Reference · built on requestCommons signalHazards & public safety

NCEI Storm Events Database

NWS Storm Data since 1950 from NOAA NCEI: severe weather events with deaths, injuries, and damage, as monthly bulk CSV files.

About

The Storm Events Database holds Storm Data as the National Weather Service enters it: storms and other significant weather phenomena with enough intensity to cause loss of life, injuries, significant property damage, or disruption to commerce, plus a partial record of other events such as record temperatures tied to another event. NOAA's National Centers for Environmental Information publishes it as three CSV tables linked by event ID: details, locations, and fatalities. Between 1950 and 1954 only tornadoes were recorded; from 1955 tornado, thunderstorm wind, and hail; since 1996, 55 event types as defined in NWS Directive 10-1605. Each monthly release lands about 75 days after the end of the data month.

Through Fibric, an operator reads the events in your counties or forecast zones and proposes annotations, benchmarks, and hazard-history sections for your approval.

This is a reference listing. It documents what Fibric would read from NCEI Storm Events Database and what it could propose, based on the publisher's published interfaces. Fibric builds it under a managed deployment when you request it; selecting it here installs nothing.

Inputs

  • Event details from StormEvents_details-ftp files: episode_id, event_id, state, cz_type, cz_name, wfo, event_type, and begin and end date-times
  • Impact fields per event: injuries_direct, injuries_indirect, deaths_direct, deaths_indirect, damage_property, and damage_crops with K, M, and B suffixes
  • Magnitude fields: magnitude and magnitude_type for wind in knots and hail in inches, tor_f_scale, tor_length, tor_width, and flood_cause
  • Event locations from StormEvents_locations-ftp files: location_index, range, azimuth, location, lat, and lon, several per event
  • Fatalities from StormEvents_fatalities-ftp files: fatality_type D or I, fatality_date, fatality_age, fatality_sex, and fatality_location codes
  • episode_narrative and event_narrative text written by the National Weather Service
  • The same tables as CSV and Parquet in the NOAA Open Data Dissemination bucket on Google Cloud Storage, for queries without a full download

Proposed actions

Read-only. This feed informs operators; it changes nothing.

Proposed actions are target capabilities. Every action runs propose-first and needs a validated deployment and the appropriate permissions.

What you can build

  • Explain a bad month with the storms that hit it

    When a region's deliveries, calls, or attendance fall in a month, the operator lists the events in its counties for that month, with type, narrative, and damage, and proposes the storm as a cause or rules it out.

    With Root Cause

  • Annotate the weekly review with storm events

    As each monthly file lands, the operator matches new events to your counties and proposes annotations on the affected weeks so the numbers carry their weather.

    With Change Annotation

  • Benchmark sites by storm exposure

    Across sites, the operator counts events by type over a span of years, sums damage_property, and proposes a ranked table for the site review.

    With Site Benchmark

  • Add a hazard history to a board pack

    Before a site or investment review, the operator drafts the storm-history section from the details table: the largest events by damage, tornado EF ratings, and flood causes for the county.

    With Board Pack

Requirements

  • The state and county FIPS codes or NWS forecast zone numbers for each site, matched against state_fips, cz_type, and cz_fips
  • A monthly job that fetches the newest file per data year; file names carry the data year as dYYYY and the creation date as cYYYYMMDD
  • The gzip-compressed CSV parsed against Storm-Data-Bulk-csv-Format.pdf, which documents each column
  • A citation of the dataset when used as a source, as the NCEI metadata record requires
Authentication
No key or account. The CSV directory and the NODD bucket answer plain HTTP requests.

Limits

  • Events land about 75 days after the data month ends, up to 120 days per NCEI's metadata record; this is history, not a live feed
  • Period of record differs by event type: tornado only for 1950 to 1954, three types through 1995, all 55 types only since 1996
  • The NWS does not guarantee accuracy or validity; entries from media, law enforcement, or the public may not have been checked by the NWS
  • Damage values are estimates, the source field is free text, and NCEI notes the category field has never held a value

Access and pricing

Reference listing. Public feed access is separate from Fibric integration. If you request this feed, your quote covers the build, usage, and support.

Request NCEI Storm Events Database ↗

Questions and answers

How soon after a storm does it appear in the database?
NCEI states the most recent NWS Storm Data bulk release is published approximately 75 days after the end of a data month, and its metadata record notes up to a 120 day delay. NCEI updates the current yearly file each month.
Which event types are recorded?
Since 1996, 55 event types defined in NWS Directive 10-1605, including Tornado, Thunderstorm Wind, Hail, Flash Flood, Flood, Coastal Flood, High Wind, Wildfire, Winter Storm, Heat, Drought, and Hurricane (Typhoon). Between 1950 and 1954 only tornadoes; from 1955 to 1995 tornado, thunderstorm wind, and hail.
How are the three CSV files linked?
By event_id, the primary key the NWS assigns to each event within an episode. The details file holds one row per event, the locations file up to several lat and lon points per event indexed by location_index, and the fatalities file one row per fatality with fatality_id.
Ask about NCEI Storm Events Database

Ask about the capabilities and requirements in this listing.

For project-specific requirements, contact Fibric.