Short description and objective of the project:
Intro / background
Climate change is increasing the risk of large floods. Both due to extreme rainfall events (pluvial) and due to extreme river discharge (fluvial), threatening dike breaches. As a result, there is a growing need for fast and reliable flood hazard maps. Flood simulations are an important tool to assess the consequences of different flood scenarios and to support the creation of these maps. These simulations are based on numerical methods that solve physics-based mathematical equations describing fluid flow. However, a major drawback is that such simulations can be computationally expensive, especially for large-scale or high-resolution domains. Machine learning models offer a promising way to reduce computation time, but challenges remain regarding reliability and physical interpretability of the results generated by these models.
Objectives
The goal of this project is to build and train a machine learning model for 2D flood simulations.
Methodology / Scope
We foresee roughly the following scope, but please note that the scope/methodology can be adapted to the student’s interests.
A central question is how we can use machine learning models to support our numerical modelling activities. As a first opportunity we want to apply this to 2D numerical flood modelling. The first part of the project focuses on building and testing a simple machine learning model that predicts inundation levels after a dike breach using fictitious training data. The second part of the project aims to use the principles and lessons learned and apply this to a real world case, using flood simulation data of the River Rhine and River IJssel area.
Who are you and who are we?
You are motivated to do a fun but challenging MSc project combining machine learning and numerical modelling. You have experience with programming in Python (or are willing to learn) and numerical (hydrodynamic) modelling. You are motivated to work on climate adaptation topics and motivated to learn more about Artificial Intelligence and data science. You have strong analytical and communication skills. We offer committed guidance from our River, Coast and Sea team. The project will be carried out in our offices in Amersfoort and/or Zwolle.
Figure: Simulated inundation depths with D-Hydro after a dike breach in the River Rhine and River IJssel area.
Head graduation committee: tbd
Daily supervisors:
