What this is:

The WELL records of GSA020K: 140,992 rows, one per gas well, keyed on the six-digit RRC gas well identification number that is unique across Texas.

How it is laid out:

The Railroad Commission publishes this as EBCDIC (IBM cp037). IBM mainframe encoding (cp037). Opening it as text gives you nonsense; it has to be transcoded first. We decode it into a table you can open anywhere. It has 70 columns and about 141K rows.

The columns you will most likely want:

  • district RRC district the well is in, as the ledger tape writes it -- 07B, 07C, 08A with the leading zero, where the G-10 file and PDQ write 7B, 7C, 8A.
  • field_no The eight-digit RRC field number the well produces from.
  • operator_no P-5 organization number of the well's operator.
  • gas_rrc_id The gas well's six-digit RRC identification number, which uniquely identifies it statewide: 140,992 distinct values in 140,992 rows.
  • well_no The operator's well number -- tract, well number and suffix -- right-justified in six bytes, so 123,566 of the 140,992 rows carry leading blanks.
  • lease_name The lease name from Form P-4, up to 32 characters. Personal names are written last name first.

…and 64 more, all documented on the full page.

What we add:

Find the gas wells a given pipeline gathers for, read the deliverability and pressures behind a well's allowable, and follow the six-month balancing accounts that decide whether a well must underproduce to make up an overage.

  • Decoded out of the Commission's original format into something Excel opens.
  • Filterable before you download, so you can take one county instead of the whole state.
  • The Commission's single-letter codes translated into words.

Worth knowing before use:

gas_rrc_id is the join key to everything else: PDQ carries it as lease_no with oil_gas_code = 'G' and the G-10 file as rrc_id. It changes when a well is worked over significantly, so it identifies a completion rather than a hole in the ground; well_no survives the workover instead.

The full page lists every trap in this dataset.

Get it:

People who want this usually want: