Potential environmental release indicator

GA oil/hazmat spill & release incident reports from the U.S. Coast Guard National Response Center (NRC) — FEDERAL SUBSTITUTE for Integration 101 (SPILLS/GA/ga_dnr), since GA EPD's state Emergency-Response 'Environmental Releases' DB is not openly bulk-available (query-only/internal). fetch.py downloads NRC's yearly XLSX workbooks (nrc.uscg.mil/FOIAFiles/CY<YY>.xlsx, CY1990–CY2025), reads INCIDENT_COMMONS, filters LOCATION_STATE='GA', derives decimal lat/lon from DMS → 19,679 incidents. conform.py maps ds_ address from LOCATION_STREET1/NEAREST_CITY/STATE/ZIP and ds_lat/lon from the derived coords; ds_SPILLS=true. Located mainly by city+county (ds_city 97.7%, ds_state 100%); ds_street 20.1% and ds_lat/lon 2.6%.

Cumulative Alabama subset of National Response Center (NRC) oil/HazMat spill-release incident reports, 1990-2024 = 20,823 incidents. Sourced from NRC because Alabama DEM (the Integration's assigned provider) publishes NO public spill-incident dataset — ADEM's Emergency Response page only gives reporting phone numbers and directs reporters to the NRC. NRC publishes one XLSX per year (nrc.uscg.mil/FOIAFiles/CY<YY>.xlsx); fetch.py reads the INCIDENT_COMMONS sheet, filters LOCATION_STATE='AL', dedups by SEQNOS, converts DMS lat/lon to decimal. ds_street 77.9% (from LOCATION_ADDRESS) / ds_city 91.7% / ds_lat-lon 9.0% → Passable (street-driven). Provider = us_uscg_nrc (federal) — mismatch with al_adem, by necessity.

Arizona subset of the National Response Center (NRC) incident-report archive, 1990-2025 = 6,933 oil/chemical/hazmat release reports. PROVIDER DIVERGENCE: Integration #97 is keyed to az_deq, but Arizona DEQ publishes no comprehensive public spill dataset (Hazardous Material Incident Logbook frozen Nov 2001; legacy.azdeq.gov blocked) and explicitly redirects post-2001 records to the federal NRC. Source owned by us_uscg_nrc and linked to the az_deq Integration. Built from per-year NRC workbooks (nrc.uscg.mil/FOIAFiles/CY<YY>.xlsx), INCIDENT_COMMONS sheet filtered to LOCATION_STATE='AZ'. Single-attribute (SPILLS). Address-described, coordinate-sparse: ds_street 81.8% (LOCATION_ADDRESS), ds_city 97%, ds_state=AZ, ds_lat/lon 3.2% (DMS-assembled) -> Passable on ds_street.

Kentucky subset of the National Response Center (NRC) incident-report archive, 1990-2025 = 16,411 oil/chemical/hazmat release reports. PROVIDER DIVERGENCE: Integration #105 is keyed to ky_dep, but Kentucky EEC publishes no public spill/incident dataset (ER Branch page is contact-only; eSearch exposes only Agency Interest/permits/documents; EEC Data Access Tool lists no spill route; no EEC ArcGIS layer). Source owned by us_uscg_nrc and linked to the ky_dep Integration — same pattern as us_uscg_nrc__incident_reports_az (#97). Built from per-year NRC workbooks (nrc.uscg.mil/FOIAFiles/CY<YY>.xlsx), INCIDENT_COMMONS filtered to LOCATION_STATE='KY'. Single-attribute (SPILLS). Address-described, coordinate-sparse: ds_street 83.7% (LOCATION_ADDRESS), ds_city 97%, ds_state=KY, ds_lat/lon 1.8% (DMS-assembled) -> Passable on ds_street.

Louisiana subset of the federal National Response Center (NRC) oil/chemical/hazmat incident-report archive (USCG). Full archive 1990-2025, filtered to LOCATION_STATE='LA' from the per-year NRC FOIA workbooks (INCIDENT_COMMONS sheet). 144,173 reports. Provider divergence: Integration 106 is keyed to la_deq, but Louisiana DEQ's Single Point of Contact (SPOC) is an intake-only system with no public dataset, so the authoritative source is the federal NRC (us_uscg_nrc). Each record: incident date, type, cause, description, location address/city/county/zip, and (where reported) lat/lon assembled from NRC DMS components. NRC posts initial, unvalidated reports.

Minnesota subset of the U.S. Coast Guard National Response Center (NRC) incident-report archive — petroleum and hazardous-material spill/release reports, 1990–2025 (per-year FOIA workbooks, INCIDENT_COMMONS sheet, filtered to LOCATION_STATE='MN'). 8,633 reports. Federal-substitute source: the Minnesota Pollution Control Agency's own Emergency Response / MERTS spill-incident database is not published for bulk download, so this Source is owned by us_uscg_nrc and linked to the mn_pca Integration #107 (provider divergence). Single-attribute: SPILLS.

Missouri subset of the federal National Response Center (NRC) oil/chemical/hazmat incident-report archive (USCG). Full archive 1990-2025, filtered to LOCATION_STATE='MO' from the per-year NRC FOIA workbooks (INCIDENT_COMMONS sheet). 12,972 reports. Provider divergence: Integration 108 is keyed to mo_dnr, but MO DNR's Environmental Emergency Response tracking systems (MEERTS ≤Oct 2021; MERLIN since Nov 2021) are search-only web interfaces with no bulk download/API, so the authoritative downloadable source is the federal NRC (us_uscg_nrc). Scope caveat: NRC captures the federally-reported subset of MO's EER universe (which also logs state-spill-line-only reports), so this ~13k runs below the ~25k full-state-DB prediction. Each record: incident date, type, cause, description, location address/city/county/zip, and (rarely) lat/lon. NRC posts initial, unvalidated reports.

Cumulative New Jersey subset of National Response Center (NRC) oil/HazMat spill-release incident reports, 1990-2024 = 20,792 incidents. Sourced from NRC because the NJDEP-native spill data (Communications Center / Environmental Action Line in NJEMS DataMiner) is not programmatically sourceable: DataMiner is behind Imperva Advanced Bot Protection AND its record-level report times out/resets on export, with no ArcGIS/Open-Data mirror. NRC publishes one XLSX per year (nrc.uscg.mil/FOIAFiles/CY<YY>.xlsx); fetch.py reads INCIDENT_COMMONS, filters LOCATION_STATE='NJ', dedups by SEQNOS, converts DMS lat/lon to decimal. ds_street 87.8% (from LOCATION_ADDRESS) / ds_city 97.4% / ds_lat-lon 3.4% → Passable (street-driven). Provider = us_uscg_nrc (federal) — substitution for nj_dep, same as AL #96 / AZ #97.

Pennsylvania subset of the National Response Center (NRC) incident-report archive, 1990-2025 = 28,451 oil/chemical/hazmat release reports. PROVIDER DIVERGENCE: Integration #113 is keyed to pa_dep, but PA publishes no broad public hazmat-incident dataset (PEIRS is PEMA/internal; eFACTS is storage-tank-only; no incidents on data.pa.gov; no PASDA incidents layer). Source owned by us_uscg_nrc and linked to the pa_dep Integration — same pattern as AZ #97 / KY #105. Built from per-year NRC workbooks (nrc.uscg.mil/FOIAFiles/CY<YY>.xlsx), INCIDENT_COMMONS filtered to LOCATION_STATE='PA'. Single-attribute (SPILLS). Address-described, coordinate-sparse: ds_street 86.2% (LOCATION_ADDRESS), ds_city 97.8%, ds_state=PA, ds_lat/lon 1.8% (DMS-assembled) -> Passable on ds_street.

South Carolina subset of the federal National Response Center (NRC) oil/chemical/hazmat incident-report archive (USCG). Full archive 1990-2025, filtered to LOCATION_STATE='SC' from the per-year NRC FOIA workbooks (INCIDENT_COMMONS sheet). 12,596 reports. Provider divergence: Integration 114 is keyed to sc_des, but SC DES's Oil & Chemical Emergency Response division takes reports by phone and publishes no public spill dataset (no data.sc.gov dataset; gis.des.sc.gov Emergency_Response folder has only a physiographic map; SC DHEC AGOL org has only aggregated count layers), so the authoritative downloadable source is the federal NRC (us_uscg_nrc). Each record: incident date, type, cause, description, location address/city/county/zip, and (rarely) lat/lon. NRC posts initial, unvalidated reports.

Cumulative Wisconsin subset of National Response Center (NRC) oil/HazMat spill-release incident reports, 1990-2024 = 6,754 incidents. Sourced from NRC because WI DNR's own spill data (BRRTS activity code 04) is not programmatically sourceable: the full archive lives only in the BRRTS bulk zip (apps.dnr.wi.gov, US-geo-locked + binary, unreachable/un-proxyable here), WI DNR publishes no spills GIS layer (entire RR_Sites_Map ArcGIS enumerated — only ERP/LUST activity points), and the open-spills web table is open-only + county-level. NRC publishes one XLSX per year (nrc.uscg.mil/FOIAFiles/CY<YY>.xlsx); fetch.py reads INCIDENT_COMMONS, filters LOCATION_STATE='WI', dedups by SEQNOS, DMS lat/lon to decimal. ds_street 85.5% / ds_city 97.9% / ds_lat-lon 2.8% → Passable. BELOW predicted band (13,500-46,500) by design: NRC = federally-reported subset, not the full BRRTS state archive. Provider = us_uscg_nrc (federal) — substitution for wi_dnr, same as AL #96 / NJ #110 / AZ #97.

Nationwide, unfiltered archive of U.S. Coast Guard National Response Center (NRC) oil/hazardous-material release-notification reports, CY1990-CY2025 (1,032,643 rows). Same upstream yearly workbooks (nrc.uscg.mil/FOIAFiles/CY<YY>.xlsx, INCIDENT_COMMONS sheet) that back the 12 state-filtered us_uscg_nrc__incident_reports_<state> SPILLS sources, kept here unfiltered by state/territory to serve Integration 331 (ERNS/US). Built after confirming no independent EPA ERNS dataset still exists separately from the NRC's own database (EPA's ERNS was structurally just EPA's copy of NRC-reported notifications; no live Envirofacts table or downloadable archive remains). Known upstream limitation: the CY2017-CY2025 yearly files are each truncated by the NRC server to Jan 1-Sep 9 of that year (not a fetch.py bug) — see the .md for details.