Page 226 - Fundamentals of Ocean Renewable Energy Generating Electricity From The Sea

Ocean Modelling for Resource Characterization Chapter | 8 215

waves of different frequencies and directions, are placed in two corners along
one side of a tank of constant water depth. The resulting waves create a diamond
pattern of crests and troughs, which has its own wave length, speed, and
direction. This diamond pattern would interact with a third-wave component,
if this third wave had the same wave length, speed, and direction as the diamond
pattern. This is the triad wave-wave interaction, which redistributes wave energy
within the spectrum due to resonance. Although each of the individual wave
components can gain or lose energy, the sum of the energy at each point in
the tank would remain constant. In deep water, it is not possible to meet these
resonant conditions (i.e. matching of wave speed, length, and direction), and
so triad wave-wave interactions cannot occur in deep water. However, in deep
water it is possible for a pair of wave components to interact with another pair
of wave components in a quadruplet wave-wave interaction.
Quadruplets transfer wave energy in deep water from the peak frequency
to lower frequencies, whereas triads transfer energy from lower to higher
frequencies, and transform single-peaked spectra into multiple-peaked spectra
as they approach the shore. Both are included as source terms in third-generation
wave models, and it is noted that both are computationally expensive. Triads, in
particular, are often omitted in wave model simulations, whereas quadruplets
are often included. For example, in the SWAN wave model, quadruplets are
activated by default in third-generation mode, whereas triads are not included
by default.

8.5 VALIDATION
According to the statistician George Box, ‘all models are wrong, but some
are useful’. Models, however sophisticated, are a representation of reality.
To provide confidence in how accurately a model has simulated reality, it is
necessary to perform model validation. Of course, validation depends upon the
availability of suitable in situ data. Although it is often desirable for such data
to be focussed in the region of interest, for example, at the approximate location
of a proposed wave or tidal energy array, this is not compulsory. If a model
performs well in one region, or under one set of conditions (e.g. during a spring
tide), this gives confidence in model performance in another region, or under
another set of conditions (e.g. during a neap tide). This is provided that the
model parameterizations (e.g. drag coefficient and eddy viscosity) are physically
realistic, and the model has not been excessively tuned to fit data in one region
or under one unique set of conditions.

8.5.1 Validation Metrics
Various metrics can be used to quantify model validation (e.g. [24]), and some
of the most popular metrics are presented in this section.