Abstract The velocity dip in river systems occurs due to the three-dimensional nature of flow and the presence of secondary currents. These secondary currents are influenced by the complex river bathymetry and the roughness, causing the maximum velocity to occur below the water surface due to the cross-momentum transfers. The evaluation of dip positions, while estimating the vertical velocity profiles, is essential for accurately estimating the flow rate across the river cross-section. This work presents a dip-informed coupled quasi-three-dimensional hydrodynamic model for predicting velocities in an ungauged natural river section with limited datasets. The model is capable of providing the longitudinal and lateral velocity (ux, uy) along with the dip-integrated vertical velocity profile. Information entropy is utilized to estimate the initial dip positions across the river cross-sections with the Principle of Maximum Entropy (POME), considering the dimensionless dip position as the random variable. An iterative framework is developed in conjunction with the predicted hydrodynamic model outputs to estimate the dip positions in the velocity profile across four river sections in the Brahmaputra River, Assam, India. The stability of the model is ensured by the Courant–Friedrichs–Lewy (CFL) criterion. With the observed survey datasets, the model results are validated. The estimated results indicate satisfactory robustness and stability of the model, with a mean percentage error ranging from 2.24 to 4.99% in estimating the depth-averaged velocities. The model's efficacy is also investigated by comparing the model-predicted velocity dip with the observed data, and improved accuracy is observed with respect to the existing approach present in the literature. Research highlights <UnorderedList Mark="Bullet"> <ItemContent> Developed a quasi-3D model integrating 2D hydrodynamic and entropy-based model for predicting dip-integrated vertical velocity profiles in data-scarce regions. </ItemContent> <ItemContent> The model could predict the velocities in a natural river in x and y direction along with the vertical velocity profile across flow depth. </ItemContent> <ItemContent> The location of occurrence of maximum velocities along the flow depth throughout the cross section is evaluated. </ItemContent> <ItemContent> The velocity dip triggered by the difference in bed level is estimated by an iterative technique. </ItemContent> </UnorderedList>

Estimating flow velocities in data-scarce rivers using a dip-informed quasi-3D hydrodynamic model

期刊论文
发表日期：

2025年5月10日

查看全文

Anurag Handique,
Anupal Baruah,
Arup Kumar Sarma,
Rajib Kumar Bhattacharjya

摘要

Abstract

The velocity dip in river systems occurs due to the three-dimensional nature of flow and the presence of secondary currents. These secondary currents are influenced by the complex river bathymetry and the roughness, causing the maximum velocity to occur below the water surface due to the cross-momentum transfers. The evaluation of dip positions, while estimating the vertical velocity profiles, is essential for accurately estimating the flow rate across the river cross-section. This work presents a dip-informed coupled quasi-three-dimensional hydrodynamic model for predicting velocities in an ungauged natural river section with limited datasets. The model is capable of providing the longitudinal and lateral velocity (u_x, u_y) along with the dip-integrated vertical velocity profile. Information entropy is utilized to estimate the initial dip positions across the river cross-sections with the Principle of Maximum Entropy (POME), considering the dimensionless dip position as the random variable. An iterative framework is developed in conjunction with the predicted hydrodynamic model outputs to estimate the dip positions in the velocity profile across four river sections in the Brahmaputra River, Assam, India. The stability of the model is ensured by the Courant–Friedrichs–Lewy (CFL) criterion. With the observed survey datasets, the model results are validated. The estimated results indicate satisfactory robustness and stability of the model, with a mean percentage error ranging from 2.24 to 4.99% in estimating the depth-averaged velocities. The model's efficacy is also investigated by comparing the model-predicted velocity dip with the observed data, and improved accuracy is observed with respect to the existing approach present in the literature.

Research highlights

Developed a quasi-3D model integrating 2D hydrodynamic and entropy-based model for predicting dip-integrated vertical velocity profiles in data-scarce regions.

The model could predict the velocities in a natural river in x and y direction along with the vertical velocity profile across flow depth.

The location of occurrence of maximum velocities along the flow depth throughout the cross section is evaluated.

The velocity dip triggered by the difference in bed level is estimated by an iterative technique.