us-county-airdata-trends 2026.5.26

U.S. county-level 5-year AirData emission trends from EPA AirData annual AQI summaries 2020-2024. Sensitivity-gated change class with facility-count floor, top-1-exclude robustness, and 4 petrochemical-corridor display-hold counties. 994 counties.

us-county-airdata-trends

U.S. county-level 5-year AirData emission trends, derived from EPA AirData annual AQI summaries 2020-2024.

Sensitivity-gated 5-year change classification for 994 US counties, with explicit display-hold flags for counties below the facility-count floor (368 counties) and the 4 petrochemical-corridor counties that surface a methodology-review notice instead of a trend value.

Produced by ZipCheckup Trend Layers v1. Display methodology underwent a 4-agent legal pre-build review (FTC compliance + defamation/stigmatization + statistical methodology + feeds feasibility) — see methodology page.

---

Install

pip install us-county-airdata-trends

Quick start

import us_county_airdata_trends as airdata

# Look up a single county
la = airdata.get_county("06037")
# {
#     "county_fips": "06037",
#     "county_name": "Los Angeles",
#     "state": "CA",
#     "airdata_change_class": "stable",
#     "airdata_pct_change": -5.9,
#     "cycles_used": 5,
#     "facility_count": 331,
#     "sensitivity_robust": None,
#     "skip_reason": None,
#     "source_attribution": "EPA AirData annual AQI summaries 2020-2024",
#     "petrochemical_corridor": False,
#     "aqi_latest_year": 2024,
# }

# All counties for a state
texas = airdata.filter_counties(state="TX")

# Counties classified as decreasing (and display-eligible)
decreasing = airdata.filter_counties(change_class="decrease", exclude_skipped=True)

# Display-eligible vs display-held — for methodology dashboards
eligible = airdata.get_display_eligible_counties()
held = airdata.get_display_held_counties()

# Coverage metadata
print(airdata.meta)
# Meta(generated='2026-05-25', total_counties=994, license='CC-BY-4.0')

# Field schema
print(airdata.schema())

Field schema

Field	Type	Description
county_fips	str	5-digit US county FIPS code
county_name	str | None	County name without "County" suffix
state	str | None	2-letter state abbreviation
airdata_change_class	"decrease" | "increase" | "stable" | "insufficient_data" | None	Sensitivity-gated 5-year class
airdata_pct_change	float | None	Signed percent change between earliest and latest cycle
cycles_used	int | None	Number of reporting cycles (target 5, minimum 3)
facility_count	int | None	Distinct facilities in latest cycle (display gate ≥5)
sensitivity_robust	bool | None	True if direction stable when top-1 facility excluded
skip_reason	str | None	"facility_count_below_threshold" | "petrochemical_corridor" | "cycles_below_threshold" | None
source_attribution	str	Required attribution string for any public render
petrochemical_corridor	bool	True for 4 corridor counties (Harris TX, Calcasieu LA, Iberville LA, Kanawha WV)
aqi_latest_year	int | None	Latest reporting year included in window

Full machine-readable schema: ship via airdata.schema().

Coverage (snapshot)

Class	Counties
decrease	67
increase	176
stable	376
insufficient_data (display held)	375
Total	994

Of which display-held:

• facility_count_below_threshold: 368
• petrochemical_corridor: 4

Methodology

Source: EPA AirData annual AQI summaries (annual_aqi_by_county_{YEAR}.zip), years 2020-2024, retrieved via aqs.epa.gov/aqsweb/airdata/download_files.html.

Pipeline:

1. Aggregate facility-reported emissions to county rollups per year.
2. Compute percent change between earliest and latest cycle.
3. Apply sensitivity gates:
   • ≥3 reporting cycles in window
   • ≥5 reporting facilities in latest cycle
   • Top-1-exclude robustness: direction unchanged when largest facility excluded
4. Bucket as decrease / increase / stable (|pct_change| < 10%) / insufficient_data.
5. Hold display for 4 petrochemical-corridor counties pending methodology review (Harris TX 48201, Calcasieu LA 22019, Iberville LA 22047, Kanawha WV 54039) — see petrochemical_corridor boolean.

Full methodology: zipcheckup.com/methodology/airdata-trend/.

Caveats

• This is facility-reported emissions data, not ambient air quality. The trend reflects what facilities reported to EPA — not what residents breathe.
• AirData updates may reflect reporting methodology changes between cycles. Disclose this in any user-facing render.
• 4 petrochemical-corridor counties surface as skip_reason: "petrochemical_corridor" with a methodology-review notice; do not silently treat them as insufficient_data.
• This package is updated periodically. The meta.generated field surfaces the build date — check it before using for time-sensitive analysis.

License

CC BY 4.0. Free to use with attribution.

Citation

Akulov, A. (2026). U.S. County AirData 5-Year Emission Trends [Data set].
ZipCheckup. https://doi.org/10.5281/zenodo.20382474

Or BibTeX:

@dataset{zipcheckup_us_county_airdata_trends,
  author    = {Akulov, Artem},
  title     = {U.S. County AirData 5-Year Emission Trends},
  year      = {2026},
  publisher = {ZipCheckup},
  doi       = {10.5281/zenodo.20382474},
  url       = {https://doi.org/10.5281/zenodo.20382474}
}

Related

• us-water-quality-data — ZIP-level water quality dataset (sibling package).
• us-housing-risk-data — Housing risk and home values (sibling package).
• ZipCheckup Public API — REST API including /v1/county/{fips}/airdata-trend/.

Source repository

github.com/artakulov/waterbyzipcode (this package lives under packages/us-county-airdata-trends-python/).