Input Reduction Analysis for Long-term Morphodynamic Simulations

Abstract

Predictions of coastal morphology evolution are necessary to assess engineering solutions as well as understand coastal systems behaviors. Among the tools used to predict morphological evolution are the process-based models that make use of physical laws and empirical knowledge. Such models account for a considerable range of coastal processes and are rather complex, hence demanding substantial computational time. Usually, when using complex process-based models, reducing the size of the input parameters, named input reduction, is made necessary in order to reduce the computational effort.
The scope of the present study is to understand the influence of reduced wave climate on simulated morphological evolution. Input reduction algorithms, sequencing methods, number of cases and duration of the reduced wave climate were investigated and evaluated with a 1D (cross-shore) brute force simulation of 3.3 years in Noordwijk, Netherlands. Noordwijk is a wave-dominated sandy beach characterized by a double sandbar system that has an inter-annual net offshore migration. The assessment of the methods was carried out through cumulative skill scores, temporal evolution of profile perturbations (bars and troughs) and profiles at the end of the simulation.
Furthermore, the findings on the Dutch coast were validated with a 1 year-long, 1D (cross-shore) brute force simulation in Anmok beach, South Korea. Anmok is a wave-dominated sandy beach with crescentic bars in the nearshore morphological evolution.
One of the conclusions of the present study is that input reduction methods that have more control on the definition of bins such as pre-definition of wave height and wave direction bins perform better than methods that are based fully on the statistical properties of the wave dataset. Also, the order of the wave cases in the reduced wave climate must resemble at its best the natural variability of the full wave climate. Finally, a less robust reduced wave climate (with less representative wave conditions) applied in a smaller timescale performs better in terms of morphology than a more robust reduced wave climate (with more representative wave conditions) applied in longer time-scales.