US Hail History by Location — Hail Reports from NOAA NEXRAD Radar

Check the hail history of any U.S. location: enter a latitude/longitude and get every date on which NOAA's NEXRAD weather radars detected hail nearby — with the maximum estimated hail size in inches, hail probability, number of detections, and distance from your point. One flat JSON row per location and hail day, back to 1995, usually up to 2 days ago.

• Hail history by location: one row per hail day, with max hail size, probability of hail and of severe hail, number of radar detections and storm cells, closest distance and exact detection times (UTC).
• Hail map data for many points: up to 100 locations per run (homes, roofs, branches, car lots, insured properties), each with its own label.
• Official source: NOAA National Centers for Environmental Information (NCEI), Severe Weather Data Inventory (SWDI), NEXRAD Level-III hail signatures. No API key, no login.
• Cheap per-location hail reports: pay only for hail-day rows found; "no hail found" and errors are free.

What can it be used for?

1. Roofing and exterior contractors — storm leads and hail verification. Check which neighborhoods, streets or customer addresses were hit by 1"+ hail after a storm, or confirm the hail history of a property before you quote a roof replacement.
2. Insurance claims and adjusters. Get a date-of-loss candidate list for a property: every day with radar-detected hail within a few miles, the largest estimated hail size and how close it came. Useful for claim triage, fraud checks and supporting documentation (see the caveat below — this is not an insurance determination).
3. Auto dealers, real estate and risk analysis. Screen car lots for hail damage risk, add "hail days in the last 3 years" to property reports, or compare hail exposure across sites and portfolios. AI agents can answer "Did it hail at this address last month, and how big?" in one call.

How to use it

Only locations is required. Run it with the default input to get the last 12 months of hail history for Dallas, TX.

Input	Type	Default	What it does
locations	list	["32.7767,-96.7970"]	Points to check as "latitude,longitude" strings, or objects {"latitude": 32.7767, "longitude": -96.797, "label": "Dallas office"}. Up to 100 per run.
radiusMiles	number	10	Count radar hail detections within this distance of each point (0.1–50 miles).
daysBack	integer	365	Search the last N days when startDate is empty (1–1098).
startDate / endDate	YYYY-MM-DD	empty	Exact period (1995-01-01 or later, max 3 years per run). endDate empty = today.
minHailSizeInches	number	0	Ignore detections smaller than this, e.g. 1 for quarter-size hail or larger.
timeZone	IANA name	America/Chicago	Time zone used to assign detections to calendar dates (evening storms can fall on the next day in UTC).
maxResults	integer	1000	Maximum number of hail-day rows to return.

This Actor takes coordinates, not street addresses. To get a hail report by address, geocode the address first (for example with the free U.S. Census Bureau geocoder or your CRM's coordinates) and pass the latitude/longitude.

Input example

Two locations, May–June 2026, 10-mile radius:

{
  "locations": [
    { "latitude": 32.7767, "longitude": -96.797, "label": "Dallas office" },
    "39.7392,-104.9903"
  ],
  "startDate": "2026-05-01",
  "endDate": "2026-06-30",
  "radiusMiles": 10,
  "timeZone": "America/Chicago"
}

Output example

Real output from the input above (run on 2026-09-19): 12 hail-day rows — 7 for Dallas, TX and 5 for Denver, CO. The first row:

{
  "status": "ok",
  "location": "32.7767,-96.797",
  "label": "Dallas office",
  "latitude": 32.7767,
  "longitude": -96.797,
  "radiusMiles": 10.0,
  "date": "2026-06-19",
  "timeZone": "America/Chicago",
  "maxHailSizeInches": 1.25,
  "maxHailProbabilityPct": 100,
  "maxSevereHailProbabilityPct": 60,
  "signatureCount": 23,
  "stormCellCount": 6,
  "closestDistanceMiles": 1.1,
  "closestHailSizeInches": 0.5,
  "closestRadar": "KFWS",
  "largestHailDistanceMiles": 7.11,
  "largestHailLatitude": 32.874,
  "largestHailLongitude": -96.837,
  "largestHailTimeUtc": "2026-06-19T11:28:38Z",
  "radars": "KDYX,KFWS,KGRK",
  "firstDetectionUtc": "2026-06-19T11:22:32Z",
  "lastDetectionUtc": "2026-06-19T13:16:01Z",
  "source": "NOAA NCEI Severe Weather Data Inventory (SWDI), NEXRAD Level-III hail signatures (nx3hail)",
  "sourceUrl": "https://www.ncei.noaa.gov/swdiws/csv/nx3hail/2026060105:2026070105?bbox=-96.9875,32.6145,-96.6065,32.9389"
}

Summary of all 12 rows from the same run:

Location	Date	Max hail size (in)	Detections	Closest (mi)
Dallas office	2026-06-19	1.25	23	1.1
Dallas office	2026-06-06	1.25	1	9.93
Dallas office	2026-06-02	1.75	5	8.42
Dallas office	2026-05-26	0.75	3	3.28
Dallas office	2026-05-24	1.25	9	6.15
Dallas office	2026-05-23	1.5	6	3.76
Dallas office	2026-05-19	1.25	11	7.33
39.7392,-104.9903 (Denver)	2026-06-26	1.25	4	9.79
39.7392,-104.9903 (Denver)	2026-06-24	1.75	24	3.18
39.7392,-104.9903 (Denver)	2026-06-08	1.75	37	1.73
39.7392,-104.9903 (Denver)	2026-06-01	2.25	24	0.36
39.7392,-104.9903 (Denver)	2026-05-28	0.75	25	6.49

Locations with no hail detections, and invalid input, return one free row that says what to change (real output):

{"status": "not_found", "location": "32.7767,-96.797", "label": null, "latitude": 32.7767, "longitude": -96.797, "radiusMiles": 1.0, "startDate": "2026-01-01", "endDate": "2026-01-31", "error": "No NEXRAD hail signatures within 1 miles between 2026-01-01 and 2026-01-31; widen \"radiusMiles\" or the dates (radar data usually lags about 2 days, and no detection does not guarantee there was no hail)."}
{"status": "error", "error": "Location #1 ('Dallas, TX') must be \"latitude,longitude\" in decimal degrees (e.g. \"32.7767,-96.7970\") or an object like {\"latitude\": 32.7767, \"longitude\": -96.797, \"label\": \"Dallas\"}."}

Output fields

Field	Meaning
status	ok, not_found or error
location, label, latitude, longitude, radiusMiles	The point you asked about, your label, and the search radius
date, timeZone	Local calendar date of the hail day, and the time zone used
maxHailSizeInches	Largest radar-estimated hail size within the radius that day (inches)
maxHailProbabilityPct, maxSevereHailProbabilityPct	Highest radar probability of any hail / of severe hail that day
signatureCount, stormCellCount	Number of radar hail detections and distinct storm cells within the radius
closestDistanceMiles, closestHailSizeInches, closestRadar	Nearest detection to your point, its estimated size and the radar that saw it
largestHailDistanceMiles, largestHailLatitude, largestHailLongitude, largestHailTimeUtc	Where and when the largest hail estimate was detected
radars	NEXRAD radar stations that detected hail (comma-separated)
firstDetectionUtc, lastDetectionUtc	First and last detection time (UTC)
source, sourceUrl	Dataset name and the exact NOAA request the row was built from
startDate, endDate, error	On not_found / error rows: the period searched and one sentence explaining what to change

Pricing

This Actor uses Pay per event pricing — you pay for hail-day rows found, not for compute time:

Event	Price	When
Result row (result-row)	$0.005 per row ($5 per 1,000)	Each ok row = one location × one hail day
Actor start (apify-actor-start)	$0.005 per run	Once at the start of each run

• not_found and error rows are free.
• Example: the default run (Dallas, TX, last 12 months, 14 hail days) costs about $0.075.
• A 3-year hail history for one location (about 50–60 hail days in hail-prone Texas) costs about $0.30 — a fraction of a typical paid per-address hail report.
• 100 customer locations, last 12 months: about $7 in a hail-prone area like Dallas (about 14 hail days per location), less elsewhere.
• Use the maximum cost per run option when you start the Actor to cap spending. If the limit is reached, the Actor stops cleanly and says so in the status message.

Use via API & MCP (AI agents)

Apify API (HTTP). Run the Actor and get the rows in one call:

curl -X POST "https://api.apify.com/v2/acts/japan-open-data~us-hail-history/run-sync-get-dataset-items?token=<YOUR_APIFY_TOKEN>" \
  -H "Content-Type: application/json" \
  -d '{"locations": ["32.7767,-96.7970"], "daysBack": 90, "minHailSizeInches": 1}'

Python (apify-client):

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("japan-open-data/us-hail-history").call(run_input={
    "locations": [{"latitude": 32.7767, "longitude": -96.797, "label": "Customer 1042"}],
    "startDate": "2026-05-01",
    "endDate": "2026-06-30",
})
for row in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(row["status"], row.get("date"), row.get("maxHailSizeInches"), row.get("closestDistanceMiles"))

MCP (Claude, ChatGPT, Cursor and other AI agents). Add this Actor as a tool through the Apify MCP server (https://mcp.apify.com), for example with ?tools=japan-open-data/us-hail-history. Only locations is required, so an agent can call it with {"locations": ["<lat>,<lon>"]}.

Recommended agent settings:

• "Did it hail at X on date D?" → {"locations": ["<lat>,<lon>"], "startDate": "D", "endDate": "D", "radiusMiles": 3}.
• "Hail history of this property" → daysBack: 1095 (3 years), radiusMiles: 1–3, minHailSizeInches: 1 to keep only damaging-size hail.
• Pass the local timeZone of the property (e.g. America/Denver) so evening storms get the right date.
• Always check status; on not_found/error show the error sentence to the user. Remind users that results are radar estimates, not confirmed hail on the ground.

Data source, attribution and caveats

• Source: NOAA National Centers for Environmental Information, Severe Weather Data Inventory (SWDI) — dataset nx3hail (NEXRAD Level-III hail signatures). https://www.ncei.noaa.gov/products/severe-weather-data-inventory
• License: public domain (CC0 1.0, U.S. federal government data, per the data.gov catalog). NOAA asks that the dataset be cited when used as a source — every row includes source and sourceUrl.
• Radar-derived estimate, not ground truth. Hail signatures are computed automatically from radar data and describe conditions that were likely; they are not observations of hail on the ground. Actual hail size at a specific address may be smaller, larger, or absent.
• Absence of data does not mean no hail. Radar coverage gaps, beam height, outages and algorithm limits mean hail can fall without a detection.
• Not an insurance or legal determination. Use this data as supporting information for inspections and claims, not as proof of damage or a claim decision.
• Not affiliated: this Actor is an independent project and is not affiliated with, endorsed by or sponsored by NOAA, NCEI or the National Weather Service. It does not use NOAA logos.
• Access: the Actor calls only NOAA's public SWDI web service, sequentially and politely (at most 300 requests per run), and identifies itself in the User-Agent.
• No personal data: only radar station IDs, detection coordinates, times, hail estimates and the points you provide.

Known limitations

• Coordinates only. No built-in address geocoding (see "How to use it").
• Coverage. The continental U.S. and other areas covered by NEXRAD radars. Data from 1995; usually up to about 2 days ago.
• Period. At most 3 years per run; longer periods need several runs. Periods are fetched in 31-day blocks (the NOAA limit).
• Many far-apart locations. Nearby locations are fetched together; if the list spans a very large area, each location is fetched separately, which is slower. A run makes at most 300 NOAA requests — if more would be needed, the Actor returns one free error row asking you to shorten the period or split the list.
• Row cap. When maxResults is reached, remaining locations are skipped (the status message says "capped").
• Cost limit. If your maximum cost per run is lower than the price of one row, the dataset stays empty and the status message explains why.

More from Japan Open Data

• Japan Postal Code & Address in English — look up Japanese postal codes (zip codes) and addresses in English/romaji, kanji and kana, from Japan Post's official data.
• TSA Checkpoint Travel Numbers (Daily, Weekly & YoY) — daily U.S. TSA airport checkpoint traveler counts as clean JSON, with Monday-Sunday weekly averages (calculated like Kalshi's TSA market) and year-over-year change.
• Mexico Gas Prices API by Station (Official CNE) — today's regular, premium and diesel prices at every Mexican gas station from the official CNE feed, with coordinates and distance.
• US Storm Reports API: Hail, Wind & Tornado (NWS) — U.S. hail, wind, tornado and flood reports from official NWS Local Storm Reports within hours (last 7 days), filterable by state or radius.

Feedback

Found a wrong value or need another field (e.g. address geocoding, grid-based hail maps)? Open an issue on the Actor's Issues tab with your input and what you expected.