Schematisation of discharge time series in morphodynamic river models

Abstract

River discharge is an essential parameter in morphodynamic modelling. It proves to be highly variable in both time and space. In this thesis the impact of the schematisation of time-dependent discharge series on morphodynamic change is studied. The impact is governed by several factors such as the long-term discharge statistics, and both short- and long-term sequences of discharge stages. A variable river discharge itself does not necessarily result in morphodynamic change. Changes in conveyance area and roughness are dominant sources of bed fluctuations and bed waves. Moreover, backwater dominated segments are subject to mild fluctuations in the river bed. The combination of natural variation of hydrological processes in the upstream river catchment and the absence of a significant correlation among statistical characteristics of subsequent years makes it hard for river engineers to construct discharge time series for river models that simulate morphodynamic development. The limited predictability of future discharge time series is an important source of uncertainty in model predictions. A way to estimate this uncertainty is by means of a time-consuming Monte Carlo approach, resulting in a mean long-term morphodynamic trend and the associated uncertainty. An alternative method is the application of a deterministic series for which the long-term discharge statistics are translated into a cycled annual hydrograph (CAH). It is expected to yield similar morphodynamic changes with respect to the mean trend from the Monte Carlo approach without the need for a large number of computations. However, as is demonstrated in this study, the CAH-method lacks in performance in simulating average long-term development and the amplitude of fluctuations in the river bed, especially at locations where strong sediment transport gradients are experienced. Using a simplified two-dimensional model that represents a locally widened floodplain, the impact of the included bandwidth, short-term and long-term sequences on morphodynamic change is investigated. The aim is to improve the deterministic hydrograph schematisation and to find a more convenient way to get insight into uncertainty in simulated morphodynamics. These findings result in a set of recommendations for future schematisations of discharge time series in morphodynamic river models. An improved deterministic approach is proposed. Using historical measurements of daily discharge data, years with a similar statistical maximum, mean and standard deviation can be classified. The classified years are translated into multiple cycled annual hydrographs (MCAH). By using these hydrographs, synthetic time series are constructed. Compared to the CAH-approach, the proposed MCAH-method yields a significant reduction of the root mean square error with respect to the long-term average morphodynamic trend from the Monte Carlo simulations. Moreover, the MCAH-series result in amplitudes of bed fluctuations that are closer to the response to natural discharge time series. Finally, the MCAH-series can be used to give a rough indication of the uncertainty in future morphodynamics.