SCI-DEM: A Method for Generating Digital Beach Surfaces from Tide and Waterline Timeseries
Abstract
We present SCI-DEM, a method for deriving Digital Elevation Models (DEMs) of intertidal beach surfaces, profiles and beach widths from time-series waterlines using low-cost coastal CCTV cameras and modelled water levels. SCI-DEM builds on two established components: a waterline extraction method built on a semantic segmentation model, and a camera-pose-estimated homography, used to georeference each waterline. The SCI-DEM-specific contribution is the spatiotemporal aggregation of the resulting waterline time series: over a two-week moving window, waterlines are cleaned, intersected with fixed cross-shore transects, tagged with modelled water elevation (astronomical tide plus sea-level anomaly), and fit with shape-constrained monotonic regression under automated quality control; profiles passing quality control are combined into intertidal DEMs, mean-high-water (MHW) shoreline contours, and per-transect beach-width estimates. We validate SCI-DEM against RTK-GNSS beach-profile surveys at two camera deployments in Southern California, Torrey Pines and Oceanside Harbor South Jetty, and demonstrate its beach-width product at a third, Ehukai Beach Park (Hawaii), a microtidal beach with no survey programme. At all three sites an independently produced CoastSat satellite-shoreline record is shown alongside the camera series. Camera-derived and surveyed profiles agree to a mean absolute vertical residual of 0.09 m (1 October 2024), 0.15 m (20 May 2025) and 0.10 m (6 October 2025) at Torrey Pines and 0.11 m (1 October 2024) and 0.18 m (6 October 2025) at Oceanside Harbor South Jetty. The resulting beach-width time series reproduce the in-situ MHW crossings to within 3.0 m at Torrey Pines and 2.2 m at Oceanside Harbor South Jetty, and follow the same seasonal trends as the tidally corrected CoastSat record. SCI-DEM is currently running on over 60 live cameras. The method offers a cost-effective route to continuous, high-temporal-resolution beach-profile and beach-width monitoring that complements (and densifies) episodic physical surveys.
Related
- From noise to signal: a data-driven framework for identifying hot moments in water quality timeseries
- Biodiversity credits hold promise, yet credit-generating protocols require safeguards to realize potential
- PrecipCorrDiff: A Deep Learning Framework for Generating 0.0625∘ High-Resolution IMD Gridded Precipitation Data
- Neutrosophic reliability assessment of smart grid power generation system using universal generating function under Weibull distribution
- Generating blue carbon credits through conservation of Posidonia oceanica: experience from the LIFE SeaForest project
BibTeX
@misc{howey2026scidem,
title = {{SCI-DEM: A Method for Generating Digital Beach Surfaces from Tide and Waterline Timeseries}},
author = {Ben Howey and Natalie Tripp Foulds and Michael Reinwald and Ketron Mitchell-Wynne and Zane D'Amico and Mercedes Stringer and Benjamin Freeston},
howpublished = {EarthArXiv},
year = {2026},
month = oct,
doi = {10.31223/x59v4b},
url = {https://eartharxiv.org/repository/view/15773/},
note = {CC BY-NC 4.0},
}
RIS
TY - UNPB
TI - SCI-DEM: A Method for Generating Digital Beach Surfaces from Tide and Waterline Timeseries
AU - Ben Howey
AU - Natalie Tripp Foulds
AU - Michael Reinwald
AU - Ketron Mitchell-Wynne
AU - Zane D'Amico
AU - Mercedes Stringer
AU - Benjamin Freeston
PY - 2026
DA - 2026/10/27
PB - EarthArXiv
DO - 10.31223/x59v4b
UR - https://eartharxiv.org/repository/view/15773/
AB - We present SCI-DEM, a method for deriving Digital Elevation Models (DEMs) of intertidal beach surfaces, profiles and beach widths from time-series waterlines using low-cost coastal CCTV cameras and modelled water levels. SCI-DEM builds on two established components: a waterline extraction method built on a semantic segmentation model, and a camera-pose-estimated homography, used to georeference each waterline. The SCI-DEM-specific contribution is the spatiotemporal aggregation of the resulting waterline time series: over a two-week moving window, waterlines are cleaned, intersected with fixed cross-shore transects, tagged with modelled water elevation (astronomical tide plus sea-level anomaly), and fit with shape-constrained monotonic regression under automated quality control; profiles passing quality control are combined into intertidal DEMs, mean-high-water (MHW) shoreline contours, and per-transect beach-width estimates. We validate SCI-DEM against RTK-GNSS beach-profile surveys at two camera deployments in Southern California, Torrey Pines and Oceanside Harbor South Jetty, and demonstrate its beach-width product at a third, Ehukai Beach Park (Hawaii), a microtidal beach with no survey programme. At all three sites an independently produced CoastSat satellite-shoreline record is shown alongside the camera series. Camera-derived and surveyed profiles agree to a mean absolute vertical residual of 0.09 m (1 October 2024), 0.15 m (20 May 2025) and 0.10 m (6 October 2025) at Torrey Pines and 0.11 m (1 October 2024) and 0.18 m (6 October 2025) at Oceanside Harbor South Jetty. The resulting beach-width time series reproduce the in-situ MHW crossings to within 3.0 m at Torrey Pines and 2.2 m at Oceanside Harbor South Jetty, and follow the same seasonal trends as the tidally corrected CoastSat record. SCI-DEM is currently running on over 60 live cameras. The method offers a cost-effective route to continuous, high-temporal-resolution beach-profile and beach-width monitoring that complements (and densifies) episodic physical surveys.
ER -