David Duncan, Scientist, Microwave Radiance Observations; Niels Bormann, Principal Scientist and Team Leader, Microwave Radiance Observations
On 10 July 2025, ECMWF started operational assimilation of data from the Arctic Weather Satellite (AWS), a trail-blazing small satellite launched by the European Space Agency (ESA) that provides new microwave sounding measurements from space.
The data are combined, together with lots of other observations, with a short-range forecast constrained by previous observations to obtain the best possible estimate of the current state of the Earth system. That estimate, called the analysis, is used as the initial conditions on which weather forecasts are based.
ECMWF is the first centre to use these new observations, and they provide a robust forecast improvement. The microwave radiometer carried by the satellite complements similar sensors on much larger platforms provided by organisations such as the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT), the US National Oceanic and Atmospheric Administration (NOAA) and the China Meteorological Administration (CMA).
For the first time, AWS also measures in a spectral band known as ‘sub-mm’, with wavelengths smaller than 1 mm, giving new information on ice clouds. The satellite demonstrates that good quality and highly impactful passive microwave observations can be made from a small and comparatively economical satellite.
The AWS is a precursor to the EPS-Sterna constellation proposed by EUMETSAT, which consists of six such satellites in polar orbit and promises strong forecast benefits in the future.
The Arctic Weather Satellite of the European Space Agency. © ESA/Mlabspace
New space approach
Since the beginning of the satellite era in meteorology, most instruments have been flown on car- or bus-sized satellites but more recently, technological developments have made it possible to put high-performing instruments onto satellites small enough to fit into the boot of a car.
ESA has embraced these so-called ‘new space’ ideas with the AWS. The satellite is about the size of a washing machine (125 kg and 1 cubic metre) and was launched into polar orbit on a ride-share rocket in August 2024. Despite its size, the payload of AWS is a microwave radiometer with performance suitable for assimilation in numerical weather prediction (NWP).
The AWS radiometer uses three key absorption bands to measure atmospheric profiles of temperature, humidity, and clouds. Like the heritage instruments ATMS, MHS, and AMSU-A that are already assimilated at ECMWF, AWS exploits the 50 and 183 GHz bands for temperature and humidity profiling, respectively.
In addition, AWS has a novel set of channels at the 325 GHz band, which provides similar humidity sensitivity as 183 GHz whilst providing more information on cirrus clouds and snow. AWS is the first mission to measure at 300+ GHz, also known as the ‘sub-mm’ band.
Improved forecasts
The beneficial impact of AWS assimilation can already be seen in several fields. The figure below shows the normalised change in root-mean-square error (RMSE) for winds at one- and three-day lead times, compared to ECMWF's analysis, for a six-month experiment. We observe improved wind forecasts at high latitudes that persist through forecast day three.
Normalised change in wind RMSE at forecast lead times of 24 and 72 hours caused by adding AWS to the assimilation system. Blue colours indicate improvements (i.e. reduced forecast error). Hatching shows statistically significant changes at the 95% confidence level. Experiments cover 1 January to 30 June 2025, using Cycle 49r1 of ECMWF’s Integrated Forecasting System (IFS) with a 27 km model resolution.
These improvements are also reflected in humidity, temperature, and geopotential fields. Independent observations also indicate that short-range forecasts of temperature, humidity, and winds are improved due to AWS assimilation.
We assimilate channels from all three sounding bands (50, 183, 325 GHz) on AWS. This marks the first time that sub-mm channels have been assimilated in a global NWP model.
An example of observations assimilated in a 12-hour period can be found in the next figure, showing assimilated data from a date in mid-July. AWS is assimilated in ‘all-sky’ conditions, meaning that observations from clear skies to heavy precipitation are all used in the assimilation system. This enables us to make use of the new sub-mm observations, with their unique sensitivity to ice clouds.
Observations minus a short-range forecast (background) for AWS channel 13 (180.31 GHz), showing points assimilated in the operational long-window data assimilation, centred on 13 July 00 UTC. Observations have been bias-corrected.
Previous work has shown that the addition of microwave sounders to data assimilation continues to improve forecast skill. This experience with AWS is consistent with that earlier work, despite the small size of AWS.
In addition, the results here are consistent with earlier simulation studies that focused on an AWS-class instrument, indicating that a constellation of small satellites like AWS could greatly improve forecasts.
New ‘sub-mm’ observations
A novel and very exciting aspect of AWS is the advent of higher frequency (shorter wavelength) channels that have greater sensitivity to ice clouds. The information from these new channels can better constrain the distribution of ice clouds in atmospheric models.
In a way, the 325 GHz channels on AWS are a preview of the Ice Cloud Imager (ICI) from EUMETSAT, which will feature even greater cirrus sensitivity, obtained from its higher frequencies. The first ICI is expected to be launched in 2026.
In the last figure, the increased cloud sensitivity from the new sub-mm channels can be seen in a tropical cyclone. Compared to previously available frequencies at 89 and 183 GHz, the new channel at 325 GHz observes much colder brightness temperatures, due to greater scattering of radiation by snow and cloud ice. In contrast, 89 GHz has little sensitivity to cloud ice and instead primarily sees larger drops lower down in the atmosphere.
Observed brightness temperatures from AWS channels at 89 GHz (top), 180 GHz (middle), and 325±6.6 GHz (bottom) on 9 November 2024 south of Hong Kong. This was Typhoon Yinxing, which had passed by the Philippines and later made landfall in Vietnam.
Looking ahead
The success of AWS demonstrates the potential of small-satellite constellations for improving forecast skill. Simulation studies had indicated this potential, but it was not clear whether in-orbit performance from a small satellite would be sufficient to achieve positive impacts in NWP until AWS was able to prove this.
EUMETSAT has proposed a constellation of small microwave sounders like AWS in a mission to complement the main system of polar-orbiting backbone satellites, EPS-Sterna. If this mission goes ahead, there would be six AWS-type satellites in orbit at any given time, providing unprecedented spatiotemporal coverage of all-weather atmospheric sounding to drive better weather forecasts.
Top banner image: © ESA/Mlabspace