Analysis of the advisability of applying the estimated rates of runoff by rating curve methods and fixed submerged Doppler sensor

In practice, there are several techniques used to estimate the rates of runoff in urban and rural areas, each technique having its advantages and limitations. It is up to the administrators of water resources to assess the advisability of applying these technologies, according to their accuracy in various contexts relating to their costs of acquisition and maintenance. The purpose of this article is to analyze the deviation between the flows estimated by means of traditional rating curve and those measured directlybythe ArgonautSW(shallow water) type Doppler. The proposed methodology consistsfirst offiltering and validating the measures of heights from the standard water level gauge and the measures of speed and height from the Doppler SW. This database, so validated, is split into two parts. First, part of the data is used to calibrate a rating curve that connects the series of validated levels obtained from the standard water level gauge to the flows obtained from the SW Doppler. Secondly, the other part of the data is used to estimate the capacity of the rating curve already obtained, and to estimate the flows measured by the SW Doppler by using the levels from the standard water level gauge. It was found that the rating curve calibrated for the study site reproduces the data from theSWDoppler withanerror that varied accordingtothe hydrographs analyzed. The relative error on the estimation of the complete flood hydrograph varies between 2.51[%] and 9.87[%] according to the years, whereas the average relative errors for the peak flows and the flood volumes are 1.44[%] and 2.11[%], respectively. Analysisofthe measures shows that the standard water level gaugeismore reliable than theSWDoppler because the average percentagesofmissing values are 4.64[%] and 20.75[%], for the standard water level gauge and the SW Doppler, respectively. However, the SW Doppler has the advantage of accurately generating a large number of flow measures over a wide range of variations, thus calibrating a reliable rating curve. There is therefore clearly a synergy between the two flow estimation approaches.