A revised Earth system model-based analysis of glacial-interglacial changes in ocean δ13C

Malin Ödalen, Bolin Centre for Climate Research, Stockholm, Sweden; Stockholm University, Department of Meteorology, Stockholm, Sweden, Carlye Peterson, University of California Riverside, Earth Sciences, Riverside, CA, United States, Andy Ridgwell, University of Bristol, School of Earth Sciences, Bristol, United Kingdom; University of California, Riverside, Department of Earth Sciences, Riverside, United States, Kevin I. C. Oliver, University of Southampton, Ocean and Earth Science, National Oceanography Centre, Southampton, United Kingdom and Paul J Valdes, University of Bristol, Bristol, United Kingdom
Across the latest glacial-interglacial (G-I) cycle, climate change caused rearrangements in ocean circulation and carbon reservoirs. Glacial climate was characterized by small atmospheric and terrestrial carbon reservoirs, and increased ocean carbon storage, compared to interglacials. Ocean δ13C, as recorded by benthic foraminifera, reflects these changes in ocean circulation and carbon storage, and thereby in atmospheric CO2. However, sparse data coverage makes inference of ocean changes directly from the proxy records difficult. In this study, we use a model to extend the records of benthic δ13C from Peterson et al. (2014), from the Holocene (HOL, 0-6 ka) and the Last Glacial Maximum (LGM, 19-23 ka), to areas where data are absent. We apply appropriate boundary conditions for each time slice, and run ensembles to find the optimal combination of wind stress scaling (ws), freshwater forcing (fw), and brine rejection (br), for reproducing the HOL and LGM ocean δ13C.

Differences in fw are found to be central for distinguishing our best representations of HOL and LGM δ13C patterns. These simulations reproduce well the deglacial weakening of the surface-to-deep ocean δ13C gradient. For both time slices, good fits to the proxy data are however achieved for several combinations of br and fw. To account for uncertainty in the ensemble simulations, we correct the ensemble for model-data bias and weight them by model skill. From this, we compute an ensemble average difference in whole-ocean δ13C between LGM and HOL of 0.22 – 0.34‰. This translates to an addition of 340 – 520 Pg C with a terrestrial δ13C signature to the LGM ocean. In the next step, we expand the study by exploring the effect of interactive sediments on these results.