This repository contains the output data files from the LES-DPM numerical model utilized for the analyses presented in DeVoe et al. (2025). Please note that the model is a coupled two-phase model and that this repository contains both the fluid and particle data, as well as the necessary files to convert data from their native binary format to ASCII format.

The LES-DPM was developed by Hagan (2018) and validated by Hagan et al. (2023). Note: This dataset was generated using the model's governing equations and framework as presented by Hagan (2018) and Hagan et al. (2023), but is not the same dataset used for the analyses in those two manuscripts. This dataset has a higher sampling frequency (100 Hz) and a shorter duration than the datasets used in Hagan (2018) and Hagan et al. (2023). Please refer to the README and DeVoe et al. (2025) for additional information about the model parameters.

Corresponding Author for Dataset & LES-DPM Inquiries: Yves Dubief, University of Vermont.

Corresponding Author Email: ydubief [at] uvm [dot] edu.

*DeVoe, S. R., Wengrove, M. E., Foster, D. L., & Hagan, D. S. (2025). Characterization of the spatiotemporal distribution of shear stress and bedload flux within a mobile, rippled bed. Journal of Geophysical Research: Oceans. DOI: doi.org/10.29/2025JC022369

Written by Savannah DeVoe, Aug. 2024

Edited by Savannah DeVoe, July 2025:

• Grouped individual data files into .zip folders in groups of 100 (e.g., fluid000001-fluid000100.zip) to meet new Dryad limit of 100 files per repository
• Merged particle and fluid repositories together (originally, they were separate due to file size and number limitations)
• Incorporated article title and reference edits following paper acceptance/JGR reviewer feedback
• Additional edits to .f90, .txt, and README files for further clarity.

File Descriptions

• README: This file. Contains general information about the dataset properties, including, but not limited to, a description of the model's boundary conditions and initial conditions, the particle and fluid variables output by the model, the sampling rate, the length of the time series, the oscillation period, wave zero-crossing time steps, etc.
• fluid000XXX-fluid000XXX.zip: Zip files containing individual binary fluid data files for time step XXX (e.g., fluid000001-fluid000100.zip contains fluid files fluid.000001, fluid.000002, ... etc., where fluid.000001 is the fluid data for time step 1). For this dataset, XXX = 1-903. After zip files have been extracted, fluid.000XXX binary files can be opened in any text editor application but they are not human-readable until after they have been converted. The following zip files are included:
  •   fluid000001-fluid000100.zip
  •   fluid000101-fluid000200.zip
  •   fluid000201-fluid000300.zip
  •   fluid000301-fluid000400.zip
  •   fluid000401-fluid000500.zip
  •   fluid000501-fluid000600.zip
  •   fluid000601-fluid000700.zip
  •   fluid000701-fluid000800.zip
  •   fluid000801-fluid000900.zip
  •   fluid000901-fluid000903.zip
• particle000XXX-particle000XXX.zip: Same as fluid000XXX-fluid000XXX.zip files, but for the binary particle data files. The following zip files are included:
  •   particle000001-particle000100.zip
  •   particle000101-particle000200.zip
  •   particle000201-particle000300.zip
  •   particle000301-particle000400.zip
  •   particle000401-particle000500.zip
  •   particle000501-particle000600.zip
  •   particle000601-particle000700.zip
  •   particle000701-particle000800.zip
  •   particle000801-particle000900.zip
  •   particle000901-particle000903.zip
• extract_fluid.f90: Fortran script for converting binary fluid data files into ASCII format .DATA files for further processing. Can be opened in any text editor application. Contains instructions for installing specific open-source packages required for binary file conversion, and instructions for either single file conversion (in Terminal/WSL prompt) or batch conversion using the binary_textfileconv_fluid_stats routine (see below).
• extract_particles.f90: Same as extract_fluid.f90, but for the binary particle data files. Batch conversion performed using the binary_textfileconv_particle_stats routine (see below).
• binary_textfileconv_fluid_stats.txt: Simple example script for batch processing to extract multiple binary fluid files at once with the extract_fluid.f90 routine. Written in MATLAB language but can be adapted to other programming languages. To open in MATLAB, simply change the file extension from .txt to .m, or copy/paste the contents of the file into a new MATLAB script.
• binary_textfileconv_particle_stats.txt: Same as binary_textfileconv_fluid_stats.txt, but for the binary particle data files.
• matfileconv_fluid_stats.txt: Example script for importing converted fluid000XXX.DATA files into MATLAB, parsing columns into particle variables, and saving the variables in a fluid000XXX.mat file format. Written in MATLAB language but can be adapted to other programming languages.
• matfileconv_particle_stats.txt: Same as matfileconv_fluid_stats.txt, but for the converted particle000XXX.DATA files.

Users should begin by reading this file (README), then follow the instructions provided in extract_fluid.f90 and extract_particles.f90 to install required packages for conversion. Batch processing can then be done following the example in binary_textfileconv_fluid_stats.txt and binary_textfileconv_particle_stats.txt, and finally, data can be prepped for post-processing following the MATLAB example in matfileconv_fluid_stats.txt and matfileconv_particle_stats.txt.

*Please note that binary conversion and preparation for post-processing may take several minutes if you choose to process all 903 fluid or all 903 particle files at once. You can check the progress of conversion in your output folder as new .DATA or .mat files are created.*

LES-DPM Model Description

The LES-DPM data is derived from a coupled Large Eddy Simulation (LES) and Discrete Particle Model (DPM). The model's governing equations were developed by Hagan (2018) and validated against laboratory data by Hagan et al. (2023). The model simulates oscillatory flow over a movable, rippled bed comprised of individual sediment grains. The LES solves large scales of the fluid flow and models the small scales, while the DPM resolves the position and velocity of individual particles. See Hagan et al. (2023) for more information about the governing equations and numerical methodology.

Model Domain

The domain measures 0.60 m x 0.15 m x 0.30 m (LxWxH). The top and bottom boundaries have a no-slip condition, while the streamwise and spanwise boundaries are periodic. The bottom wall is covered by a rippled bed comprised of individual sediment grains. There are 6 sinusoidal ripples in the domain with a wavelength of 0.10 m and a height of 0.02 m. A total of 1,540,800 individual particles exist in the model. The fluid in the domain is water. The coordinate system for the DATA defines x in the streamwise, z in the spanwise, and y in the vertical directions, respectively. (Note this differs from the coordinate system adopted in DeVoe et al., 2025.) The fluid grid system is comprised of 256 x 128 x 64 nodes (streamwise, vertical, spanwise, respectively) with a uniform dx = dz = 0.0023 m and a varying dy (smallest near the top and bottom boundaries and larger towards the center of the domain). Particles are simulated on a Lagrangian reference frame and can exist at any (x,y,z) position within the domain. See Hagan et al. (2023) or DeVoe et al. (2025) for more information about the model domain.

Model Initial Conditions and Forcing

Particles in the model are first allowed to settle under gravity in a face-centered cubic configuration (solid volume fraction = 0.74) in the still fluid. After settling, the domain is subjected to fluid oscillations driven by an oscillating pressure gradient with period, T = 2.4 s. The maximum fluid velocity in the free stream reaches 0.23 m/s. Quasi-equilibrium is reached after approximately 3 wave periods (Hagan et al., 2023). The bed is then subjected to 11 additional oscillations. The LES-DPM datasets used by Hagan (2018) and Hagan et al. (2023) contained these 11 oscillation periods with a model output of approximately 10 Hz. The dataset used in DeVoe et al. (2025) contains only 3 full oscillations in the set of 11 oscillations, and is output at a higher frequency (100 Hz). See Hagan et al. (2023) or DeVoe et al. (2025) for more information about the initial conditions and forcing.

Time Series

The dataset contains 903 fluid files and 903 particle files, with each file pertaining to a single instant in time (i.e., fluid.000001 and particle.000001 are the fluid data and particle data for time step 1, respectively). The data is output at 100 Hz, meaning each oscillation is comprised of 240 samples. The wave zero crossings (where the free-stream fluid velocity is approximately zero) occur at time instances 1, 241, 481, and 721. The remaining files (722-903) are for a partial fourth oscillation period.
NOTE: Time step 586 contains anomalous fluid velocity data. As such, it should be replaced by interpolation of the fluid variables between time steps 585 and 587.

Binary Data Files

Data is divided into binary fluid and particle files for each time instant. E.g., particle data for time step 1 is contained in the binary file particle.000001, and fluid data for time step 1 is contained in fluid.000001. Binary files are compressed into zip files that must be downloaded and extracted (because Dryad limits repositories to 100 files or less). After extraction, the individual binary files must be converted from to ASCII format using the extract_fluid.f90 and extract_particles.f90 scripts via Terminal/WSL prompt. Refer to those documents for instructions on how to convert from binary to ASCII (.DATA file). Example code is provided to execute the binary file conversion in a loop (for batch processing multiple binary files at once) and to then import the converted ASCII file into MATLAB for further processing. These example codes are provided as .txt files so they can be opened, viewed, and subsequently converted to other programming languages if the user does not have access to MATLAB or wishes to use another program.

Once the binary files are converted, the data in the resulting .DATA file is divided into the following columns with units shown in square brackets:

Fluid Variables

• x - fluid grid x position (streamwise) [m]
• y - fluid grid y position (VERTICAL) [m]
• z - fluid grid z position (SPANWISE) [m]
• p - pressure [N/m2]
• visc - dynamic viscosity [kg/m-s]
• epsf - fluid volume fraction [unitless]
• rho - density [kg/m3]
• pcrit - P criterion, first invariant of the velocity gradient tensor (sum of rank one determinants on diagonal)
• qcrit - Q criterion, second invariant of the velocity gradient tensor (sum of rank two determinants on diagonal)
• rcrit - R criterion, third invariant of the velocity gradient tensor (sum of rank three determinants on diagonal)
• u - x-component of fluid velocity (streamwise) [m/s]
• v - y-component of fluid velocity (VERTICAL) [m/s]
• w - z-component of fluid velocity (SPANWISE) [m/s]

Particle Variables

• n - row number [unitless] (not physically relevant to the model, reflects binary conversion order only)
• id - unique particle identification number [unitless]
• diam - particle diameter [m]
• x - particle x position (streamwise) [m]
• y - particle y position (VERTICAL) [m]
• z - particle z position (SPANWISE) [m]
• u - x-component of particle velocity (streamwise) [m/s]
• v - y-component of particle velocity (VERTICAL) [m/s]
• w - z-component of particle velocity (SPANWISE) [m/s]
• wx - x-component of particle angular velocity (streamwise) [m/s]
• wy - y-component of particle angular velocity (VERTICAL) [m/s]
• wz - z-component of particle angular velocity (SPANWISE) [m/s]

REFERENCES

DeVoe, S. R., Wengrove, M. E., Foster, D. L., & Hagan, D. S. (2025). Characterization of the spatiotemporal distribution of shear stress and bedload flux within a mobile, rippled bed. Journal of Geophysical Research: Oceans. DOI: https://doi.org/10.29/2025JC022369.

Hagan, D., Wengrove, M., Dubief, Y., Desjardins, O., Frank-Gilchrist, D., & Calantoni, J. (2023). Particle based Large Eddy Simulation of vortex ripple dynamics using an Euler–Lagrange approach. European Journal of Mechanics-B/Fluids, 97, 53-69. https://doi.org/10.1016/j.euromechflu.2022.09.003.

Hagan, D. S. (2018). Large eddy simulation of oscillatory flow over a mobile rippled bed using an euler-lagrange approach (Order No. 10747462). Available from ProQuest Dissertations & Theses Global. (2025980859). Retrieved from https://unh.idm.oclc.org/login?url=https://www.proquest.com/dissertations-theses/large-eddy-simulation-oscillatory-flow-over/docview/2025980859/se-2