Beyond Water Surface Profiles: A New Iterative Methodology for 2D Model Calibration in Rivers Using Velocity Data from Multiple Cross-Sections

Authors: RIVERA-TREJO, FABIAN - Juarez Autonomous University Of Tabasco; SOTO-CORTES, GABRIEL - Metropolitan Autonomous University; KONSOER, KORY - Louisiana State University; Langendoen, Eddy; PRIEGO-HERNANDEZ, GASTON - Juarez Autonomous University Of Tabasco

Submitted to: Water
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/27/2025
Publication Date: 1/30/2025
Citation: Rivera-Trejo, F., Soto-Cortes, G., Konsoer, K.M., Langendoen, E.J. 2025. Beyond Water Surface Profiles: A New Iterative Methodology for 2D Model Calibration in Rivers Using Velocity Data from Multiple Cross-Sections. Water. 17(3):377. https://doi.org/10.3390/w17030377.
DOI: https://doi.org/10.3390/w17030377

Interpretive Summary: With the increase in computing capabilities, two-dimensional (2D), depth-averaged computer model simulations of open-channel flow are becoming more widely used to study a broad range of river engineering problems. At the same time, advances in acoustic measurement technology allow us to collect water velocity and river bathymetric data at high spatial resolutions. However, tools to calibrate and verify 2D open channel flow computer models using these high resolution data sets are lacking. ARS researchers from Oxford, MS, in collaboration with researchers from Louisiana State University and universities in Mexico developed a novel methodology that automates the calibration of the widely-used 2D open channel flow computer model TELEMAC, and classifies the error between simulated and observed flow. The methodology offers a robust 2D calibration process and provides guidance on setting values for the channel boundary roughness and downstream boundary water surface elevation. The developed tool allows practitioners to objectively qualify their 2D simulations of river hydrodynamics and morphodynamics; nowadays, such simulations frequently contribute to the design of river engineering measures.

Technical Abstract: It is common to use observed longitudinal water surface profiles to calibrate and assess the accuracy of computer models of river hydrodynamics. This practice implies two basic assumptions about the water surface elevation (WSL) and the velocity field within a cross-section: 1) the WSL is constant in a lateral direction, and 2) a cross-sectionally uniform velocity distribution. Though this practice is adequate for 1D hydrodynamic model calibration as it agrees with 1D model assumptions, it has limited value for 2D model calibration. When 2D flow measurements are available, model performance is often conducted visually or quantitatively using only select cross-sections. A quantitative evaluation is required to determine how well the simulations capture the reality of the flow. One way to address this is by applying statistical methods when comparing 2D simulations against 2D measurements. We use high-resolution bathymetric and hydrodynamic data, measured with a multibeam echosounder (MBES) and acoustic Doppler current profiler (ADCP), and develop a new methodology to improve the 2D model calibration process and assess model performance. The methodology uses cross-correlation to align the ADCP measurement cross-sections with the MBES bathymetric data and then compares observed and simulated flow velocities at matching spatial locations. For the qualification of the hydrodynamic modeling, we apply relative mean absolute error and root-mean-square error measures. We test the methodology using MBES and ADCP measurements and results from a TELEMAC-2D computer model of an actively evolving neck cutoff on the White River, Arkansas, USA. The proposed methodology is a new way to examine the level of accuracy of 2D hydrodynamic simulations because it offers a robust 2D calibration process and provides guidance on setting values for the channel boundary roughness and downstream boundary water surface elevation.