Using Machine Learning Techniques for the Retrieval of Cloud Information from Polar, Geostationary and Deep Space UV/VIS/NIR Instruments

TROPOMI on board Sentinel-5 Precursor (polar, launched in 2017), Sentinel-5 (polar, to be launched in 2024) and Sentinel-4 (geostationary, to be launched in 2024) are atmospheric Copernicus missions dedicated to air quality monitoring and focus on trace gas, greenhouse gas, aerosol and cloud retrievals in the UV/VIS/NIR spectral region. EPIC/DSCOVR (launched in February 2015) is a NASA mission located at the Lagrangian point L1 at a distance of 1.5 million km and comprises 10 wavelength bands in the UV/VIS/NIR spectral region, which we use to extract cloud information. The Korean GEMS mission (geostationary, launched in 2020) covers the UV/VIS and forms, together with Sentinel-4 and TEMPO, a geostationary constellation for air quality monitoring that covers large parts of the globe. For a precise retrieval of trace- and greenhouse gases as well as aerosol properties, a good knowledge about the presence and characteristics of clouds is required. Furthermore, detailed knowledge about cloud conditions and their long-term behavior is an important contribution to quantifying the Earths radiation budget, and hence, their impact on climatological applications. In this contribution, we summarize the algorithms for retrieving cloud properties from the aforementioned sensors. These algorithms are called OCRA (Optical Cloud Recognition Algorithm) and ROCINN (Retrieval Of Cloud Information using Neural Networks) and both have their heritage with GOME/ERS-2 and GOME-2 MetOp-A/B/C, where they have already been successfully implemented in an operational environment. Radiometric cloud fractions are determined using a color space approach, and cloud-top height, cloud optical thickness and cloud albedo are retrieved from NIR measurements in and around the oxygen A-band. The cloud parameters are provided for two different cloud models: one which treats clouds more realistically as layers of scattering water droplets (clouds-as-layers, CAL), and one which treats clouds as simple Lambertian reflectors (clouds-as-reflecting boundaries, CRB). The latest improvements of the OCRA and ROCINN algorithms are presented along with recent results from TROPOMI.