HR: 1340h
AN: A13D-1512    [Abstracts]
TI: A Geostatistical Approach for Gap-filling XCO2 Using Non-Stationary Spatial Covariance Functions
AU: * Alkhaled, A A
EM: alanood@umich.edu
AF: Department of Civil and Environmental Engineering, The University of Michigan, 155 EWRE Bldg. 1351 Beal Ave., Ann Arbor, MI 48109-2125, United States
AU: Kawa, S
EM: stephan.r.kawa@nasa.gov
AF: NASA Goddard Space Flight Center, Code 613.3, Greenbelt, MD 20771, United States
AU: Michalak, A M
EM: amichala@umich.edu
AF: Department of Civil and Environmental Engineering, The University of Michigan, 155 EWRE Bldg. 1351 Beal Ave., Ann Arbor, MI 48109-2125, United States
AU: Michalak, A M
EM: amichala@umich.edu
AF: Department of Atmospheric, Oceanic and Space Sciences, The University of Michigan, 183 EWRE Bldg. 1351 Beal Ave., Ann Arbor, MI 48109-2125, United States
AB: The Orbiting Carbon Observatory (OCO) will measure global column-averaged CO₂ dry air mole fraction (X_CO2). However, geophysical sampling limitations (e.g. clouds, aerosols) are expected to cause large gaps in the satellite data product. To reduce the percentage of contaminated soundings, OCO is designed to measure at high resolutions (3km² at the Nadir). Although this high measurement resolution will improve the clear sky sounding probability, the retrieved data product will still have large data gaps. The availability of a high precision gap filled product, together with an accurate assessment of its associated uncertainty, would allow for improved estimations of CO₂ sources and sinks. Furthermore, gap-filled global maps would provide much needed information for validating both biospheric and atmospheric transport models of CO₂ over gap- filled areas. This work examines the feasibility of producing high precision X_CO2 fields in the presence of geophysical sounding limitations. First, global high resolution maps of geophysical sampling limitations are constructed using measurements from satellites, such as MISR and MODIS. These maps are superimposed on an OCO global sounding grid of 0.5° resolution to identify clear sky locations. Second, a geostatistical spatial interpolation approach (i.e. kriging) is adopted to evaluate the expected precision of future satellite X_CO2 gap-filled maps, in the presence of realistic sounding retrieval errors. Filling X_CO2 gaps caused by the geophysical sounding limitations raises a number of challenges including the required high resolution and the global scale of the analysis, as well as the expected differences in X_CO2 variability over various geographical regions. To handle these challenges, the adopted spatial interpolation approach emphasizes the characterization of local variability and the computational feasibility of the analysis. Simulated X_CO2 concentrations are used to characterize the spatial variability of X_CO2 using flexible non-stationary covariance functions. This representation of the covariance structure aims to capture regional changes in X_CO2 variability. An optimal spatial interpolator (i.e. kriging) is then used together with the non-stationary covariance functions to gap-fill the X_CO2 product over areas contaminated by clouds and/or aerosol, on a 0.5° resolution grid, and to provide an estimate of the interpolation precision. Overall, the presented work provides both an evaluation of the effects of geophysical limitations on the precision of future satellite X_CO2 data products and a geostatistical gap-filling approach that is able to estimate the uncertainty associated with the resulting spatiotemporal distribution of X_CO2.
DE: 0428 Carbon cycling (4806)
DE: 0480 Remote sensing
DE: 0490 Trace gases
DE: 3252 Spatial analysis (0500)
DE: 3275 Uncertainty quantification (1873)
SC: Atmospheric Sciences [A]
MN: 2007 Fall Meeting