Impact of different weather generator scenarios on extreme flood estimates in Switzerland

Abstract

The estimation of extreme floods using long continuous simulations is linked to uncertainties which are inherent in different components of the modeling chain. The main objective of this study was to investigate the role of precipitation input data from a weather generator for extreme flood estimates. A hydrometeorological modeling chain consisting of a multi-site weather generator (GWEX) at an hourly time scale, a rainfall-runoff model (HBV) and a hydrologic routing model (RS Minerve), was implemented, using different parameterizations of GWEX. While the sensitivity to the altered precipitation inputs was not uniform across the selected catchments due to their different physiographic characteristics, we found that the uncertainty of flood estimates increased with increasing return period. In addition, the flood peaks were strongly affected when a bootstrapping of precipitation was performed and to a lesser extent when weather types (WT) were used to condition the parameters of GWEX. However, the latter seemed to reduce the spread of the uncertainty both in generated precipitation and simulated floods. Therefore, results suggested that precipitation inputs strongly contribute to the uncertainties of extreme floods. Accounting for uncertainty information enhances the usefulness of long continuous simulations and is essential as a context for applications including hydraulic engineering, spatial planning and safety assessments.

Bibliographic Information