Hoofdstuk 3: Irregular waves

Introduction
The linear wave theory gives a description of regular waves that can be
described with a sinusoidal profile and constant wave characteristics. These
characteristics consist of 3 parameters:

- Wave height H
- Wave period T or wave length L if water depth d is known (L = f(d,T))
- Wave direction θ

Due the constant wave characteristics, monochromatic waves are obtained and
the parameters are studied with an deterministic approach.

So, for a regular, linear wave in time and space we can apply the linear wave
theory. This means that a number of equations can be used that are only valid
for this kind of waves.

In reality, the sea state consist of waves with varying (in time) of wave height
and periods moving in different directions. For that reason, a stochastic
approach of the different parameters will be used. Further, it will become
necessary to treat the characteristics of the sea surface in statistical terms.
This will make the analysis more complicated, but a more realistic description of
the sea surface is obtained.

Opm. a statistical description of the waves will not be done based on the normal
distribution because it’s not good enough. For the waves, a Rayleigh distribution
of individual wave heights will be used.

The name irregular waves will be used to describe the natural sea states where
the wave characteristics have a statistical variability in contrast to
monochromatic waves (= has constant properties).
Irregular waves
The following figures shows a typical time series of a water surface elevation,
measured at 1 location of an irregular sea state. Here, it’s clear that we have a
significant variability in wave height H and period T from wave to wave.

For that reason, as an engineer we need to search for a representative wave

characterises of wave height H and wave period T. Therefore, we need to take
the following thins into account:

- Making use of a statistical approach for stochastic process

- Simple indicator for accurate description of the severity of the complete
wave conditions
- Smaller and larger waves are also present
- Analyse can be done in time domain or in frequency domain

Time domain analysis (1st option)

The time domain analysis can be done with the so-called “wave-by-wave analysis”.
Therefore, a wave record with a duration of T r is needed where we will make an
identification of N individual waves (i = 1,…,N) with each a wave height H i and
wave period Ti .

In other words, we look for an algorithm (the so-called Zero-Down-Crossing ZDC

method) that can identify the individual waves in the time domain. This algorithm
goes as followed:

- When the water surface elevation is crossing the zero-line in a downwards

branch, we can identify the start of each individual wave.
- For each individual wave, a wave height Hi and wave period Ti can be found.
- We obtain a table with n waves, where a statistical analysis is possible.

Zero-line = still water level, or in other words the level where we have a water
surface elevation of 0.
The result we obtain with the first option, the time domain analysis by using the
Zero-Down-Crossing algorithm, will be different than the result that would be
get when a frequency domain analysis would be used.

Representative wave height parameter(s)

With the Zero-Down-Crossing algorithm, we find N number of individual waves in
the wave record. These waves can then be ranked from highest to lowest, where
we then can find different representative wave height H parameters:

- Significant wave height H1/3

- Mean wave height Hm or H
- Root-mean-square height Hrms
- Maximum wave height Hmax

Significant wave height H1/3

The significant wave height is the mean value of the highest 1/3 of all the
waves. A first reason to make use of this significant wave height H 1/3 is that we
are kicking out the small waves, which are not interesting for the design of
coastal structures (they are not damaging it).

A second reason to use it, is because it’s very simi to the estimated visual height
by an experienced observer.

Mathematical, the significant wave height H1/3 can be calculated as:

Root-mean square height Hrms

Mathematical, the root-mean square wave height H rms can be calculated as:

Maximum wave height Hmax

Sometimes, damage to a coastal structured is not caused by the highest wave,

but due to a train of waves that pass. Therefore, not only the maximum wave
height is interesting.
Opm. We can also make use of an average height of the highest 1/n of all the
waves in record. This will then be noted as H1/n . For example, H1/10 is the average
height of the highest 10% of all waves in the record.

Energy in sea state

In a sea state with N individual waves, the energy E per unit surface area can be
calculated as:

Hereby, we make use of the root-mean-square height H rms which is also

proportional to the average energy E per unit area:

 E ~ Hrms

Probability distribution for individual wave height Hi

The probability that a wave height Hd is going to be exceeded can be written as:

 P(H > Hd) = m/N

Hereby, m is the number off waves higher than Hd . And N is the amount of
individual waves of the used wave record.

Further we can say the following things:

- E(x) = exceedance probability of the stochastic variable x (here H d)

- F(x) = 1 – E(x), is the so-called probability of non-exceedance or the

CumDistrFunc CDF.

- dF(x)/dx = p(x), is the so-called probability density function or the

ProbDensFunct pdf

The probability distribution P(H > Hd) can be fitted by a theoretical distribution,
here the so-called Rayleigh distribution (blue) is used. This is not a normal
distribution (red), but a skewed distribution because there are more small than
large waves.
Opm. Rayleigh distribution is applied on wave heights, and not on wave periods.

In more general notation, we replace the notation H d with “H = x” . This will

result in the following equations:

 EX(x) = Prob( X > x )

 x or H = the independent stochastic variable.

Assumptions of the Rayleigh distribution for individual wave heights provides

useful relationships between H1/3 and other representative wave heights:

These relationships are interesting for design, because for vertical breakwaters
we need the (very) high waves, but only H1/3 is known. By accepting Rayleigh
distribution and these relationships, we can find other values of H 1/n .

We also see in these relationships that H max is almost equal to 2x the significant
wave height H1/3 :

 Hmax ≈ 2 ∙ H1/3
 A very important thing to take into account with the Rayleigh distribution, is the
validation in deep and shallow waters:

- For deep water waves, Rayleigh distribution is very good approximation

- For shallow water waves, the Rayleigh distribution becomes an
overestimation of the wave heights, therefore it’s not longer valid. Reason
therefore is that the largest waves break in the shallow water, this means
that we have more smaller wave heights. For that reason, the Rayleigh
distribution makes an overprediction for the larger wave heights.

The “x” are a wave record plotted

in the graph. Here, we see that
small waves have a high exceeding
probability E(x). For the high Special scale to obtain
waves we see a difference a linear relation
between the measured and
theoretical Rayleigh value. Reason
is that measurement happened in
a shallow water. So, Rayleigh is an
overestimation for the high
waves, due braking of these
waves in shallow water.

Representative wave period parameter(s)

With the Zero-Down-Crossing algorithm, we find N number of individual waves in
the wave record. This gives us the opportunity to speak about different
representative wave period T parameters:

- Significant wave period T1/3

- Zero crossing period TZ or TGTZ
- Tmax
- Tm = T

For these periods, there is no possibility to make use of a fitting with the
theoretical (Rayleigh) distribution of wave periods. The Rayleigh is only possible
for wave heights, and not for wave periods.

Significant wave period T1/3

The significant wave period T1/3 is the mean value of the periods of the highest
1/3 of all the wave heights.
 Zero crossing period TZ

The zero crossing period TZ can be calculated as followed:

 TZ = T r / N

Related wave periods

Also for the wave periods, a relationship for the different wave periods exist.
This relationship can be used as a rule of thumb:

 Tmax ≈ T1/3 ≈ (1,1 - 1,3) ∙ T

Frequency domain analysis (2nd option)

As a second option, we can make use of a frequency domain analysis. This will
result in a solution (a frequency domain with a frequency spectrum) which is
different from the time domain analysis.

For the frequency domain analysis, we make a conversion of the time series of
the wave record into a wave spectrum by making use of Fourier analysis (FFT).
This wave spectrum, shows the energy for each individual spectral wave
component.

The wavey energy spectral density E(f) [m²s], or in other words wave spectrum
can directly be obtained from a continuous time series of the surface η(t) by
using the Fourier analysis. With the Fourier analysis, a wave profile can be
written as an infinite sum of sinusoids with an amplitude a n and angle θn :

 η ( t )=∑ a n ∙ cos ( ϑ n )

And each spectral wave component given as:

2
a
 E ( f n )= 1 ∙ n
2 ∆f

S(f) = spectrum, which can be given as:

2
1 an
 S(f) = a ; S(f) = a² ; S(f) = ∙
2 ∆f

This one will be used here

 = m0 = surface area under the spectral energy density

The parameter of is called the moments of a spectrum. This can be obtained by:

 Moments of a spectrum m0 =

Hereby, we can further define the 0th- moment “m0”, which is the total area
under the wave energy density spectrum. This parameter m 0 can be used to
define the significant wave height for a given E(f) .

For the parameter m0 , 2 types of definitions can be used:

- Mathematical: m0 is the variance of the time series σ²n , and so will be the
total area under the spectrum

- Physics: m0 is the energy in the sea state per m² (E/ρg)

Besides the moment of order 0 (= m0), it’s also possible to define a moment of
the order i (= mi):


 Definition of significant wave height
For a frequency domain analysis, it’s also possible to define a significant wave
height Hm0 as followed:

Usually, as an approximation, the following relationship will be used:

Out of this formulas, it’s clear that the area under the spectrum m 0 is
proportional to the significant wave height Hm0 of a frequency domain analysis.

Definition of peak wave period Tp

The peak wave period Tp is found by defining it as the frequency where we find
the highest amplitude of the wave spectral density E(f n) . In other words, at
that point of highest amplitude, we have the highest amount of energy.

On the figure, we see different wave spectra of a fully developed sea for
different wind speeds. For heavier winds/storms, more and more long waves with
more energy are created.

Here, highest amplitude, so there the

highest amount of energy. Means that at
this point, we find peak wave period Tp
Here, we have a high frequency.
Here, we have a small frequency. This means a low period, or in
This means a high period, or in other words short waves.
other words long waves.

Parametric spectrum models

The sea state, and thus the spectrum at a location, depends on different local
conditions:

- Wind characteristics (duration D, speed uW and direction θ)

- Fetch length F
- Water depth d
- Presence of coast (resulting in wave transformations)
- Start, maximum or end of the storm
Due, to this local conditions, the wave spectrum has no specific mathematical
formulation. However, empirical expressions are available, which fit the
observed spectra worldwide. The reason for the development of these empirical
expressions are the engineering applications that exist.

The development of the empirical expressions will happen by making use of

parametric spectrum models. Therefore, we have 2 different options:

- Pierson-Moskovitz (PM) spectrum, can be used for fully developed sea.

This means that only 1 parameter is important, the wind speed U s . the
fetch length F and duration D are here unlimited.

- Jonswap spectrum, can be used for fetch-limited sea, or in other words

for not fully developed waves. Examples are situations with a short fetch
length, like in the Join North Sea Wave Project.

- Many other parametric spectrum models are available.

Depending on the used parametric spectrum model, expressions are available as

a function of the significant wave height Hs and the peak wave period Tp .

Besides other parameters spectrum models that are available, it’s also possible
to expand these 2 models (PM and Jonswap spectrum):

- Deformations in shallow water

- Directional spectra
Jonswap experiment

Following picture shows a development of wave spectra of a developing sea for

increasing fetches. These are measured at different locations, but for a same
wind speed. In other words, the fetch length of the wind is variable.

For a long fetch length, the peak wave period becomes higher, because the wind
has a longer opportunity to put energy into the waves. Therefore, the amount of
energy (= amplitude of the peak) becomes higher.

Example

Following picture shows a measured at a certain location. Here, we see 2 swells

with a different frequency at 1 location. The first swell is developed somewhere
elsewhere, and is almost vertical. Meaning that it’s almost a regular wave.

The second swell will be a local record because we have more spreading of the
frequency, meaning that it’s a irregular wave.

More spreading, so an
irregular wave period of a
local record.

Almost vertical, meaning a

regular swell wave coming
from somewhere else
Significant wave height “Hs”
The significant wave height, coming from a time or frequency domain analysis, is
generally noted as Hs . So, the value of this significant wave height H s can be
found in 2 ways, depending on the type of analysis that is used:

- In time domain: the calculated value of H1/3 is be used to define the

significant wave height Hs .

- In frequency domain: the estimated value of Hm0 is be used to define the

significant wave height Hs .