Synergies in a Constellation of Greenhouse Gas Observing Satellites

With the successful launch of the Orbiting Carbon Observatory 2 (OCO-2) and the existing Greenhouse Gas Observing Satellite (GOSAT), as well as other greenhouse gas (GHG) measurement missions that may be launched in the next 5 years (GOSAT-2, OCO-3, ASCENDS, CarbonSat), the quantity of useful measurements in promises to revolutionize our understanding of the carbon cycle. In addition to the Low Earth Orbit (LEO) measurements already in the pipeline, a geostationary passive measurement of GHGs, entitled geoCARB, has been recently proposed to the EV-I program and potentially other avenues to space.

With numerous observing platforms in space, we must ask how these instruments can best be used together to take advantage of their strengths, while avoiding each instrument's weaknesses. With this in mind, a series of instrument simulators has been created with assumptions made so as to be as consistent with reality and each other as possible. The instruments simulated in this pilot study include OCO-2, GOSAT-1 (and potentially -2), and geoCARB.

1 Instrument Descriptions

1.1 GOSAT

GOSAT has been operating in space since 2009, and the ACOS team (and others) have been improving algorithms in the interim for retrieving XCO2 from GOSAT measurements. This data is freely available online (http://disc.sci.gsfc.nasa.gov/acdisc/data-holdings/acos-data-holdings), and the location and error information for a year's worth of GOSAT observations is used to create pseudo-data for the OSSEs. The initial study neglects the impacts of systematic errors, which are thought to be significant.

1.2 OCO-2

OCO-2 was launched in early July 2014, and is rapidly moving towards data collection. For the present, we must make assumptions about how OCO-2 will perform, based on previous modeling studies and mission specifications. Relative to GOSAT, about 10 times as many observations are expected, with much higher precision. In addition, the work characterizing GOSAT biases is expected to greatly improve the quality of OCO-2 retrievals. The CALIPSO satellite track, with nighttime portions removed, was used as a surrogate for the OCO-2 flight track in our experiments, since both are flying in the A-train. Individual soundings were aggregated to a 1° grid, and the total number possible was inflated to account for the shorter averaging time of OCO-2 than the 5km CALIPSO averaging time. In addition, CALIPSO observations of clouds at 70m resolution were averaged to the nominal 2km for OCO-2 in order to derive statistics to properly account for cloudy skies as a function of space and time. Nominal error values of 1 and 2 ppm for the gridded column integrated CO2 (XCO2) were selected, and these values were inflated by the amount of cloudy scenes each month.

1.3 geoCARB

GeoCARB is a geostationary GHG observing satellite recently proposed to the EV-I program. One potential orbital slot is at 110° E, geoCARB would measure CO2, CH4, CO and O2 with an average 5km footprint. Though coarser in spatial resolution that OCO-2, the ability to scan the entire FOV twice per day (depending on the season) would enable geoCARB to make transformative measurements for monitoring emissions from natural and anthropogenic sources. The instrument simulator takes into account the nominal scanning times and strategies for placement at 110° East, 60° East and 90° West, and individual soundings are aggregated to a 1° grid. Cloudiness statistics are calculated from a nearby geostationary satellite with a visual band, and the number of cloudy scenes is used to inflate the nominal error, which is assumed to be 2-3 times larger than OCO-2 with current configurations.

2 Flux Estimation

The preceding instruments are used to sample a 4D concentration field that is generated using the TM5 transport model driven by a baseline set of emissions that includes CASA-GFED biosphere and fire emissions, CDIAC fossil fuel emissions, and Takahashi air-sea exchange values for the ocean. The prior is a statistical perturbation of the truth fluxes with an assumed background error covariance. Observations are assimilated using the 4DVAR framework for TM5. Experiments include the impact of each satellite instrument alone, as well as the impact of multiple instruments being in space simultaneously. Results will be presented as error reductions, which are a measure of how much information the perturbed observations contain about the true fluxes, relative to the prior.