How it works

About the stations

From sensor to forecast — how weather data is produced

Météo-France’s Global Observation Network

While primarily focused on mainland France through the RADOME and ÉTENDU networks, Météo-France observations extend far beyond it. Météo-France operates and contributes to a broad global observing network to ensure the quality and accuracy of international forecasts, because the atmosphere has no borders.The organization maintains automatic stations across its overseas territories (Caribbean, French Guiana, Réunion, French Polynesia, New Caledonia, etc.), totaling several hundred sites overall (around 1,150 automatic surface stations across all French territories).These stations are vital nodes of the WMO’s Global Observing System (GOS). Their strength lies in their ability to provide consistent and standardized data (in formats such as BUFR), enabling global numerical models to map the weather reliably and anticipate phenomena that will later affect mainland France.

The Core of Forecasting: Automatic Weather Stations

To produce reliable forecasts, meteorologists need an accurate picture of the atmosphere right now. That is the role of Météo-France’s Automatic Weather Stations (AWS).

1. Manufacturers: High-End French Engineering (example: STERELA)

Unlike a smartphone or a car, a weather station doesn’t have a single consumer-facing “brand”. Météo-France relies on highly specialized companies to design and integrate these instruments.

Manufacturer
One of Météo-France’s major suppliers is the French company STERELA.
Technical name
STERELA offers ranges such as Mercury stations, adapted to different use cases (synoptic stations for general forecasting, road weather, agrometeorology).
What they provide
These companies don’t just install sensors: they deliver a complete system (sensors, an acquisition unit, and communications).

2. Météo-France Networks: RADOME and ÉTENDU

The names RADOME and ÉTENDU refer to the networks used to organize these stations and ensure efficient coverage.

RADOME
Extended Meteorological Data and Observation Acquisition Network (historical name of the main network).
The core stations providing baseline variables (temperature, wind, pressure, humidity) at high frequency (as often as every 6 minutes for some). It forms the backbone of Météo-France surface observations.
ÉTENDU
A broader term referring to the full set of Météo-France automatic surface stations.
It includes RADOME, as well as partner stations or stations dedicated to specific uses (road weather, mountain sites, etc.).

In short: when a measurement comes from one of these networks, it is certified and standardized according to national and international requirements.

3. What the Station Measures (Sensors)

A station is a set of high-precision instruments that measure key atmospheric variables. Accuracy is critical, which requires specialized sensors.

Temperature
Typically a platinum resistance probe (Pt100) placed in a ventilated shelter (to avoid solar heating) at 2 meters above ground, following WMO standards.
Wind
Increasing use of ultrasonic anemometers (no moving parts). More accurate and robust than cup anemometers, mounted on 10-meter masts in open exposure.
Humidity
High-precision sensors (capacitive polymer) to measure relative humidity and compute the dew point (useful for fog/dew).
Precipitation
Tipping-bucket rain gauges measuring volume with fine resolution (e.g., 0.2 mm per tip), to track totals and real-time intensity.
Pressure
Very stable pressure sensors for atmospheric pressure—fundamental for tracking highs and lows (anticyclones and depressions).

4. The Process (from sensor to app)

  • Acquisition: each sensor sends electrical signals to the acquisition unit (the station enclosure).
  • Processing: the unit converts these signals into physical values (using calibration curves) and stores them.
  • Transmission: every 6 minutes (RADOME network), the unit sends data via GPRS, 4G, or satellite to Météo-France’s collection systems in Toulouse.
  • Distribution: after validation, data is fed into forecast models and used by forecasters, then made available to users, such as on kstl.fr.