Uncertainties and variability in the global carbon cycle

Sammanfattning: Anthropogenic emissions of carbon dioxide (CO2) are perturbing the radiation balance of the atmosphere that ultimately determines the climate for all life on Earth. The prediction of future climate change and the explanation of observed changes are contingent on us quantitatively understanding the link between CO2 emissions and atmospheric concentrations that is provided by the biogeochemical carbon cycle. Our knowledge of key processes is currently incomplete and there is a perceived imbalance in the atmospheric carbon budget. In this thesis, the imbalance and its uncertainty are calculated from preindustrial times to 1990 and it is shown to be statistically significant from the 1950s to the present day. The implications for future CO2 projections are demonstrated by calculating atmospheric concentrations to 2100 assuming that different mechanisms (CO2 fertilisation and temperate forest regrowth) are responsible for the budget imbalance. The different balancing mechanisms result in divergent concentration projections, thereby illustrating the importance of resolving the issue. A step in this direction is taken by simulating terrestrial carbon exchange and atmospheric CO2 concentrations on-line in an atmospheric general circulation model, whose advanced land surface parameterisation calculates carbon fluxes at each time step. After identifying important strengths and weaknesses in an evaluation of the model, the climate induced interannual variability in terrestrial carbon exchange and atmospheric CO2 concentrations is examined. This variability represents carbon cycle "noise" that is superimposed on the long-term anthropogenic trend. It is shown that fluctuations driven by unforced variability in the atmosphere are likely to be non-negligible relative to perturbations from specific phenomena like El Niño and volcanic eruptions. Continental regions exhibit large variations, and yet they are poorly sampled by the current CO2 monitoring program. The effects of El Niño are investigated and isolated from these unforced variations in an ensemble integration. Land areas are simulated to release CO2 to the atmosphere during El Niño events as a result of meteorological perturbations driven by the sea surface temperature anomalies. The release is sufficiently large to explain why atmospheric CO2 concentrations are observed to increase during El Niño, even though the usual CO2 outgassing from the equatorial Pacific is diminished.