Machine Learning for Geoscience glossary

Clear, one-line definitions of the Machine Learning for Geoscience terms used across the OgbonLab textbooks. Each entry links to the interactive sections where the idea is taught.

forward operator

The mapping F from model parameters m to predicted data d = F(m); the simulator that the inverse problem inverts.

lithology prediction

Predicting rock-type labels along a borehole from log measurements (GR, NPHI, RHOB, …) using classification or sequence models.

seismic facies classification

Clustering or supervised labelling of seismic traces by waveform shape or attribute vectors to delineate depositional units.

variety

One of the big-data 4 Vs: the diversity of data types and sources (tabular, image, text, sensor) that must be combined.

veracity

One of the big-data 4 Vs: the reliability and uncertainty of the data, reflecting noise, missing values, and provenance issues.

volume

One of the big-data 4 Vs: the sheer scale of a dataset, often too large for a single machine to hold or process in memory.

See: Sum the Volumes, Bulk-Volume Water

well log

A continuous measurement along a borehole (gamma ray, density, neutron, resistivity, sonic, …); the primary feature stream for ML-geophysics tasks.

See: Blocking Well Logs, Cores and Well Logs