Editorial

Corrections

Every figure on this site is a value from a federal dataset, so a wrong figure is either a wrong join, a wrong reading of the source, or the source itself being wrong. All three are worth telling us about, and the third is worth telling EPA about.

Write to hello@umwelt.com with the page and the figure. What comes back is either a fix or an explanation of why the number is right; corrections are applied against the source dataset, never negotiated.

What counts as a correction

A correction is a figure this site published that was wrong: a bad join, a mis-parsed column, a status matched by the wrong string, a unit conversion error. Those get logged below, dated, with what broke and what fixed it.

Not a correction is a number that changed because the source changed. The datasets refresh quarterly and figures move with them. Logging routine refreshes as corrections would bury the real ones, so refreshes show up only as the retrieval date on every page.

Also not a correction is a disagreement about what a record means. If EPA lists a site and the operator disputes it, this site reports the listing and links the file. The place to change the record is the agency.

The log

2026-09-03
The home page reported 0 sites on the final Superfund NPL. The filter matched the phrase 'Final NPL', but EPA's status vocabulary is exactly 'NPL Site', 'Proposed NPL Site' and 'Deleted NPL Site'.
Corrected before launch to match EPA's exact status values; the figure is 1,337. The same wrong match was fixed on the Superfund index, county tables and report.
2026-09-03
County joins dropped rows where EPA and Census disagree on spacing in a county name, such as Du Page against DuPage, and multi-county Superfund sites matched no county at all.
County matching now compares names with spacing removed, and a multi-county site is attributed to its first listed county while its own page names all of them. NPL county matching went from 1,763 to 1,805 of 1,840; the remainder are territories, which have no county pages.
2026-09-03
About 100 tribal water systems showed an EPA region number where a state code belongs, because that is what UCMR5's State column carries for them.
Their state is now recovered from the counties EPA records them as serving. Systems that still cannot be placed in a state get no page of their own rather than a page filed under a number.

The first three entries are defects found and fixed during the build, before launch. They are logged because a corrections page that starts empty tells the reader nothing about whether anyone is checking.