Observations of Atmospheric Carbon Dioxide and Methane

From the late 1950s through mid-1990s, measurements of atmospheric CO₂ and CH₄ concentrations were mostly targeted at understanding variations in “background” marine air, remote from the complex signals found over continents. Motivated largely by the finding of Tans et al. (1990) that Northern Hemisphere extratropical land regions were very likely a significant CO₂ sink, new attention was placed on understanding the role played by terrestrial ecosystems. New measurement sites were established on land, with an emphasis on platforms extending well into the daytime planetary boundary layer or higher, in an attempt to capture signals of regional (approximately 1,000 km) surface exchange (Gloor et al., 2001). This effort included observations on towers extending far above the ecosystem canopy (typically >300 m above ground level) and from light aircraft flying well into the free troposphere (typically >6 km above sea level).

The availability of calibrated, comparable observations of atmospheric CO₂ mole fractions on a common scale has made it possible to estimate surface exchange via inversion of atmospheric transport. Studies including Enting and Mansbridge (1991), Fan et al. (1998), and the ensuing Atmospheric Tracer Transport Model Intercomparison Project (TransCom) model intercomparisons (e.g., Baker et al., 2006; Gurney et al., 2002) reported widely ranging values of mean sinks for continental-scale land regions. These results demonstrated that, in the face of highly variable surface fluxes, uncertainties and biases in atmospheric transport models (e.g., Stephens et al., 2007), coupled with the sparseness of available observations, render the estimation of mean surface fluxes strongly underconstrained. In the context of a common estimation methodology, interannual variability in surface fluxes can be strikingly coherent between inversion models (Baker et al., 2006; Peylin et al., 2013), suggesting that standing biases in transport models may drive differences in the mean flux estimated by global inverse models.

At the time of the First State of the Carbon Cycle Report (SOCCR1; CCSP 2007), there was agreement within large uncertainty bounds between “bottom-up” estimates from terrestrial biomass inventories and “top-down” atmospheric studies (Pacala et al., 2001; see Ch. 2 and Ch. 3 in SOCCR1) on the size of the terrestrial CO₂ sink in North America. Atmospheric inverse modeling was discussed in SOCCR1, but the final fluxes reported for North America excluded estimates from those techniques. These estimates were brought together for the first time at the continental scale for the North American Carbon Program (NACP) interim regional synthesis project (Hayes et al., 2012; Huntzinger et al., 2012).