U.S. Environmental Protection Agency

Texas AFFF sites (AQUEOUS_FOAM) = 35 federal (DoD/DOE) installations with PFAS presence (from Aqueous Film Forming Foam use), from EPA's PFAS Analytic Tools 'Federal Sites' layer (PAT_Federal_Sites, FeatureServer/13) filtered to State='TX'. Sourced federally because TX TCEQ maintains NO dedicated AFFF inventory (confirmed: no PFAS/AFFF layer on gisweb.tceq.texas.gov, none on data.texas.gov). Same source family as the Verified us_epa__pfas_afff / us_epa__pfas_npl_sites / us_epa__pfas_tsca sources. Coordinate-only (installation centroids; layer has no street/city/zip): ds_lat/lon 100% (all in TX bbox) → Fit; ds_state='TX' constant. Composition: Air Force 12, Army 13, Navy 7, DOE 3; PFAS_Presence Known 16 / Suspected 14 / Pending 2 / blank 3. Within predicted band 15-85. DoD-only — excludes civilian airport/fire-training AFFF sites (TX publishes none).

EPA's quarterly Facility Registry Service Interests download exposes the SEMS layer as an ESRI file geodatabase containing active, non-archived Superfund sites nationwide. The integrated product emits normalized address and NAD83 coordinate columns, a SEMS indicator, and the underlying facility identity, location, NPL status, federal ownership, EPA region, and program-interest metadata. It can be consumed as a complete bulk layer or queried through the documented ArcGIS FeatureServer fallback.

EPA's PFAS Analytic Tools Superfund layer identifies National Priorities List sites with reported PFAS detections. The integrated product filters the ArcGIS layer to NPL_Site=Yes and emits 464 site records with EPA identifiers, names, addresses, coordinates, NPL status, site type, program context, and profile links. It preserves the 18 fetched ArcGIS attributes and adds seven normalized location and PFAS indicator columns.

EPA's SEMS Archived Site Inventory identifies former CERCLIS-NFRAP sites removed from active Superfund tracking where no further federal remedial action is planned unless new information changes that determination. The integrated Envirofacts API product emits nationwide site identifiers, names, address and jurisdiction fields, limited coordinates, NPL and non-NPL status, federal-facility and Superfund Alternative Approach indicators, archive dates, standardized address columns, and a SEMS_ARCHIVE flag.

EPA's FOIA-12 full-file report identifies Superfund property-lien activity by site. The integrated output flattens the XLSX attachment embedded in the SEMS PDF into one row per lien, emitting site identifiers, lien lifecycle dates, available property-address parts, associated parties, and an explicit NPL_LIENS indicator. It contains 383 liens at 352 sites.

EPA's weekly FRS ICIS-Air ArcGIS layer is the current successor feed for legacy AIRS/AFS and covers stationary air-pollution-source facilities across the United States, territories, and tribal areas. The integration queries the complete layer and emits facility identity, program identifiers, air-interest classification and active status, address, EPA region, and NAD83 coordinates. It also adds normalized address and location columns plus an AIRS/AFS membership flag that is true for every row.

EPA's nationwide CORRACTS extract identifies facilities subject to Resource Conservation and Recovery Act corrective-action authority under sections 3004(u) and 3004(v). The primary CIMC extract emits facility identifiers and names plus EPA region, state, county, city, and ZIP code; the integrated CSV adds normalized address columns, enriched coordinates, and a CORRACTS indicator while preserving the source columns. The pre-verified integrated file contains 3,973 facility records and is released quarterly.

EPA's Federal Agency Hazardous Waste Compliance Docket is a quarterly public master list of federal facilities that manage or have managed hazardous waste, or where hazardous substances have been released. The FEDFacts Qlik Engine API emits one row per docket facility with EPA identifier, facility and agency identity, location, reporting mechanism, docket status, NPL/non-NPL cleanup classifications, and date added. The integrated conformed output also emits standardized address columns and a FEDERAL_FACILITY flag for all 2,392 records.

A Maryland-only, facility-grain extract of EPA's national UST Finder tool, filtered server-side to tanks with an Installation_Date before 1997-01-01 (the frozen legacy HIST_UST registry MDE stopped maintaining once the 1997 re-registration effort moved surviving tanks into the current UST system). Built as a two-source join against the same ArcGIS FeatureServer item already used by the existing us_epa__ust_finder Source: the USTs table (tank grain, 42,391 rows / 14,423 distinct Facility_IDs) supplies the date filter, and the Facilities layer supplies address/lat-lon, joined on Facility_ID and deduplicated to one row per facility. The join resolves only 10,418 of 14,423 facilities (72.23% fill on both ds_street and ds_lat/ds_lon), a Strong Concern fit rating under the agent contract's thresholds. This was confirmed to be a genuine coverage gap between the historical tank registry and the current Facilities snapshot, not a fetch defect: the older US_Underground_Storage_Tank_2019 FeatureServer item was checked as an alternative join target and resolves the exact same 10,418/14,423 facility set, so the gap is not fixable by switching EPA data-source vintages. Accepted by the developer with the Strong Concern rating on record.

EPA UST Finder ArcGIS FeatureServer, two layers: Facilities (layer 0, UST_FINDER/Integration 299) and Releases (layer 1, UST_FINDER_RELEASE/Integration 300). Static 2018-2021 EPA snapshot, intentionally distinct from the ~19 existing Verified per-state UST/LUST sources — nationwide baseline for states without an individual per-state Source. Deduped on full-row content per layer (not the declared PK alone) after investigating shared-key groups: Facilities dropped 8 true duplicates of 108 shared Facility_ID pairs (100 genuinely distinct facilities kept); Releases dropped 146 true duplicates of 466 shared (Facility_ID, LUST_ID) groups (370 genuinely distinct events kept). Tank-level USTs layer (2.3M rows, no own address/coords) and the two *_by_County polygon-aggregate layers deliberately out of scope — no clean Attribute mapping.

EPA's national e-Manifest system (RCRAInfo HWIP) bulk CSV export of the core EM_MANIFEST table -- one row per hazardous-waste manifest (shipment) tracked since the system's June 2018 launch. Public, unauthenticated, refreshed weekly. fetch.py resolves the current weekly-rotating download URL via EPA's own JSON export/summaries API rather than hardcoding a dated path. ds_street/city/state/zip are derived from the GENERATOR's PHYSICAL LOCATION address (not mailing, not the designated facility). CRITICAL: ds_lat/ds_lon fill is 0% -- this dataset has no coordinates at all. Records must be geocoded downstream using the Generator Physical Address.

Arizona drycleaner facilities from EPA RCRAInfo (hazardous-waste handlers), served via the ECHO REST API and filtered to the drycleaning/laundry NAICS family 8123 — 295 facilities statewide. Used because Arizona has no dedicated drycleaner program and ADEQ publishes no accessible drycleaner facility list (azdeq.gov + its GIS server block external access; the ADEQ open-data hub has no drycleaner layer). EPA RCRAInfo is the authoritative accessible registry and is how EPA's own dry-cleaning sector program identifies PCE dry cleaners (by drycleaning NAICS among hazardous-waste generators). Carries facility name, handler ID, address, ZIP, county, and EPA WGS84 coordinates.

Delaware drycleaner facilities from EPA RCRAInfo (via ECHO), filtered to drycleaning NAICS 812310/812320 (coin-op + non-coin-op drycleaners; excludes linen/industrial launderers). Statewide; 86 facilities; EPA-provided addresses and coordinates; all solvent types.

Massachusetts drycleaner facilities from EPA RCRAInfo via the ECHO REST API. All 35,593 MA RCRA hazardous-waste handlers are fetched (get_facilities -> QueryID -> get_download CSV) and filtered client-side to the drycleaning/laundry NAICS 8123 family -> 400 facilities (364 core drycleaning 812320/81232, plus coin-op, industrial launderers, linen supply). Recorded under provider us_epa rather than ma_dep because MassDEP publishes no accessible drycleaner roster: the ERP PERC drycleaners (310 CMR 70/72) file via eDEP (a filing system, not a published list), MassGIS exposes no dry-cleaner/air-ERP layer, and www.mass.gov is bot-protected (403). Mirrors the Verified us_epa__rcra_az_drycleaners / us_epa__rcra_md_drycleaners / us_epa__rcra_de_drycleaners sources. ds_ fill 100% across address + coordinates (Fit); EPA WGS84 coords, never geocoded.

Maryland drycleaning facilities sourced from EPA RCRAInfo via the ECHO REST API — all MD RCRA hazardous-waste handlers in the drycleaning/laundry NAICS family (8123), 97% of which are core drycleaning (812320). MDE publishes no accessible registered-drycleaner list (its Air Emissions ArcGIS layer holds only 3 dry-cleaning facilities; the ARMA roster is PDF-only), so this federal mirror recorded under us_epa is the authoritative accessible registry — mirroring the Verified us_epa__rcra_az_drycleaners. 737 facilities, fetched 2026-06-16; 100% address & coordinate fill (Fit).

Missouri drycleaning facilities sourced from EPA RCRAInfo via the ECHO REST API — all MO RCRA hazardous-waste handlers in the drycleaning/laundry NAICS family (8123), 95% of which are core drycleaning (812320). MO DNR publishes no accessible current registry (the Drycleaning Environmental Response Trust Fund expired in 2017; dnr.mo.gov returns HTTP 403 to us), so this federal mirror recorded under us_epa is the authoritative accessible registry — mirroring the Verified us_epa__rcra_az_drycleaners. 557 facilities, fetched 2026-06-22; 100% address & coordinate fill (Fit).

Mississippi drycleaner/laundry facilities derived from EPA RCRAInfo (served via the public ECHO REST API), filtered to MS hazardous-waste handlers classified as drycleaners by authoritative industry code. MS DEQ publishes no accessible dedicated drycleaner list (only the enSearch Online portal, no bulk download), so EPA RCRAInfo is used as the accessible authoritative registry, recorded under us_epa consistent with the AZ (Integration 159) and DE drycleaner sources. Selection is code-only: NAICS 8123 family OR drycleaning/laundry SIC family (7211-7219, excl. 7217 carpet/upholstery). 31 facilities, fetched 2026-06-17; 100% street and lat/lon fill (Fit). NOTE: partial registry by design — MS RCRAInfo leaves ~55% of NAICS and ~81% of SIC blank, so MS drycleaners EPA never coded are not captured; comprehensive coverage would need an authoritative MS-native list.

EPA's TSCA Chemical Data Reporting (CDR) Manufacture-Import Information file (2024 reporting cycle) — site-level records of manufacturers and importers of chemicals on the TSCA Chemical Substance Inventory, including production volume, site NAICS/industry classification, and site address/coordinates per plant site. Published every four years as a non-confidential (non-CBI) CSV extract from EPA's Access CDR Data portal. Approximately 14% of rows have the entire site identity (name, address, coordinates) redacted under a Confidential Business Information (CBI) claim; the chemical/production-volume data for those rows is still present.

National inventory of wastewater treatment plants from EPA's Clean Watersheds Needs Survey (CWNS) 2022, published April 2024. 16,454 treatment plants across 56 states and territories, built by filtering FACILITY_TYPES.csv to FACILITY_TYPE='Treatment Plant' (deduplicated on CWNS_ID: 22,787 rows collapse to 16,454 distinct facilities) and joining PHYSICAL_LOCATION.csv for address and coordinates plus FACILITIES.csv for name, owner and Superfund cross-reference. Coordinates are true plant locations at 100% fill; street 89.9% (rural plants are often described by legal/landmark descriptions rather than postal addresses) - Fit on the better-of-two rule. CWNS surveys publicly owned wastewater infrastructure, so it is a census of POTWs (Public 16,394 / Private 47 / Federal 13) rather than of private industrial treatment plants.

EPA ICIS-NPDES facility data via the public ECHO CWA REST API, filtered per-state (fetch.py loops over a STATES list: AK, CT). Both AK DEC and CT DEEP report permit actions into this federal system under their delegated APDES/NPDES primacy, and neither state publishes a comparably complete NPDES permit list on its own site. Same endpoint generalizes to additional states by appending to STATES.

EPA's PFAS Transfers methodology reimplemented against public bulk e-Manifest CSVs (EM_MANIFEST + EM_WASTE_LINE join) -- one row per manifest whose waste-line description matches a PFAS keyword (PFAS/PFOA/PFOS/PERFL/AFFF/GENX). 10,335 manifests nationwide. Only 2 of the 5 fields named in EPA's own methodology exist in the public bulk export (the other 3 appear internal-only), so this Source systematically undercounts relative to EPA's closed PFAS Analytic Tools dashboard -- documented as a known completeness ceiling.

EPA ECHO's national RCRA Facilities bulk file (RCRA_FACILITIES.csv, extracted from the rcra_downloads.zip bundle). One row per RCRA handler nationwide, keyed on ID_NUMBER. Multi-attribute source: FED_WASTE_GENERATOR and HREPORT_UNIVERSE_RECORD together determine generator-status/TSDF classification, satisfying five Attributes (RCRA_TSDF, RCRA_LQG, RCRA_SQG, RCRA_CESQG, RCRA_NLR) across Integrations 185/186/187/188/335. 1,613,224 rows as of 2026-08-03 fetch. Fill rates: ds_street 99.9%, ds_lat/ds_lon 93.5% (Fit rating: Fit).

EPA's Coal Combustion Residuals (CCR) Comprehensive Compliance Report, obtained by a FOIA request to EPA's National FOIA Office (Case No. 2026-EPA-06323, "Granted in Full") and delivered as Comprehensive_Compliance_Report_20250611.xlsx (783 CCR units nationwide, one row per unit, 150 compliance columns). Filtered to the Customer's scope — Unit Type = Surface Impoundment AND Hazard Potential Classification = High — yielding 49 high-hazard surface impoundments at 34 plants. This is the current authoritative equivalent of EPA's retired archive.epa.gov High-Hazard list (50 units / 32 facilities). One row per unit; coordinates (Latitude/Longitude) 100%, no site street address published. Single-Attribute (COAL_ASH_EPA) source. Refresh = a new FOIA request.

EPA Facility Registry Service (FRS) state single-file CSV, filtered to petroleum bulk-storage industry classifications, serving AST integrations in states that operate no AST registry. Filter: NAICS 424710 (petroleum bulk stations & terminals), 454310 (fuel dealers) or 324110 (petroleum refineries), OR SIC prefix 5171/5172/5983/5984/2911. Illinois slice = 1,575 facilities from 186,053 IL FRS rows. Scoped to STATES = ["IL"]: the same recipe yields only 81 rows for AZ and 429 for MD, both far below their integration bands, so this is NOT a general national AST answer. Federal proxy for a state attribute — same provider-divergence pattern as us_doe__petroleum_terminals serving SC AST #257.

EPA Superfund "Noticed Parties at Sites in SEMS" report (FOIA-11) — the agency's public listing of Potentially Responsible Parties (PRPs): every party that received a general or special notice letter for a site tracked in SEMS. Published as a single large PDF in SEMS collection 33301 that wraps an embedded XLSX (documentId resolved dynamically each run; 406302 as of 2026-06-11). One row per site × party × notice-action: ~144,983 rows across 4,819 Superfund sites and 78,589 distinct parties, all 10 EPA regions. ds_ address/location columns are mapped from the contaminated SITE columns (≈100% street fill); the party's own mailing address is preserved as raw columns.

EPA's Fifth Unregulated Contaminant Monitoring Rule (UCMR 5) occurrence data, filtered to Texas public water systems. A federal, EPA-run PFAS + lithium monitoring program covering community water systems and large non-transient non-community systems during a single 12-month window between 2023 and 2025. Sourced from EPA (not TCEQ) because TCEQ's own Drinking Water Viewer was confirmed to have zero PFAS sample data loaded for any Texas system. Yields 214,143 analyte-sample rows (206,808 PFAS-analyte rows + 7,335 lithium rows) covering 1,153 distinct Texas PWSIDs, with 6,346 quantified PFAS detections.

Facilities reporting Total Phosphorus (NPDES parameter code 00665) on Discharge Monitoring Reports, from EPA's ECHO Water Pollutant Loading Tool (ICIS-NPDES). Harvested at the monitoring-period grain (get_custom_data_monitor_pd) state by state and rolled up to one row per NPDES permit, which yields a genuine per-facility count of TP reports (tp_dmr_record_count, range 1-888). 22,227 facilities across 53 states from 185,782 TP DMR records for reporting year 2024. Address and coordinates arrive on the same response with no join needed; both axes are 100% filled (Fit). One record = one NPDES-permitted facility that actually submitted TP data that year, so this is a reporting universe rather than a permit universe.

EPA's 2017 'Lead Smelter Strategy Summary Report' (Office of Land and Emergency Management), Appendix A -- the full 464-site universe of historic secondary lead smelters tracked under EPA's 2012 Lead Smelter Strategy (Eckel et al. 2001 study). Extracted from PDF via bounding-box table parsing. Only City+State location granularity is published upstream -- no street address or coordinates exist in the source document.

EPA's PFAS Analytic Tools 'Fire Training Sites' spreadsheet -- a facility-name keyword search over ECHO permitted facilities, identifying sites where firefighting foam may have been used in training exercises. 459 facilities nationwide. Static file, current as of January 2021; confirmed no live ArcGIS equivalent exists (the live PFAS_Analytic_Tools_Layers Industry field has zero Fire Training records). No street address or ZIP published upstream.

The Water Quality Portal's PFAS ambient sampling results (joint USGS/EPA/state/tribal warehouse), filtered via WQP's own characteristicType=PFAS preset. 387,690 sampling results across 9,778 monitoring locations nationwide, joined with Station data for coordinates. 4 of 155 distinct analytes found not to be genuine PFAS and explicitly forced to ds_PFAS_WQP=false.

EPA CLU-IN 'Drycleaner Site Profiles' database — a curated set of 216 detailed remediation case-study profiles for dry cleaner contamination sites across 18 states, originally compiled under the now-defunct State Coalition for Remediation of Drycleaners (SCRD). Each profile covers site description, contaminants, remediation technologies, costs, and lessons learned, sourced from a static XML feed. SCRD's own site (drycleancoalition.org) is dead; this CLU-IN page is the only surviving machine-readable SCRD dataset. No street address or coordinates are published for any profile.

Superfund (SEMS) sites where Total Petroleum Hydrocarbons is a named contaminant of concern, from EPA Envirofacts. A single dmapservice request joins sems.envirofacts_contaminants to sems.envirofacts_site, so each contaminant record arrives with its site address and coordinates. Selection is on the authoritative preferred_contaminant_name containing 'PETROLEUM HYDROCARBON', capturing the whole TPH analyte family (TPH, TPH-Diesel, TPH-Gasoline, TRPH, TEPH); deliberately not broadened to 'PETROLEUM', which would add petroleum products rather than the TPH analyte class. 156 contaminant records roll up to 61 sites across 21 states, with a per-site count (tph_contaminant_record_count, range 1-19). Both address and coordinate axes are 100% filled (Fit). The universe is small by construction, not by sourcing failure: CERCLA section 101(14)'s petroleum exclusion keeps TPH out of most federal Superfund decision documents, so TPH-as-contaminant-of-concern lives overwhelmingly in state LUST and voluntary-cleanup data. Treat as the complete federal picture and a supplement to per-state TPH coverage.

EPA's quarterly, nationwide ArcGIS dataset identifies Superfund sites that are listed on, proposed for, or deleted from the National Priorities List. The integrated product emits NPL, PROPOSED_NPL, and DELISTED_NPL status indicators for approximately 1,840 records and supplies WGS84 coordinates and city/state location context, but no street address or postal code. EPA states that the upstream layer is sourced from Hazard Ranking System and NPL listing documentation; the candidate notes that the integrated source is sourced from SEMS.