Sensitivity, Uncertainty and Refinement of a Global Flood Model

  • Cain T Moylan

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)


Floods are a major natural hazard, causing significant damage and fatalities across the world. The modelling of floods is possible, but has been historically performed by developed nations, with good access to data, allowing them to build and test hydraulic models which can predict the frequency and severity of flooding events. However, in recent years, satellite applications have led to a proliferation of remotely sensed data products, hence making it feasible to run global scale flood models. However, while feasible, the new generation of global flood models face serious issues, as the datasets are coarse and there are significant assumptions required for consistent methodologies. It is therefore necessary to understand the effect that these assumptions have on flood hazard predictions. In this thesis, one of the new generation of global flood models is analysed. First, the uncertainty of the model’s parameters is assessed, which codify assumptions into the modelling methodology. From a pool of 36 parameters, it is found the vast majority of the model’s output variance can be represented with only 7 parameters. Following this, the model is tested more rigorously at a single case-study location, the Po river basin in north Italy. Here the model is assessed for its ability to represent local conditions, and it is found that even a global scale model requires some information at the local scale to appropriately constrain the uncertainty of the predictions. Once done, there are modelling outputs of reasonable skill. The final chapter focusses on the prediction of peak flows, found to be the most uncertain component in the methodology. The regionalization scheme of the model was refined to incorporate uncertainty estimation. This thesis has therefore started the long process of incorporating uncertainty into global flood modelling and serves as a guide for further work in the field.
Date of Award18 May 2021
Original languageEnglish
Awarding Institution
  • The University of Bristol
SupervisorJeff Neal (Supervisor) & Jim E Freer (Supervisor)

Cite this