Flood risk management decisions contain uncertainties from a variety of sources. The information on the flooding system is often scarce and subject to change. The models used to predict the behaviour of flooding systems are incomplete representations of reality and often are accompanied by rather limited, or no, validation information. Flood management decisions are driven by multiple values and objectives. Under these circumstances it is important the uncertainties are, as far as possible, accurately represented and communicated to decision-makers.
To support assessment of the influence and impact of uncertainty, methods of uncertainty characterisation, propagation and representation are required. Substantial work on uncertainty handling in flood risk models has been funded by Member States, the EU and elsewhere internationally in recent years. Over the past few years, projects have been carried out that seek to understand uncertainties in flood risk management and improve the extent to which uncertainties are reflected in decision-making. This task aims at developing new, more advanced techniques, and integrating their use in the context of flood modelling. In particular this tasks aims to:
- develop an approach and prototype software for propagating integral (total) uncertainty through integrated flood models. The approach will make use of data mining and computational intelligence techniques, in particular for emulating complex computer codes. To enhance resampling methods for parameter uncertainty analysis (Activity 1).
develop an approach and prototype software supporting the modular definition and execution of complex flood risk analysis studies including rigorous uncertainty analysis. The approach will consider both the human and computational aspects of this problem, and will be generic across system component models and analysis tools (Activity 2).
test, demonstrate, and disseminate the developed framework through application to pilot studies in the context of both policy and forecasting situations (Activity 3).