Measuring carbon dioxide concentrations from the Orbiting Carbon Observatory: radiative transfer challenges & solutions

The remote sensing of greenhouse gases, such as carbon dioxide and methane, from hyperspectral near-infrared measurements in the presence of atmospheric scattering represents a major radiative transfer challenge. In this presentation we describe some radiative transfer advances made in support of the recently-announced reflight of the Orbiting Carbon Observatory (OCO) mission, currently slated to launch in 2013. Using data from the Total Column CO₂ Observing Network (TCCON) as well as the Greenhouse gases Observing SATellite (GOSAT) mission, spectroscopic advances have been made with respect to carbon dioxide absorption near 1.6 and 2.0 µm, as well as oxygen absorption near 0.76 µm. Most importantly, line mixing and collisionally induced absorption in both the O₂ A band as well as the strong CO₂ band near 2.0 µm greatly increase the agreement between modeled spectra and observations.

We also describe a number of radiative transfer approximations used to yield fast yet accurate simulations of instrumental radiances and their derivatives with respect to atmospheric parameters. These simulations are used as part of an optimal-estimation retrieval algorithm for XCO₂, or column-integrated average of CO₂ concentration. OCO will take approximately 100,000 clear sky soundings per day, meaning that to process the full data set, retrievals must be extremely fast (on the order of tens of seconds per retrieval per CPU). They must also include at least an approximation of the polarization state of the radiation, as both OCO and GOSAT are polarization-sensitive. Typical “brute force” approaches, even on modern CPUs, can easily take several hours per retrieval. Our methodology relies on an optimal sampling of spectral points, reducing the number of vertical layers in the atmosphere, and most importantly a novel method termed “Low-Streams Interpolation” which enables fast yet accurate calculations of atmospheric radiances.