Because of IMF participation in Copernicus atmosphere missions, the future course of the team's work in the areas of radiative transfer, scattering, and retrieval is mapped out until well into the 2030s. The sensors employed will have more sophisticated features than the passive spectrometers available in the past. This includes spatial resolutions higher by a factor of 100 and the capability to use geostationary orbits.
Higher spatial resolution requires managing a higher data rate while meeting the same or even more stringent specifications on processing time. In addition, effects become relevant that could be ignored in coarser ground pixels, such as incomplete pixel illumination or partial cloud cover. A geostationary orbit for monitoring Europe means new line-of-sight geometries and variable illumination conditions.
Our radiative transfer and retrieval expertise must be further developed to meet all these new requirements. Algorithms and procedures must be significantly accelerated and will in the future involve more use of machine learning methodologies. Techniques for the compressed simulation of spectral range data and for porting radiation models to fast hardware components will also be needed. By expanding current procedures in order to handle multidimensional radiative transfer, take polarisation into account, and connect atmosphere and waterbody factors in atmospheric correction, the effects of spatial inhomogeneity can be better identified. In addition, the scattering processes developed at IMF for retrieving aerosol and cloud characteristics are to be incorporated. This will effectively take into account scattering processes involving aerosol particles and cloud droplets, something which will be more important for future sensors.
The Next Generation Retrieval team was established in order to do justice to these atmosphere retrieval requirements for the Big Data era in earth observation, Their work is intended to lay the groundwork for the algorithms and processors that will become part of operational systems.