Toward ultra-efficient high-fidelity prediction of bed morphodynamics of large-scale meandering rivers using a novel LES-trained machine learning approach

Aug 6, 2026

Flood-induced deformation of the bed topography in fluvial meandering rivers can drive riverbank displacement, threaten infrastructure, and modify scour and deposition patterns. Sediment transport assessment in large-scale meanders is thus a critical environmental and engineering concern. High-fidelity coupled flow–morphodynamics simulations are accurate but computationally expensive due to two-way coupling between turbulence and bed morphodynamics. This study introduces an LES-trained machine learning method to predict bed shear stress and equilibrium morphology at a fraction of the cost, using a CNNAE model trained and evaluated on coupled LES-morphodynamics results.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages34

Reading Time1 h 26 min

AccessFree to download

Introduction

Background and motivation

Related AI and numerical modeling work

Study objective

Toward ultra-efficient high-fidelity prediction of bed morphodynamics of largescale meandering rivers using a novel LES-trained machine learning approach Zexia Zhang 1, Mehrshad Gholami Anjiraki 1, Hossein Seyedzadeh1, Fotis Sotiropoulos2, and Ali Khosronejad 1* 1Civil Engineering Department, Stony Brook University, Stony Brook, NY 11794 , USA 2Mechanical and Nuclear Engineering, Virginia Commonwealth University, Richmond, VA 23284, USA *Corresponding author, Email: [ali.khosronejad@stonybrook.edu](ali.khosronejad@stonybrook.edu) Abstract Flood-induced deformation of the bed topography of fluvial meandering rivers could lead to riverbank displacement, structural failure of the infrastructures, and the propagation of scour or deposition features. The assessment of sediment transport in large-scale meanders is, therefore, a key environmental issue. High-fidelity numerical models provide powerful tools for such assessments. However, high-fidelity simulations of large-scale rivers using the coupled flow and morphodynamics modules can be computationally expensive, owing to the costly two-way coupling between turbulence and bed morphodynamics. This study seeks to present a novel machine learning approach, which is trained using coupled large-eddy simulation (LES) and morphodynamics results. The proposed machine learning approach predicts bed shear stress and equilibrium bed morphology of large-scale meanders under bankfull flow conditions at a fraction of the cost of coupled LES-morphodynamics. We developed and evaluated the performance of a convolutional neural network autoencoder (CNNAE) algorithm to generate high-fidelity bed shear stress and equilibrium morphology of large-scale meandering rivers. The CNNAE algorithm utilizes instantaneous shear stress distribution and change of bed elevation of high-fidelity simulation results, along with geometric parameters of meanders as inputs to predict mean bed shear stress distribution and equilibrium bed elevation of rivers. The results demonstrated the feasibility, accuracy, and efficiency of the proposed CNNAE algorithm. 1. Introduction Predicting the interaction between turbulent river flow and bed topography in the field-scale meandering rivers is essential for various river engineering and geoscience problems. For example, such predictions allow researchers and practicing engineers to gain insight into the evolution of the riverbeds and, thus, to design effective flood management and mitigation strategies. Also, sediment transport affects the deformation and migration of river channels, which is important for navigation and maintaining infrastructure such as bridges and dams. A commonly utilized engineering tool for such predictions at large-scale rivers is the high-fidelity numerical models that are based on computational fluid dynamics (CFD) simulations (Behzad et al. 2023; Bigdeli et al. 2023; Bourgoin et al. 2021; Flora et al. 2021; Flora and Khosronejad 2021, 2022; Khosronejad et al. 2019b, 2020b; c, a; e; Khosronejad and Sotiropoulos 2017, 2020; Nian et al. 2021b; a; Yazdanfar et al. 2021) . Compared to the high-fidelity models, the emerging artificial intelligence (AI) based machine learning algorithms seem to provide a more efficient approach for the prediction of bed deformation in large-scale meandering rivers. This study attempts to develop, train, and validate the machine learning algorithms that can generate high-fidelity bed topography of large-scale meandering rivers under bankfull conditions and at a fraction of the cost associated with the high-fidelity CFD models. Jain (2001) was among the first to employ artificial intelligence (AI) methods to predict river sediment dynamics. He adopted a feedforward neural network to connect the dynamics of river stage, discharge, and sediment concentration in the Mississippi River. Building on this work, Rai and Mathur (2008) developed artificial neural networks (ANN) models to predict event-based and time-dep

FAQs

What is the main problem addressed by this study?

Flood-induced deformation of bed topography in large-scale meandering rivers affects bank stability, infrastructure safety, and scour/deposition evolution, making accurate sediment transport and morphology prediction essential.

How does the proposed approach reduce computational cost?

It uses a machine learning model trained on coupled large-eddy simulation (LES) and morphodynamics outputs to predict bed shear stress and equilibrium morphology much faster than running expensive coupled simulations.

What inputs does the CNNAE algorithm use to predict bed morphodynamics?

The CNNAE uses instantaneous shear stress distribution and changes in bed elevation from high-fidelity simulations, along with geometric parameters of the meanders, to predict mean bed shear stress and equilibrium bed elevation.

Aug 6, 2026

Upload by Ava Thompson