EPA ECHO exporter data: one compliance row per regulated facility, delivered

The record is not merely big; it is unusually well built. It scores a perfect 10 on Datadory's rubric - one of only 145 perfect scores across the 1,744 datasets we catalog - and EPA is the heaviest publisher in this slice, contributing four of the sixteen primary Environmental & Facilities Services records. When one agency supplies that much of the shelf, the differentiator stops being location and becomes shape: who hands you typed, decoded, joinable rows.

Flat, wide, and readable without a decoder ring. One facility row, then its handler-side companion:

# echo exporter facility row - each value is the documented example for its field
REGISTRY_ID   : 110000490166
FAC_NAME      : ACME LANDFILL LLC
FAC_ADDRESS   : 1200 WEST CITY RANCH ROAD, PALMDALE, CA 93550  (LOS ANGELES county)
FAC_LAT/LONG  : 34.589, -118.100   (NAD83 decimal degrees)
RCRA_FLAG     : Y          RCRA_IDS: CAD982042076     PERMIT_TYPES: TSDF

# five-year compliance window riding on the same row
INSPECTIONS   : 4          INFORMAL ACTIONS: 1        FORMAL ACTIONS: 2
PENALTIES     : 25000      LAST PENALTY: 2021-06-15
COMPL STATUS  : V          SNC FLAG: N                3YR QUARTERS: VVV_

# handler-side rcrainfo tables, joined on ID_NUMBER
ID_NUMBER              : VTD000509174
FED_WASTE_GENERATOR    : 1            (large quantity generator)
OPERATING_TSDF         : LIT          (land disposal, incinerator, treatment)
VIOLATION_TYPE         : 265          (TSDF standards, 40 CFR part 265)
VIOLATION DETERMINED   : 2019-03-04

Three layers, one join path. Identity and geography ride the Facility Registry Service spine, so coordinates arrive in a single datum (NAD83) and the same physical site keeps one registry ID across programs. The five-year window compresses inspections, escalation and money onto the same row - RCRA_3YR_COMPL_QTRS_HISTORY reads VVV_ as three compliant quarters followed by a blank current quarter. And the handler tables hang off ID_NUMBER, so a facility screen and an event-level audit share one key instead of a reconciliation project.

Every value above is the documented example for its field - the definitions ship with the delivery, so nothing has to be guessed from a column name.

Identity comes first, enforcement follows. Ten fields do most of the analytical work:

The informal/formal split is the quiet advantage. A facility with two formal actions reads very differently from one with eight letters, and most summaries flatten that distinction away. Here it survives as two countable columns, which is what makes drift-over-time features possible without re-deriving anything.

Two grains, two jobs. The Exporter answers "what does this facility's compliance posture look like" - one row, readable in a screen. The handler-level tables answer "show me every record" - one row per evaluation, violation and enforcement action, which is what hazardous waste prospecting and litigation-grade risk work actually needs.

The handler side classifies generators in FED_WASTE_GENERATOR: 1 is a Large Quantity Generator, 2 a Small Quantity Generator, 3 Conditionally Exempt, N not a generator. OPERATING_TSDF marks operating treatment, storage and disposal sub-universes - L land disposal, I incinerator, B boiler or industrial furnace, S storage, T treatment, H solid waste management. Violations carry the CFR area violated in VIOLATION_TYPE (Part 262 for generators, 263 for transporters, 264 and 265 for TSDFs) plus the date the violation was determined.

Practical rule: screens start wide, prospect lists start deep, and both start from the same keys. Anyone building a generator-class target list begins in the handler tables precisely because they keep the events instead of summarizing them - and because the two grains meet on ID_NUMBER, moving between them is a join rather than a rebuild.

Four workflows recur across the teams we deliver to.

Facility-level risk panels. Join the five-year window onto handler IDs and coordinates to score industrial assets on inspections, action counts, penalties and noncomplier status - the backbone of ESG screens and counterparty review (data scientists).

Enforcement-trend screening. Watching escalation rather than presence: an operator drifting from letters to orders to penalties shows up as a sequence, not a flag - an operating-risk signal quant investors read ahead of the filings (investors and quants).

Territory planning for compliance services. More than 1.5 million regulated facilities segmented by program, compliance status and geography rank outreach territories for consulting, testing, remediation and waste-handling vendors (sales and growth teams).

Due diligence and accountability journalism. Violations, penalty dates and determinations attach to named target assets ahead of transactions - the fastest way to find the landfill history nobody volunteered, and a citable federal record when the story publishes.

One join rule governs all four: identity is stable, geography is self-contained, and time is explicit. Rows arrive shaped for panels, not for eyeballing.

The Exporter is the compliance core, not the whole picture, and the rest of the pool fills in around it by question rather than by accident.

When a job needs slices instead of the full universe - one program family, one state, one year - the EPA Envirofacts Web Services record covers the same program universe at query grain across TRI, SEMS and ICIS tables. When the question turns to what facilities emit rather than how they behave, the EPA Greenhouse Gas Reporting Program data adds annual facility-level emissions for roughly 8,000 reporters across 32 industry types, reporting years 2010 through 2023. For contaminated-site diligence, EPA Superfund data and reports (SEMS) exposes 27,201 document records with NPL inventories and site boundaries. And across the Atlantic, the UK Defra Data Services Platform carries the equivalent permits, waste operations and carrier registers - so transatlantic screens need both stores.

The layers answer different questions: ECHO fixes behavior, GHGRP fixes emissions, SEMS fixes legacy liability, Defra fixes the UK half. None substitutes for another, which is why they land joined on shared keys rather than as four separate projects.

Files, feeds, or straight into your warehouse. Daily, weekly, or hourly - your call.

The shaping work happens once, upstream of you: resolving program codes onto readable labels, keeping every coordinate in one datum, stitching registry and handler IDs so facility views and event views agree, and flagging the pre-2000 records whose quality was never assessed instead of shipping them looking clean. The wide 130-plus-column record gets reshaped into something a BI tool tolerates, and the field dictionary travels alongside every pull, so the schema in your sample is the schema you ship against. Start with a sample cut to your own segments and states - real rows before anything recurring starts.