A41I-0182
Evaluation of Cloud and Aerosol Screening of Early Orbiting Carbon Observatory-2 (OCO-2) Observations with Collocated MODIS Cloud Mask

Thursday, 17 December 2015
Poster Hall (Moscone South)
Thomas Taylor1, Christopher O'Dell2, Heather Q Cronk3, Phil Partain3, Christian Frankenberg4, Annmarie Eldering5, David Crisp5, Michael R Gunson5, Robert R Nelson6, Albert Chang4, Brendan Fisher5, Gregory B Osterman5, Harold R Pollock5, Andrey Savtchenko7 and Emily J Rosenthal8, (1)Colorado State University, Atmospheric Science, Fort Collins, CO, United States, (2)Colorado State University, Fort Collins, CO, United States, (3)Cooperative Institute for Research in the Atmosphere, Fort Collins, CO, United States, (4)NASA Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, United States, (5)NASA Jet Propulsion Laboratory, Pasadena, CA, United States, (6)Colorado State University, Atmospheric Sciences, Fort Collins, CO, United States, (7)NASA Goddard Space Flight Center, ADNET, Greenbelt, MD, United States, (8)Millersville University of Pennsylvania, Millersville, PA, United States
Abstract:
Effective cloud and aerosol screening is critically important to the Orbiting Carbon Observatory-2 (OCO-2), which can accurately determine column averaged dry air mole fraction of carbon dioxide (XCO2) only when scenes are sufficiently clear of scattering material. It is crucial to avoid sampling biases, in order to maintain a globally unbiased XCO2 record for inversion modeling to determine sources and sinks of carbon dioxide. This work presents analysis from the current operational B7 data set, which is identifying as clear approximately 20% of the order one million daily soundings. Of those soundings that are passed to the L2 retrieval algorithm, we find that almost 80% are yielding XCO2 estimates that converge.

Two primary preprocessor algorithms are used to cloud screen the OCO-2 soundings. The A-Band Preprocessor (ABP) uses measurements in the Oxygen-A band near 0.76 microns (mm) to determine scenes with large photon path length modifications due to scattering by aerosol and clouds. The Iterative Maximum A-Posteriori (IMAP) Differential Optical Absorption Spectroscopy (DOAS) algorithm (IDP) computes ratios of retrieved CO2 (and H2O) in the 1.6mm (weak CO2) and 2.0mm (strong CO2) spectral bands to determine scenes with spectral differences, indicating contamination by scattering materials. We demonstrate that applying these two algorithms in tandem provides robust cloud screening of the OCO-2 data set.

We compare the OCO-2 cloud screening results to collocated Moderate Resolution Imaging Spectroradiometer (MODIS) cloud mask data and show that agreement between the two sensors is approximately 85-90%. A detailed statistical analysis is performed on a winter and spring 16-day repeat cycle for the nadir-land, glint-land and glint-water viewing geometries. No strong seasonal, spatial or footprint dependencies are found, although the agreement tends to be worse at high solar zenith angles and for snow and ice covered surfaces.

<< Previous Abstract | Next Abstract >>