Project Funding: 2015 - 2018

NRA: 2014 NASA: OCO-2 Science Team for the OCO-2 Mission   

Funded by NASA

Abstract:
This proposal seeks continuing funding for studies related to OCO-2 radiative transfer and uncertainty analyses for partly and fully cloudy observations. A large fraction of OCO-2 observations fall into these categories (~60% over water and ~45% over land). In order to constrain CO2 sinks, the OCO-2 carbon dioxide column retrievals must be highly accurate with a dry air mole fraction XCO2 better than 1 ppm after spatial averaging. Achieving this goal will only be feasible if potential systematic errors caused by cloud contamination can be minimized. The operational OCO-2 retrieval (ACOS, O'Dell et al. [2012]) employs a rigorous spectral-based screening of clouds to avoid potential systematic errors in retrieved XCO2. Over the last two years our group has established a modeling system that allows assessing three-dimensional radiative transfer errors under partly cloudy conditions and interfaces with OCO-2’s ACOS retrieval. We have already used this system to study the accuracy of the aforementioned OCO-2 pre- and post-screening theoretically for nadir observations over land [Merrelli et al., 2014]. Here we propose to extend this study in the following manner: 1. We will continue using our modeling system to study glint mode observations and to quantify other three-dimensional radiative transfer effects, such as cloud adjacency effects and shadowing. These theoretical studies will inform about possible biases in XCO2 observations when clouds directly or indirectly affect the retrieval. This modeling exercise will also help refine spectral observation error covariances within the full physics XCO2 retrieval that are caused by unresolved subscale inhomogeneities not limited to clouds (e.g. subscale variations in surface albedo). 2. We will study the impact of sub-pixel broken clouds on XCO2 retrievals from actual OCO-2 observations. We will use MODIS observations collocated with OCO-2 to calculate sub-scale cloud fraction for each OCO-2 pixel. For this effort, we will use the MODIS 250 meter cloud mask, which provides a high enough spatial resolution to quantify subscale cloudiness. These observational studies will help reconcile actual XCO2 retrievals and retrieval errors with the theoretical insights gained under 1. 3. Since we are already running the ACOS retrieval as part of our modeling environment, we will evaluate a first attempt at XCO2 retrievals over clouds initially for completely cloudy OCO-2 fields of view. Earlier studies performed in our group show the feasibility of this idea [Vidot et al., 2009]. We are now in a position where we have the actual OCO-2 observations as well as the modeling system to fully explore this idea. If meaningful XCO2 retrievals over clouds can be obtained, a limited atmospheric profiling would be possible, in which the above-cloud retrieved XCO2 is subtracted from nearby cloud-free retrievals to obtain boundary layer CO2 mole fraction. We have assembled a group of scientists with a long track record of collaborative work with high relevance to OCO-2. The project will be performed in close collaboration with the OCO-2 algorithm team. We anticipate this project will have a substantial positive impact on algorithm development, error characterization, and will potentially lead to a significant increase in the number of useable OCO-2 observations.