The Orbiting Carbon Observatory Cloud Screening Algorithm

NASA's OCO satellite mission will provide values of atmospheric carbon dioxide column (XCO2) at high temporal and spatial sampling via an inversion of radiances measured at 0.76um (Oxygen-A band) and 1.61 and 2.06 (weak and strong CO2 bands, respectively). Due to the sensitivity of radiances to small amounts of cloud and aerosol relative to XCO2 at these channels, it is necessary to provide some classification of the scenes with regards to optical thickness. Here we present a review of the OCO cloud screening process which consists of five distinct components; 1.) a test of the spatial variability provided by the high spectral resolution color slices, 2.) a comparison of OCO surface reflectances to those measured by the MODIS satellite, 3.) a clear/cloudy classification based on a simple apparent optical path difference (AOPD) fitting routine, 4.) a retrieval of cloud liquid and ice optical depths and effective scattering heights as well as column water vapor via a neural network and 5.) a photon path length PDF analysis. Together these five tests will provide information to the OCO scene selection data base which will be used to determine a select number of soundings (2% of the total) to be processed by the OCO full-physics retrieval algorithm.