GEOINT
Investors

Eight layers, from acquisition to decision

Select a layer to see what it does, what goes in and what comes out.

  1. Acquisition
  2. Data fabric
  3. Processing
  4. Physics reconstruction
  5. AI
  6. Uncertainty
  7. 3D/4D twin
  8. Decisions

A · Acquisition

Concept
Purpose
Bring in measurements from seabed nodes, fibre, towed arrays and potential-field surveys.
Inputs
Raw recordings, survey geometry and instrument metadata.
Outputs
Checked, time-stamped raw data with its source.
Status
Concept stage. This layer is planned and has not been built.

B · Data fabric

Concept
Purpose
Store, version and describe every dataset so that it can be found and trusted.
Inputs
Checked raw data and its licences.
Outputs
Immutable, catalogued datasets with their lineage.
Status
Concept stage. This layer is planned and has not been built.

C · Processing

Concept
Purpose
Remove noise and artefacts while preserving the physics.
Inputs
Catalogued raw data.
Outputs
Processed data and the parameters that were used.
Status
Concept stage. This layer is planned and has not been built.

D · Physics reconstruction

Concept
Purpose
Estimate subsurface properties by fitting wave and field physics to the data.
Inputs
Processed data and a starting model.
Outputs
Property models and their misfit.
Status
Concept stage. This layer is planned and has not been built.

E · AI

Concept
Purpose
Assist where physics alone is slow or ambiguous, under the safe-AI rules.
Inputs
Models, data and held-out tests.
Outputs
Suggestions with physics residuals attached.
Status
Concept stage. This layer is planned and has not been built.

F · Uncertainty

Concept
Purpose
Estimate how well the data constrain each part of the model.
Inputs
Models, noise estimates and prior knowledge.
Outputs
Ranges of plausible models and coverage checks.
Status
Concept stage. This layer is planned and has not been built.

G · 3D/4D twin

Concept
Purpose
Hold the model of the subsurface in space and, later, through time.
Inputs
Models with their uncertainty.
Outputs
A navigable model in which every value is traceable.
Status
Concept stage. This layer is planned and has not been built.

H · Decisions

Concept
Purpose
Present results so that people can weigh them against their uncertainty.
Inputs
The twin and its evidence chain.
Outputs
Reviewed interpretations and recorded decisions.
Status
Concept stage. This layer is planned and has not been built.

Every number traces back

The evidence chain, from a decision down to the raw measurement.

Every number traces backA vertical chain of five links from a decision down to the raw measurement, each resting on the one below.DecisionInterpretationModelProcessing parameters and code versionRaw measurement

Every number on screen can be traced to the data and code that produced it.

Reproducibility and provenance

Four habits that let anyone repeat and check a result.

Concept

Immutable raw data

Raw recordings are never edited; every change creates a new, linked version.

provenance

Concept

Versioned pipelines

Each processing step is tied to the exact code version that ran.

reproducibility

Concept

Checksummed reruns

A result can be rerun and compared byte for byte with the original.

reproducibility

Concept

Recorded negative results

Failed attempts are kept with their reasons, so they are not repeated.

provenance

Integration with standard formats

Targets on the roadmap, not features that exist today.

Concept

SEG-Y

Seismic traces and headers, the common exchange format for seismic data.

seismic imaging

Concept

miniSEED

Continuous seismograms from seabed and fibre sensors.

DAS

Concept

NetCDF and HDF5

Gridded models and their metadata.

digital twin