Copulas-new risk assessment methodology for dam safety

Abstract

Consideration of a broad range of hydrological loads is essential for risk-based flood protection planning. Furthermore, in the planning process of technical retention facilities it is necessary to use flood events, which are specified by several characteristics (peak, volume and shape). Multivariate statistical methods are required for their probabilistic evaluation. Coupled stochastic-deterministic simulation may be applied to generate a runoff time series, since the required amount of data is generally not available. Even the effect of complex flood protection systems may be evaluated through generation of a data base by means of stochastic-deterministic simulations with subsequent statistical analysis of the individual hydrological load scenarios. Multivariate frequency analyses of correlated random variables are useful to specify these scenarios statistically. Copulas are a very flexible method to estimate multivariate distributions, because the marginal distributions of the random variables can differ. Here a methodology for flood risk assessment is presented which was applied in two case studies in Germany. © 2011 Springer Science+Business Media B.V.