Which uncertainties matter most? New insights into complex flood risk models

Flood models rely on many uncertain inputs, for example damage estimates and flood probabilities, but how do we know which uncertainties matter most? And how do we manage complex analysis of these uncertainties to support decision making?

In 2017, we started a PhD collaboration with Prof Francesca Pianosi at Bristol University, Prof Thorsten Wagener (University of Potsdam) and Dr Georgios Sarailidis, working with Dr Kirsty Styles from one of our host companies JBA Risk Management Ltd. Georgios explored the use of formal sensitivity analysis methods in flood catastrophe modelling.

The original PhD project led to further research and case studies, supported with models and data supplied by JBA Risk Management, for large river basins in Europe and Australia. Outputs from this work have now been published in the journal NHESS at:

NHESS – Towards global sensitivity analysis of large-scale flood loss models

Risk models are often computationally expensive, so the research explored practical approaches to analyse how both individual inputs and their interactions affect model outputs. Uncertainties in damage functions were often the most influential.  However, reducing uncertainty in one area can shift importance to other areas, for example, flood probability estimates. Model design choices, such as how data is spatially aggregated, can also significantly affect results.

The research found that a structured, multi-variable approach can help prioritise which uncertainties to reduce, improving model usefulness for decision-making. It also highlights the value of sustained, collaborative research in tackling complex environmental problems.

Click here for further details