NOAA Storm Events, Damage & Weather Risk Scraper

Stream and export official NOAA Storm Events records from 1950 onward. Each result can include event type, dates, state, county or forecast zone, locations, coordinates, magnitude, tornado attributes, fatalities, injuries, property and crop damage, source and narratives.

The Actor automatically discovers NOAA's latest revised archive for each requested year, decompresses it as a stream and saves clean structured records. This design supports large exports without loading an entire annual file into memory and needs no NOAA API key.

What can you do with this Actor?

• Build historical severe-weather and catastrophe datasets for insurance or reinsurance.
• Quantify property and crop damage by event type, region and time.
• Enrich logistics, supply-chain, energy, agriculture and real-estate risk models.
• Analyze tornadoes, hail, wind, floods, winter weather, heat and other NOAA event categories.
• Map event paths and locations using available start and end coordinates.
• Create research datasets with event and episode narratives.

Input

Field	Type	Description
years	integer[]	Years from 1950 onward. Empty input uses the current and previous year.
states	string[]	Exact NOAA state or area names, usually uppercase.
eventTypes	string[]	Exact NOAA event types such as Hail, Tornado or Flash Flood.
startDate	date	Earliest event-begin date.
endDate	date	Latest event-begin date.
minPropertyDamage	number	Minimum normalized property damage in USD.
minCropDamage	number	Minimum normalized crop damage in USD.
fatalitiesOnly	boolean	Keep only events reporting at least one death.
injuriesOnly	boolean	Keep only events reporting at least one injury.
maxResults	integer	Maximum unique events, up to 500,000.

Example for damaging Texas events:

{
  "years": [2025, 2024],
  "states": ["TEXAS"],
  "eventTypes": ["Hail", "Thunderstorm Wind", "Tornado"],
  "minPropertyDamage": 1000,
  "maxResults": 20000
}

Output

One item represents one NOAA event record.

Field group	Included data
Identity	recordId, eventId, episodeId, year and month.
Classification	State, event type, county/zone type and FIPS, WFO and source.
Time and place	Begin/end timestamps, locations, coordinates and time zone.
Severity	Magnitude, magnitude type, tornado scale/length/width, flood cause and category.
Impact	Direct/indirect deaths and injuries, totals, normalized property damage and crop damage.
Context	Episode narrative and event narrative.
Lineage	Exact NOAA archive URL and scrape timestamp.

{
  "eventId": "1234567",
  "year": 2025,
  "state": "TEXAS",
  "eventType": "Hail",
  "countyZoneName": "TRAVIS",
  "beginDateTime": "2025-05-28T18:15:00",
  "magnitude": 2.25,
  "magnitudeType": "EG",
  "deaths": 0,
  "injuries": 0,
  "propertyDamage": 250000,
  "cropDamage": 0
}

Damage normalization

NOAA bulk files commonly encode values with suffixes such as K, M and B. The Actor converts them to numeric dollars: 25K becomes 25000 and 1.5M becomes 1500000. Missing damage is normalized to zero for consistent threshold filtering. These are reported estimates and should not be treated as audited loss totals.

Annual revisions and scale

NOAA can revise historical files. The Actor reads the directory at run time and selects the highest revision stamp for each requested year. This avoids hard-coded file names becoming stale. Requested years are processed newest first, records are deduplicated by year and event ID, and result batches are written continuously.

For a full-history warehouse, split runs into year ranges so each run stays under the 500,000-item limit. For operational monitoring, schedule the current year regularly because recent records may be added or revised.

Integrations and API

Use the Apify Console, REST API, JavaScript client or Python client. Export datasets as JSON, JSONL, CSV, Excel, XML or RSS and connect them to schedules, webhooks, Make, Zapier, Google Sheets, databases or BI systems.

curl -X POST "https://api.apify.com/v2/acts/logiover~noaa-storm-events-bulk-scraper/runs?token=YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"years":[2025],"states":["FLORIDA"],"maxResults":10000}'

Through Apify MCP, AI agents can request structured storm-event evidence instead of relying on unverified web summaries.

FAQ

Which NOAA source is used? The official NCEI Storm Events bulk CSV archive.

Does it need an API key? No.

What years are available? NOAA publishes records from 1950 onward; event coverage and reporting practices vary by era.

What does empty input do? It processes the current year and previous year, newest first, up to the default result limit.

Why is a recent year incomplete? NOAA reporting and quality control lag real-world events and files are revised.

Are all coordinates populated? No. Location fields depend on the original report.

Can I filter multiple event types? Yes; use exact NOAA labels.

Are fatalities and injuries combined? Direct and indirect values are preserved and summed into convenient totals.

Can I download more than 500,000 items? Split the export by year, state or event type.

Responsible use

This Actor is not affiliated with or endorsed by NOAA or NCEI. Storm Events data contains reported estimates and historical inconsistencies. Verify life-safety, legal, actuarial and financial decisions against authoritative current sources.

Changelog

2026-08-04

• Initial official NOAA Storm Events bulk exporter.
• Added live latest-revision discovery and memory-efficient gzip/CSV streaming.
• Added year, state, event type, date, damage, fatality and injury filters.
• Added normalized damage amounts, stable IDs, coordinates, impact totals and narratives.