EPA Envirofacts Web Services (DMAP REST + GraphQL API)

It is the queryable layer over the databases behind EPA's Envirofacts system, served through the agency's Data Management and Analytics Platform. Two interfaces reach the same tables: a DMAP-EF RESTful data service that takes URL-patterned requests shaped [table]/[column][operator][value]/[join]/[first]:[last]/[sort]/[format], and a GraphQL-style query service supporting field selection, aliases, filtering, joins, subqueries, pagination, ordering, grouping and variables.

Nine program families sit behind those interfaces. TRI covers toxic-release facilities; ICIS covers compliance activity reports; SEMS covers Superfund sites; RCRAInfo covers hazardous waste handlers; GHG covers greenhouse gas reporting; NEI covers the emissions inventory; NGGS, SDWIS and RADNET round out groundwater, drinking water and radiation monitoring. Tables are addressed as [program].[table] - tri.tri_facility, sems.envirofacts_site, icis.icis_activity_report - so one mental model spans all nine.

The operators matter as much as the tables: equals, notEquals, lessThan, greaterThan, beginsWith, endsWith, contains, like, in and notIn, combinable with /and/ and /or/, with case-insensitive text comparison and left joins between tables. In practice that means server-side filtering - you ask for Virginia ZIP codes or Region 05 source-test activities, not the whole nation. A single pull tops out at 100,000,000 records under a hard 15-minute cutoff, which bounds any one extract while leaving room for very large sweeps.

Each record family has its own shape, and the three below were captured live during research:

_table           : tri.tri_facility
tri_facility_id  : 24210TRTBN14378
facility_name    : TRI-TUBE INC
street_address   : 14378 ENTERPRISE RD
city / state     : ABINGDON / VA 24210
coordinates      : 36.740556 / 81.900556

_table           : sems.envirofacts_site
site_id          : 0406480
epa_id           : FLD984216630
name             : WASHAC INDUSTRIES
city / state     : ST. AUGUSTINE / FL
npl_status       : Not on the NPL
coordinates      : 29.957778 / -81.350556

_table           : icis.icis_activity_report
activity_name    : Region Code: 05 Action Type: 37 SOURCE TEST OBSERVED
status_desc      : (null)
begin_date_fy    : (null)

Read them together and the design shows: the TRI row is a manufacturing facility pinned to a street corner in southwest Virginia, the SEMS row is a Florida industrial site whose NPL status answers the first question any acquirer asks, and the ICIS row is a federal source-test observation tagged to Region 05. Same corpus, three regulatory lenses.

Sixteen fields carry the analytical weight across the three sampled families, and every definition above was checked against live row output - the dictionary describes what actually arrives. Three structural facts are worth knowing before you model it.

First, identifiers stack rather than duplicate: the TRI facility ID keys release reporting, the FRS epa_registry_id links a site across EPA programs, and SEMS carries both an internal site_id and an epa_id. Joins between program families run on those identifiers, not on fuzzy name matching. Second, geography is uniform: city, state, ZIP and decimal-degree coordinates appear across the TRI and SEMS families, so radius queries and territory cuts work without per-program gymnastics. Third, nulls are information - the ICIS activity row above carries empty status and fiscal-year begin fields, which is normal on some activity types rather than a data fault.

Geography - United States-wide, with state, county, ZIP and coordinate columns running through the program tables. The same facility can be located three ways, which matters when you are screening by drive-time rather than by political boundary.

Temporal - program-dependent by design. ICIS and enforcement tables carry fiscal-year date fields such as actual_begin_date_fy; SEMS carries NPL and cleanup status dates; the TRI tables hold reporting years from 1987 to the present. No single timeline covers all nine programs - which is precisely why delivery cadence is a subscriber decision here: daily if you watch enforcement activity break, weekly for portfolio reviews, hourly for event-driven work.

Granularity - one row per facility, per site, per activity, per chemical or monitoring observation. Nothing arrives pre-aggregated, so national totals, state rankings and site-level timelines all come out of the same rows. For scale: the companion ECHO Exporter alone spans more than 1.5 million regulated facilities, and this corpus reaches the program databases underneath that universe.

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

• Site due diligence. NPL status and non-NPL cleanup language turn 'is there a Superfund problem here' into a query instead of a site visit. Coordinates make a radius screen around a target parcel a one-line filter.
• ESG and supply-chain exposure. parent_co_name rolls facility-level releases up to corporate groups, so a holding company's environmental footprint aggregates without manual entity mapping. The FRS registry ID ties the same site across EPA programs.
• Enforcement monitoring. ICIS activity reports - region-coded, action-typed, fiscal-year dated - give compliance teams a stream of who got inspected, tested and cited, filterable server-side to their own sectors and states.
• Insurance and lending models. NPL proximity, cleanup status and coordinates feed liability scoring at address level rather than ZIP level.
• Market sizing for remediation services. Counting sites by NPL status, state and cleanup type sizes the addressable market for state-lead and other non-NPL cleanup work directly from the status fields.
• Journalism and academic replication. Row-level public-regulatory records mean any published total can be recomputed independently - the same rows EPA's own tools summarize.

Data scientists get row-level regulatory records clean enough to model on - see environmental & facilities services data for data scientists. Developers and builders get a documented query surface over nine program families without building nine integrations - see developers builders use cases. Journalists and academics get reproducible official numbers straight from program tables - see journalists academics use cases. Competitive intelligence teams track which facilities and corporate parents appear in new compliance activity - see competitive intel product teams use cases. Market researchers size remediation and compliance-service demand from status and geography fields - see market researchers use cases. Sales and growth teams build territory lists from program, state and ZIP filters - see sales growth teams use cases.

Because the useful test is a join across program families, not a glance at one table. A sample comes back cut to the states, ZIPs, programs or corporate parents you name, with identifiers intact so you can verify immediately that a TRI facility resolves to its FRS registry record and that SEMS sites near your targets carry the NPL statuses you expect. Test whether your site list, customer accounts or portfolio holdings key cleanly onto EPA's identifiers before anything reaches production. Start small: get a sample of this dataset scoped to the slice you will actually use, or browse the best environmental facilities services datasets for companions.