Publikationsdatum:
2023-08-02
Beschreibung:
The benefit of continuous modelling in hydrological studies is widely recognized, indeed it is particularly promising for estimating the design hydrological input for a variety of practical applications and there is a clear tendency to overcome the concept of design hydrograph in favor of the design runoff simulation. Recently, it was underlined the possibility and the opportunity to apply the continuous framework also in the challenging case of ungauged basins. However, while promising, this approach is still not commonly adopted in practice mainly because it needs as input a simulated rainfall time series, that still is not user-friendly task. In this contribution we summarize the recent steps forward of the continuous modelling in ungauged basins and the residual procedural bottlenecks, moreover we explore its use for designing flood early warning systems. We illustrate a framework based on hydrological-hydraulic synthetic scenarios for selecting the best-performing machine learning model for forecasting discharges. Finally, feature importance measures are introduced for discerning the most influential sub-basins where the measurement instrumentations should be installed enabling the implementation of a cost-effective flood early warning system.
Sprache:
Englisch
Materialart:
info:eu-repo/semantics/conferenceObject
