Where to lay a submarine cable is not, first and foremost, a question of depth — it is a question of seabed stability. A route that looks impeccable on the bathymetry map can cross sediments that are set in motion by every storm, exposing the cable to scour. We built a tool that quantifies this risk cell by cell, then derives the least-cost corridor between two landing points.
Why bathymetry is not enough
Depth tells you where a cable can physically go, not whether it will stay stable there. Two phenomena govern the fate of a laid or buried cable: sediment mobility — the fraction of time the seabed is set in motion by waves and currents — and extreme scour, the localised erosion during rare events. Fine sand may be mobile 8 % of the year where the coarse seabed next to it is mobile only 0.6 % of the time; within a few hundred metres, the risk varies by an order of magnitude. That information, absent from a depth chart, is what the atlas reconstructs.
The backbone: heterogeneous data on a single grid
Everything starts from public sources reprojected onto one 500 m reference grid (EPSG:3035). Each variable comes from its natural source:
- Near-bed currents — GLORYS12 (Copernicus Marine reanalysis).
- Sea states (height, period, near-bed orbital velocity) — RESOURCECODE / HOMERE hindcast, hourly over more than 30 years.
- Bathymetry, seabed nature (Folk classes), traffic density, exclusion zones — EMODnet.
Raw extracts are cached once and reused; the heavy physics runs at the hindcast wave nodes rather than over thousands of cells, and only the statistics are interpolated onto the grid. The result is a single data cube where waves, currents, seabed and constraints align in time and space — the foundation everything else reads.
Measuring mobility: bottom shear stress
Mobility follows from the combined wave + current bottom shear stress, compared with the sediment threshold of motion. We apply Soulsby's (1997) formulation for the bottom stress and the Soulsby & Whitehouse criterion for the critical stress, calibrated on the d50 inferred from the seabed nature. A cell's mobility is then the fraction of time the stress exceeds that threshold.
On our proof tile (North Sea, January 2020), the seabed is clearly wave-dominated — the ratio of wave stress to current stress is around 8 — with a 90th-percentile maximum shear near 0.18 Pa and a 99th percentile near 0.33 Pa. Each cell is assigned a class (stable / intermediate / mobile) and a number of events per year, exported as a georeferenced map.
Quantifying extreme scour with extreme-value analysis
Design risk comes not from the mean but from rare events, which we handle with extreme-value analysis (POT-GPD). From the stress series, the peaks-over-threshold method fits a generalised Pareto distribution and produces the 1-, 10-, 50- and 100-year return levels, each with a 95 % confidence interval.
Concretely, the atlas answers the design engineer's question: what stress does the seabed experience once a century at this location, and with what uncertainty? A sensitivity analysis (robustness to ±1 Folk class on d50) accompanies each result so the assumptions stay visible.
From risk to route: the least-cost corridor
The risk layers are merged into a cost surface, on which an A* algorithm traces the optimal corridor between two landing points. Each cell's cost combines, with adjustable weights, mobility, extreme scour, anchoring risk (derived from traffic) and regulatory exclusions (Natura 2000, prohibited zones). Where a landing falls on a barred cell, it is automatically snapped to the nearest passable cell.
The tool proposes not a single route but three Pareto variants — min-risk, balanced, min-length — so the decision stays with the client. Replaying a set of weights recomputes the corridor in under a second, making it a discussion aid rather than a black box.
A zone-agnostic method
Changing study area means changing a configuration file, not the code. Extent, resolution and sources are parameters; nothing about an area is hard-coded. The same pipeline, validated on the North Sea, now runs on a tile in the English Channel — with its hourly tidal currents and its own constraints — without rewriting. That is what makes the study reproducible and transferable from one project to the next.
The physics and the routing solver are isolated, unit-tested modules, read and validated by a human — not components delegated blindly. That is the condition for signing off an engineering result.
What the client receives
The core deliverable is a self-contained atlas (an HTML report with provenance, methodology, assumptions and results) together with a web service: an API and a cartographic viewer in read-only mode, where you query a point (mobility, 100-year return level, cost) or redraw the corridor by moving weight sliders. The layers, stored in Zarr format, deploy equally on a single machine or on S3-compatible object storage.
Planning the lay or burial of a cable, umbilical or pipeline? Tell us about your area and we will scope the leanest risk atlas that answers your decision. For choosing the sources upstream, see also our guide to ocean data.
Frequently asked questions
What data is needed to assess submarine cable stability?
Near-bed currents (GLORYS12), a long wave hindcast (RESOURCECODE or HOMERE), and bathymetry, seabed nature, traffic and exclusion zones (EMODnet), all reprojected onto a common reference grid.
How is sediment mobility estimated?
We compute the combined wave + current bottom shear stress (Soulsby formulation) and compare it to the sediment threshold of motion (Soulsby & Whitehouse). Mobility is the fraction of time this stress exceeds the critical threshold.
Can the method be applied to any area?
Yes. The pipeline is zone-agnostic: the extent, resolution and sources are parameters in a configuration file. A new area = a new config file and a run, with no code rewrite.