Stochastic modelling of river morphology
Uncertainties in the prediction of morphological effects due to civil engineering activities in rivers have generally received very little attention. Therefore, Delft Hydraulics and Delft University of Technology have embarked on a four-year research programme on methods for the stochastic modelling of morphological processes in rivers. Herein, the goal is to express morphological predictions in statistical properties of river bed topography, such as expected bed levels, variance and probabilities. The attention focuses on the reliability of the predictions due to uncertainties in future discharges.
| Figure 1: Result of Monte Carlo simulation, after a dynamical equilibrium has been reached |
|
Figure 2: Convergence of Monte Carlo results with increasing number of realisations |
To illustrate this effect of uncertainty in the river discharge, a case study was performed on scouring in a straight alluvial channel containing a constriction.
|
Figure 1 shows some results of a Monte Carlo simulation, in which the individual simulations are performed with the 1D morphological model SOBEK-Rivers. The results show an uncertainty in the predicted bed level of the same order as the mean bed level changes, which emphasises how important it is to take account of uncertainties in predicting morphological processes.
|
|
Figure 2 shows the convergence behaviour of the Monte Carlo results with an increasing number of SOBEK calculations. It is clear that more than 200 realisations were required to justify some confidence in the convergence of the results. This illustrates an important disadvantage of the Monte Carlo method, namely the large computation time needed. This motivates the forthcoming activities within this research programme, in which other stochastic prediction methods for river morphological processes will be investigated.
|
more information
For more information please contact Hanneke van der Klis.
Copyright © 2010, Deltares.
|
