The Effect of Spatial and Cross-Profile Data on Morphodynamic Modelling

Morphodynamic modelling relies on different types of riverbed surveys. Surveys are essential as the basis of the evaluation of temporal river bed development, mesh creation, and model calibration. Spatial data, for example, obtained by topo-bathymetric airborne laser scanning (ALB) or sonar surveys results in a dense point cloud, providing detailed information on the river bathymetry. However, data gaps can occur due to restrictions in data acquisition (e.g. high water turbidity or water depth for ALB, low water for boat-mounted sonar). In contrast, cross-profiles contain only limited information on the bathymetry strongly dependent on the cross-profile and point spacing.

To assess the effect of the two survey data types on river bed development and morphodynamic predictions, the temporal evolution of a river stretch in the upper Danube at Donauwörth was analysed. The study area contains homogeneous river sections and sections with complex river geometry due to scours, bridge foundations, and river mouths.  Spatial sonar and ALB surveys were conducted from 2013 to 2020 and give detailed documentation of the river bed development. Cross-profiles with a cross-profile spacing of 200 m were derived from the spatial data. The spatial and cross-profile datasets show continuous river bed erosion. However, in this case, cross-profile data overestimate the overall erosion compared to spatial data. The geometry of homogeneous river stretches is depicted very similarly in the two datasets. For cross-profile data two cases exist for reaches with more complex river bed geometry: (i) The geometry lies in between two cross-profiles and it is missed entirely. (ii) The geometry is covered by a cross-profile and the resulting geometry is smeared in between the cross-profiles due to the interpolation process. Both possibilities result in an unsatisfactory depiction of the riverbed geometry.

To analyse the effect of morphological developments two morphodynamic models based either on the spatial or cross-profile datasets were set up. The models were calibrated against the datasets from 2013 to 2020 by adjusting the Strickler value for river sections with a length of 200 m. The Strickler values differ over the entire river stretch and not only in sections where complex river bed geometry occurs, meaning that the calibration errors propagate through the entire study area. Consequently, the deviations in calibration outcomes affect the model predictions, which simulate 7 years. In this case, the general shape of the predicted riverbed is similar, but due to the overestimation of riverbed erosion by cross-profile data, the morphodynamic model overestimates the erosion compared to the spatial data. However, the obtained error is for river reaches with low local variability within an acceptable range. If a project demands a highly accurate depiction of the river bed and the river geometry is known for having complex features, the use of spatial data is strongly advised.