You are on page 1of 4

MORPHOLOGICAL INPUT REDUCTION

A fundamental problem with performing process-based morphological modelling is that


morphological evolution of coastal features of interest to engineers, managers, and the public at
large usually occurs at time scales several orders of magnitude larger than the time scale of the
hydrodynamic fluctuations driving the sediment transport. (Lesser, 2009).As discussed by de
Vriend et al. (1993), this potential separation of scales also provides a basis for a range of
morphological acceleration approaches. They consequently discuss three distinct approaches to
accelerating long-term morphodynamic modelling:
1. Input reduction, which is based on the idea that the residual (long-term) effects of smaller-scale
processes can be obtained by applying models of those smaller-scale process forced with reduced
“representative” inputs. (wave schematization, morphological tide)
2. Model reduction, which is based on the idea that the model itself can be reformulated at the
scale of interest without describing the details of the smaller-scale processes. (morphological
acceleration factor-morfac)
3. Behaviour-oriented modelling, which attempts to model the phenomenon of interest without
attempting to understand or describe the underlying processes. (shoreline sediment processes,
undertow, wetting-drying etc.)
INPUT REDUCTION
The first thing to do in creating representative time series or climate of conditions is to select the
basic input parameters. A first list of these parameters in a typical coastal model would be:
(1) tidal amplitude or phase within the spring-neap cycle;
(2) offshore wave height, period, direction and spectral shape at some representative location;
(3) wind speed and direction;
(4) river discharge;
(5) surge level;
It is important to realize that some of these parameters are correlated while others are not:
(1) The tidal boundary conditions are not related to the other parameters, although the tidal motion
in a coastal model is modified by the surge, waves and wind;
(2) The surge level is often correlated with the wind, though with considerable scatter; it is also
correlated with the wave height for a given directional sector.
(3) The wind wave direction is correlated with the wind direction, especially for strong wind
conditions; swell wave direction generally is not correlated much with the local wind;
(4) The wind wave height for a given direction is related to the wind speed;
(5) The wind wave period is correlated with the wind wave height
Before we can make sensible simplifications to the input conditions, we have to study these
relations based on the available data and an analysis of the situation. (Roelvink,2011)
1. MORPHOLOGICAL TIDE
Guidelines for selecting the morphological tide are presented in Roelvink (2011) and Lesser
(2009).
The recommended procedure for selecting a morphological tide is to calculate the required M2
and C1 harmonics using Equation 5.13. Over-tides of M2 should also be included if they are
significant. The optimum factor f2 should then be determined through trial and error. A brute-force
simulation including typical levels of other important forcing processes, such as waves, should be
simulated for at least the duration of a neap-spring cycle. The simulation should be repeated using
morphological tides consisting of f2 × M2 + C1 + M2 overtides using a range of f2 factors. These
simulations could be accelerated to save computational effort. The optimum factor will typically
be close to that given by Equation 5.16 if non-tidal residuals are significant, and closer to 1.0 if
they are weak. Choose the factor which most closely matches the pattern of the neap-spring
bathymetry changes (e.g. minimises MSE). The gradient of the correlation line should be used as
an additional time-scale factor if it significantly deviates from 1.0.

2. WAVE SCHEMATIZATION
Various wave input reduction techniques have been presented in literature. (Benedet et al,2016 ;
Roelvink, 2011)
The objective of wave input reduction is to define a limited number of offshore wave classes which
together produce the same residual sediment transport patterns and rates as the full time series of
offshore wave conditions over the time period of interest.
Five different methods of wave climate schematization are presented in Benedet et al. (2016), they
are named:
 ‘Fixed Bins Method’ (FBM),
 ‘Energy Flux Method’ (EFM),
 ‘Energy Flux with Extreme Wave Conditions Method’ (EFEM),
 ‘CERC Method’ (CERC)
 ‘Opti-Routine Method’ (Opti).
Four of these methods are directly based on the wave climate characteristics. The Opti-Routine
Method is based on transport patterns resulted from modeling simulations.
The outline of the latter method, called ‘OPTI’, is as follows (Roelvink, 2011):
• Set up a (coarse-grid if necessary) flow, wave and morphology model that runs over one tidal
cycle, computing the tide-averaged sedimentation/erosion pattern
• Run this model for each of the wave/wind/tide conditions (for a short time only)
• Compute the ‘target’ sedimentation/erosion pattern as the weighted average over all
sedimentation-erosion patterns, taking into account the probability of occurrence or ‘weight’ of
each condition
• Find a reduced set of conditions and weight factors that produce the same sedimentation-erosion
pattern as the ‘target’ one
The optimization procedure developed is as follows:
(1) We start with the given set of weight factors. If we add up all the individual patterns multiplied
by these weight factors we get the target pattern. By the way, this adding up does not cost any
significant computation time.
(2) Create a large number of ‘mutations’, typically in the order of 1000, where we vary the weights
randomly within a certain range. For each of these mutations we compute the weighted average
pattern and compute some error statistics in this pattern compared to the target pattern.
(3) Select the mutation with the smallest error and keep track of error parameters
(4) Remove the condition with the smallest contribution to the average pattern
(5) Return to 2
What happens in this process is that weights gradually grow or reduce until they become either
dominant or extinct, and that the overall pattern remains almost unchanged until we’re down to
less than 10 or even 5 conditions. When we plot the error statistics against the number of
conditions remaining we usually see a point where the error increases sharply; obviously we want
to stay away from this point.
REFERENCES
Benedet L., Dobrochinski J.P.F., Walstra D.J.R., Klein A.H.F., Ranasinghe R., 2016, A
morphological modeling study to compare different methods of wave climate schematization and
evaluate strategies to reduce erosion losses from a beach nourishment project. Coastal Engineering
112, pp. 69–86
de Vriend H.J., Capobianco M., Chesher T., de Swart H.J., Latteux B. and Stive M.J.F, 1993,
Approaches to long-term modelling of coastal morphology: A review. Coastal Engineering 21, pp
225-269
Lesser, G.R., 2009, An approach to medium-term coastal morphological modelling. PhD thesis,
UNESCO-IHE & Delft Technical University, Delft.
Roelvink J.A. and Reniers A.J.H.M., 2011, A Guide to modeling Coastal Morphology. Advances
in Coastal and Ocean Engineering: Volume 12, Word Scientific

You might also like