A Coupled Approach with Stochastic Rainfall-Runoff Simulation and Hydraulic Modeling for Extreme Flood Estimation on Large Watersheds

Abstract:

The SCHADEX method aims at estimating the distribution of peak and daily discharges up to extreme quantiles. It couples a precipitation probabilistic model based on weather patterns, with a stochastic rainfall-runoff simulation process using a conceptual lumped model. It allows exploring an exhaustive set of hydrological conditions and watershed responses to intense rainfall events. Since 2006, it has been widely applied in France to about one hundred watersheds for dam spillway design, and also aboard (Norway, Canada and central Europe among others).

However, its application to large watersheds (above 10 000 km²) faces some significant issues: spatial heterogeneity of rainfall and hydrological processes and flood peak damping due to hydraulic effects (flood plains, natural or man-made embankment) being the more important.

This led to the development of an extreme flood simulation framework for large and heterogeneous watersheds, based on the SCHADEX method. Its main features are:

• Division of the large (or main) watershed into several smaller sub-watersheds, where the spatial homogeneity of the hydro-meteorological processes can reasonably be assumed, and where the hydraulic effects can be neglected.
• Identification of pilot watersheds where discharge data are available, thus where rainfall-runoff models can be calibrated. They will be parameters donors to non-gauged watersheds.
• Spatially coherent stochastic simulations for all the sub-watersheds at the daily time step.
• Identification of a selection of simulated events for a given return period (according to the distribution of runoff volumes at the scale of the main watershed).
• Generation of the complete hourly hydrographs at each of the sub-watersheds outlets.
• Routing to the main outlet with hydraulic 1D or 2D models.

The presentation will be illustrated with the case-study of the Isère watershed (9981 km), a French snow-driven watershed. The main novelties of this method will be underlined, as well as its perspectives and future improvements.