Quantifier l'effet de choix du site dans l'incertitude des jaugeages ADCP par transects

HMEM, Durham, New Hampshire, USA, 09-/07/2017 - 12/07/2017; International audience; Stage-discharge rating curves are used to relate streamflow discharge to continuously measured river stage readings to create a continuous record of streamflow discharge. The stage-discharge relationship is estimated and refined using discrete streamflow measurements over time, during which both the discharge and stage are measured. There is uncertainty in the resulting rating curve due to multiple factors including the curve-fitting process, assumptions on the form of the model used, fluvial geomorphology of natural channels, and the approaches used to extrapolate the rating equation beyond available observations. This rating curve uncertainty leads to uncertainty in the streamflow timeseries, and therefore to uncertainty in predictive models that use the streamflow data. Many different methods have been proposed in the literature for estimating rating curve uncertainty, differing in mathematical rigor, in the assumptions made about the component errors, and in the information required to implement the method at any given site. This study describes the results of an international experiment to test and compare streamflow uncertainty estimation methods from 7 research groups across 9 institutions. The methods range from simple LOWESS fits to more complicated Bayesian methods that consider hydraulic principles directly. We evaluate these different methods when applied to three diverse gauging stations using standardized information (channel characteristics, hydrographs, and streamflow measurements). Our results quantify the resultant spread of the stage-discharge Quantifying the uncertainty of discharge measurements (or "gaugings") is a challenge in the hydrometric community. A useful tool to empirically estimate the uncertainty of a gauging method is the field inter-laboratory experiment (Le Coz et al., 2016). Previous inter-laboratory experiments conducted in France (in 2009, 2010, 2011 and 2012) showed that the expanded uncertainty (with a probability level of 95%) of an ADCP gauging made of six successive transects is typically around 5% under optimum site conditions (straight reach, uniform and smooth streambed cross-section, homogeneous flow, etc.) and may be twice higher under poorer site conditions. In practice, the selected cross-section does not always match all quality requirements which may result in larger uncertainty. However, the uncertainty due to site selection is very difficult to estimate with predictive equations. From 9 to 10 November 2016, 50 teams from 8 different countries, using 50 ADCPs simultaneously, conducted more than 600 discharge measurements in steady flow conditions (~14 m3/s released by a dam). 26 cross-sections with various shapes and flow conditions were distributed over 500 meters along the Taurion River at Saint-Priest-de-Taurion, France. A specific experiment protocol, which consisted of circulating every team over half of the cross-sections, was implemented in order to quantify the impact of site selection on the discharge measurement uncertainty. Beyond the description of the experiments, uncertainty estimates are presented. The overall expanded uncertainty of a 6-transect ADCP gaugings (duration around 720 seconds) is estimated to be around 6%.The uncertainty of the discharge measurements varies among the cross-sections. These variations are well correlated to the expert judgment on the cross-section quality made by each team. First results seem to highlight a relation between uncertainty computed for each cross-section and criteria such as flow shallowness and measured discharge ratio. Further investigations are necessary to identify the criteria related to error sources that are possibly meaningful for categorizing measurement conditions and site selection. Moreover, experimental uncertainty and the uncertainty predicted by analytical methods such as QRev, QUant, OURSIN, RiverFlowUA or QMSys software will be compared.