Steady theme

fault faults

One of 50 themes found in 20,416 EGU water abstracts (2024 to 2026). Its most distinctive words are below — they are what separates this body of work from the rest of the corpus.

faultfaultstectonicmantlegeothermalfracture

357

abstracts in this theme

1.7% of the water corpus · 32 of 50 by size

1.4%

use machine learning

11.8 points below the 13.2% corpus average

+0.02

points of share, 2024 to 2026

23 of 50 by trend

#46

of 50 for AI penetration

among the least AI-touched themes

How to read this theme

This theme holds 357 of the 20,416 water abstracts, making it the 32th-largest of the 50 themes. Between 2024 and 2026 its share barely moved, by 0.02 points.

1.4% of its abstracts were flagged as using machine learning, 11.8 points below the 13.2% corpus average. AI adoption here sits close to the corpus norm.

Find data for this theme

Search the directory for datasets matching this theme's defining terms.

fault faults tectonic mantle geothermal fracture

Semantically nearest themes

Themes that sit closest to this one on the semantic map — the work that reads most like it.

fluids calcite

fluids, calcite, rocks, carbonate, mantle, minerals

349 abstracts · 0% use AI · +0.16 pp

speleothem holocene

speleothem, holocene, miocene, speleothems, cave, carbonate

501 abstracts · 1% use AI · -0.59 pp

earthquake resistivity

earthquake, resistivity, interferometry, insar, fault, bedload

286 abstracts · 7% use AI · +0.29 pp

fracture porous

fracture, porous, pore, permeability, solute, fractures

289 abstracts · 4% use AI · +0.23 pp

rainfall-induced slow-moving

rainfall-induced, slow-moving, lews, deep-seated, insar, geotechnical

451 abstracts · 19% use AI · +0.55 pp

Themes are clusters found by grouping abstracts on their sentence embeddings, so their boundaries are statistical rather than official EGU categories. "Uses AI" comes from a keyword detector validated at precision 0.95 and recall 0.9. Point estimates carry roughly half a point to a point of uncertainty from the clustering. Full method on the main trends page.