Storm Tracker API Integration Guide
Storm Tracker API Integration Guide
Storm Tracker is an OpenWeather API that returns everything you need to draw active tropical cyclones (hurricanes, typhoons, tropical storms) on a map in a single GeoJSON response.
For every active storm, one request gives you:
You do not need to download or decode raw model data, resolve storm identifiers or average models yourself. The service merges the sources, drops outlier forecasts, normalizes model names and returns a ready-to-draw result.
Basin code | Region | Local name |
|---|---|---|
AL | Atlantic | hurricane |
EP | Eastern Pacific | hurricane |
CP | Central Pacific | hurricane |
WP | Western Pacific | typhoon |
IO | Indian Ocean | cyclone |
SH | Southern Hemisphere | cyclone |
A storm stays in the response while its last observed position is no older than 3 days. Storm IDs follow the standard pattern, e.g. al052024 — the 5th Atlantic storm of the 2024 season.
The service is based on forecast models and official advisories only; it does not use radar or satellite imagery.
How to get the data
All data comes from a single endpoint. Pass your OpenWeather API key in the appid parameter. The default call returns the consensus product for every active storm:
GET https://api.openweathermap.org/data/2.5/storms/geojson?appid={API key}Use the feature_types parameter to request only the layers you draw. This keeps the payload small.
GET https://api.openweathermap.org/data/2.5/storms/geojson?appid={API key}&feature_types=current_position,consensus_track,uncertainty_cone,forecast_coneRequest | What you get | Typical use |
|---|---|---|
| Consensus track, cones, current position and wind zone for all storms | Default product map |
| Only the listed feature types | Lightweight map layers, mobile |
| Everything, including each model's own track | Expert / comparison views |
| One model only (also AVNX, UKM, HWRF, …) | Model comparison, debugging |
In single-model mode, storms that this model does not forecast are omitted. Single-model mode is not the consensus — it shows that model's raw track only.
The response is a standard GeoJSON FeatureCollection. The metadata block tells you when the snapshot was built and how many storms are active; each feature carries a feature_type property that tells you how to draw it.
{
"type": "FeatureCollection",
"metadata": {
"generated_at": "2024-09-10T18:05:00Z",
"sources": ["atcf", "ecmwf", "nhc"],
"active_storms": 3,
"mode": "consensus"
},
"features": [ ... ]
}Coordinates are [longitude, latitude] with longitude in the −180…180 range, as in any GeoJSON.
The underlying forecasts update every 6–12 hours and the service rebuilds its snapshot about once an hour. Polling once every 30–60 minutes is enough; there is no benefit in polling more often. Check metadata.generated_at to see whether the snapshot has changed.
If the service is rebuilding, you may briefly receive the previous snapshot instead of waiting — this is expected and safe to display.
How to integrate the data into your map
Every feature in the response has a feature_type property. Switch on it to decide how to draw the feature. A typical product map uses the first six rows; the rest are optional.
feature_type | Geometry | What it shows | Suggested rendering |
| Point | Where the storm center is now | Marker / eye icon |
| Polygon | Zone with winds ≥ 34 kt right now | Filled area around the marker |
| LineString | Most likely path (OpenWeather consensus) | Main forecast line |
| Point | Forecast position at +24 h, +48 h, +72 h ... | Dots on the line with hour labels |
| Polygon | How much the models disagree | Translucent cone around the line |
| Polygon | Official NHC cone (Atlantic / Pacitic only) | Separate toggleable layer |
| LineString | Each model's own track (feature_types=all) | Thin colored lines, off by default |
| Line / Point / Polygon | Single-model track, points and wind zone (?model=...) | Expert views only |
al052024), name and basin — group all features of one storm by its ID.tau_h): hours after the model run start. Forecast points are available at 0, 6, 12, 24, 36, 48, 72, 96, 120, 168 and 240 hours.wind_kts) and category: see the tables below.radii_ne_nm, radii_se_nm, radii_sw_nm, radii_nw_nm): how far the 34 kt wind zone extends in each quadrant, in nautical miles. The polygon is already built for you; these numbers are for labels or tooltips.forecast_cone has source: "nhc", so you can label it as the official product.Tropical meteorology uses nautical units. Convert them for display if your users expect metric.
Unit | Equals | Example |
Knot (kt) | = 1.85 km/h = 0.51 m/s | 64 kt = 119 km/h, the Category 1 threshold |
Nautical mile (nm) | = 1.85 km | Wind radius 120 nm = 222 km |
The category is derived from the maximum sustained wind in knots. Use it to color markers and tracks.
Category | Wind (kt) | Meaning |
-1 | unknown | No wind data |
0 | < 64 | Tropical storm or depression |
1 | 64-82 | Hurricane, Category 1 |
2 | 83-95 | Category 2 |
З | 96-112 | Major hurricane, Category 3 |
4 | 113-136 | Category 4 |
5 | ≥ 137 | Category 5 |
uncertainty_cone and forecast_cone can look alike but are independent products. Show them as separate layers with different styling and labels.
feature_types your map needs.forecast_cone, uncertainty_cone, wind_radii, consensus_track, consensus_point, current_position.tau_h and wind speed.metadata.generated_at changes.Frequently asked questions
Where does the forecast come from?
Why does a storm appear with a new ID?
Why is there no forecast_cone for some storms?
Why has the response not changed in the last hour?
Why are some forecast points missing at longer lead times?
Are 50 kt and 64 kt wind zones available?
Can I get the data for one model only?