Changes to Carbon Isotopes in Atmospheric CO2 Over the Industrial Era and Into the Future

Abstract In this “Grand Challenges” paper, we review how the carbon isotopic composition of atmospheric CO2 has changed since the Industrial Revolution due to human activities and their influence on the natural carbon cycle, and we provide new estimates of possible future changes for a range of scenarios. Emissions of CO2 from fossil fuel combustion and land use change reduce the ratio of 13C/12C in atmospheric CO2 (δ13CO2). This is because 12C is preferentially assimilated during photosynthesis and δ13C in plant‐derived carbon in terrestrial ecosystems and fossil fuels is lower than atmospheric δ13CO2. Emissions of CO2 from fossil fuel combustion also reduce the ratio of 14C/C in atmospheric CO2 (Δ14CO2) because 14C is absent in million‐year‐old fossil fuels, which have been stored for much longer than the radioactive decay time of 14C. Atmospheric Δ14CO2 rapidly increased in the 1950s to 1960s because of 14C produced during nuclear bomb testing. The resulting trends in δ13C and Δ14C in atmospheric CO2 are influenced not only by these human emissions but also by natural carbon exchanges that mix carbon between the atmosphere and ocean and terrestrial ecosystems. This mixing caused Δ14CO2 to return toward preindustrial levels in the first few decades after the spike from nuclear testing. More recently, as the bomb 14C excess is now mostly well mixed with the decadally overturning carbon reservoirs, fossil fuel emissions have become the main factor driving further decreases in atmospheric Δ14CO2. For δ13CO2, in addition to exchanges between reservoirs, the extent to which 12C is preferentially assimilated during photosynthesis appears to have increased, slowing down the recent δ13CO2 trend slightly. A new compilation of ice core and flask δ13CO2 observations indicates that the decline in δ13CO2 since the preindustrial period is less than some prior estimates, which may have incorporated artifacts owing to offsets from different laboratories' measurements. Atmospheric observations of δ13CO2 have been used to investigate carbon fluxes and the functioning of plants, and they are used for comparison with δ13C in other materials such as tree rings. Atmospheric observations of Δ14CO2 have been used to quantify the rate of air‐sea gas exchange and ocean circulation, and the rate of net primary production and the turnover time of carbon in plant material and soils. Atmospheric observations of Δ14CO2 are also used for comparison with Δ14C in other materials in many fields such as archaeology, forensics, and physiology. Another major application is the assessment of regional emissions of CO2 from fossil fuel combustion using Δ14CO2 observations and models. In the future, δ13CO2 and Δ14CO2 will continue to change. The sign and magnitude of the changes are mainly determined by global fossil fuel emissions. We present here simulations of future δ13CO2 and Δ14CO2 for six scenarios based on the shared socioeconomic pathways (SSPs) from the 6th Coupled Model Intercomparison Project (CMIP6). Applications using atmospheric δ13CO2 and Δ14CO2 observations in carbon cycle science and many other fields will be affected by these future changes. We recommend an increased effort toward making coordinated measurements of δ13C and Δ14C across the Earth System and for further development of isotopic modeling and model‐data analysis tools.


Contents of this file
Text SM1 and SM2 Figure S1 Table S1  Table S2 Text SM1. Simulations for "no bombs" and "no fossil" scenarios Simulations for Δ 14 C and δ 13 C of atmospheric CO 2 were conducted following Graven (2015). The model has one atmosphere box and 43 ocean boxes in a one-dimensional box diffusion model. Graven (2015) used a one-box terrestrial biosphere but here we have changed the model so that the biosphere is represented by three boxes representing fast, medium and slow turnover pools. There is no representation of carbon transport from terrestrial boxes to ocean boxes via rivers.
To run the simulation with no 14 C from nuclear weapons tests, atmospheric CO2 concentration and Δ 14 CO 2 are computed prognostically, accounting for fossil fuel emissions and natural 14 C production. To run the simulation with no fossil fuel emissions, the input of 14 C from the nuclear weapons tests that is consistent with the observations, within the structure of the simple model, is first diagnosed with a historical run forced with observed atmospheric Δ 14 CO2. The annual change in total 14 C inventory of all the carbon pools in this run is then specified as the 14 C from the nuclear weapons tests. The total production is 634 x10 26 atoms through 1980, which is comparable to the estimate of Naegler and Levin [2006]: 598-632 x 10 26 atoms between 1945 and 1980, both including 14 C production by the nuclear power industry. Then a simulation is run with the specified annual input of 14 C from nuclear weapons tests from the previous simulation through 2005 and then with constant 14 C production from natural and anthropogenic sources for 2005-2015, but with no fossil fuel emissions. Land use change emissions are included in both simulations. For both simulations, the model is run with one set of parameters (Table S1) chosen from the sets of calibrated parameters as described in the next section.

Text SM2. Simulations for SSP-based scenarios
Simulations for Δ 14 C and δ 13 C of atmospheric CO 2 were conducted following Graven [2015].
Simulations were conducted for scenarios that are based on the SSP framework, the four Tier 1 ScenarioMIP scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5) and two of the Tier 2 scenarios (SSP1-1.9 SSP5-3.4-Overshoot) [O'Neill et al., 2016]. These six scenarios span the full range of radiative forcing and CO 2 emissions included in the larger set of SSP-based scenarios.
The simulations use historical data and data for the six SSP-based scenarios. Historical atmospheric CO 2 concentration was specified by global annual mean data from Meinshausen et al. [2017]. Historical fossil fuel CO2 emissions data were given by the Community Emissions Data System (CEDS) [Hoesly et al., 2018]. Historical land use CO 2 emissions data were given by C4MIP (CMIP6_C4MIP_landuse_emissions.nc available at http://c4mip.net/index.php?id=3455). Global annual atmospheric CO2 concentration data for the SSP scenarios were retrieved from input4mips (https://esgfnode.llnl.gov/projects/input4mips/). The SSP fossil fuel and land use CO 2 emissions and the SSP bioenergy with carbon capture and storage (BECCS) data were gathered from the SSP database hosted at the International Institute for Applied Systems Analysis (https://tntcat.iiasa.ac.at/SspDb/), and accessed in October 2018.
The δ 13 C of fossil fuel emissions was specified by Andres et al. [2016] for 1850 to 2013 and then kept fixed at the 2013 value from Andres et al. [2016], -27.73 per mil, for the period 2013 to 2100 in all SSPs. Due to limitations in the data available about the SSPbased scenarios it was not possible to accurately specify δ 13 C in fossil fuel emissions in the scenario projections, i.e. CO 2 emissions or time-varying emission factors (amount of CO 2 emitted per unit energy produced) for each fuel type were not reported in the SSP database, despite each fuel type having a specific 13 C fingerprint. Sensitivity tests estimating future changes in δ 13 C of fossil fuel emissions showed that the potential impact on simulated atmospheric δ 13 C is likely to be small, less than ~0.4 per mil for simulated δ 13 C in 2100.
Simulations for 1850-2005 were conducted following a spinup period of 11850 years. Atmospheric Δ 14 C and atmospheric δ 13 C were specified by historical data from Graven et al. [2017] for 1850-2005 and then atmospheric Δ 14 CO 2 and δ 13 CO 2 were simulated prognostically for 2005-2100.
Three changes were made to the model setup in comparison to Graven [2015].
First, a sensitivity of photosynthetic 13 C discrimination to atmospheric CO 2 concentration was included, following Schubert and Jahren [2015] and Keeling et al. [2017]. The formulation for discrimination in Schubert and Jahren [2015]'s equations 3 and 4 was used, beginning from 13 C discrimination of 17 per mil in 1850. This leads to discrimination of 18.7 per mil in 2100 for SSP1-1.9 and 22.5 per mil in 2100 for SSP5-8.5.
Second, sea surface temperature (SST) changes were included for both the historical and future period. The historical data used were annual, global data SST anomalies from HadSST.3.1.1.0 from 1850 to 2018 (https://www.metoffice.gov.uk/hadobs/hadsst3/data/download.html, accessed October 2018). A reference SST of 18°C in 1850 was assumed. Future SST was estimated from global mean temperature data for the SSPs retrieved from the internal SSP database. SST anomalies in the SSPs for 2020 to 2100 were estimated from global mean temperature by applying a scaling factor of 0.85, which derives from a regression of HadSST.3.1.1.0 SST anomaly data and HadCRUT.4.6.0.0 global annual near surface temperature anomaly data (https://www.metoffice.gov.uk/hadobs/hadcrut4/data/current/download.html, accessed October 2018). SST is used in the model in the calculation of the ocean carbonate system and in the specification of fractionation factors related to air-sea gas exchange.