GroundTruth / methodology
Four EPA datasets, each downloaded in bulk, archived immutably (content-addressed, so every fact traces to the exact federal file), and re-snapshotted monthly: Superfund/NPL site list (SEMS), Safe Drinking Water Act systems and violations (SDWA via ECHO), Toxics Release Inventory basic files (latest complete reporting year), and ECHO compliance data (from which we index ONLY facilities carrying a current significant-noncompliance flag). EPA publishes these on its own cadences (ECHO weekly, SDWA quarterly, TRI annually), so the monthly snapshot bounds our staleness; every page and API response states the snapshot date it came from.
Great-circle distance from the queried coordinates to each record's published coordinates, in km. Records without coordinates cannot appear in radius results (a stated coverage gap).
Systems are matched by the county/city they report serving. Service-area boundaries are not public data, so this is approximate by construction; the biggest matching system is listed first as the most likely server of an address, and your water bill is authoritative.
The address search's Superfund check distance-weights every federal cleanup (NPL) site and every toxic-release (TRI) facility within 5 miles of the point. NPL sites weigh most. A TRI facility weighs more when its reports include carcinogens, PBT chemicals (ones that persist and build up in living things), or PFAS.
Note, dated 2026-08-27: the chemical-flag weights are active as of 2026-08-27. Earlier scores did not include them; a parsing defect kept the flags off. We fixed the defect and re-scored.
Composite scores, rankings, and "is it safe" verdicts. TRI pounds are not toxicity; an enforcement flag is not a conviction; an NPL listing is a cleanup-process fact.