A hydraulic study on the applicability of flood rating curves Giuliano Di Baldassarre and Pierluigi Claps ABSTRACT Several hydrological studies have shown that river discharge records are affected by significant uncertainty. This uncertainty is expected to be very high for river flow data referred to flood events, when the stage–discharge rating curve is extrapolated far beyond the measurement range. This study examines the standard methodologies for the construction and extrapolation of rating curves to extreme flow depths and shows the need of proper approaches to reduce the uncertainty of flood discharge data. To this end, a comprehensive analysis is performed on a 16 km reach of the River Po (Italy) where five hydraulic models (HEC-RAS) were built. The results of five topographical surveys conducted during the last 50 years are used as geometric input. The application demonstrates that hydraulically built stage–discharge curves for the five cases differ only for ordinary flows, so that a common rating curve for flood discharges can be derived. This result confirms the validity of statistical approaches to the estimation of the so-called ‘flood rating curve’, a unique stage–discharge curve based on data of contemporaneous annual maxima of stage and discharge values, which appears insensitive to marginal changes in river geometry. Key words 9 cross-sections, flood discharge, hydraulic models, observation uncertainty, rating curve INTRODUCTION Hydrological models often disregard the fact that river flow data are affected by a significant uncertainty (e.g. Clarke 1999). This is despite the fact that it is well known that river discharges are almost never directly measured, as opposite to the water stage. Usually, observed river stage values are converted into river discharges by means of a stage–discharge relationship, the so-called rating curve (World Meteorologi- cal Organisation 1994). The main sources of uncertainty that affect river dis- charge data, obtained using the rating curves, are: (1) errors in the individual stage and discharge measurements; (2) errors induced by the presence of unsteady flow condi- tions; and (3) errors induced by the extrapolation of the rating curve beyond the range of measurements used for its deriva- tion. Depending on the specific case study, additional sources of uncertainty can be significant. These include the presence of relevant backwater effects (caused by downstream con- fluent tributaries, lakes and regulated reservoirs) and tem- poral changes in the hydraulic properties governing the stage–discharge relationship (e.g. scour and fill, vegetation growth and ice build-up during cold periods). Concerning the measurement uncertainty (case 1), Pelletier (1987) reviewed 140 publications and concluded that the overall uncertainty in a single determination of river discharge can be more than 8% at the 95% condence level. More recent studies reported errors around 5–6% (e.g. Le ´onard et al. 2000) that could possibly be reduced by using appropriate discharge measurement techniques (Lintrup 1989; European ISO EN Rule 748 1997). The errors induced by the presence of unsteady flow (case 2) can be relevant in very mild river slope conditions, where the variable energy slope leads to the formation of a Giuliano Di Baldassarre (corresponding author) Department of Hydroinformatics and Knowledge Management, UNESCO-IHE Institute for Water Education, Westvest 7, PO Box 3015, 2601 DA Delft, The Netherlands E-mail: g.dibaldassarre@unesco-ihe.org Pierluigi Claps Dipartimento di Idraulica, Trasporti ed Infrastrutture Civili, Politecnico di Torino, Torino, Italy doi: 10.2166/nh.2010.098 10 & IWA Publishing 2011 Hydrology Research 9 42.1 9 2011