Effect of water level uncertainty on critical dune retreat distance : a probabilistic approach

Dutch law requires that the primary water defences be assesses every six years. For the assessment of dunes in the 2006-2011 round of assessments, a semi-probabilistic dune-erosion prediction method was developed in which uncertainties in input parameters of the physical dune model, DUROS+, were taken into account. However, the uncertainty in the extreme water level distribution input was not considered. Results have shown that the water level is by far the most influential parameter affecting erosion. Furthermore, the estimated distribution of extreme water levels known to contain large statistical uncertainty. Therefore, the exclusion of this uncertainty in the probabilistic model was expected to affect the critical retreat distance. In the current study, the probabilistic model was augmented to include the uncertainty in the water level distribution, and the impact of this inclusion on the critical retreat distance (i.e. the retreat distance with n exceedance frequency of 10-5 per year) was computed at five dune locations along the Dutch coast. Further, the probabilistic computation method FORM was employed to determine the relative contribution of the water level uncertainty (as well as the remaining stochastic variables) to the critical retreat distance. An analyses was carried out to estimate and parameterize the water level uncertainty; additionally, a reduction in uncertainty was imposed to estimate the sensitivity of the critical retreat distance to the estimated uncertainty and to compute what value a reduction in uncertainty can offer. The impact on the retreat distance was substantial with increases in critical retreat distance ranging from 34% to 93% of the original estimate. The reduced uncertainty showed a more subtle impact, with increases in critical retreat distance ranging from 10% tot 26% of the original estimate. The relative importance analysis showed that the water level uncertainty has a strong influence, with the relative importance ranging from 10% tot 23% for an exceedancs frequency of 10-5 per year.