Extending coupled hydrological-hydraulic model chains with a surrogate model for the estimation of flood losses

Abstract

In comparison to a local-scale flood risk analysis, modeling flood losses and risks at the river basin scale is challenging. Particularly in mountainous watersheds, extreme precipitation can be distributed spatially and temporally with remarkable variability. Depending on the topography of the river basin and the topological characteristics of the river network, certain rainfall patterns can lead to a synchronization of the flood peaks between tributaries and the main river. Thus, these complex interactions can lead to high variability in flood losses. In addition, flood inundation modeling at the river basin scale is computationally resource-intensive and the simulation of multiple scenarios is not always feasible. In this study, we present an approach for reducing complexity in flood-risk modeling at the river basin scale. We developed a surrogate model for flood loss analysis in the river basin by decomposing the river system into a number of subsystems. A relationship between flood magnitude and flood losses is computed for each floodplain in the river basin by means of a flood inundation and flood loss model at sub-meter resolution. This surrogate model for flood-loss estimation can be coupled with a hydrological-hydraulic model cascade, allowing to compute a high number of flood scenarios for the whole river system. The application of this model to a complex mountain river basin showed that the surrogate model approach leads to a reliable and computationally fast analysis of flood losses in a set of probable maximum precipitation scenarios. Hence, this approach offers new possibilities for stress test analyses and Monte-Carlo simulations in the analysis of system behavior under different system loads.