Isotherm
Climate data platform

Temperature is a field, not a number.

Isotherm fuses 41,000 ground stations, satellite passes, and reanalysis models into one continuous temperature surface, so you can see where heat is gathering hours before the monthly average admits it.

Get an API key Read the method
+1.48°C
Global mean anomaly, base 1991 to 2020
41,207
Stations ingested
1km
Grid resolution
52min
Median latency
ISO-9 temperature field / heat dome, 34.2°C peak
Isotherm contour field Nested temperature contour lines forming a heat dome, warm orange at the core fading to teal at the cooler edges, drawn like a scientific weather chart with labeled isotherms and observation stations.
How the surface is built

Most dashboards give you dots. We give you the whole field.

A station is a pinprick. The weather between two of them is a guess. Isotherm makes that guess explicit, gridded, and queryable, then tells you how sure it is.

01 / interpolation

One surface, not scattered dots

Every station and satellite pass is kriged onto a single 1 km grid, not snapped to its nearest neighbor. Where coverage is thin, cells carry honest uncertainty instead of a silent zero.

kriging1 km gridhourly
02 / anomalies

Every cell carries its own baseline

Each grid cell remembers its 1991 to 2020 normal, so you read departures in warm and cool at a glance. A mild-sounding 24°C in February still lights up when it should.

per-cell basewarm / coolz-score
03 / access

Built to be queried

Pull any cell, any hour, back to 1880 through a typed API. Isotherm answers in GeoJSON, Arrow, or CSV, with the provenance of every reading attached to the value.

GeoJSONArrowCSV
isotherm / field.http
# any point, any moment
GET /v1/field?lat=48.85&lon=2.35&at=now

{ "temp_c": 24.1, "anomaly_c": +3.2,
  "base": "1991-2020", "grid_km": 1,
  "src": ["stn","sat","era5"] }
The signal

144 years, one glance.

Each ridge is a decade of global surface temperature, drawn back to front. The cool teal band at the back is the late 1800s. The warm crest pushing forward is now. Nobody needs the trend explained twice.

Decadal temperature ridgeline A stacked ridgeline chart. Fourteen overlapping ridges run from back to front, cool teal at the top for the earliest decades warming to bright orange at the front for the most recent, with each crest sitting higher than the one behind it.
coolerwarmeranomaly vs 1991 to 2020
  • 01The baseline is the argument. We anchor every reading to the 1991 to 2020 normal, the same window the field standardized on, so the crest is not a chart trick. It is the departure.
  • 02No smoothing you cannot switch off. Every ridge is a real decadal mean you can pull back out as numbers. The picture and the table are the same data.
  • 03Read it at any scale. The same surface that draws this global ridge answers a query for one city block. Zoom is a parameter, not a different product.
Start reading the field

Point a query at any spot on Earth.

The free tier includes 10,000 cell queries a month and the full archive back to 1880. No card, no sales call, no throttling on the history.

Get an API key Browse the live map

curl https://api.isotherm.dev/v1/field?lat=..&lon=..