Methodology & Data Sources

Data Sources

PlainHazard builds its natural hazard profiles from three official U.S. government sources:

• FEMA OpenFEMA API, Disaster Declarations Summaries v2: Official record of all federally declared disasters including major disaster declarations (DR), emergency declarations (EM), and fire management assistance declarations (FMAG). Data includes disaster type, declaration dates, and affected areas at the county level.
• NOAA NCEI Storm Events Database: The National Centers for Environmental Information database of severe weather and natural hazard events. Our data covers 2015–2025 and includes event counts, fatalities, injuries, and property/crop damage estimates for 50+ event types.
• FEMA National Risk Index (NRI): FEMA's county-level expected-annual-loss and risk-rating dataset, used on the homepage hazard-loss summary and shown as separate context on county results in the Historical Activity tool described below. NRI values are not inputs to PlainHazard's score. Current ingested release: December 2025.

Processing Pipeline

1. Retrieve FEMA disaster declaration records from the OpenFEMA API
2. Download NOAA Storm Events CSV bulk data files from NCEI for years 2015–2025
3. Aggregate FEMA records by state and county: total declarations, declaration types, and most common disaster types
4. Aggregate NOAA records by state and county: event counts by type, total fatalities, total injuries, and property/crop damage in nominal dollars
5. Organize the FEMA and NOAA datasets side by side in state and county profile pages, the two ARE kept separate there, not merged into a single number, because they measure different things (see below)
6. Load into our geographically indexed database

Why FEMA and NOAA Stay Separate on Profile Pages

FEMA and NOAA measure different things. FEMA declarations capture disaster-level events that triggered a presidential or agency response. NOAA Storm Events capture individual weather events including many that never result in a federal declaration. Combining them into a single number on a state or county PROFILE page would conflate these different measurement frameworks, so profile pages show both datasets side by side, not merged, so readers can assess historical exposure from multiple angles.

Separately, our Historical Activity tool and the matching ScoreCard on every state and county profile compute an explicit retrospective PlainHazard index. State scores deliberately mix two disclosed windows: OpenFEMA declaration frequency across the full 1953–present ingest (30 points), NOAA storm-event frequency in the 2015–2025 extract (25), NOAA property and crop damage in that same 2015–2025 extract (25), and declared-hazard diversity (20). Those FEMA and NOAA inputs are not the same population - different agencies, event definitions, and coverage years, and the ScoreCard labels each dimension with its window. County scores use FEMA declaration frequency (30), declaration density versus the state county average (25), Major Disaster share (25), and declared-hazard diversity (20), because this database has no county-level NOAA damage. A measured zero on a factor scores 0 points; a truly unavailable factor drops out and its weight redistributes across the remaining factors so the index stays on a 0–100 scale. Bands are Low / Moderate / High / Very High historical activity, not A–F letter grades, because an A–F frame would read as a safety or livability rating on this registry. The result is a historical index, not a future-risk forecast or agency rating, and FEMA NRI values shown on county results remain separate context rather than an input. Any published ratio that divides NOAA damage by FEMA declarations (for example the research frequency-vs-cost page) must use OpenFEMA unique disasters with fiscal year inside the NOAA extract window, never all-time declaration totals as the denominator.

Damage Figures

Property and crop damage figures are reported in nominal dollars as provided by NOAA, they are not inflation-adjusted. Fatality and injury counts include both direct and indirect causes as classified by NOAA.

Data Vintage and Update Frequency

PlainHazard's complete OpenFEMA declaration snapshot covers 1953 to the present. New FEMA records are incorporated after they are available from the source and the pipeline is re-run. NOAA Storm Events data on PlainHazard covers 2015 through 2025, providing a decade of detailed severe weather event records. The FEMA National Risk Index release currently loaded is December 2025. NOAA says Storm Data and Storm Events reports are generally available 90 to 120 days after an event, following reporting and verification. PlainHazard refreshes its database within 30 days of each major FEMA or NOAA data release, and each refresh is recorded in the source-release log that powers the FEMA declaration change report.

Coverage

• FEMA declarations: OpenFEMA declaration records from 1953 to present, covering states, U.S. territories, and tribal nations
• NOAA Storm Events: 2015–2025, covering 50+ event types including tornadoes, hurricanes, floods, winter storms, wildfires, droughts, and hail events
• Geographic resolution: State and county level for both datasets
• Metrics: Event counts, fatalities, injuries, and property/crop damage estimates

Accuracy Commitment

PlainHazard preserves FEMA and NOAA source figures within their stated scopes, while clearly labeling its own aggregates, rankings, and comparisons. The two datasets remain separate on state and county profiles because FEMA declarations and NOAA Storm Events measure different phenomena. The optional Historical Activity tool is separately labeled as a retrospective PlainHazard calculation, not an agency risk rating or forecast. When data is incomplete or unavailable for a specific county or time period, this is shown transparently.

Limitations

• Historical hazard frequency does not predict future risk. Climate patterns are changing, and past event frequency may not reflect future exposure for any given location.
• NOAA storm event damage estimates are preliminary and subject to revision. Not all events have complete damage data, particularly for events where property damage was difficult to assess or unreported.
• NOAA Storm Events counts are influenced by how many people are present to observe and report a storm, not purely by how often severe weather occurs. Densely populated areas tend to show more recorded events for a given hazard than sparsely populated ones with genuinely similar meteorological exposure, because there are more people, spotters, and media outlets to notice and log an event. Treat state and county event counts as a record of what was documented, not a precise measure of true storm frequency.
• FEMA declarations reflect political and administrative processes as well as physical severity, a state governor must request a declaration, and approval involves assessment of state and local capacity alongside disaster impact.
• County-level aggregation may mask variation within large counties. Urban and rural areas within the same county may face very different hazard exposures.
• Do not use this data as the sole basis for personal safety decisions, emergency planning, or insurance purchasing. Consult local emergency management agencies for community-specific risk assessments.
• PlainHazard is not affiliated with FEMA, NOAA, or any government agency.

Data Reuse and Licensing

Everything PlainHazard publishes is designed to be reusable. Every figure is rendered as extractable HTML text and semantic tables (no image-only data), the national statistics report carries machine-readable Dataset markup with a downloadable county summary, each report page includes a stable citation block, and new FEMA data releases are announced as dated entries in the RSS feed. PlainHazard's derived presentation is released under CC0; the upstream federal datasets (FEMA OpenFEMA, NOAA Storm Events, FEMA NRI) are U.S. government works that retain their own terms. If you reuse our derived tables or figures, an attribution link to the source page is appreciated but not required. Download the open county extract at plainhazard-county-summary.csv; for a custom bulk extract or API needs, contact us.

Contact

Questions about our methodology? Contact us.