Uncertainty Quantification for the Orbiting Carbon Observatory-2 Remote Sensing Retrieval

• Activity Number: 24
• Type: Topic Contributed
• Date/Time: Sunday, August 9, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Jonathan Hobbs* and Amy Braverman and Noel Cressie and Robert Granat and Michael Gunson
• Companies: California Institute of Technology and Jet Propulsion Laboratory and University of Wollongong and Jet Propulsion Laboratory, California Institute of Technology and Jet Propulsion Laboratory, California Institute of Technology
• Keywords: inverse problem ; Bayesian inference ; carbon dioxide ; surrogate model ; uncertainty quantification ; remote sensing

Abstract

The Orbiting Carbon Observatory-2 (OCO-2) satellite launched in 2014, and the OCO-2 mission aims to provide estimates of atmospheric carbon dioxide at fine spatial and temporal resolution with global coverage. These remote sensing data present a unique opportunity for understanding the carbon cycle, but they also present an important challenge for scientific inference because the measurements are indirect. The measurements of reflected sunlight must be combined with an understanding of the underlying physical processes of radiative transfer to infer the atmospheric state. A retrieval algorithm automates this process of transforming observed radiances into estimates of geophysical variables, and several sources of uncertainty are introduced in the process. We present a simulation framework with a physically-based surrogate model and investigate the impact of uncertain algorithm inputs on the bias and covariance of the retrieval error.