EU countries are required under the Floods Directive (FD) to map flood hazards and risks every 6 years. Witteveen+Bos have recently prepared the flood hazard maps for the Water Board Hollandse Delta. Hydrodynamic models have been prepared for for 7 dike rings and dike breach calculations have been carried out (1 per compartment) for different return periods of outside water levels (1/100, 1/1,000, 1/10,000 and 1/100,000). A total of approximately 700 breach calculations have been carried out.
The flood hazard maps and hydrodynamic models can be used for policy purposes and to calculate the effects of flood mitigation measures. However, these models are not suitable for use in a disaster situation because the calculation time is too long (order of day to days). The hydrodynamic models would have to be simplified for this to reduce the calculation time. The disadvantage of this is that the resolution of the flood information is reduced and choices about where to simply the model have to made.
The large dataset of breach calculations has the advantage that a Machine Learning (ML) model can be trained with it. The advantages of ML compared to a simplified hydrodynamic model are that they have very fast calculation times (order of minutes), and require no difficult choices about simplifications in the schematization. The goal of the research is to determine if it’s possible to generate a Machine Learning model (in combination with hydrological/hydraulic model) to predict inundation extent and depths. Challenges include finding appropriate ML algorithms, and applying the best performing one to generate inundation depths in locations other than the locations in the training set.
References
- Besseling, L. S., Bomers, A., & Hulscher, S. J. M. H. (2024). Predicting Flood Inundation after a Dike Breach Using a Long Short-Term Memory (LSTM) Neural Network. Hydrology, 11(9), 152. https://doi.org/10.3390/hydrology11090152
- Bentivoglio, R., Isufi, E., Jonkman, S. N., and Taormina, R. (2023). Rapid spatio-temporal flood modelling via hydraulics-based graph neural networks. Hydrol. Earth Syst. Sci., 27, 4227–4246, https://doi.org/10.5194/hess-27-4227-2023
Supervision
Are you interested in this assignment? Contact the Master thesis coordinator.
