Embedded Files
What is data assimilation?
What is data assimilation?
In order to make weather forecasts we need an accurate estimate of the current state of the atmosphere, in the form of a 3D distribution of temperature, pressure, humidity, wind and clouds. This is achieved by data assimilation: ingesting observations from satellites, weather balloons, surface stations and aircraft into a weather forecasting model. Satellite do not directly measure the atmospheric properties we need, so the observations (typically the radiance emerging from the atmosphere at a particular wavelength) are simulated from the model fields, and where they differ from the actual observations, the data assimilation system adjusts the atmospheric properties to get a better fit.
Naturally, clouds and rainfall are a crucial part of a weather forecast, but the only cloud-sensitive observations that are routinely assimilated into most weather models are microwave radiances, which lack information on the detailed vertical structure of clouds and precipitation. This is where EarthCARE shines: its radar and lidar measure clouds with a vertical resolution as high as 100 m, and both the level-1 observations and level-2 retrievals are available in near-real-time. And it's not just clouds: the lidar observations of aerosols are well suited for assimilation into air-quality forecasts.
Why assimilate EarthCARE data?
Why assimilate EarthCARE data?
The European Centre for Medium-Range Weather Forecasts (ECMWF) began assimilation of EarthCARE's Cloud Profiling Radar (CPR) on 10 June 2026 as illustrated in the image above. This is the first time that a weather forecasting centre has assimilated cloud radar observations operationally. The process involves simulating the CPR radar reflectivity from the model fields, accounting for the detailed scattering properties of the various cloud and precipitation particles, and the attenuation of the signal as it propagates through the atmosphere. This is demonstrated below for a 1700 km slice through thunderstorms in the vicinity of Mexico City on 1 July 2026. We see that before any data were assimilated, the simulated radar reflectivity from the model shows far too widespread cloud and rainfall. After assimilation of EarthCARE and other data, the simulated radar reflectivity features are aligned much more with the actual CPR observations indicating a much better placement of thunderstorms in the model. It has been found that EarthCARE observations have a modest but positive impact on forecast skill.
Work is underway to exploit more of the EarthCARE observations in data assimilation and to assimilate EarthCARE data into other forecast models. A scientific working group is being formed to share experience and code between different forecasting centres.
See also
See also
The system to assimilate cloud radar and lidar data at ECMWF was developed through a series of joint ECMWF-ESA projects led by Marta Janisková. The method is described in the two-part paper Fielding and Janisková (2020) and Janisková and Fielding (2020), with testing performed using CloudSat and CALIPSO data.
Page updated
Google Sites
Report abuse