Cloud top height and optical depth from Sentinels 5p, 4 and 5 using ROCINN

Kurzfassung

Cloud information from UV/VIS/NIR spectrometers aboard spacecraft platforms is highly valuable for the construction of cloud property climatologies, since it is complementary to the information retrieved using IR spectrometers, imagers and active sensors (cloud radars and lidars). Clouds, when present, are the main modulators of the radiation transport across the atmosphere. They are very diverse in terms of composition (ice, liquid water), thickness (thin cirri to thick cumulonimbi), variability (homogeneous stratocumuli to inhomogeneous cumuli), altitude, etc. Consequently, the modeling of the cloud-radiation insteraction is one of the most challenging tasks in the atmospheric science. The distance that photons travel across the atmosphere before being captured by a spaceborne instrument can be greatly affected by clouds. Absorption by atmospheric molecules is not only proportional to the molecular number density but also to the photon path legth. Therefore, the effects of clouds on the photon path legth is of particular interest in the atmospheric remote sensing. During the retrieval processing, changes in the absorption bands within the spectra are analyzed. These changes can be linked to variations in the molecular number density but also to variations in the photon path legth. Therefore, precise cloud information to correct for variations in the photon path length is mandatory for the accurate retrieval of atmospheric trace gases. The ROCINN algorithm retrieves cloud top height (pressure), cloud optical depth and cloud albedo rom measurements in and around the oxygen A-band (~760nm) taking as input the cloud fraction computed with the OCRA algorithm (a color space approach based on broadband reflectances). This paper presents the latest version of the ROCINN algorithm as implemented to process Sentinel-5p, Sentinel-4 and Sentinel-5 data. There are two variants of the ROCINN algorithm: one that treats clouds as reflecting boundaries (CRB)) (i.e. Lambertian equivalent reflectors) and a second one that treats louds as scattering layers (CAL). The ROCINN-CRB algorithm has been successfully used operationally for GOME/ERS-2 and GOME-2/MetOp-A and -B instruments since over one decade. The CAL algorithm treats clouds in a more realistic way and provides cloud properties closer to reality. ROCINN V3.0 in the CRB and CAL variants have been included into the latest UPAS operational processor and is the baseline algorithm for the generation of cloud products from the atmospheric Sentinels. In this work, we present and analyze preliminary ROCINN cloud properties from S5P/TROPOMI, in addition to those from GOME-2 on both MetOp-A and Metop-B. A quantitative comparison with independent cloud products, e.g. from VIIRS on NASA/Suomi-NPP, from AVHRR on MetOp and MODIS on NASA-EOS/Terra is forseen.