The RUHM framework for rapid rating curve uncertainty estimation: comparison to power-law methods and potential using drone-derived data

Climate change, together with other natural and anthropogenic drivers lead to changes in streamflow patterns that are now occurring with increasing frequency. At the same time traditional streamflow monitoring methods are time-consuming and costly so that it typically takes many years of significant field efforts to establish reliable streamflow data for a new location or for stations with major temporal changes to the stage—discharge relation. To provide timely and reliable streamflow data to tackle these changes to the hydrological regime and their impacts on society’s water management requires new cost-effective monitoring methods that can rapidly produce data with low uncertainty. Hydraulically modelled rating curves are a promising alternative to traditional power-law methods as they need much fewer calibration gaugings, but they are associated with additional uncertainty sources in the hydraulic knowledge and these need to be assessed.
We present the Rating curve Uncertainty estimation using Hydraulic Modelling (RUHM) framework which was developed to rapidly estimate rating curves and their uncertainty. The RUHM framework combines a one-dimensional hydraulic model with Bayesian inference to incorporate information from both hydraulic knowledge and the calibration gauging data. In this study we compare RUHM and the Bayesian power-law method BaRatin in application to a Swedish site using nine different gauging strategies associated with different costs. We compare results for the two methods in terms of accuracy, cost and time required for establishing rating curves. 
We found that rating curves with low uncertainty could be modelled with fewer gaugings for RUHM compared to BaRatin. As few as three gaugings were needed with RUHM if these gaugings covered low and medium flows, whereas high flow gaugings were not necessary. This makes the RUHM method both cost effective and time efficient as low and medium flows occur more frequently than high flows. When using all gaugings (i.e., a high-cost gauging strategy), the uncertainty for RUHM and BaRatin was similar. The results for this Swedish site show that hydraulic rating curve uncertainty estimation is a promising tool for quickly estimating rating curves and their uncertainties. Finally, we discuss the potential of using RUHM together with drone-derived data to make field efforts even more efficient.