DrillerDB opens 10 million water well logs to researchers

DrillerDB is releasing a standardized North American dataset built from more than 10 million water well logs and 53 million rock-layer intervals, free for non-commercial research. The company says the resource could help researchers study groundwater, drought, geothermal siting and private-well risks across the U.S. and Canada.

Why it matters: - The dataset gives researchers a standardized view of the shallow subsurface across state and provincial lines. - It covers the zone where drinking water wells, geothermal systems and foundations are built. - More than 40 million Americans rely on private wells, making groundwater and well construction data a public-interest issue. - DrillerDB says the release can support research on groundwater trends, drought response, aquifer characterization and public health.

What happened: - DrillerDB opened free record-level access to its Underground Lithology Grid for non-commercial research. - The company standardized more than 10 million water well logs and more than 53 million depth intervals. - The release includes the interactive public map at the well map. - The announcement says the dataset is North America's largest standardized well-log dataset. - DrillerDB says the records are available under a one-page data use agreement at research access.

The details: - The database draws from more than 18 million state well completion reports across the U.S. and Canada. - Nearly 1.4 million domestic wells were used in an analysis that found deeper new wells in several states over time. - In Colorado, the average new domestic well in the 2010s reached 360 feet, or 152 feet deeper than in the 1970s. - California's average new domestic well depth rose by 135 feet. - Maine's average new domestic well depth rose by 139 feet. - In Michigan and Wisconsin, average well depths changed little over five decades. - In California, all 38 counties with enough records in both decades showed deeper new wells. - DrillerDB coded 51.3 million intervals with a single controlled lithology vocabulary. - The dataset preserves the original driller wording alongside more than 99% of intervals. - Color and grain size are stored separately when drillers recorded them. - The company says 96.5% of intervals were coded with a curated, versioned dictionary. - Less than 1% were assigned by AI, and 1.7% were left uncoded rather than guessed. - An independent audit of 1,000 sampled coded intervals found 96% matched the original description, 3% were partially correct and fewer than 1% were wrong. - The dataset includes 14.5 million well records that name the contractor who drilled the well. - It also includes more than 80,000 state-issued driller licenses.

Between the lines: - The dataset is built from public records filed under state law, but those records were never designed to be compared across jurisdictions. - DrillerDB is positioning the product as a translation layer, not a claim of ownership over the underlying logs. - The company says its approach is intentionally broader and faster than the federal U.S. Geological Survey's National Water-Well Database, which has published Minnesota so far. - The same breadth also limits precision, because drillers describe cuttings at the rig, well locations are not a scientific sampling grid and terminology is not laboratory-grade. - The release blends deterministic coding with limited AI use, which matters for researchers who want scale without losing traceability to the original log.

What's next: - Researchers can apply for free access through DrillerDB's research program. - The company says the dataset is intended for work on groundwater levels, long-term water-level trends, scarcity, drought response, well construction trends, groundwater development, supply planning, aquifer characterization, geothermal siting and public health. - DrillerDB says it would rather see the dataset drive research than remain internal.