A vine copula-based approach for constructing nonparametric synthetic design flood hydrographs

<p>Reliable flood risk management needs to correctly estimate and design the size of volumes in reservoirs, spillways of dams and flood levees. To design secure and well-serving hydraulic structures, we often need to use design flood hydrographs that allow a sufficient description of the impacts of flood events in many cases.</p> <p>In this study, a methodology is proposed based on using both empirical and statistical approaches for constructing nonparametric synthetic design flood hydrographs. It is based on flood hydrographs that are observed in the hourly discharge time series, in which is respected the dependence among the peaks, volumes and duration of a set of observed seasonal flood hydrographs. The method consists of seasonality analysis of floods, sampling of seasonal flood hydrographs, normalization of the hydrographs into flood fragments, dependence modelling of peaks, volumes and durations using the vine copulas, rescaling of hydrograph fragments with the appropriate design flood into synthetic design hydrographs and determining the joint conditional return period of the flood volume and the duration conditioned on the flood peak for each synthetic hydrograph.</p> <p>By that, the designer is furnished with a set of design flood hydrographs, which have diverse shapes, volumes, and durations for a selected design discharge with a known joint conditional return period of the volumes and durations for flood risk analysis. &#160;The method was tested and carried out on gauged discharge data from the Horn&#233; Ore&#353;any reservoir in the watershed of the Parn&#225; river in Slovakia. Using flood regionalization approaches can be this method also applicable to ungauged catchments.</p> <p>&#160;</p> <p><em>Acknowledgeme</em><em>nts:</em></p> <p><em>This study was supported by PhD student project SYLUETI.&#160; The study was also supported by the Slovak Research and Development Agency under Contract No. APVV-20-0374.</em></p>