May 6-8, PROBABILISTIC SENSITIVITY ANALYSIS OF DUNE EROSION CALCULATIONS 4 th International Symposium on Flood Defence, Toronto, Canada Kees den Heijer, Jan van de Graaff and Pieter van Gelder Section of Hydraulic Engineering
May 7, Contents Introduction Model description Probabilistic model Dune erosion model Approach Results Conclusions
May 7, Introduction approximately 350 km Dutch coastline 254 km dunes 34 km sea dikes 38 km beach flats 27 km of boulevards, beach walls etc. Design conditions very extreme (10 -5 year -1 ) No prototype data for these extreme situations Policy to maintain coastline of 1990
May 7, Introduction Current method for Dutch dune safety assessment: Empirical model Applied in semi-deterministic way Conditions based on probabilistic investigations (WL | Delft Hydraulics, 2007)
May 7, Objective This study aims at more insight in the sensitivity of the rate of dune erosion for the stochastic characteristics of various variables, currently taken into account. Remarks: year -1 probability of exceedance DUROS-plus dune erosion model Simple cross-shore profile
May 7, Model description Probabilistic model Prob2B (former 'Probox'; Courage & Steenbergen, 2007) FORM method Easy to couple with other software
May 7, Model description Dune erosion model (DUROS-plus)
May 7, DUROS-plus Within McTools (Van Koningsveld, Stive and Mulder, 2005), recently the DUROS-plus code has been re- written: Efficient iteration procedure Modular structure
May 7, Approach Stochastic characteristics of reference situation are mainly based on the probabilistic investigation of WL | Delft Hydraulics (2007) The stochastic characteristics have been changed one by one, keeping the others the same
May 7, Overview Reference situation ParameterMean valueUncertainty/varianceDistribution type Water levelRelated to P exceedance -Conditional Weibull Wave heightRelated to water level0.6 mNormal Wave periodRelated to wave height1 sNormal Grain size 225 m10% of mean (22.5 m) Normal Profile fluctuation060 m 3 /m 1 Normal Surge duration010 % * ANormal Model accuracy015 % * ANormal
May 7, Reference situation
May 7, Design point VariableValueRelative contribution [%] Water level5.48 m89.53 Model accuracy58.4 m 3 /m Grain size 208 m 3.28 Surge duration26.0 m 3 /m Wave height7.92 m1.26 Profile fluctuation-23.4 m 3 /m Wave period12.71 s0.19 Retreat distance78.8 m
May 7, Overview of investigations VariableMean valueUncertainty/variance Water levelReference situation + [-0.5, -0.25, 0, 0.25, 0.5] m- Wave heightReference situation + [-0.5, -0.25, 0, 0.25, 0.5] m[0, 0.3, 0.6, 0.9 and 1.2] m Wave periodReference situation + [0, 1 and 2] s[0, 0.5, 1.0, 1.5 and 2.0] s Grain size [200, 225, 250, 275 and 300] m[0, 5, 10 and 15] % of 225 m Profile fluctuation[-20, 0 and 20] m 3 /m 1 [0, 30, 60, 90 and 120] m 3 /m 1 Surge duration[0, 5 and 10] % * A[0, 5, 10, 15 and 20] % * A Model accuracy[0, 5 and 10] % * A[0, 5, 10, 15 and 20] % * A
May 7, Results: Sensitivity water level
May 7, Sensitivity mean grain size
May 7, Standard deviation grain size
May 7, Standard deviation peak wave period
May 7, Conclusions Water level and grain size distribution are the most important variables in current safety assessment method for the Dutch dune coast Proper field data of these variables are crucial The other variables do have their contribution, but changing their stochastic char. does not have so much influence
May 7, Water level
May 7, Wave height
May 7, Peak wave period