modeis) one can draw a map of that specitic sea, or ocean or part of it. A map with isolines related to the tide, so it is a tidal map: the main lines are the co-tidal ones, sbowing the points of equal phase, then the co-range ones are given.
Thi
s
introduces a scientific way to understand tbe real world of tides, that we confined before to an essentially empirie analysis. Seeing tbe real tides on a large scale gives us again the possibility of an insight that immediately surprises the newcomer: most maps show a circular dynamics around certain points in the sea tbat are called amphidromic (i.e. run around). All this is due to tbe eartb rotation and the consequent sideways deviation of motion (in terms of currents, tbis willhe
discusserl shortly). At an amphidromic point the tidal range for that particular component will be zero and it will increase as long as we go away from tbe point. Delicate experiments are sometimes required to solve ambiguities for tidal pattems in tbeocean.
m
Tidal eurrents
A.CONSIDERATIONS ON FLUID FLOW
Among general considerations concerning currents, one sbould never forget tbe deflecting force of tbe eartb's rotation. Indeed, we want to consider tbe motion relative to tbe earth, i.e. not with respect to an absolute coordinate system but relative to a rotating one.
The laws of mecbanics tben rernain valid only if one adds tbe accelerations ax ,ay and
az
to all moving masses, as given by (x pointing to tbe cast,y
to tbc north) ax=-msin , Uy - 21lcos , Uzay=
2llsin cjlu,
az=2ncos cjlU
xwhere
n
is tbe angular velocity of tbe earth, cjl is tbe geograpbical latitude (negative to the Soutb) and Ux,Uy and Uz are the components oftbe velocity. If, out of these. only the horizontaI ones are considered, one speaks of tbe deflecting force of tbe earth rotation. Ir always acts perpendicular to the motion, towards tbc right in the Northem Hemispbere when one looks in tbc direction of the current. To avoid saying "right" and32 ALBERTO TOMASIN
"I
eft"
,
the suggestionwas acceptedby oceanographersto speak of "cum sole
"
and
"
contra
solem
"
respe
c
ti
v
ely
,
i
.
e.
in ac
c
o
r
dance
o
r aga
i
nst the apparent motion of the sun in the
sky.Th
i
s term
i
nol
ogy,
describingthe d
i
rect
i
onoftuming, corresponds to the expressions
ant
ic
yclonic and cyclonic
,
us
e
d in mete
o
rology.
By the way
,
there
i
s no more need to
specifythe hemispherewe refer to
.
Under the above influence
,
a water particIe moving with horizontal velocity U and not
subjectto any other force in the horizontal des
c
ribes a
ci
rcular trajectory
"
cum sole" with
radius U
/ (
2U sin
cp
).
Since in order to go along a stra
i
ghttrajectory a partiele should be
subjec
t
t
o
a fo
rce o
pp
o
s
i
n
g
and balan
cing
th
e
d
evi
ati
ng
fo
r
ce
, i
t i
s e
a
sy t
o d
e
duce that all
other forces
(
not exactly calibrated
)
will entrainthe partiele in cycloidal trajectory, a kind
a c
i
rcularmotion plus a drift
:
these are
i
ndeedthe currentsobserved in the large seale
.
B.
CURRENTOBSERVATIONS
Mapping the water motion in the sea is a common task for oceanographers and
engineers.
However, we are faced with serious problems since the motions are
complicated and any sampling turns out to be very expensive. In the past
,
there were
three groupsof methodsfor current measurements:
I) Indirect methods
,
when an estimate of the currents
i
s given following the
hydrodynamiclaws
.
The measured surface inclinat
i
on
,
the pressure d
i
stribution or the
map of other hydrographic properties (salinityand temperature)have been good tools to
deduce the currents
,
mainly in the ocean
.
The present use of numeri
c
modeis
,
also for
small areas, is in the same line
.
2) Drift measurements,tracking the trajectory of by a float.
3) Current measurements at a fixed point. Such measurements have the same
relationship with the previous ones following
,
that the Eulerian point of view has with
the Lagrange one
.
According to the latter
,
the problem is solved when we know the
trajectory of each particle, whilst for Euler one has to know the current direction and
velocitya
t
each point.
To this distinction of methods we should add today the remote sensing, that gives the
possibilityto know the complete(Eulerian) distributionof currents
ina certain area from
observationstaken from a few points
,
taking advantage of acoustic waves.
Sometimes
one gets only integrated information
(
say, the diseharge in the English Channel) using
electromagneticproperties(like the GEK)
.
The current directionis always g
i
ven by the direction
i
n which
(
o
r
towards which) water
is moving
,
in contrast to meteorology wherethe wind direction is given by the one from
which the wind is blowing.
C
.
TIDAL CURRENTSIN SEAS AND ESTUARlES
TIDE AND CURRENTS
33
lt
is well known that the most important currents in the oceans are not related to tides
:
think of the Gulf Stream and many
si
milar phenomena
.
Tidal currents exist all over but
there are much less measurementsof them than of tidal levels
,
except perhaps for places
of interest for shipping
.
The first
t
h
i
ng one comrnents
i
n t
i
dal current observations is that
the direct
i
on of the current varies
i
n the horizontal plane
,
in addition, of course
,
to
variations in intensity. Thus an extra degree of freedom is added with respect to level
observations
:
It
is obvious that the vert
i
cal component of the current is negligible
:
of course, being a
wave, the tide should show
,
for the motion of a single water particle, an elliptic pattern in
the vertical plane
.
But the special wave we are considering, thousands of kilometers
long
,
g
i
ves origin to a very compressedellipse
.
If we add the variations in the horizontal plane that we mentionedbefore
,
we realize that
(except for channels
,
river mou
t
hs or similar configurations) the current figures
,
as
described by the tip of the current vector, deviate from a straight line
.
Like tides, tidal currents can be analyzed by harmonie components. The simplest way is
to take the two perpendicular projections of the current vector (say, north and east
components)and separatelyrun the harmonie analysis.
A simple analysis shows that the maximum current U in a place should be related to the
tidal amplitudeC (thinking of a single sinusoid) by the relation
U
= (gIh)1I2C
where h is the depth and g the gravity acceleration
.
This holds in the open sea, where
one assumes the tidal wave to be a progressive one, i.e
.
, having the current maximum
correspondingto a maximum(or minimum) level.
IV
.
Storm Surges and Extreme High Tides
A. ATMOSPHERICEFFECTS
,
WIND AND PRESSURE ACTION
The wind blowing on the sea has an obvious effect on it
,
since it pushes it by shearing
on the surface
,
g
iv
ing o
ri
gin to a mot
i
on that propagates down to a certain depth
.
Without entering in deta
i
l
,
that are object of many stud
i
es due to the difficulty in
describing precise
l
y such a fuzzy phenomenon
,
we will better mention the general
problems conceming the wind action
.
34 ALBERTO TOMASIN
Let us consider firstof all the relevanee of the fetch, or the size of the area where the wind actson the sea. Every surface unit adds up for the resulting action. In conneetion with the wind wave studies (which differ from the present ones for the different wavelength of interest) one can think of a different dragging effectiveness of the wind over a smooth sea surface or over weil developed waves. This would force us to use.
dragging coefficients depending on the history, so to say,ofthe surface.
Another consideration is interesting, conceming a kind of steady state in an enclosed sea,a lake or perhaps just a coastal area. Suppose the wind is blowing towards the shore,
then we think of the surface water moving, giving origin to a certain slope. Will this slope, in principle, grow indefinitely as long as the wind keeps blowing? No, it is reasonable to expect that the bottom water will hydrostatically react to the surface slope by flowing in a bottom countercurrent: in this simplified scherne we have circulation maintaining a steady state. This, in fact, is not very far from what happens.
The barometer (or inverse-barometer) effect of the atmosphere on the ocean is weIl known. To a certain extent, an assumption of meteorological uniformity was implicit when we described the tendency of the ocean to match the gravitational surface.
Hydrostatic considerations suggest that corresponding to a low pressure area there will
,be a rise in the water level in order to maintain the same pressure over any ideal horizontal surface in the sea. Needless to say,a meteorological map with lows and highs cannot be stabIe and will evolve, but the favorable condition for our study is that at any time the barometer effect gives the instantaneous modification of the equilibrium surface at which all the sea is aiming.
B.PREDICT ABILITY OF STORM SURGES
The problem arises of how easily and - above all - how effectively can storm surges be predicted. In fact, people are not concerned with the level a certain storm surge will attain. Indeed, the actual level is of concern, the one caused by the astronomical dynamics PLUS the surge. The difference is subtIe. There are many cases reported where storm surges of relatively short duration (say, six hours) occurred with ebb tide,
and since the two facts were of the same order, nothing really happened for what concerns floods,i.e. the real threat. Another surprise: are there only"positive" surges or also "negative" ones? Literature reports that frequently disasterous waves happened to
TIDE AND CURRENTS 35
hit a coastal area immediately after a sudden retreat of the sea, but this was just a phase trick.
A
really negative surgecan
certainly occur, with unpleasant situations for sailors and coastal power plants: both wind and inverted barometer effects can act oppositely to the way we considered (wind from the coast and sudden pressure increase) and give origin to negative surge.To predict these phenomena requires a good atmospheric knowledge (and forecasting)
and ability to compute the air-sea interaction. The first point concerns the availability of weather data(from the sea:islands, ships, satellites and obviously coastal areas). Suppose now that
the
observational data are available, and suppose they give enough information aboutthe
atmospheric phenomena of interest as tbey are now. Then a suitable modellingcan
predictthe sea
response a few hours ahead, using the significant time it takes to tbesea
to respond to the present conditions.For a longer term prediction, an eflicient weather forecasting is required, and we all know the international effort in this direction.
The use of observed
and
predicted meteorological data to estimate the effect on the sea(and hence the possible surges) will be considered below:the models are the main tooi to do it.
C. EXTREME HIGH TIDES
Technically speaking, extreme high tide is equivalent to storm surge: tbe daily tide greatly affected by atmospheric factors. This compels us to consider possible interactions between tides and surges: indeed, it is frequently wrong to estimate the surge by subtracting the predicted astronomical tide from the observed heights. There are more sophisticated methods to perform this separation effectively.
It should be clear, from the above considerations, that this difficulty is due to nonlinear terms in the equations describing the motion: it sounds staggering to attribute a physical fact to a mathematical formulation, but this only derives from our necessity to describe tbings by the tools we have.
Statistical analysis turns out to be significant when applied to the extreme tides as tbey are, i.e., including the astronomical tides (which have no interesting statistical behaviour) and to the differences between the observed values and the table forecast. One
can
see that there isn't any essential differenee if the sample is appropriate.Obviously what these methods give are statistical results, sinee they concern distributions and probability: they cannot give any prediction for tomorrow or, say, for the storm approaching now. But they give the probability for a certain level to
he
attained36 ALBERTO TOMASIN
(under exceptional conditions) during a certain time (say, ten years). The methods by
Gumbel, best known for the run-off of rivers, have been adapted (or completely
reformulated) for the present purposes and appear to be very useful in planning dikes and other sea defenses.
v.
Models and Computing MethodsA.
BASIC EQUATIONS
The basic hydrodynamie equations will
be
shown here, without too many comments.Tbey are adequate to study storm surges but without great difliculty one can use them for
tides, in particular when the propagation is considered in a gulf or in a bay. Tbe
equations can be written as:
dQx/dt =f Qy -
gh
Bz/lix+
'tsx/p - 'tbxlp -
gh lipJBx dQyldt = - fQx - gh Bz/liy+
'tsylp - 'tbylp -
gh lipJByBz/ot
+ liOx/Bx + liOy/liy
=0Here z is the water level, h is the local depth,
Q
is the transport, i.e., depth timesspeed
(averaged top to bottom). The symbols 't are stresses, eitber acting on the surface (s) or on tbe bottom (b), i.e. due to wind or bottom friction; p is the atmospheric pressure. The symbols g,t,x,y have the standard meaning, whilst fis defined as the Coriolis parameter, originated by the earth rotation.Who is not familiar witb tbese relations will appreciate the hint that the first two of them (the momentum equations) simply say that tbe acceleration equals tbe total applied force per unit mass. The tbird one is the continuity equation, expressing the conservation of water. Instead, who knows them will remark that the rainfall contribution is omitted, tbat many terms could
be
better developed and that by writing tbings this way we restrain from investigating tbe details in the water column, so tbat this modelling is barotropie and not baroclinic.It was anticipated before tbat the
direct
tidal motion cao be studied by tbese equations (i.e., in addition to surges or simple propagation). Indeed, one should first of all evaIuate the distribution of gravitational potential (or tbe equilibrium tide): it requires a certain effort, but the available literature helps. Tben one should consider the surface slope (tbe second term in tbe right hand side of tbe momentum equations) and reaIize that it contributes to motion whenever the surface is not horizontaI. But the effect of theTIDE AND CURRENTS
37
gravitational perturbation consists in changing the horizontal reference: this way (provided one knows the equipotential surface) the tide can be really used as a driving agent for water flow.
B. FRICTION AND STRESSES
Very important for a good calculation, friction and stresses appear to
be
the most elusive terms in the equations: so to say, the "dirty" factors.Friction is always present whenever water moves, whilst wind is dominant for surge
generation. By the way, one should observe that surge has the conventional meaning of a level perturbation occurring offshore, but as soon as it approaches the coast it will undergo special interactions with the bottom (in particular, when the shore is gently declinating to the sea) and the"wave setup" phenomena.
Conceming stress, one promptly realizes that a variety of conditions could occur; in the open sea, one thinks of a column of water, moving for some reason: it will stress the bottom in the direction of the column motion. Think now of the kind of equilibrium that was figured about the wind: after a certain time, a wind blowing inshore will determine a bottom stress in the opposite direction.
Many cases could
be
explored, to conclude that wind stress is a really difficultparameter. Yet people working on models find successful results in assuming a simp Ie form for the bottom stress, by adopting in general the above choice of the moving column of water, with the transport
Q,
affected by a bottom friction~=kIQlQ
The similar doubts conceming the wind stress are even worse (as mentioned above), but again the most generaI way people try to gel rid of these difficulties is to adopt a quadratic relationship (
I Y I
wind speed)Is =k' IYI Y
Many precise measurements have been done on the stress, mostly in microscale
experiments: it is not that obvious that their use in the modelling is fruitful, whicb suggests substantial differences at different scales.
C. FINITE DIFFERENCE METHODS
The basic equations we saw before can
be
solved eitber analytically or numerica1ly. In the fust cast the basin of interest (the North Sea, a small bay, an estuary ...) is assumed to38 ALBERTO TOMASIN
have a regular shape, sometimes not very simple but allowing a geometrical description which isadequate to condition the equations, given a proper boundary condition. Also tbe meteorological input needs a "regular" description: in spite of these limitations, interesting conclusions have been obtained in many cases.
Tbe numeri cal approach is by far preferred to the analytical one, due to the case
i
n
adapting its schemes to all possible situations, both for morphology, extemal input and different hypothesis on delicate points like friction and stresses.Tbere is a variety oftechniques proposed. We are considering here the finite difference method, widely used in the sea in tbe last forty years (and much longer for the atmosphere by meteorologists).
Finite differences are what one uses to replace the differential quantities in our equations. Tbis way the continuity is lost in space and in time and one has to decide roughly how the topography has to be reduced by this quantization. Typical finite difference schemes use a space grid, regular and orthogonal, in the horizontal plane. At each knot the depthis given and, in the calculation, current and level will be estimated at each time step. Tbe grid is intended to help in subsituting spacial derivatives with differences between adjacent points. It would take a long time to enumerate all the refinements tbat have
been
adopted.They
are necessary to avoid the troubles that a straightforward use of tbe above ideas would induce.Here we make some simple remarks: tbe grid usually lies in a horizontal plane, but earth is a sphere. It is not difficult to
he
more correct, since there are relatively simple formulae to keep curvature into account, which is important when we cover a rather large area.The grid itself has usually to be more complicated, with separate points where to estimate different variables.
Tbere is an interesting point conceming the proposed schemes. Suppose we do exactly what was described above, namely: at a certain time we assume to know tbe relevant quantities (level and current) at all points of the grid. Since we presume to know all the space derivatives in a eertam point of the grid through its neighboring points, we are able to estimate the time derivatives, hence also the new quantities at the new time, i.e.after a time step. But at a time step L\t a sudden perturbation originating at a certain point cannot travel more than one grid step (or mesh) L\s. This defines a kind of numeri cal velocity that, in order to describe properly what happens, should never be less than tbe velocity of the real facts at sea. We can expect that if this requirement is not satisfied, OUT simulation will soon be unsuccessful. Since from our basic equations one could easily derive wave equations witb celerity (g h)l/2 , one gets tbe so-called "Courant-Friedrich-Lewy criterion" (or requirement)
(g h)l/2 < As/L\t
TIDE AND CURRENTS 39
(with some numeri cal factor depending on the details of calculation).
Meteorologists had trouble for many years before this criterion was adopted. What it means is that, given a certain grid, one has to reduce the time step At until the above criterion is satisfied: it might mean a much longer computing time than one could hope.
Tbe operating scheme proposed so far is perhaps the most simple because each step has a clear meaning: it is called the explicit scheme. We have just tackled tbe amount of dîfficultiesthat one faces as a toll for simplicity. If,instead of solving step by step, point after point, one writes huge systems of simultaneous equations, it tums out tbat many difficulties can be encompassed and usually much longer time steps can he adopted. This is the implicit metbod, another of the many fruitful approaches that appeared in tbe field.
Another inconvenience of ordinary finite differences is the use of an ortbogonal grid, where, in addition, Ax is the same all over, and so is Ay. Now, the important features of a given area could go exactly the other way: for any possible orientation of the grid, we frequently find significant streamlines in oblique directions and the array-like scheme has obvious difficulties in handlin them. Further, it is usual to have certain parts tbat clearly require a much finer description (and hence smaller meshes) than other very uniform areas. Solutions and improvements have
been
proposed, with finite differences, but we will instead consider substantially different methods.D.MISCELLANEOUS METHOOS
Tbere are many approaches to our problem, as we know: a good neighbour to finite differences is the one using finite elements: usually a net of triangles, instead of rectangles, covers the map; each triangle has arbitrary size, provided tbat each side is shared only by two adjacent "elements" (the triangles). In each element the relevant variables vary Iinearly: this is the transposition of the corresponding discretization by finite differences, but with a much better ease to fit the topography with its different relevanee for flow. Usually a system of conditions is written (like in the implicit method) and special techniques are available to handle tbis type of equations.
Another essentially different approach will be mentioned now, i.e. the statistical (or
40 ALBERTO TOMASIN