A Cloud Detection Method Combining ATMS Measurements and CrIS Hyperspectral Infrared Data at Double Bands

It is important to ensure the accuracy of cloud detection in the infrared hyperspectral measurements before they are used for geophysical retrievals or data assimilation; otherwise, larger errors will occur. Therefore, this paper introduces a method for cloud detection by using the Crosstrack Infrared Sounder (CrIS) hyperspectral radiances in the longwave (709.5-746.0cm^-1) and shortwave (2190-2250cm^-1) channels and the Advanced Technology Microwave Sounder (ATMS) measurements. Specifically, nine field of views (FOVs) within one field of regard (FOR) are preliminarily assigned as clear or cloudy by comparing the measured IR radiances and simulated IR clear radiances which generated from the MWretrieved geophysical state vector based on a physical inversion method. The number of cloud formations within the CrIS fieldof-regard (FOR) is then estimated using the principal component analysis (PCA). Next, CrIS radiances at longwave channels and shortwave bands are used to evaluate the thermal contrast within the FOR. Based on the above-mentioned cloud information extracted from the observations, each FOR will finally have a cloud mask classification assigned to it. Further, depending upon the cloud mask designation, a common cloudclearing method is performed on the FOR, which estimates clear radiance by using multiple FOV observations, for improved radiance in geophysical retrieval or data assimilation.