Tidal prediction - OXP2003 Lecture 5 The harmonic method of prediction

• Practical methods of tide prediction are based on Laplace’s principle (1775) which • Mathematically, the tidal elevation at any place can be written as below. (It is
states that, for every constituent of the equilibrium tide there exits a harmonic conventional to use the Greek letter zeta, ζ, for tides. Sorry….η for waves but ζ for
constituent in the real tide with the same frequency but with different amplitude and tides.)
phase (e.g. the M2 constituent may be bigger, high water may occur later than the N

equilibrium tide, but the period is still 12.42 hours). ζ = Z 0 + ∑ H n cos (ω n t − α n − g n )

• Any tidal curve is the sum of a large number of these constituents. This method was
first used practically by Sir George Darwin (Charles’ nephew) around 1880. • N refers to the sum of all constituents, Hn is the amplitude of the nth constituent, ωn is
the angular speed of that constituent, αn is its phase relative to lunar transit of the
• Predictions for standard ports typically use the sum of 60 constituents (the “standard
equilibrium tide and gn is the phase lag of the real tide to the equilibrium tide.
harmonic method” due to Doodson, 1921) but to accurately describe the tide in many
estuaries involves the summation of 100 constituents (the “extended harmonic • The term Z0 is the difference between mean sea level and chart datum (approximately
method”). the level of Lowest Astronomical Tide, or LAT). This ensures that the predicted
value of ζ is always positive. This is preferred by mariners.
• Reasonably accurate predictions can be made just using M2, S2, K1 and O1
• A further correction is required for the 18.61 yearly regression of the lunar nodes.
• Try this on a spreadsheet using the following angular frequencies
The nodal factors h and ε on the next slide complete the practical equation that is used
– M2 2π / 12.42 = 0.5059 radians/hour for tide prediction worldwide.
– S2 2π / 12.00 = 0.5236 radians/hour
– O1 2π / 26.87 = 0.2338 radians/hour

Harmonic analysis Least squares tidal analysis

• To illustrate the method we will consider just the M2 constituent. The
ζ = Z0 + ∑ hn H n cos(ω n t − α n − g n + ε n ) theoretical tide (ignoring nodal corrections - that would require a 19 year time
series) in this case is given by:
• In the above equation, all the frequencies and constants relating to the equilibrium tide
are known. All that is required to make prediction possible is analysis of the values of
ζM2 = Z0 + H M 2 cos(ω M 2t − g M 2 )
H, g and Z0 , at every location where they are required.
• This is done through observations of tidal elevation using accurate tide gauges (e.g. • Least squares fitting subtracts the theoretical value from the observed value, at
stilling wells or pressure recording tide gauges such as pneumatic gauges). every point where a reading is available and then minimises the sum of the
• The length of record required depends on the number of constituents that need to be squares. It can be thought of as a multiple regression approach.
identified. A 29 day analysis allows 14 astronomical constituents to be deduced
(Doodson, 1954). Over half of the standard ports in the UK use a full year of data in • Using a series of observations, z(t), the difference between the observed tide and
their tidal analyses. the predicted tide is (z - ζM2).
Fourier analysis is a mathematical technique for deducing the amplitudes and phases The sum of squared error
∑ (z (t ) − ζ (t ))
• •
=
2
of tidal constituents from observations. But, an uninterrupted time series at regular S
intervals is needed.
• Least squares fitting is more forgiving and allows data gaps and shorter time series. • . The value of S is then minimised to obtain Z0, HM2 and gM2

Shallow water constituents Shallow water constituents

• Shallow water distortion of the M2 constituent
• Making tidal predictions on the basis of frequencies present in the equilibrium tide
does not allow for
gives rise to new constituents M4 (twice the M2 M2
frequency), M6 (three times ), etc.
– meteorological perturbations T = 12.42 h
– shallow water constituents • In certain areas the shallow water effects take
the form of double high water (Southampton) or
• When the tidal wave enters shallow water (less than 50 m), several physical processes double low water (Portland, Hook of Holland).
act to modify and distort the shape of the waveform. Typically, the time between low
water and high water shortens giving the tidal curve a sharp rise and a slower fall. • The double HW at Southampton is NOT caused
The processes responsible are: by two separate waves being split by the Isle of T = 6.21 h
– Bottom friction Wight. It is a combination of M2 + M4 + M6 M4
– Non-linearity in the equations constituents.
– The wave amplitude is no longer small compared to the depth • For these effects to be significant the amplitude
• The result is that higher harmonics appear in the equations. These have frequencies of M4 needs to be large compared to M2 . For
that are simple multiples of the fundamental frequencies. The new shallow water double high waters, the ratio of M4 to M2 T = 4.14 h
amplitude needs to be greater than 0.25. This is M6
terms are sometimes called “overtides”. They are generated in shelf seas, and do not
exist in the equilibrium tide. the case around southern England where there is
a degenerate M2 amphidrome.
• These shallow water constituents must be included for accurate tidal predictions in
shelf seas and estuaries.
 Generating the shallow water constituents M6 species from the friction term
• The M4 shallow water constituent is usually the largest of the shallow water Friction = U|U| Fourier Series
constituents. This originates through non-linearities in the governing
equations (which is when one of the variables is multiplied by another).
• An example is the change in water column height that arises in the continuity U = cos ωt
equation Solid line = U|U|
1 ∂ζ
( Z 0 + ζ ) ∂t
0 90 180 270 360
• Terms like M4 that oscillate with twice the semidiurnal frequency are called
quarter diurnal, or (sometimes) species 4
• It can be shown that the quadratic drag law for friction at the bed , -kU2 (as in
the first practical) gives rise to harmonics of species 6 and 10. (i.e. six cycles Dashed line = A cos ωt + B cos 3ω
ωt + C cos 5ωt
per day and ten cycles per day). M2 M6 M10

Shallow water interactions Double high waters (low waters)

• Interaction between constituents can also occur (called “intermodulation”). For • Tidal predictions for most UK ports can be made fairly accurately by including shallow
example, interaction between M2 and S2 gives rise to a shallow water constituent MS4 water constituents up to species 6.
whose frequency is the sum of the original frequencies (0.5059 + 0.5236 rad h-1). • Tides become more severely distorted in estuaries. Prediction requires a more complex
• Non-linear interactions between the tidal constituents (M2, S2, O1, K1, etc) can generate treatment (100 terms in the extended harmonic method) with the inclusion of shallow
a whole range of shallow water terms. water constituents up to species 10.
• In a semidiurnal regime, only the even harmonics are important (M2 M4 M6, etc) but • A qualitative picture of the double high and low waters that occur at certain ports (e.g.
in a mixed regime there can be interactions between diurnal and semidiurnal tides Southampton and Portland) can be obtained by constructing a tide comprising just the
giving rise to odd harmonics (e.g. in Anchorage, Alaska, species 5 tides are observed). M2 and M4 constituents.
• To appreciate the relative size of shallow water constituents, at Liverpool:
ζ = A2 cos ωt + A4 cos (2ωt - λ)
– M2 305 cm
M2 M4
– M4 22 cm
– M6 6 cm • In the following graphs the M4 amplitude (A4) is taken as 0.4 times the M2 amplitude
– M8 3 cm (A2). Different tidal curves result (the bold lines), depending on the relative phase, λ ,
– MS4 12 cm of M4 to M2.
• The magnitude of the M4 tide is proportional to the square of the magnitude of the M2 • NB. For the double high or low waters to occur, A4/A2 ≥ 0.25
tide (M4 ∝ M22 ). Shallow water distortion is therefore most severe at spring tides.

Double low water (λ = 0) Slow rise, rapid fall (λ = 90)

Double high water (λ = 180) Fast rise, slow fall (λ = 270)