Thirty-year analysis of a changing river system using Bayesian rating curves

The River Waal is the largest downstream distributary of the Rhine. It has a long history of human intervention and serves important economical, ecological and recre­ ational functions. For the past fifteen years, large-scale flood mitigation interventions have been carried out to lower flood risk at extreme discharges. Using a thirty-year dataset of water levels and discharges derived from ADCP and Ott-mill flow velocity measurements, we ana­ lyze the trends in water levels as a function of discharge. The number of datapoints is limited by the number of measurement campaigns carried out, as well as limited by the range of dis­ charges occurring in a given year. We take this data scarcity, as well as variance in the data-points, into account using multistage Bayesian rating curves. We analyze thirty years of data to detect trends in water levels. Results clearly show a linear trend attributed to autonomous bed degradation. However, expected deviations from the linear trend resulting from human intervention were not found. We discuss various explanations for the resulting trends, as well as future work to reduce the uncertainty of rating curves.