Balancing accuracy and efficiency in centennial-scale 2D morphodynamic river modelling

Abstract

Engineered river systems, such as the Lower Rhine, are currently facing pressures from historical channelisation and projected climate change, which, together, exacerbate issues such as channel bed incision and skewed discharge partitioning at bifurcations (Ylla Arbós, 2021; Blom, 2024). In response to these challenges, Rijkswaterstaat is currently exploring various large-scale intervention strategies under the new Room for the River 2.0 programme. To determine how robust these interventions remain under future climate scenarios characterised by increased hydrograph variability and sea level rise, river management requires simulating large-scale measures, such as Longitudinal Training Walls (LTWs) with or without side channels, and floodplain lowering, over a centennial timescale. While one-dimensional (1D) models have been successfully employed for long-term simulations of the Lower-Rhine (Ylla Arbós, 2023; Chowdhury, 2025), they rely on nodal-point relations, of which the formulation contains challenges to adequately capture critical 2D/3D flow structures, lateral morphological changes, and complex bifurcation feedback mechanisms that influence the discharge partitioning in the system. As an alternative, we consider two-dimensional (2D) depth-averaged morphodynamic modelling. However, applying high-resolution 2D models over large spatial domains (~400 km) for century-long periods presents a computational bottleneck.