Lead Authors:
Andrew R. Jacobson, University of Colorado, Boulder, and NOAA Earth System Research Laboratory
John B. Miller, NOAA Earth System Research Laboratory
Contributing Authors:
Ashley Ballantyne, University of Montana
Sourish Basu, University of Colorado, Boulder, and NOAA Earth System Research Laboratory
Lori Bruhwiler, NOAA Earth System Research Laboratory
Abhishek Chatterjee, Universities Space Research Association and NASA Global Modeling and Assimilation Office
Scott Denning, Colorado State University
Lesley Ott, NASA Goddard Space Flight Center
Science Lead:
Richard Birdsey, Woods Hole Research Center
Review Editor:
Nathaniel A. Brunsell, University of Kansas
Federal Liaison:
James H. Butler, NOAA Earth System Research Laboratory

Observations of Atmospheric Carbon Dioxide and Methane

8.6.1 Findings from Atmospheric Inversions and Related Analyses

The present collection of atmospheric CO2 inversions shows no clear trend in the boreal North American sink, but it does suggest the possibility of an increasing sink in temperate latitudes. A more robust feature of atmospheric inversions is that they show that the North American CO2 sink is more highly variable and sensitive to drought and temperature stress than bottom-up biosphere models (King et al., 2015; Peters et al., 2007). Inversions also produce a larger mean sink and a deeper annual cycle than terrestrial biosphere models. Significant uncertainty remains about the magnitude of the mean North American carbon sink, in part because models disagree about the partitioning of the net sink between northern and tropical land regions. The mechanisms behind the land sink cannot be understood fully without more agreement on its location. Notably, distinguishing between a potentially short-lived sink due to recovery from past land-use practices (mainly a temperate Northern Hemisphere phenomenon) and a longer-term sink due to CO2 fertilization remains elusive. Moreover, the role of carbon-climate feedback processes in North America, both negative (e.g., extended growing seasons and tree-line migration) and positive (e.g., permafrost carbon release and insect outbreaks), is poorly understood at present. Atmospheric measurements can impose significant constraints on these processes (e.g., Sweeney et al., 2015), and continued and expanded measurements, especially in sensitive Arctic and boreal regions, will be critical moving forward.

Inventories suggest that fossil fuel CO2 emissions are stabilizing and even decreasing for certain regions and sectors of the global and North American economy. This finding is difficult to verify given the ad hoc nature of the GHG observation network, lack of integration among programs, and sparse measurements of anthropogenic emissions tracers such as Δ14CO2 and CO.

Individual atmospheric CH4 inversions consistently show no trend and little interannual variability in total CH4 emissions (natural and anthropogenic) for both the temperate (largely the United States) and boreal regions and the continent as a whole (see Figure 8.3). These results suggest that North American emissions have not contributed significantly to the global upward trend that started in 2007. Increasing oil and gas production in North America could result in increased CH4 emissions, a result apparently confirmed by Turner et al. (2016) on the basis of comparing inverse model estimates from different time periods. This conclusion has been called into question by Bruhwiler et al. (2017), who argue that robust trend detection is limited by interannual variability, the sparse in situ measurement network, and biased satellite CH4 retrievals. Recent increases in atmospheric ethane and propane suggest increased CH4 emissions from fossil fuel production, although there is uncertainty in this conclusion due to poorly quantified emissions ratios (Helmig et al., 2016). As with CO2 though, little reliable spatial information is available from the current suite of CH4 inverse models. This limitation hampers attribution to specific mechanisms including CH4-climate feedbacks, especially in the boreal zone where permafrost degradation plays a key role in changing CH4 and CO2 fluxes (McGuire et al., 2016; see also Ch. 11: Arctic and Boreal Carbon)

8.6.2 Future Atmospheric Measurement Challenges and Strategies for North America

Compatibility Among Networks

As the community expands research into new domains and with new measurement strategies, new challenges are emerging. Compatibility of measurements among existing and future networks is a concern, as there is ample history of calibration difficulties from the decades of in situ measurement experience (e.g., Brailsford et al., 2012). This challenge is being addressed by careful attention to calibration and participation in laboratory and field intercomparison activities (Masarie et al., 2011; www.esrl.noaa.gov/gmd/ccgg/wmorr/). Much more challenging is linking ground- and space-based remote-sensing measurements to each other and to the calibrated in situ networks. Concentrations derived from any remote-sensing gas measurement, whether ground- or space-based, cannot be formally calibrated because the measurement instrument cannot be “challenged” by a reference sample with a known concentration. Thus, identification and correction of biases remain a significant challenge. With the OCO-2 and GOSAT programs, the primary strategy has been to compare the satellite-based retrievals with TCCON retrievals. The TCCON retrievals of column CO2 are themselves remote-sensing products that have been statistically linked to the World Meteorological Organization CO2 calibration scale using aircraft in situ partial column CO2 and CH4 extrapolated to the top of the atmosphere (Wunch et al., 2011). This linkage remains uncertain due to the limited number of in situ profiles used and their limited maximum altitude. A limited number of nearly total column AirCore (Karion et al., 2010) measurements also have been compared with TCCON columns.

Bias correction of satellite retrievals remains challenging due to the limited number of TCCON stations (currently less than 20) and because estimates of the TCCON site-to-site bias of 0.4 ppm (one-sigma; Wunch et al., 2016) are significant for carbon cycle studies. As an example of the importance of small biases, Reuter et al. (2014) demonstrated that a gradient of 0.5 ppm in column CO2 across Europe was associated with a change in flux over that region of about –500 Tg C per year. This increased sink over Europe using a regional model is consistent with the inversion intercomparison of Houweling et al. (2015), who found that assimilating GOSAT column CO2 retrievals in global inversion models caused an increase of about 700 Tg C per year in the European sink, with a compensating increase in the northern Africa source of about 900 Tg C per year. These shifts in emissions were associated with degraded agreement with unassimilated in situ observations from both surface observation sites and aircraft campaigns. For comparison, the in situ assimilation models collected for this chapter estimate a modest sink of 219 ± 405 Tg C per year in Europe and a negligible source of 13 ± 281 Tg C per year in northern Africa over the 2004 to 2013 period. These uncertainties, which comprise both interannual variability and intermodel differences in the inversions, are relatively large but still appear inconsistent with the GOSAT-driven flux increments reported in Houweling et al. (2015). In the relatively short time that GOSAT and OCO-2 have been collecting data, significant progress has been made in identifying and correcting biases in those datasets. Progress also is needed in understanding the time and space scales of remote-sensing data least susceptible to bias and how to assimilate these retrievals jointly with in situ data having less bias. Moving forward, more measurements will be key, including expansion of AirCore (Karion et al., 2010) and commercial aircraft observations (Basu et al., 2014) that will enable better assessment and utilization of both ground- and space-based total column CO2 and CH4 remote-sensing data.

Next-Generation Measurements

Atmospheric measurements will play an important role in addressing these critical questions on the present and future state of both anthropogenic and biogenic components of the North American carbon cycle. The following is a list of potential, yet achievable, atmospheric measurement approaches that could dramatically change the current view of the North American (and global) carbon cycle.

  1. Commercial Aircraft CO2 and CH4 Observations. The Comprehensive Observation Network for Trace gases by Airliner (CONTRAIL) program has measured GHGs from commercial aircraft for nearly two decades (Matsueda et al., 2008). A similar European effort, In-service Aircraft for a Global Observing System (IAGOS) project (Filges et al., 2015), is not yet fully operational for GHG measurements. The technology exists for unattended, high-accuracy airborne CO2 and CH4 measurements (Karion et al., 2013), and deploying instruments aboard 40 domestic U.S. commercial aircraft could result in approximately 500 vertical profiles per day, radically changing CO2 and CH4 data density over North America.

  2. Greatly Expanded Δ14CO2 Measurements. Recently, Basu et al. (2016) demonstrated that expanding the U.S. network of Δ14CO2 measurements from about 800 per year to 5,000 per year, as recommended by the U.S. National Research Council (Pacala et al., 2010), could allow for atmospherically based determination of U.S. fossil fuel CO2 emissions to within 5%, complementing official U.S. EPA inventory-based estimates. In addition to 14CO2, other tracers such as CO, non-methane hydrocarbons, halogenated species, and 14CH4 (for fossil CH4 identification) can serve as powerful constraints on emissions, both in total and by sector.

  3. Upcoming Satellite-Based CO2 and CH4 Sensors. These sensors, including GOSAT-2, OCO-3, TanSat (China), Geostationary Carbon Cycle Observatory (GeoCARB; NASA), MERLIN (France and Germany), TROPOMI (European Space Agency), and others (Ciais et al., 2014) likely will enable dramatically increased spatial coverage of total column CO2, CH4, and other gases. For the utility of these data to be maximized, existing challenges associated with aerosols, characterization of the ocean and land surface, clouds, daylight, and, more generally, the linkage to formal gas concentration scales must be overcome. GOSAT and OCO-2, and particularly their planned successors, also will yield information on chlorophyll fluorescence (SIF), which has potential as a marker of time and space patterns of plant photosynthesis.

  4. NEON. If built out as planned, NEON (National Science Foundation) will provide calibrated CO2 measurements on towers over a variety of North American biomes that will add significantly to the North American CO2 observational dataset.

  5. Additional Gas Tracers. As with anthropogenic ancillary tracers (see B), numerous gases can serve as tracers of terrestrial ecosystem processes. Gross primary production fluxes are closely linked to atmospheric gradients in COS and Δ17O (anomalies in the 18O:17O ratio of CO2; e.g., Campbell et al., 2008; Thiemens et al., 2014). Atmospheric δ13CO2 is sensitive to the impact of regional-scale moisture stress on terrestrial photosynthesis (Ballantyne et al., 2010) and can distinguish C3 and C4 plant productivity. Schwietzke et al. (2016) showed the potential for δ13CH4 observations to distinguish fossil fuel CH4 emissions from other sources. Measurements of the δ18O of CO2 reflect both biospheric processes and changes in the hydrological cycle (Ciais et al., 1997; Flanagan et al., 1997; Miller et al., 1999).

  6. Measurements to Improve Atmospheric Transport Simulation. Such measurements are critical for fully extracting the information content of atmospheric CO2 and CH4 data. Better understanding and parameterizing of atmospheric transport are critical. Near-surface GHG concentrations are a sensitive function of the planetary boundary-layer mixing height, wind speed, and wind direction. Measurements of the vertical wind structure and boundary-layer depth using rawinsonde, LIDAR, and radar, and assimilating these data into atmospheric transport models, can improve atmospheric transport significantly (Deng et al., 2017). Simulated CO2 transport is sensitive to boundary-layer mixing, convective cloud transport, synoptic weather patterns, and the surface energy balance, all of which can be difficult to simulate with the high accuracy and precision required for atmospheric inversions. Fortunately, decades of weather forecasting research provide a strong foundation for improving the meteorological reanalyses used in atmospheric inversions. Observational programs that merge meteorological measurements with high-density GHG data (e.g., ACT-America) are aimed at advancing this aspect of atmospheric inverse modeling. In addition, measurements of tracers such as water vapor isotopic ratios, sulfur hexafluoride (SF6), and even 14CO2, where emissions are relatively well known (Turnbull et al., 2008), also can constrain simulated transport (Denning et al., 1999; Patra et al., 2011; Peters et al., 2004).

See Full Chapter & References