A national inventory of well lithology and hydraulics for the conterminous United States (v1.0)

Harmonized, openly-harvested inventory of per-well lithology and hydraulics for the conterminous United States, accompanying the manuscript “A national inventory of well lithology and hydraulics for the conterminous United States, with a vision-language method for reading scanned driller's logs” (Rafiei, 2026). v1.0 contents: gw_lithology_v1.parquet — 23,951,153 depth-resolved lithology intervals across 3,391,633 wells in 17 states (AL, CA, IA, IL, KY, MA, MI, MN, MO, MT, NM, OH, OK, TN, TX, VT, WI); and gw_hydraulics_v1.parquet — per-well structured hydraulics across 18 states, of which 2,827,965 wells carry genuine hydraulic measurements (static water level, depth, yield, drawdown, derived specific capacity, screen interval, and where available measured transmissivity/conductivity and pump-test duration), with surface elevation and completion date for additional wells. This release contains the portion recovered from public, structured or born-digital state sources — the reproducible, free-to-extract half of the inventory. Lithology locked in scanned driller's logs is characterized in the manuscript by validated sampling (yield, extraction accuracy, token cost) and is recoverable with the companion code rather than shipped here as bulk data. Because the upstream public sources mutate and disappear, this frozen snapshot is citable independently of whether the live sources later reproduce it. Records use each state's own primary key; depths are in feet. See DATA_DICTIONARY.md and README.md. Released under CC-BY-4.0; underlying records are public US state well data.