ADCP-Based Proxy Sediment Transport Monitoring – Data and Code Archive

OverviewThis release provides a citable, reproducible snapshot of the revised workflow (v3) used for ADCP-based preprocessing and ML/DL experiments to predict the apparent bed velocity (v_a), the bottom-track velocity corrected for platform motion in the field, from ADCP-derived hydraulic, acoustic, and measurement-geometry features. v_a is an ADCP-based proxy for near-bed sediment activity, not a measured bedload transport rate. The archive accompanies the manuscript “An ADCP-Based Data-Driven Framework for Proxy Sediment Transport Monitoring: From Controlled Flumes to Natural Rivers” (Geoscientific Model Development, egusphere-2025-6190). ContentsData/ Processed_LAB_Data_with_BS_v3.xlsx Processed_Field_Data_with_BS_v3.xlsx Processed_Field_Data_Corrected_v3_filtering_stats.xlsx Raw Data/Lab_raw/: 15 flow-rate .mat files Raw Data/Lab_raw/rsqmb/: laboratory RS5 binary files for BS_rel extraction Raw Data/Field_raw/: 46 transect .mat files from 7 campaigns Raw Data/Field_raw/rsqmb/: field RS5 binary files for BS_rel extraction Notebooks/ LAB_Preprocessing_v3.ipynb, FIELD_Preprocessing_v3.ipynb LAB_ML_v3.ipynb, FIELD_ML_v3.ipynb LAB_DL_v3.ipynb, FIELD_DL_v3.ipynb Model_analysis_v3.ipynb Top level: README.md, requirements.txt Reproducibility Recommended environment: Python 3.12, Jupyter (see requirements.txt) Run order: Preprocessing → ML → DL → Model analysis All paths are relative (../Data/) GitHub: https://github.com/thtuhin488/ADCP_BT_Vel_ML_DL Relation to the previous versionThis version supersedes the earlier versions of this record (e.g. https://doi.org/10.5281/zenodo.18716833). The field target of the previous version was computed directly from the bottom-track velocity, which in the RS5 export is the negative of the platform velocity over ground, so it represented the survey-platform speed. In this version the platform motion is subtracted using the GNSS track. Laboratory results are unchanged.