Physics first
A model must obey wave physics before any learning is added to it.
wavefield · FWI
Research stage · Concept
See deep beneath the seabed. Show how sure we are.
We study how sound, electrical, gravity and magnetic measurements can be combined to image deep geological structures beneath the sea floor, beginning with the Mediterranean Basin. Each image is meant to carry a statement of how certain it is.
Synthetic — illustrativeThe seabed is imaged, not seen
Why every picture of the subsurface is an inference.
Waves are the only practical probe
Light does not travel through rock, but sound and fields do. Deep structure can only be reached through what passes through it.
Every image is an inference
Different structures can explain the same measurements, so an image is a best estimate and never a photograph.
The uncertainty is usually hidden
Most images show a single answer. We treat the range of plausible answers as part of the result.
Four principles behind the program
The commitments that shape every method we study.
Uncertainty as a feature
Each result is paired with the range of answers the data still allows.
uncertainty · posterior distribution
Evidence you can trace
Every number on screen should lead back to the data and code that produced it.
provenance · reproducibility
Safe AI
Learned components are tested against held-out geology and cannot certify science.
synthetic data · ground truth
How it works
The path from a measurement to a decision, in five steps.
Measure
Sensors record how waves and fields behave around the area.
Process
Noise and artefacts are removed while the physics is preserved.
Invert
A model of the ground is adjusted until it explains the data.
Quantify uncertainty
The range of models that still fit the data is estimated and kept.
Decide
People weigh the image together with how certain it is.
Three research programs
Seismic imaging, multi-physics intelligence and next-generation sensing.
A · Advanced seismic imaging and scientific computing
Wave-equation imaging, velocity model building and learned methods, tested against physics and with uncertainty attached.
FWI · RTM
B · Multi-physics geophysical intelligence
Electromagnetic, gravity, magnetic and passive data fused with seismic, kept only when each adds information.
CSEM · gravimetry
C · Nano, quantum and experimental sensing
A research workstream on new sensor concepts. Each must first prove it can work in the marine setting.
DAS · calibration
Where it could matter
One primary application and five secondary ones the research is framed around.
Secondary applications
A roadmap measured in evidence
Eight phases, each closed by a test rather than a date.
- P0FramingCurrent phase
- P1Synthetic benchmarks
- P2Solver validation
- P3Open-data inversion
- P4Uncertainty tests
- P5Joint inversion
- P6Independent review
- P7Pilot readiness
Where to start
Three paths, depending on what you need from us.
Scientists
Methods, evaluation criteria and the limits we state up front.
Industry partners
Ways to collaborate, from shared data to pilot preparation.
Investors
The thesis as testable hypotheses, and the risks we gate.
Build the evidence with us
The program advances only as fast as its tests. If you hold data, methods or questions worth testing, we would like to hear from you.
Become a partner