Modeling Concepts - Coastal Lagoons: Ecosystem Processes and Modeling for Sustainable Use and Development - page 298

gradients in the velocity field should be chosen, because a constant eddy formulation
will not give correct results in areas with such varying depth.
The wind friction factor (drag coefficient) is used as a calibration parameter
when a simulation includes wind forcing, for example, surge level set-up.
The bathymetry is by far the most important calibration parameter. Initial cali-
bration attempts must always focus on the model bathymetry before trying to vary
other calibration parameters. In fact, several changes may have to be made in such
areas where the bathymetry has been schematized in order to represent details that
are finer than the grid dimensions.
The boundary conditions can also be used to calibrate the model. For example,
the introduction of a sea-level slope along an open boundary to simulate geostrophy
across that boundary may significantly affect the model results.
A soft start (or ramping time) can be applied at the beginning of the simulation
to avoid initial numerical instabilities and to get rid of oscillations. If a simulation
includes, for example, wave-generated radiation stresses along a shoreline, it is
necessary to have a very long soft start to avoid shock waves generated by the
instantaneous application of high long-shore current as normal boundary conditions.
A model should be calibrated for all different situations that are relevant to the
study . In other words, calibration coefficients for simulations of seasonal processes
could differ from those required for simulations over daily time scales. For instance,
if tidal propagation in a region is expected to be highly variable over time and space,
the calibration may be performed separately for neap and for spring tides. Or, if the
hydrodynamic conditions vary with the seasons, e.g., due to monsoon influence, the
calibration should be performed for each part of this seasonal cycle.
A basic rule to follow is that all parameters should be tuned during calibration,
but one at a time . If more than one parameter is changed at one time, it may not be
possible to clearly identify which parameter caused what changes.
Finally, we can ask what level of accuracy should be reached before the model
calibration is completed? There is no clear answer to this question. It will depend
on the type of study performed, the quality of the boundary data and measurements,
the time allocated to the modeling investigation, and of course the required
accuracy.
In any case, the required accuracy must be established before the study is started.
An example of acceptance criteria for a 2D hydrodynamic model calibration and
verification can be: 37

• 5% on average (but not less than 0.1 m) for the tidal range
• 30% on average (but not less than 0.1 m/s) for current speeds
•45

° on average for current directions

However, the required accuracy of a solution can never exceed the accuracy of
the field measurements with which it is compared. One also may ask how represen-
tative the model solution is of the field measurements. For example, a current velocity
measured at a certain point above the bottom is not necessarily representative of the
vertically averaged velocity simulated by a 2D model. In this case, some additional
transformations are required. Assuming a logarithmic current profile in the vertical,