HailScore Data Coverage · Methodology v6Hailscore LLC · Updated hourly from production · 7.4M+ total records

What's in the arsenal

7.4M+ records, four government sources.

Every address lookup, every score, every API call resolves against the four data sources below. All sources are public-domain. Counts refresh hourly directly from production.

§I

Master totals by source

NEXRAD SWDI

radar_hail

NOAA NEXRAD Level-III hail-signature records (nx3hail) via the Severe Weather Data Inventory, filtered to ≥0.75″ severe threshold.

2015 → present

5,512,903

records

NCEP MRMS

mrms_mesh

Multi-Radar Multi-Sensor MESH-MAX-1440min product. Operational hail-size estimates from the National Centers for Environmental Prediction.

Live (sub-2hr latency, 2026 → present)

1,558,287

records

NOAA Storm Events DB

storm_events

Verified hail occurrences with NWS narrative text, magnitudes, and event identifiers. The carrier-grade ground-truth source.

2000 → present (all 50 states + territories)

303,457

records

CoCoRaHS

cocorahs_reports

Community Collaborative Rain, Hail and Snow Network — ground-truth reports from volunteer spotters.

2015 → present

10,161

records

§II

Top hail states — NOAA Storm Events

State-level distribution of verified hail occurrences in the NOAA Storm Events database. All 50 states + territories are covered; the top 20 by event count are shown below. Severe (≥1.0″) counts are a subset of the total.

#StateHail events≥1.0″Max size
1Texas34,04024,8057.02
2Kansas28,15218,5367.75
3Nebraska18,31512,9017.00
4Oklahoma16,64311,0076.00
5Colorado15,24510,6805.25
6Missouri14,9109,1086.00
7South Dakota13,5609,1498.00
8Iowa12,3367,7355.50
9Minnesota11,7896,9816.00
10Illinois9,3835,5094.75
11Arkansas7,4544,4245.00
12North Dakota7,4094,9325.00
13Wisconsin7,0444,0765.50
14North Carolina7,0293,6644.50
15Ohio6,5933,2165.00
16Montana6,4794,4864.50
17Georgia5,7923,2164.25
18Indiana5,6743,3684.50
19Alabama5,5103,3205.38
20Mississippi5,4743,3324.88
§III

Temporal coverage

NOAA Storm Events go back to 2000 (with earlier scattered records). Radar-derived data starts at the operational SWDI cutover in 2015. Live MRMS ingestion is current to within 2 hours.

Year 2015

9,398

verified hail events

Year 2016

9,407

verified hail events

Year 2017

10,398

verified hail events

Year 2018

7,863

verified hail events

Year 2019

8,992

verified hail events

Year 2020

7,677

verified hail events

Year 2021

6,268

verified hail events

Year 2022

7,180

verified hail events

Year 2023

11,761

verified hail events

Year 2024

8,938

verified hail events

Year 2025

9,208

verified hail events

Year 2026

185

verified hail events

Shown: last 12 years of NOAA Storm Events. Complete 2000–present available via API at /v1/property/history. Live data freshness per source documented at /status.

§IV

Source attribution

All HailScore data ingestion is sourced from publicly accessible government archives. No proprietary or unattributed data is incorporated.

NOAA NCEI
National Centers for Environmental Information. Storm Events Database + Severe Weather Data Inventory.
ncdc.noaa.gov/stormevents
NCEP MRMS
National Centers for Environmental Prediction. Multi-Radar Multi-Sensor operational hail product.
nssl.noaa.gov/projects/mrms
NWS
National Weather Service. Active Alerts API (warning polygons + VTEC strings).
weather.gov/documentation/services-web-api
SPC
Storm Prediction Center. Local Storm Reports (preliminary spotter reports).
spc.noaa.gov/climo/reports
CoCoRaHS
Community Collaborative Rain, Hail and Snow Network. Volunteer ground-truth reports.
cocorahs.org
§V

What we don't cover

Honest scope statement, not a marketing one.

  • ·NOAA Storm Events pre-2000 is sparse (scattered records back to 1955 for some states). Modern ingestion starts at 2000 because that's when NOAA's electronic record-keeping became consistent.
  • ·Radar-derived data starts in 2015, when NEXRAD SWDI became operationally stable. Pre-2015 radar exists in NOAA's archive but isn't in the same product format.
  • ·Live MRMS data starts 2026 with our operational cutover. Historical MRMS (pre-2026) isn't backfilled; the SWDI archive serves the historical use case.
  • ·NWS Alert + SPC LSR per-event citations are listed in the methodology as referenced sources, but per-event publisher identifiers are not yet exposed via the v1 API. Both are roadmap items.
  • ·Non-hail perils (wind, tornado, flood, fire) are out of scope. HailScore is a hail-specific product. NOAA exposes these in the same Storm Events feed; we filter to event_type='Hail' at ingestion.
  • ·International addresses are out of scope. HailScore covers US addresses only (50 states + territories). Non-US queries fail at geocoding.

─── Maintained

Coverage counts on this page refresh hourly directly from the production database. For live data-freshness per source see /status. For the methodology white paper see /methodology. For carrier API integration see /api or /for-carriers.

Source data is public-domain US government records. HailScore's contribution is the ingestion pipeline, address-level spatial indexing, calibrated scoring, and API surface — none of which is incorporated by reference into the counts above.