Methodology
Every grade is computed from named government datasets using the fixed rules below. This page is generated from the same constants the data pipeline uses, so what you read here is what the code does.
The unit of analysis
The state is divided into H3 hexagons at resolution 9 (each about 0.1 km², roughly 350 m across). A report is made for every hexagon that touches a 2020 Census block with at least one resident: 1,902,144 hexagons covering about 14,384,741 homes. Distances and point lookups use the center of the hexagon, not the exact address, so a report describes the immediate neighborhood. Distances are geodesic, measured on the WGS84 ellipsoid, and given in miles.
Grades compare homes with other California homes. Each hexagon's percentile is computed across all populated hexagons in the state, weighted by its estimated number of homes from 2020 Census housing-unit counts. Without that weighting, empty desert and forest would dominate the comparison. Ties share the middle rank.
| Grade | Percentile among California homes |
|---|---|
| A | 0 to under 50 |
| B | 50 to under 75 |
| C | 75 to under 90 |
| D | 90 to under 97.5 |
| F | 97.5 and above |
Where an official classification exists, it can cap a grade: the grade can be no better than the cap, and it can still be worse. If the data needed for a grade is missing, the section says Not enough data. It is never given a default grade. There is no overall score.
Nearby pollution
The pollution index is a weighted count of official records near the hexagon's center. Nearer records count more. Deleted Superfund sites are listed but not counted, because EPA has determined that no further response is needed there.
| Record | Points each |
|---|---|
| Toxics Release Inventory facility within 1 mile | 3 |
| Toxics Release Inventory facility 1–3 miles away | 1 |
| Final or proposed Superfund (NPL) site within 1 mile | 5 |
| Final or proposed Superfund (NPL) site 1–3 miles away | 2 |
| Facility with recent formal enforcement or significant violations within 1 mile | 2 |
A hexagon with no such records nearby gets A. Otherwise the grade comes from the index's percentile.
- TRI facilities are those that filed Toxics Release Inventory reports in the last 5 reporting years. For the 5 nearest, we show the latest year's total on-site releases and top chemicals.
- ECHO facilities are active facilities with a formal enforcement action in the last 5 years, current significant noncompliance or a high-priority violation, or a significant-violation quarter in the last 3 years. A plain-English line translates EPA's compliance fields.
- Facilities just across the state line (Oregon, Nevada, Arizona) are included.
Flood
The flood percentile is the worse of two percentiles: the share of the hexagon inside FEMA's 1%-annual-chance floodplain (the Special Flood Hazard Area), and paid National Flood Insurance Program claims per 1,000 homes in the census tract.
- Official cap: if the hexagon's center is in a FEMA high-risk zone (A*, V*), the grade can be no better than D.
- If the center is in a moderate-risk zone (shaded X: the 0.2%-annual-chance floodplain, or an area protected by a levee), the grade can be no better than B.
- A hexagon with no floodplain, no paid claims in its tract and an unshaded X zone gets A.
- No digital FEMA map, or a zone with undetermined hazard (D, AREA NOT INCLUDED, OPEN WATER): Not enough data.
This is not a flood zone determination. For a building's official zone, use the FEMA Flood Map Service Center or a licensed professional.
Wildfire
The grade comes from the percentile of the USDA Forest Service's Risk to Potential Structures ("Risk to Homes" on wildfirerisk.org) at the hexagon's center. The report also shows the annual burn probability, exposure type, and the Forest Service's Wildfire Hazard Potential class, all with the datasets' own labels.
Official caps: CAL FIRE Fire Hazard Severity Zones are California's official classification. A Very High zone caps the grade at D. A High zone caps the grade at C.
Earthquake
The grade comes from the percentile of the USGS 2023 National Seismic Hazard Model's chance of potentially damaging shaking (MMI VI or stronger) in 100 years, interpolated at the hexagon's center. There is no official zone for homes, so no cap applies. Nearly all of California has high earthquake hazard by national standards. Each report says how the location compares with the continental U.S. and shows the shaking intensity with a 2% chance of being exceeded in 50 years.
| MMI | USGS description of shaking |
|---|---|
| V | Moderate |
| VI | Strong |
| VII | Very strong |
| VIII | Severe |
| IX | Violent |
| X | Extreme |
Disaster history (not graded)
Counts of federal disaster declarations (major disasters, emergencies and fire management) for the county over the last 30 years, by incident type, plus the 5 most recent. Statewide declarations count for every county.
Area context (not graded)
Median home value, median rent and median year built for the census tract come from the American Community Survey. They are shown as plain facts only. They are never used in any grade, color, ranking or AI-written summary, and neither is any other demographic data. An automated test enforces this, in keeping with the Fair Housing Act.
AI-written summaries
Short summaries of the pollution records, plain-English descriptions of facility violations, and area overviews are written by an open-weight language model running on our own hardware. The model receives only the structured records shown on the page, returns a fixed format, and never assigns grades. Every number, facility name and chemical in its output is checked automatically against its input, along with tone. If any check fails, the text is replaced with a plain template sentence.
How grades are distributed
Share of California homes at each grade after caps are applied.
| Section | A | B | C | D | F | N/A |
|---|---|---|---|---|---|---|
| Pollution | 50.3% | 25% | 14.5% | 7.8% | 2.4% | — |
| Flood | 39.6% | 22.3% | 17.1% | 12.1% | 4.4% | 4.5% |
| Wildfire | 61.3% | 10.3% | 13.5% | 12.4% | 2.5% | 0% |
| Earthquake | 49.9% | 24.8% | 16.3% | 5.7% | 3.3% | — |
Data vintages
acs: ACS 2020-2024 5-yearcensus_blocks: 2020 Census (P.L. 94-171 block counts)disasters: OpenFEMA declarations through 2026-08-04echo: ECHO Exporter (published 2026-09-24)fhsz: CAL FIRE FHSZ: SRA map 2023-09-29 (effective 2024-04-01); LRA map 2025-03-24nfhl: NFHL_06_20260910 (effective through 06/10/2026, latest LOMR 09/11/2026)nfip: NFIP claims 1978–2026 (as of 2026-09-24)npl: EPA NPL feature service (downloaded 2026-09-24)nshm: USGS 2023 50-State NSHM (data release 2023)tiger: TIGER/Line 2025tri: TRI reporting years 2020–2024whp: WHP 2023 (4th edition, published 2024)wrc: WRC 2nd edition (2024; LANDFIRE 2020 fuels)
Full citations, licenses and known limitations are on the sources page. Data was rebuilt on 2026-09-25.
Limitations
- Records can be out of date or wrong. Each section links to the official source; please report errors.
- Being near a facility or site is not evidence of exposure or harm. The records describe where facilities are and what they reported, not what reaches any home.
- Each report describes a hexagon about 350 m across, measured from its center. Homes at the edge of a hexagon may differ.
- Homes built after the 2020 Census may fall in a hexagon without a report. We then show the nearest populated hexagon and say so.
- The wildfire and earthquake layers are official modeled hazard products published by federal agencies. We show them as published and do not model anything ourselves.