Flood risk mitigation by reservoirs - application of multivariate statistical methods for risk assessments downstream

摘要

A flood is an extreme hydrological event which is characterized by several correlated variables (flood peak, volume and duration). The efficiency of flood reservoirs is strongly affected by the event-specific combination of the two correlated characteristics flood peak and flood volume. As flood events with unusual combinations of these characteristics can be more critical than a single design flood, growing emphasis is given to multivariate frequency analysis to consider this uncertainty. Here a new approach to assess the efficiency of a reservoir for downstream flood risk mitigation which considers interactions of tributaries is discussed. It is shown that coincidences of floods which are originating from different sub-basins have a large impact on the efficiency of a reservoir. As the interaction of tributaries depends strongly on the spatial and temporal distributions of rainfall, the event-specific flood risk mitigation by a reservoir stays uncertain. This uncertainty can be described by multivariate statistical analyses. Here it is shown how multivariate statistics can be used to locate reservoirs within river basins under consideration of their capacity to mitigate floods downstream.