Implications of stage-discharge rating curve uncertainty for model evaluation and hydrological characterisation at the national scale

Quantifying and understanding the sources of observational uncertainties is essential for robust and meaningful hydrological research and its applications. Discharge records derived from rating curves form the basis of most hydrological analyses and are one of the key uncertainties in hydrological modelling. Given the demands for comparative hydrology and modelling to be conducted over regional and national scales, there is therefore a need to benchmark the information content of discharge data not only for a single catchment but for multiple catchments.
The aim of this study is to explore the spatial and temporal variability of discharge uncertainty and its implications for model evaluation and hydrological characterisation. To achieve this aim, we analyse rating-curve data and stage-discharge measurements for over 700 gauging stations across England and Wales. First we calculated the percentage deviations of the stage-discharge measurements from the rating curves, and analysed how these deviations varied across the flow range and over time for each station. Secondly, we compared the magnitudes of the deviations for different stations, investigating patterns in the uncertainties related to gauge type, region, flow range, and catchment characteristics. Lastly, we investigated different methods to estimate discharge uncertainty such as fuzzy linear regression of rating curves and envelope curves of deviations for similar stations. We discuss how this methodology can be applied nationally, and we discuss a number of case studies that illustrate different types of rating curves and their related problems for discharge-uncertainty estimation. Finally, we address our ability to identify more/less reliable periods of the flow record and discuss the implications of the estimated discharge uncertainties for hydrological modelling, particularly for assessing floods and low flows.