National Groundwater Lithology Inventory | SWATGenX

A living, CONUS-wide inventory of groundwater well records, harmonized from the primary state agency sources — driller and geologist lithology logs, static water levels, total depths, pumping yields, and screened intervals. It exists to answer a practical question: what does the subsurface look like where you want to build a groundwater model?

Every value is as-published by the source agency — no interpolation, no gap-filling, no modeled substitutes. Where an agency only reports a well location, that is all the inventory holds; where it publishes a full lithology log, the inventory holds every depth interval. The counts below read live from the database and grow as new state sources are integrated.

The inventory right now

Live counts from the SWATGenX database (loading…) — updated automatically whenever a new source is integrated.

What each well carries

Wells are tiered by how much of the record the source agency published. The same tiers (and colors) are used by the Groundwater Wells overlay in the Watershed Explorer:

Coverage by state

Explore it on the map

The full inventory is browsable in the Watershed Explorer: turn on the Groundwater Wells overlay to see every well colored by its data tier, click a well to read its lithology log, or switch to the richness-weighted density view to see where the subsurface record is strong enough to support groundwater modeling.

How it is built

Each state publishes its well records differently — bulk downloads, ArcGIS services, per-well report pages, or scanned driller's-log documents. The inventory harvests each source with a dedicated, reproducible pipeline and harmonizes the results to one schema: well location, per-interval lithology (top depth, bottom depth, material description), and the hydraulic fields the agency reports (static water level, total depth, yield, screened interval). For states whose logs exist only as scanned documents, a validated vision-model extraction converts handwritten records into structured intervals (~91% material accuracy against human transcription).

The design rule throughout: honest harvest. The inventory reports what the agencies published, at the coverage they published it — a faithful national picture of where subsurface data exists and where it does not.

Cite the dataset

A frozen, versioned release is archived on Zenodo (CC BY 4.0). The accompanying research article describing the inventory and its methods is under peer review; the citation will be updated on acceptance.

Cite this dataset

Source agencies are credited in the dataset documentation. State groundwater programs interested in collaborating — including digitizing scanned driller's-log archives — are welcome to get in touch.