Addressing the Uncertainty Assessment for Real- Time Stage Forecasting

A Flood Monitoring and Warning System (FMWS) operating in real time is the main non-structural measure for reducing risk in flood-prone areas. During the last years significant efforts were addressed to improve the reliability of the forecast quantities mainly by assessing the forecast uncertainty in order to avoid the “illusion of certainty” for decision-makers. In this context, the purpose of this study is to assess the forecast uncertainty for a simple stage forecasting model, called STAFOM-RCM, currently operative within the FMWS of the Upper-Middle Tiber River basin (central Italy). Firstly, the model reliability is tested for three river reaches considering several floods and assuming as performance evaluation measures the Nash–Sutcliffe coefficient, the error on peak stage and time to peak and the coefficient of persistence. Then, the estimate of model uncertainty is addressed by determining the 95% confidence band (CB) on the basis of the errors between observed and forecast stages after their Box-Cox transformation. The results of this preliminary analysis show that the method furnishes the 95% CBs including, on the average, 94.6% and 93.8% of observed stages with the Box-Cox transformation factor calibrated through all and half of the selected floods, respectively.