Every federal declaration since 2000, mapped and ranked.
Since 2000, the federal government has issued thousands of disaster and emergency declarations: the formal decisions that unlock federal help after a hurricane, flood, wildfire, ice storm, or public-health crisis. The full record is public, but it lives in spreadsheets and APIs built for analysts.
Disaster Data puts that record on a map and a scoreboard. Drag the year range to watch a decade reshape the rankings, filter by hazard, and drill into any state to see its individual declarations and how often requests were denied. No account, no cost, nothing to install.
Every county, parish, borough, and tribal nation also has its own page with a full declaration history and a ready-to-use previous-occurrences table for local hazard mitigation plans. Over 3,600 jurisdictions, all searchable from the homepage.
Straight from OpenFEMA, refreshed automatically.
Every figure here is drawn from FEMA's Disaster Declarations Summaries (v2), published through the OpenFEMA program and refreshed on a schedule so the dashboard stays current.
Disaster Data uses OpenFEMA data but is not endorsed by or affiliated with FEMA. It's an independent project. Any errors or interpretations are ours, not the agency's. Where a number here disagrees with an official FEMA source, treat the official source as authoritative.
Source, definitions, and how counts are made.
Source. Every figure is drawn from FEMA's OpenFEMA datasets: Disaster Declarations Summaries (v2) for declarations and Declaration Denials (v1) for turndowns. It is rebuilt automatically each week. Uses OpenFEMA data but is not endorsed by or affiliated with FEMA.
What counts as a “declaration.” FEMA's file lists one row per county per declaration. We collapse those to one row per declaration: a unique combination of declaration type (DR / EM / FM), disaster number, and state. A single disaster affecting five states therefore counts as five declarations. Counts reflect declaration records, not dollars, damage, or loss.
Years are federal fiscal years (Oct 1 to Sep 30), matching FEMA's own fyDeclared field. FY2000 begins October 1, 1999. The headline view shows complete fiscal years only; the current, in-progress year stays one slider-drag away.
Declaration-request denial rate. The share of declaration requests turned down = turndowns ÷ (declarations + turndowns). This is the rate at which a governor or tribal leader's request for a federal declaration was denied, not a rate of individual assistance denials. Denials are tracked in a separate FEMA dataset from approved declarations, so this figure cannot be reproduced from the summaries file alone.
Fire-management (FM) declarations fund wildfire suppression under a faster, lower-threshold program. They're counted as their own type. Toggle FM off anywhere on the site to see the disaster picture underneath.
Declarations, not dollars.
Every count on this site is a count of declaration records, how many times a federal declaration was issued, not a measure of damage, cost, or lives affected. A small county flood and a multi-state hurricane each register as one declaration; the dollars behind them can differ by orders of magnitude.
A few things worth knowing before you cite a figure. Fire is the largest category by far, but that's inflated by Fire Management Assistance declarations, which a single fast-moving wildfire can generate quickly. It doesn't mean fire is the most damaging hazard. The four biggest categories (Fire, Severe Storm, Hurricane, and Flood) make up roughly five of every six declarations. And a denial reflects a governor or tribal leader's request for a federal declaration being turned down, not an individual being denied assistance under an approved declaration; it is a measure of process and eligibility at the declaration stage, not of how severe an event was.
Three kinds of declaration.
FEMA issues declarations under three programs. Disaster Data color-codes them the same way everywhere they appear.
The broadest and most common. Issued for events that overwhelm state and local resources, unlocking the widest range of federal assistance.
A narrower, faster declaration to protect lives and property, often issued ahead of, or alongside, a major disaster.
Fire Management Assistance, for wildfires threatening to become major disasters. Busy fire seasons can generate many of these in a short window.
The 25 hazard types.
FEMA tags every declaration with an incident type. Disaster Data tracks all 25 that appear in the record, grouped here into families. On the dashboard the hazard count ticks down as you narrow the years: the rarest types drop out of short windows.
A few terms.
The data behind the dashboard, yours to use.
The dashboard is one way in. If you would rather work with the data directly, the same federal record that powers this site is published as a small set of static JSON files you can fetch straight from a browser or a script. No key, no login, no rate limit, refreshed on the same weekly schedule as the rest of the site.
These are not a copy of FEMA's raw files. They are the cleaned, joined version this site runs on: declarations collapsed to one row each, independent cities and counties correctly distinguished, the same storm grouped into a single event across state lines, and Public Assistance obligations rolled up by county and disaster. It is the work you would otherwise redo from the raw OpenFEMA endpoints.
data/decl-index/{ST}.jsondata/decl-index/manifest.jsonpa-timing.jsonmap-events.jsalso available// Your live origin. Edit this one line if your domain differs. const BASE = 'https://disasterdata.io'; // List Virginia's declarations, no key required const declarations = await fetch(BASE + '/data/decl-index/VA.json').then(r => r.json()); // Discover which state files exist const manifest = await fetch(BASE + '/data/decl-index/manifest.json').then(r => r.json());
Free to use. If you publish something built on this, please credit it: Source: DisasterData.IO utilizes OpenFEMA.org, FEMA.gov.
The shape is stable. New fields may be added over time, but existing fields will not be renamed or removed without a versioned path, so anything you build today keeps working.
Same caveat as the rest of the site. This is derived from public FEMA data but is not endorsed by or affiliated with FEMA. Where a figure here disagrees with an official FEMA source, treat the official source as authoritative.
State declarations, published the same way.
A federal declaration is the end of a sequence, not the start of one. Before FEMA arrives, a governor declares. Those state emergency declarations live in fifty separate state archives, in fifty different formats, and no one publishes them as a single dataset. DisasterData Plus collects them, classifies each one by hazard from the order text, and joins it to NOAA Storm Events so you can see what was actually reported on the ground while the order was in effect.
All fifty states are collected. Counts below describe the current build: 1,381 state declarations across the states with resolved local evidence, 9,661 county and declaration pairs, covering 2,148 counties in 42 states. The other eight states are collected but have no county-resolved evidence yet, for reasons given in each one's coverage note.
state-declarations.jsonplus/coverage.jsonplus/{state}/declarations_for_join.csvplus/{state}/eo_storm_severity_summary.csvrow level// Your live origin. Edit this one line if your domain differs. const BASE = 'https://disasterdata.io'; // What is covered, and how well, state by state const coverage = await fetch(BASE + '/plus/coverage.json').then(r => r.json()); // State declarations with local evidence in Roanoke County, Virginia (FIPS 51161) const states = (await fetch(BASE + '/state-declarations.json').then(r => r.json())).states; const local = states.VA.counties['51161'];
Read the county attribution carefully. A declaration listed under a county does not mean the governor named that county in the order. Most state declarations apply statewide and name no localities at all. It means NOAA Storm Events recorded reports of the matched hazard in that county while the order was in effect. That is evidence of local impact during a declared state emergency, which is usually what you want, but it is an inference from two joined sources and not a quotation from the order. Treat it as evidence, not as a designation.
What is left out is left out on purpose. Only county typed storm reports are resolved into the county file. NOAA forecast zones do not align to county boundaries and are not folded in silently; they stay in the per state row level file. Declarations whose text matches no hazard keyword are skipped rather than guessed at. Area names that cannot be resolved to a real county are dropped and counted, not approximated. The counts above are what survived that, not what was collected.
Free to use. If you publish something built on this, please credit it: Source: DisasterData.IO, state executive order archives joined to NOAA Storm Events. The underlying state orders and NOAA records are public; the collection, classification, and join are the work being credited.
Same caveat as the rest of the site. This is derived from public records but is not endorsed by or affiliated with any state government, FEMA, or NOAA. Where a figure here disagrees with the issuing state's own archive, treat the state as authoritative.
Lead Time, a training game.
Lead Time is a free, browser-based game about the decisions an emergency manager actually makes. You work an unfolding disaster one call at a time, weighing incomplete information against limited time and resources, and your choices decide how the event turns out. It runs entirely in your browser and keeps a service record and commendations that carry across sessions, with no account or install.
Part of CompliAid.
Disaster Data is built and maintained as part of CompliAid, and shares its principles: privacy-first, no ads, nothing tracked for sale, and tools designed for the real constraints of rural and local government.
The aim is simple: make public disaster data genuinely usable by the people who live with disasters, not just the people who study them.
If something looks off, report a data issue and we will investigate.