Atmostorm REST API

A single, consistent API for severe weather data from the NWS, SPC, IEM, NEXRAD, GOES, HRRR, USGS, and more. Everything is GeoJSON-first with spatial filters, real-time-ish update cadence, and free public access.

#Base URL

All endpoints are under /api/v1:

https://atmostorm.com/api/v1/

Spatial endpoints return a GeoJSON FeatureCollection; everything else returns plain JSON. The shape is fixed per endpoint and isn't negotiable via a query parameter.

#Quickstart

Get all active NWS alerts in Oklahoma:

curl "https://atmostorm.com/api/v1/alerts?state=OK&active=true"

Observations within 50 km of Norman, OK:

curl "https://atmostorm.com/api/v1/observations?point=35.22,-97.44&radius=50km"

An HRRR point forecast with the next 12 hours:

curl "https://atmostorm.com/api/v1/models/hrrr/point?lat=35.22&lon=-97.44&hours=12"

#Authentication

The API is free and open. You can call every public endpoint anonymously. Passing an API key increases your rate limit and removes per-IP throttling.

Create a key from your account dashboard (requires a free login) and send it in the X-API-Key request header:

curl -H "X-API-Key: wx_abc123…" "https://atmostorm.com/api/v1/alerts"

#CORS & browser apps

The public data endpoints send Access-Control-Allow-Origin: *, so you can call them directly from a browser app on any origin — no proxy needed. Send your key in the X-API-Key header as usual; it works cross-origin.

The X-RateLimit-* and Retry-After response headers are exposed to cross-origin JavaScript, so your client can read its own budget and back off properly.

#Rate limits

Limits are per-minute, rolling. You'll get a 429 Too Many Requests if you exceed them.

Need more? Email hello@atmostorm.com and describe your use case — we'll work with public-good projects and operational users.

#Common query parameters

Most spatial endpoints share the same filter vocabulary. Mix and match freely — they compose as an AND.

#Errors

Errors return a JSON body with a detail field:

{
  "detail": "Invalid bbox: expected 4 numbers separated by commas"
}

#Response formats

Spatial endpoints return a GeoJSON FeatureCollection by default:

{
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "geometry": { "type": "Polygon", "coordinates": [...] },
      "properties": {
        "id": 42,
        "event": "Tornado Warning",
        "severity": "Extreme",
        "expires": "2026-04-19T22:15:00Z"
      }
    }
  ]
}

For non-spatial data (e.g. HRRR point forecasts), you get plain JSON:

{
  "lat": 35.22,
  "lon": -97.44,
  "elevation_m": 365.2,
  "run": "2026-04-19T12:00:00Z",
  "times": ["2026-04-19T13:00:00Z", ...],
  "variables": { "t2m_c": [...], "cape_jkg": [...] }
}

#Update cadence

Each source runs on its own cadence — we poll, normalize, and deduplicate server-side so you don't have to:

#Terms & attribution

Almost everything we serve is a US Government work (public domain by statute) or plain meteorological fact, so we claim no rights over it. Our output is released under CC0 1.0 — use it for anything, commercial or not, no permission needed.

Attribution is a request, not a requirement: if you display a derived product, please credit the data sources — it helps people verify what they're looking at.

CC0 covers the data. It doesn't waive these terms of use — the rate limits above, no abuse, and the disclaimer below still govern access to the service.