Methodology

What we publish per county

Each county page carries four data products: a monthly oil and gas production series (rendered as a chart and a CSV download), the county's currently producing geological formations, the operators most active in the county over the last 12 months, and recent drilling-permit activity. The monthly series is published as a structured-data Dataset compliant with schema.org/Dataset and indexed by Google Dataset Search.

Per-state source matrix

Each state's production data is sourced from that state's oil and gas regulator. We aggregate well-level monthly filings into county-level monthly totals and normalize formation/reservoir labels to a canonical set so cross-state comparisons are meaningful.

How we normalize formation names

State regulators report producing reservoirs in dozens of variants — "WOLFCAMP A", "Wolfcamp (Upper)", "WC-A 8500", "PHANTOM (WOLFCAMP)", "NEWARK EAST (BARNETT SHALE)". We canonicalize each into a single label per generalized formation (Wolfcamp, Bone Spring, Eagle Ford, Barnett, Granite Wash, etc.) so a Reeves County Wolfcamp well and a Howard County Wolfcamp well aggregate consistently.

The Texas RRC convention (and several other state regulators) encodes the producing reservoir in parentheses after the field name — "NEWARK EAST (BARNETT SHALE)" means the Newark-East field producing from the Barnett Shale. Our normalizer extracts the parenthetical first; if no parenthetical exists, it falls back to a hand-curated field-to-formation crosswalk for legacy field names like "LEVELLAND" (San Andres) and "EAST TEXAS" (Woodbine).

Normalization rules are in scripts/lib/normalize-formation.mjs and version-controlled. Changes are reviewed against a 39-county sentinel set whose expected formations are codified at scripts/data/county-sentinels.json.

Drilling permits dataset

The drilling permits dataset counts approved new-drill oil and gas permits per county and month for Texas and Pennsylvania, from the permits each state regulator issued. Its column new_drill_permits counts regulator approvals of a permit to drill a new oil, gas, combined or coalbed-methane well (undetermined and unknown types included), counted once per permit in the calendar month of its first approval and attributed to the surface county.

Texas.The Railroad Commission's drilling-permit query returns new-drill permits of every well type. Cathodic protection, injection, storage, service, exploratory test and water supply permits are excluded.

Pennsylvania.The Department of Environmental Protection's permits-issued report lists every authorization type it issued. The three new-well authorization types are counted; alterations, change-in-use, re-drill, drill-deeper and disposal authorizations are excluded.

The newest months.Permits for a month keep reaching the regulators' records after the month ends, so the 3 months before each weekly build are inside a settle window: they are not asserted and appear in no CSV. Every complete month is asserted, and a complete month with no permit in a county is a 0.

Basin figures in this dataset are sums of whole counties. Each county belongs to at most one basin within its state, assigned by a fixed county-to-basin crosswalk (basin-county-crosswalk.csv, the same table Pointer Minerals uses for its basin production series); there are no weights, no split counties and no formation filter, so a basin's value for a month is the plain sum of its member counties' values for that month. Nine basins are defined by the crosswalk; two are published here in full, one (Permian) is published under an identifier naming the one member state this version covers, and six are withheld because a member state is not in this version. A basin row appears only for months in which every member county in the version has a value; when a member state is not in a version the series is published under a separate identifier that names the states it covers, or withheld — nothing labelled with a basin's bare name is a partial sum. The crosswalk is a geographic assignment, not a formation filter: a permit in a Barnett Shale county may target another zone. The basin outlines drawn on Pointer Minerals maps are a different artifact, used for display only, and are not the source of any figure.

County drilling permits downloads →

Known limitations

• Texas (RRC): pre-1993 well-completion records are sparse in the public bulk download. Counties whose production base is dominated by pre-1993 verticals may under-report formation diversity.
• Oklahoma (OCC): SCOOP/STACK horizontal wells are pooled in pooling orders that can obscure the producing bench (Woodford vs Meramec vs Sycamore). We use the OCC pool code where available; a small number of multi-bench wells report only the dominant pool.
• California (CalGEM): the public Well Search exposes field name only; producing pool requires a join to the WellSTAR Production Injection CSV which we update monthly. Some active wells without a recent production report fall through to field-name only.
• West Virginia (DEP): the producing-formation column is sparsely populated for pre-2005 conventional wells (~70% coverage post-2010 unconventional, ~15% pre-2005). Where the column is null, we omit the well from county formation aggregation rather than guess.
• Alabama (OGB), Michigan (EGLE): regulators report formation at field-level granularity that can conflate stacked pays (Smackover/Norphlet in AL; Antrim members in MI). We surface the field-level label honestly rather than over-disambiguating.
• Pennsylvania (DEP), Oklahoma (OCC):the public well layer doesn't expose a formation column. We rely on basin-derived defaults (Marcellus/Utica in PA; Anadarko/Woodford in OK) until a join to permit/pooling tables is wired in.

Correction policy

If you spot an error on any county page — wrong formation, wrong operator, wrong production figure — email [email protected] with the URL and what you expect to see. We acknowledge correction emails within 5 business days and ship verified fixes in the next monthly data refresh (or same-week if the page is materially wrong).

For systemic disagreements (e.g., we picked a different normalization than your local convention), we'll explain the rule and either adjust it or document the divergence here in the limitations section.

Last audited

Per-county formations and operators are re-audited on every build. The full data pipeline is rebuilt monthly from the latest regulator filings. Build provenance is recorded on each county Dataset's sdDatePublished property in the page's structured data; you can read it with any JSON-LD inspector.

Browse all county pages →