Correlation of water level estimations before and after a river intervention (Source: Berends et al. 2018).
- Novel method to estimate uncertainty in water level predictions based on the correlation of modelling results before and after a river intervention.
- Good agreement with the traditiona uncertainty estimation approach of trying all possible parameter combinations at lower computational cost.
- Efficiency increases if more interventions are evaluated.
- Accuracy of uncertainty estimation is explicitly quantified.
- Method applied to an idealized river intervention for reducing high water levels.
On the right (see Figure), an example correlation of water level estimations before and after a river intervention. The confidence interval (dashed line) is calculated from a subsample (+ points in the Figure) of all possible water level estimations.
Comparison of the water level reductions and its uncertainty range for two different interventions along the flowing direction of the river. (Source: Berends et al. 2018).
Implications to practice
To analyze the expected effect of a river interventions, advisors and project managers rely on computations with hydraulic models that are physically based. The model results are sensitive to uncertain input parameters. Uncertainty estimations require calculations with all possible combinations but long model runtimes are infeasible with standard computer resources. A new method was demonstrated on an ideal 1D situation for river interventions that aim at lowering water levels such as dike relocation and vegetation removal. The application to more complex modelling approches (2D and morphological) will be studied in future work.
- Berends, K. D., Warmink, J. J., & Hulscher, S. J. M. H. (2018). Efficient uncertainty quantification for impact analysis of human interventions in rivers. Environmental Modelling & Software, 107, 50–58. https://doi.org/10.1016/j.envsoft.2018.05.021.
- Here is available the full list of related conference abstracts related to this project.
The method described in this paper including test data for the cases described is available for download at https://dataverse.nl/dataset.xhtml?persistentId=hdl:10411/GCS6HE.
A new implementation of the tool including example data is available from: https://github.com/kdberends/coral
Uncertainty quantification of flood mitigation predictions and implications for decision making
Last modified: 19/05/2019