Reconstruction of River Topography for 3D Hydrodynamic Modelling Using Surveyed Cross-Sections: An Improved Algorithm

Yunhao Song,Erik Toorman,Guolu Yang,Jinfeng Huang

doi:10.3390/w12123539

Abstract

Multidimensional hydrodynamic modelling becomes tricky when lacking the bathymetric data representing the continuous underwater riverbed surface. Light detection and ranging (LiDAR)-based and radar-based digital elevation models (DEMs) are often used to build the high-accuracy floodplain topography, while in most cases the submerged riverbed could not be detected because both radar and LiDAR operate at wavelengths that cannot penetrate the water. Data from other sources is therefore required to establish the riverbed topography. The inundated river channel is often surveyed with an echo sounder to obtain discrete cross-section data. In this context, an improved algorithm based on the classic flow-oriented coordinates transformation is proposed to generate the riverbed topography using surveyed cross-sections. The dimensionless channel width (DCW) processing method is developed within the algorithm to largely increase the prediction accuracy, especially for the meandering reaches. The generated riverbed topography can be merged with the floodplain DEM to create an integrated DEM for 2D and 3D hydrodynamic simulations. Two case studies are carried out: a benchmark test in the Baxter River, United States, with carefully surveyed channel–floodplain topographic data to validate the algorithm, and a 3D hydrodynamic modelling-based application in Three Gorges Reservoir (TGR) area, China. Results from the benchmark case demonstrate very good consistency between the created topography and the surveyed data with root mean square error (RMSE) = 0.17 m and the interpolation accuracy was increased by 55% compared to the traditional method without DCW processing. 3D hydrodynamic modelling results match the observed field data well, indicating that the generated DEM of the TGR area was good enough not only to predict water depths along the tributary, but also to allow the hydrodynamic model to capture the typical features of the complex density currents caused by both the topography of the tributary estuary and the operation rules of TGR.

Highlights

Recent decades have witnessed a significant upgrade of computing power as well as numerical schemes, which largely facilitated the practical application of two- and three-dimensional (2D and 3D)Water 2020, 12, 3539; doi:10.3390/w12123539 www.mdpi.com/journal/waterWater 2020, 12, 3539 hydrodynamic models
Motivated by the necessity to carry out 2D or 3D modelling of river systems and reservoirs where only sparsely surveyed cross-sectional data are available, on the basis of the mathematical methodology described by Legleiter and Kyriakidis [19], in this paper we propose a robust algorithm for river topography reconstruction in which the innovative dimensionless channel width (DCW)
To validate the TELEMAC-3D model based on the generated digital elevation models (DEMs) of the Three Gorges Reservoir (TGR) area, the modelled velocity profiles at the downstream site xx#1, middle reach site xx#2 and upstream site xx#3 were compared against the field acoustic doppler velocimeter (ADV) data

Summary

Introduction

Recent decades have witnessed a significant upgrade of computing power as well as numerical schemes, which largely facilitated the practical application of two- and three-dimensional (2D and 3D)Water 2020, 12, 3539; doi:10.3390/w12123539 www.mdpi.com/journal/waterWater 2020, 12, 3539 hydrodynamic models. Recent decades have witnessed a significant upgrade of computing power as well as numerical schemes, which largely facilitated the practical application of two- and three-dimensional (2D and 3D). To set up a 2D or 3D model, the first step, in general, is generating a computational mesh containing real-world topographic information of both the underwater riverbed and the dry land. The z-value, i.e., the elevation of every mesh node, is normally assigned through the direct interpolation of topographic data surveyed in situ or collected by other means. Numerous studies have indicated that the irregularity and anisotropic characteristic of riverbeds and floodplains could strongly influence the response of 2D and 3D hydrodynamic models [5,6,7]

Methods

Results

Conclusion