Quantifying the discharge forecast uncertainty by different approaches to probabilistic quantitative precipitation forecast

摘要

A probabilistic approach to flood prediction over the Reno river basin, amedium-sized catchment in Northern Italy, has been tested using twodifferent meteorological ensemble systems. The future precipitationscenarios are provided either by an analogue-based technique (statisticalapproach) or by a limited-area ensemble prediction system (dynamicalapproach), then used as different inputs to a distributed rainfall-runoffmodel. The ensemble of possible future flows so generated allows to convey aquantification of uncertainty about the discharge forecast. Theprobabilistic discharge forecasts, based on the precipitation forecastprovided by the two ensembles, are then compared to the deterministic oneobtained by the rainfall-runoff model fed on precipitation input provided bya non-hydrostatic meteorological model, run at 7km of horizontalresolution. For this case study, the dynamical approach appears to be morefeasible in providing useful discharge ensemble forecast than thestatistical one, because the observed large spread among members obtainedwith the analogue method makes difficult to issue real-time flood warnings.