Title: Geostatistical Data Fusion for Remote Sensing Applications
Primary Author: Braverman, Amy
Organization: Jet Propulsion Lab
Co-Author(s): Hai Nguyen (JPL), Noel Cressie (Ohio State U), Matthias Katzfuss (Ohio State U), Edward Olsen (JPL), Rui Wang (Ohio State U), Anna Michalak (U of Michigan), and Charles Miller (JPL)

Remote sensing data are often sparse relative to the space-time domains of geophysical processes: no instrument observes everywhere, all the time. Remote sensing data are also massive, taken over different, usually non-nested spatial footprints, and subject to measurement error biases. Our goal is to infer the true values of a spatially and temporally continuous geophysical quantity from these aggregated, noisy, and heterogeneous observations. We do this through a geostatistical model that relates the observations to the true but not directly observed variable of interest. This model accounts for spatial and temporal correlations, and relationships among different resolutions. Crucially, our estimates are accompanied by uncertainty measures so we know how reliable the estimates are. In this talk, we review the basic principles of our methodology, as they apply to estimating column carbon dioxide in the planetary boundary layer, inferred from remote sensing observations taken at multiple scales with differing sensitivities and sampling characteristics.