Uncertainty and predictability

Given the many non-linearities involved in the generation of extreme floods, predicting them in deterministic and probabilistic frameworks involves a range of uncertainties and it is difficult to gauge the limits of how well extreme floods can be predicted at different space and time scales. Representing the statistical characteristics of extreme storms in space and time and their uncertainties is still a challenge. Challenges also exist with combining the precipitation probabilities with those of other system components (catchment states, river system) as it is difficult to specify the joint probabilities. Knowledge about the event characteristics and spatial and temporal characteristics is needed, in particular for the transition from small to extreme floods.

Key science questions

  • What are the limits of predictability of the characteristics of extreme floods by probabilistic and deterministic models?
  • What uncertainties are involved in predicting river floods as a function of space and time scales?
  • What data are needed to even better constrain models of extreme river floods?
  • Are there dependencies between predictability of floods and model complexity?