NY State Oil & Gas Annual Production (Kaggle)

NY State Oil & Gas Annual Production is the record of every producing oil and gas well in New York, published annually by the New York State Department of Environmental Conservation — the agency operators report to — and carried by Datadory as structured, join-ready rows. One row per well per calendar year, starting in 2001, which makes it exactly what most state publications are not: well-level rather than lease-level, statewide rather than single-basin, and small enough that the whole history fits in one file you can open on a laptop.

Each record identifies itself twice over. There is the human-readable identity — operator company, operator-assigned well name, county, town, parent field, producing formation — and there is the machine identity: a 14-digit API well number such as 31009119450000, the industry-standard key that decomposes into state code, county code, hole, sidetrack and completion segments so that one physical borehole never masquerades as another. Well type codes distinguish an oil well (OW) from dry-hole and disposal variants (OD); status codes carry each well's regulatory standing.

Four numeric columns close out the record: production (the total), plus oilprod, gas and water as the reported calendar-year components. In the published samples all four read zero — a reminder that this is a wellstock census as much as a volume ledger, and that a zero is information about a well's standing, not missing data.

In Datadory's catalog of 1,744 datasets across 159 viable industries, this record anchors the Northeast end of the eighteen-dataset oil and gas exploration and production set. Amaze yourself with what's in it, then get the rows that matter:

Get a sample of this dataset

Two records captured during the August 2026 research pass, exactly as they arrive:

api_wellno         operator                 well       county       town      field      form      typ st  year oil gas water
--------------------------------------------------------------------------------------------------------------
31009119450000     Miller Oil Co.           Kinley 04  Cattaraugus  Allegany  Bradford   Bradford  OW  NR  2015   0   0     0
31009054960000     Solomons Resources, LLC  Weis 34    Cattaraugus  Olean     Five Mile  Bradford  OD  AC  2006   0   0     0

Read what the first row settles immediately. Miller Oil Co.'s Kinley 04 sits in Allegany, in Cattaraugus County, producing from the Bradford formation inside the Bradford field — the great Allegheny-region oilfield that has been producing since the 1870s and still defines western New York's oil country. Its API number opens with 31 (New York's state code) followed by 009 (the Cattaraugus county code), and the trailing segments pin it to hole 11945 with no sidetrack and no completion branches. Type OW says oil well; status NR marks its standing in the state's vocabulary.

The second row shows why the type code earns its keep. Solomons Resources' Weis 34 produces from the same Bradford sand but through a different structural position — the Five Mile field near Olean — and its type reads OD, the oil/dry-and-disposal family, with status AC. Two wells, two decades apart in vintage, one shared formation, and the schema keeps them cleanly separable without any interpretation on your side.

One habit worth taking away: the zero-volume years are the point, not a defect. An annual census that reports every producing well — including the ones that moved nothing that year — is the difference between measuring a basin and merely totaling it.

Eighteen documented columns carry every well-year, each shown with its example from the published samples:

• Identity spine — api_wellno, the 14-digit American Petroleum Institute well number (state code + county + hole + sidetrack + completion), plus its decomposed segments hole, sidetrck and completion, so joins survive reformatting.
• Location — cnty (county), town (town within the county), field (the named oil or gas field) and georeference, the latitude/longitude point of the well where the upstream copy includes it.
• Attribution — coname, the operating company, and well_nm, the operator-assigned well name/number.
• Classification — well_typ, the well type code (OW oil well, OD oil/dry or disposal variants), and wl_status, the well status code observed in samples as NR and AC.
• Geology — prod_form, the producing formation — the geologic unit the well actually draws from, which is how a Bradford-field well gets distinguished from a deeper target on the same acreage.
• Volumes — year (calendar year of the report), production (total reported annual volume), and the component readings oilprod, gas and water.

The label-and-code pairing runs throughout: readable place names travel with coded classifications, so a dashboard renders "Cattaraugus / Allegany / Bradford" while the analysis keys on the API number and never collides.

Three extensions ship as additional fields on request: confirmation that your delivered copy carries the georeferenced latitude/longitude point alongside each well, verified at sampling rather than asserted here; the pre-2001 companion series covering 1985–2000 under the same well-level schema, scoped and quoted when you request the sample; and derived views — per-operator well counts, per-field year-over-year volume rollups and county aggregates — computed on delivery rather than guessed in advance.

One convention matters before you build on the volumes: this is reported annual production, filed by operators with the state regulator on the state's own definitions of oil, gas, water and total. Treat the total column as the filing and the three components as its decomposition, reconcile them before deriving ratios, and remember that a zero is a reported condition of the well in that year — not a gap to be interpolated over.

• Geography: New York State, every county that holds oil and gas wells, recorded per well. Western counties dominate the published samples — Cattaraugus County's Allegany and Olean sit in the historic Bradford and Five Mile fields — but the schema is uniform wherever a well reports. Coordinates ride with each well in the upstream copy, so mapping needs no geocoding step.
• Temporal: annual reporting periods from 2001 to present, one row per well per year. Every record declares its own calendar year, so the depth of any delivered copy is self-declaring — check the maximum year and you know exactly what you hold.
• Granularity: the finest cut the state legally publishes — one record per producing well, per calendar year, identified by API number. No lease rollups, no county aggregation, nothing averaged away.

An honest caveat: field documentation was captured by inspection rather than publisher documentation during the August 2026 research pass, so the dictionary above reflects observed structure, and the exact roster of columns in any given copy is confirmed at sampling instead of promised blanket.

API, files, or your warehouse. Daily, weekly, or hourly.

Name the wells, counties, operators and horizon you want when you request the sample — the full state history, or a slice cut to your acreage. The sample ships first either way; the ongoing feed lands on whatever cadence your models run at, already shaped into the spine above.

A state well-year census earns its keep in four specific jobs:

• Operator prospecting in a legacy play — operator company names, per-operator well counts and their locations across two decades turn "who works New York's old oilfields" into a callable list, with the API numbers to prove coverage.
• Northeast basin screening — a well-level state series fills the Appalachian gap between county-level Pennsylvania reporting and national totals, and its small footprint means a screen runs in seconds rather than hours.
• Landowner and title research — because each row resolves to one hole, sidetrack and completion under one operator, a parcel's drilling history reads directly off the county, town, field and formation columns.
• ML practice on real well data — a tidy single-file table with codes already normalized is the honest middle ground between toy datasets and multi-gigabyte state archives: big enough to be real, small enough to iterate on.

Each job maps onto a persona page: sales growth teams, data scientists, market researchers consultants, and developers builders wiring the feed into a product.

Sales and growth teams get a prospecting list that builds itself — operator names and footprints drawn from mandatory state reporting, which is more reliable ground than scraped directories for a play full of small, hard-to-find companies. Journalists, academics and students get per-well annual figures back to 2001 behind any story on New York drilling, landowner rights or the state's long tail of aging wells, citable to a named state agency rather than an advocacy count. Data scientists and ML engineers get a join-ready well-year panel keyed on stable API numbers — hours of cleaning already done, and small enough to prototype against before scaling to Texas-scale archives. Market researchers and consultants get the simple state series that illustrates an Appalachian trend in one chart, without stitching three sources together.

Start from the oil and gas exploration and production data hub to see where this record sits among the eighteen-dataset set, then read the ranking of the best datasets in the industry for where a narrow state census beats and loses to continental trackers.

Provenance note — the data originates with the New York State Department of Environmental Conservation, the agency that permits wells and collects operators' annual production reports; Datadory delivers it as a packaged, analysis-ready snapshot of that single publisher's rules and schema. One agency, one state, one wellstock.

Methodology note — these are operator-reported calendar-year filings, not metered measurements. Volumes are annual sums by design, zeros are reported conditions rather than gaps, and the API well number is the only key that reliably separates one borehole from another across years. Reconcile the total column against its three components before deriving anything from ratios.

Completeness note — the field dictionary above maps to the structure observed during the August 2026 research pass, with two published sample rows anchoring it. Coordinate presence, the 1985–2000 companion series and derived rollups are confirmed at sampling rather than claimed here.

Where to go next — pair the state census with the continental layer so the wellstock and the basin totals land in one frame: the Railroad Commission of Texas Data Sets for lease-month depth at the other end of the scale, the Global Oil and Gas Extraction Tracker (GOGET) for asset-level context worldwide, and NSTA UKCS Open Data once the workflow goes offshore. The production-by-well glossary entry defines the vocabulary this page leans on.