High-resolution reconstruction of atmospheric CO2 concentrations during the last interglacial based on the EDC ice core

• Main Author: Borges da Silva, Lucas
• Format: Doctoral or Postdoctoral Thesis
• Language: English
• Published: Universität Bern
• Subjects: 530 Physics; Antarctic; Antarc*; ice core
• Online Access: http://boristheses.unibe.ch/3684/; http://boristheses.unibe.ch/3684/1/PhD_Thesis_LSilva_CC_license_noCV_adaptedSciArticle.pdf; https://doi.org/10.48549/3684

The successful reconstruction of past atmospheric CO2 concentrations from Antarctic ice cores started in the early 1980s. Each newly published record is the product of painstaking discrete measurements of hand-sized samples from ice cores that can reach more than 3 kilometers of depth. Hence, high-resolution reconstructions of CO2 are usually limited to a specific window of time of the last 800 thousand years (800 ka). Using the EDC ice core, we reconstructed atmospheric CO2 concentrations during Marine Isotope Stage 5 (MIS 5; 135–106 ka). The new dataset covers the penultimate deglaciation, the last interglacial, and the last glacial inception in unprecedented centennial resolution. Our new record shows remarkably stable CO2 concentrations for ten thousand years during MIS 5e. Simultaneously, a series of worldwide climatic changes took place, such as falling temperatures in the oceans and over the poles, growing ice sheets, generalized climate instability in the Northern Hemisphere, and changes in the Earth’s orbital parameters. The lack of marked variability in the CO2 record during this period can be explained by an unusual combination of dynamic carbon fluxes and the lack of a suitable deep ocean storage reservoir. As enigmatic as the plateau is the last glacial inception when CO2 suddenly drops from interglacial levels and resumes its coupling with Antarctic temperature. We propose that a Northern Hemisphere trigger sourced this threshold-like behavior. Despite the centennial-scale resolution achieved with the MIS 5 record, we only tentatively interpret submillennial CO2 features. While building the dataset, we realized that CO2 showed sharp oscillations between neighboring data points, too fast to be fingerprinted by true atmospheric variability. Much of the ensuing work tackled the understanding of CO2 fluctuations at the centimeter scale and how they affected our record. We concluded that while the high resolution allowed the establishment of precisely timed slope changes, individual fluctuations at the ...