Probabilistic estimation of dune erosion and coastal zone risk

Abstract

Coastal erosion has gained global attention and has been studied for many decades. As a soft sea defence structure, coastal sandy dunes protect coastal zones all over the world, which usually are densely populated areas with tremendous economic value. The coastal zone of the Netherlands, one of the most well-known vulnerable areas to coastal hazards, is mainly protected by the dunes. Therefore, the design, maintenance and assessment of the dune safety standard is extremely important for this low-lying country, especially under the circumstance of sea level rise (SLR). Generally, the safety assessment of the coastal dune includes the prediction of extreme storm events indicated in the form of return periods and the estimation of the dune erosion volume during the extreme storms. Based on the safety levels, a corresponding prudent land-use strategy should be formulated. The aim of this thesis is the prediction and extrapolation of the coastal dune erosion volume and dune retreat distance based on the limited field measurements of wave climate and coastal profile, and in the end, proposing a calculation framework to find the optimal land-use strategy. The traditional method, proposing a design storm event with a particular probability of exceedance, does not cover all the possible combinations of the wave climate parameters. Hence in practice, this method will lead to inappropriateness in estimation of small dune erosion exceedance probabilities. The Monte-Carlo technique can overcome this drawback by generating a huge number of storm events with different combinations of the wave climate parameters. The difficulty of realizing it is to maintain the statistical characteristics of every involved parameters and simultaneously make sure the dependencies between any two of them are simulated accurately. In this thesis, four statistical modelling methods are used and tested for the Dutch dataset which includes wave climate and beach profile data. All of these methods can generate synthetic time series of storm events for the purpose of dune erosion estimation The Gaussian copula method is recommended for the Dutch study site. The connection between the storm events and the consequences of the wave impacts on the dune is the dune erosion model. Considering the principle of the Monte-Carlo method, which relies on repeated random sampling millions of times, a too computationally expensive model is apparently inappropriate. In order to ensure the accuracy on the one hand, and reduce the model running time on the other hand, an empirical dune erosion model (DUNERULE), adjusted by the process based storm erosion model, XBeach model, was used. The dune erosion volumes with different return periods, ranging from 1×100 year-1 to 1×10-6 year-1 under different SLR scenarios are obtained. Three methods of reducing the simulation uncertainty are proposed to remove the physically unreal values. Besides, the exceedance probability curves of the coastal dune retreat distance by the year of 2100 were also predicted. This method was calibrated and validated in Narrabeen Beach, Australia. This application proves the validity and the applicability of the proposed method in the aspect of wave climate simulation and the dune erosion estimation. Based on the estimated erosion probabilities, economically optimal land-use strategies were evaluated and compared. To keep far from the unacceptably high erosion risk zone, a buffer zone can be created between the shoreline and the first line of the buildings along the coast. To move the erosion probability contours seaward, the beach nourishment can be carried out. Three different beach nourishment methods were analysed. Due to the many hypotheses and approximations in the computation process, the main point in this PhD research was to propose a generic computational methodology based on the maximum NPV. This thesis improves the method of probabilistic simulation of sea storms for a dune erosion exceedance probability estimation, simplifies the dune erosion model to make the Monte Carlo approach feasible, estimates the impact of projected SLR to the coastal erosion and recession, and, in the end, proposes a method to optimize land-use planning on the basis of cost benefit analysis.