Checking the coupled model
Interactions of waves and currents
Waves can cause setup, i.e. an increase or decrease of the average water level.
In general there is an interaction between average water level and current velocity
on the one hand and waves on the other.
The influence of water level and current on wave propagation is clearly due to
the presence of terms depending on the water depth and the current in the
action balance equation.
The correct procedure for the modelling of regions with strong interaction
is to make alternatively wave and flow computations, until the
Setup and wave-induced currents are considerable only in very narrow regions, because the forcing by the waves is substantial only in surf zones. It is therefore useless to compute these in large regions. Only in nested regions close to the coast this computation makes sense.
In some circumstances the computations of wave-induced current and setup can be
It is difficult to give the general conditions for this decoupling.
If the wave-induced currents are relatively small the decoupling is justified.
In such circumstances SWAN can compute the setup directly, without the need
to use a current model. It is always an approximation.
Note that it is incorrect to compute setup in SWAN if the interaction of waves and currents is computed in the way as suggested in the previous section, since the setup is already included in the water level computed by the flow model.
Checking results of the coupled model
Since the wave-induced flow and the setup have to be computed using two coupled models, checking whether the results are correct is more complicated than the two models separately. Also the transfer of data between the two models must be tested thoroughly. Questions involved are: 1. are the units of the quantities written by one model the same as the units of the quantities expected by the other model as input? 2.Are the vectors defined with respect to the same coordinate system?
Therefore the following advice: Never (I mean NEVER) start with a complicated case, always
run a few simple cases first where you roughly know what result you should obtain.