PhD candidate: ML-based prediction of flood types based on atmospheric and catchment attributesAre you interested in how floods are generated and which processes can lead to floods? Do you want to understand how extreme events manifest and how they differ in space and time? Do you want combine statistical and physical knowledge to understand patterns and coherences of these events? Then we may have the perfect opportunity for you!We are looking for a highly motivated PhD candidate to join our team (18 months with the intention for extension to up to four years, full time, fully funded) at the Hydrology and Environmental Hydraulics Group at Wageningen University. The position is based within the NWO-funded VIDI project “FutureFlood - Future-proof flood prediction” to develop novel classification methods for understanding the generating processes of flood events in both the atmosphere and catchment and to be able to predict the upcoming flood type, e.g. heavy-rainfall flood or rain-on-snow flood.
As PhD candidate you will compare several machine-learning based algorithms regarding their ability to predict the flood type based on atmospheric and catchment conditions and regarding their accuracy and sensitivity to the input data. A variety of algorithms will be tested, including ANNs, WNNs, SVM and CART, with the possible extension to fuzzy-based algorithms. You will base your work on two different flood classification approaches, namely a hydrology-based one and a hydrograph-based one and compare these regarding their ability to capture flood types and be used for prediction. Test cases will be selected around the globe to cover different climate zones and hydrologically diverse catchments, including the Netherlands, New Zealand, Germany and Brazil. You will analyse how the relevant flood-generating processes differ in space and time. In a second step, you will expand the developed classification framework to investigate how flood types will change under different climate change scenarios based climate projections.
This framework will be ultimately included in a flood prediction model, which will be developed within the VIDI project. For this purpose, you will exchange with the project team on regular basis. You will exchange with stakeholders in the case study regions to receive feedback on the applicability of the framework.
Your duties and responsibilities include: - Development of a flood classification framework for flood type prediction
- Comparison of different ML algorithms in a sensitivity study
- Communication with stakeholders
- Development of open source programme code (preferably as a software package)
- Exchange within the VIDI project to embed the results in flood prediction
You will work hereWe are the
Hydrology and Environmental Hydraulics Group (HWM), a dynamic and committed team of 15 academics, 5 support staff, 4 postdocs and 24 PhD researchers. Our focus is on hydrology and environmental hydraulics, which we combine to understand and quantify hydrological processes in catchments, aquifers, rivers and deltas. We develop novel sensing techniques, create models and perform field and lab experiments to advance process understanding.
You will be supervised by Svenja Fischer, Ryan Teuling and Claudia Brauer.