NOAA's Storm Events database, explained
What NCEI publishes
The Storm Events database is the reviewed record of severe weather in the United States. It is kept by the National Centers for Environmental Information, the archive arm of the National Oceanic and Atmospheric Administration, and it holds the entries each National Weather Service forecast office prepares for the monthly publication called Storm Data. The database and the publication are one record in two forms: the publication is the monthly document, and the database is the same entries made searchable, exportable and citable one at a time.
The Storm Events Database contains the records used to create the official NOAA Storm Data publication, documenting: a. The occurrence of storms and other significant weather phenomena having sufficient intensity to cause loss of life, injuries, significant property damage, and/or disruption to commerce; b. Rare, unusual, weather phenomena that generate media attention, such as snow flurries in South Florida or the San Diego coastal area; and c. Other significant meteorological events, such as record maximum or minimum temperatures or precipitation that occur in connection with another event.
National Centers for Environmental Information, Storm Events Database
Each entry carries an event type from a fixed list, a state and a county or forecast zone, a begin and an end time, a location where the office could fix one, a magnitude where the type has one, the source the office received it from, counts of deaths and injuries, estimates of property and crop damage, and a narrative in the office's own words. For hail the magnitude is the diameter of the largest stone. The record reaches back further for some event types than for others, and the agency states which.
Tornado: From 1950 through 1954, only tornado events were recorded.
Tornado, Thunderstorm Wind and Hail: From 1955 through 1995, only tornado, thunderstorm wind and hail events were keyed from the paper publications into digital data.
All Event Types (48 from Directive 10-1605): From 1996 to present, 48 event types are recorded as defined in NWS Directive 10-1605.
National Centers for Environmental Information, Storm Events Database: details
How a hail entry gets there
An entry begins as a report to a forecast office while the storm is moving: a trained spotter with a stone and a coin, a cooperative observer at a fixed site, a law officer on a road, an emergency manager, a broadcaster, a member of the public. The office fixes the report to a location and a time and issues it within minutes as a local storm report. That is the raw entry of the storm record, and it has been checked against nothing.
After the month closes, the office prepares its Storm Data entries under a written directive, National Weather Service Instruction 10-1605. The directive lists the event types, defines each, sets the criteria an event has to meet to be entered, and tells the office what to record. The office works through the month's local storm reports with the whole day in view: it checks reports against each other and against what it can verify afterward, merges duplicates, corrects sizes and locations where a better source turned up, drops what it cannot support, and writes the narrative. The result is signed off by the office and sent to NCEI, which publishes it as the month's Storm Data and loads the entries into the database.
- A point entry
- Filed with a location the office could fix: a town, a distance and direction from one, or coordinates. A report for an address measures from it.
- A county or zone entry
- Filed to the county or forecast zone rather than a point. Such an entry did not happen at any one address, and a report that gives it a distance is carrying a figure the source never stated.
- A survey entry
- A tornado or wind event whose rating and path came from a storm survey the office conducted afterward. The entry carries the survey's finding, not a spotter's estimate.
The lag and the revisions
The review is the database's authority, and the review is also its delay. An office has a deadline after the month closes, NCEI publishes after that, and the agency states how long the two take together.
Storm Data is typically available approximately 75 days after the end of a data month.
National Centers for Environmental Information, Storm Events Database: frequently asked questions
A storm in the first week of a month therefore waits the rest of that month and most of a season before its reviewed entry exists. In the meantime the preliminary logs carry it. A date of loss set in the weeks after a storm rests on those logs, and it should be revisited when the reviewed record arrives, because the reviewed entry can differ from the preliminary one on size, on time, on location, and on whether the event happened at all.
Publication is not the end of it. An office can correct an entry after the month is published, and NCEI loads the correction, so an entry read today can differ from the same entry read last year. This is why a report carries a fingerprint over the records it cites: a copy that rests on an entry its source has since revised no longer matches, and the reader can see that it does not. The free pages say when a recent day is still inside a source's revision window, and the day's page calls itself provisional until the window closes.
What it does not contain
The database is a record of reports that met the directive's criteria and survived the office's review. It is not a record of everything that fell. Four absences matter to a claim.
- Hail nobody reported. A stone that fell over open country, overnight, or on a roof whose owner did not call the office is not in the record. Absence from the database is not evidence that no hail fell; it is evidence that no report was entered.
- Hail below the criterion. The directive sets the sizes and effects an event has to reach to be entered, and an office may leave a small stone out. A claim about sub-severe hail is often a claim about a day the reviewed record holds nothing for and the preliminary logs do.
- Any radar-derived figure. The database holds what people and instruments reported, not what an algorithm computed for a grid cell. A detected size never comes from here, and a report that cites the database for one has mislabelled its source.
- Any opinion about a roof. An entry says that hail of a stated size was reported at a stated place and time. It does not say what the hail did, and a narrative that mentions damage is the office's summary of what it was told, not an inspection.
The record can also be certified. NCEI certifies data from its archive under the Department of Commerce seal for legal proceedings, and a certified public record authenticates itself under the federal rules of evidence, which is the footing a weather report should stand on and the subject of the guide on automated weather reports in court.
How it sits beside the SPC log and the local storm report
Three federal records describe the same hail day, and a reader who does not know which is which will count one storm three times or one source as three. They are related by descent: the local storm report is the raw entry, the Storm Prediction Center's log is the same entries gathered nationally as they were issued, and the Storm Events database is the reviewed form the office made of them after the month closed.
| Record | Made by | When | Reviewed |
|---|---|---|---|
| Local storm report | The forecast office, from whoever called it in | Within minutes, while the storm moves | No; issued as received |
| SPC storm report | The Storm Prediction Center, from every office's local storm reports | The same day, filed under the convective day | Little; kept as filed, duplicates and all |
| Storm Events entry | The forecast office, under the directive, published by NCEI | A season after the month closes | Yes; checked, merged, corrected or dropped |
The reviewed entry is the one to cite where the two disagree, and the preliminary one records what was known at the time. But the three are not one source in three copies. The Storm Prediction Center's log is a separate publication kept as filed, and an entry that reached the reviewed record by a different route from the one in the log is corroborated by it. The product counts a day carried by the reviewed record and by a preliminary log as carried by two independent sources, holds the two figures side by side rather than letting one overwrite the other, and never merges the log's duplicates, because the source did not.
What the report shows
A Titan Weather report for an address lists each reviewed entry it found with its date and time as the office filed them, its size in the office's own units, the location as filed, and its distance and bearing from the address where the entry has a point. An entry filed to a county or zone is listed with the county named and no distance. Records within 3 miles are banded as primary and records within 10 miles as contributing, and the preliminary form of the same event, where the record holds it, is listed beside the reviewed one under its own source rather than folded into it.
The sources and method section names the reviewed record as searched, with the period and the radius, and says when a day in the window is still inside its revision window. The statement of compilation says that records were reproduced as their sources published them and that no opinion is offered on the cause, extent or cost of any damage. Reading the whole document, figure by figure, is the subject of the guide on how to read a hail report.
Questions this guide answers
- What is the Storm Events database
- The reviewed national record of severe weather, published by the National Centers for Environmental Information from the entries each Weather Service forecast office prepares for the monthly Storm Data publication. Each entry carries a type, a location, begin and end times, a magnitude where there is one, and a narrative.
- How long after a storm does an entry appear in the Storm Events database
- A season, roughly. Each office has a deadline after the month closes, NCEI publishes the month after that, and an entry can be corrected after publication. A recent storm is absent from the reviewed record while the preliminary logs already carry it.
- Is the Storm Events database the same as the SPC storm reports
- No. The Storm Prediction Center log is compiled from local storm reports as they are issued, with little review, and stays as filed. The Storm Events database is the reviewed record of the same days, in which entries were checked, merged, revised or dropped. The two are independent sources for one day.
- Does an entry in the Storm Events database prove hail at my address
- No. An entry establishes that hail of a stated size was reported at a stated place and time, and some entries are filed to a county zone rather than a point. The distance from the entry to a roof, and what the hail did there, are questions the entry does not answer.
From the record
1 Nov 2026
A day the reviewed record is still being written for
- Reports on the day so far
- 13
- Shortest revision window
- 7 days
- Longest revision window
- 540 days
- Settles by
- 24 April 2028
1 November 2026 carries 13 reports so far from the preliminary logs, and at least one national source is still revising it: the longest revision window among the sources is 540 days, so the day is provisional until 24 April 2028 and its page says so.
Measured Detected, not measured
The page these figures come fromRelated
The terms this page uses and the guides that go further. Each is shown against its own example from the record.
-
Term
Storm Events database
The federal archive of reviewed storm reports, compiled by forecast offices and published by NCEI between two and three months after each month ends.
-
Term
Local storm report
A report of severe weather at a place and a time, relayed by a National Weather Service office as it comes in and before it is reviewed.
-
Term
SPC storm report
A preliminary hail, wind or tornado report collected nationally by the Storm Prediction Center from forecast offices on the day it was filed.
-
Term
Storm survey
A ground or aerial examination of storm damage by a National Weather Service team, made to fix a tornado's path, width and rating after the event.
-
Term
Independent source
A record-keeping route by which a storm reached the archive, counted once however many rows it produced, so that agreement between routes can be measured.
-
Term
Storm day
A calendar date on which hail, wind or tornado reports were filed, counted as one day however many reports it carries.
-
Guide
Automated weather reports in court: what survives a challenge
Two Florida courts excluded wind figures taken from commercial weather products because nobody could verify how they were made. This guide states what those rulings turned on, what Rule 702 asks of a figure and of the person citing it, what independent corroboration means as a count, and what a report has to say about itself to be cited.
-
Guide
How a date of loss is established, and how it is contested
A date of loss is a claim about one calendar day. Most courts have read it as the day the storm fell rather than the day the damage was found, and a policy answers only for days inside its term, so the date is tested before anything else is. This guide states how the date is set, how a carrier checks it, where it is most often wrong, and what a record has to show for the date to hold.
-
Guide
How to read a hail report: what each figure means
A hail report is a list of records with figures attached, and every figure on it came from somewhere in particular. This guide walks the document in the order a reader takes it: the date and the window searched, a measured size beside a detected one, the distance and bearing from the address, the count of independent sources, and the line at the end that says what the report does not claim.
Sources
The primary references. Each opens on the agency's own site.
- 01 National Centers for Environmental Information, Storm Events Database
- 02 National Centers for Environmental Information, Storm Events Database: details and data periods
- 03 National Centers for Environmental Information, Storm Events Database: frequently asked questions
- 04 National Centers for Environmental Information, Storm Data bulk CSV format
- 05 National Weather Service, Storm Data preparation (NWS Instruction 10-1605)
- 06 Storm Prediction Center, preliminary storm reports
- 07 National Weather Service glossary, Local Storm Report
- 08 National Centers for Environmental Information, certification of data for legal proceedings
The guide explains the record. The report is what cites it.
Everything above is explanation: what the record is, how it is read, and what a figure in it can establish. None of it says what reached one roof.
Run the address and the archive is searched at the property: the storm, the date, the measured size, the distance and bearing from the roof, and the federal record each figure came from, in a document that states its own method. Ready in minutes.
How this page is made
- The text
- Written by the editors and last revised 23 August 2026. It explains the record and how the report reads it; it contains no figure of its own.
- The example
- Drawn from the archive when the page was compiled, 1 November 2026, and rebuilt daily as the record deepens. Every figure in the panel marked From the record is computed, not written, and links to the page it came from.
- Measured and detected
- A report from a person or an instrument is measured. A radar-derived size is detected, not measured, and is marked as such, here and in every document we issue. The two are never added together.
- What this is not
- An explanation of the record and of how the report reads it. It is not a report on any property, and it states no opinion about one. The coverage page states what the record holds, by kind and by year.
About this record
Everything here is reproduced from public federal and volunteer weather records, as the agencies and observer networks that publish them released it. Titan Weather compiles the record; it does not observe the weather, and it is not affiliated with any source it cites.
No archive holds everything. Storms cross places where nobody is standing, instruments and observers cover some areas better than others, sources revise their records, and recent reports can take weeks to arrive and settle. Coverage also thins as the years go back: an early quiet stretch may describe who was watching rather than what the weather did. Where this page shows nothing, nothing was found in the sources searched, within the period and area it states. That is not a finding that no weather occurred.
A record sits where its source filed it. Reports made from the ground are commonly logged against a town, a landmark or an offset from one rather than the spot where the measurement was taken, so any distance shown for one carries that source's own location error. Radar detections and warning geometry are instrument-located and do not.
These pages describe the public record near a place or on a date. They do not establish what reached any particular property, and nothing here is insurance, legal or engineering advice.