Efficient quantile estimators for river bed morphodynamics

The large impact of civil water engineering to nature and society imposes high requirements for the precision of numerical simulations used in planning and evaluation of river engineering concepts. In particular, coupled morphodynamic - hydrodynamic simulation uses models of river bed evolution possessing uncertain parameters. The sources of uncertainty can be the natural variability, the deficient description of the physical processes in the model and the imprecision of the model parameters. The propagation of these uncertainties to the variance of the model result can be quantified with the aid of stochastic analysis. Precise evaluation of stochastic characteristics normally requires a huge amount of samples, which can be provided by surrogate-based modeling of simulation results. In this paper we present our advances in quantile estimation of morphodynamic simulations of river bed evolution. We use metamodeling of bulky simulation results with radial basis functions (RBF), quasi-Monte Carlo sampling (QMC) and efficient quantile estimator (QE). Four different quantile estimators have been tested. A realistic application case is used to demonstrate the efficiency of the approach.