Our projects
AI projects for coastal systems
Innovative solutions combining physical modeling, environmental data and artificial intelligence.
Submarine cables
Sediment mobility atlas and optimal corridor
Pre-survey screening for cable layers: where the seabed moves, and where to route.
- Soulsby bottom-boundary physics, Shields threshold
- 500 m grid, 8,871 cells, 30 years (1993–2022)
- A* corridor and depth-of-lowering guidance
- Validated on 51 SHOM points and 908 dune crests
R&DEMODnet / SHOM
AI emulation
FNO surrogate — currents of the Brest roadstead
Emulating the currents of a high-resolution CROCO model with a neural operator.
- Fourier neural operator (2D FNO), autoregressive output
- CROCO ground truth, 250 m grid, 90 days
- Median error of 17% over a full tidal cycle
- Against 29% for a harmonic baseline
R&DFNO
SeaRoute-Py
SeaRoute-Py — weather routing
Vessel trajectory optimization via A* weighted by wind, waves and currents.
- Time, fuel or balanced cost profiles
- ETA validated on 800 real AIS voyages
- Gain concentrated in heavy seas (Hs 2–5 m)
- GeoJSON, GPX and KPI outputs (ETA, fuel, CO₂)
ApplicationPython
Discover SeaRoute-Py Blend AI
Blend AI — observation and model fusion
Calibrating model output against in-situ measurements.
- Observation quality control
- Analysis of model ↔ measurement gaps
- Hs bias reduced by 20 to 40% (model vs ADCP)
- Stable forecasts during extreme events
OngoingBlending
MeteoData Hub
MeteoData Hub — unified metocean data access
Query ERA5, GLORYS12 and GFS without building the ingestion pipeline.
- A single interface for heterogeneous sources
- Extraction by point, by route or by bounding box
- Directly usable outputs (NetCDF, JSON)
- Product in beta, join the waiting list
BetaProduct
Join the waiting listLet us discuss your coastal project
For any question or diagnostic request, write to us directly. Response within 48 hours.
Request a diagnostic
Describe your need, and we will get back to you quickly.
We respond within 48 business hours.