Atlantic thermohaline changes and its implications on the carbon cycle during the Last Interglacial

Abstract

Large and abrupt changes in the Earth’s climate have been recorded in marine records and Greenland ice cores and mostly characterize the last glacial period. These sudden transitions of climate, occurring within a few decades, have been associated with abrupt modifications of the Atlantic Meridional Overturning Circulation (AMOC). While this process is well studied during the glacial period, during warmer Interglacials the AMOC is thought to be relatively vigorous on multi-millennial (equilibrium) timescales. However, recent high-resolution reconstructions reveal large and sudden variations in benthic δ13C during the Last Interglacial (LIG, 125ka - 115ka), as recorded in several North Atlantic sediment cores. The origin of these isotopic variations remains poorly understood and may point toward a strong modification in ocean interior biogeochemistry and/or ocean circulation. The main goal of this thesis is to better understand the response of AMOC to warmer boundary conditions - the LIG - and its implication on the ocean carbon cycle. To this end, the first part of this thesis explores the mechanisms behind the changes in oceanic carbon cycle dynamics by comparing two simulated quasi-equilibrium states: the early, warm LIG (125 ka) versus late, cooler LIG (115 ka). The second and third parts focus on the origin of the abrupt and large changes observed in the North Atlantic benthic δ13C and investigate how these short-lived transitions occur mechanistically through a series of model simulations. Using the Norwegian Earth system Model, we found that the ocean dissolved inorganic carbon (DIC) content is considerably reduced (314.1 PgC) during the early and warm phase of the last interglacial (125ka) relative to the latter portion (115ka). The difference between these two quasi equilibrium states is attributed to the changes in biological pump and the ocean DIC disequilibrium. This difference in ocean carbon storage is particularly marked in the Atlantic Ocean where large water mass reorganization occurs, affecting the ventilation timescales and the remineralized carbon budget. While the Southern Source Water (SSW) extends across the Equator at 115ka, it retreats further south during the warmer 125ka, reducing the volume of this DIC-rich water mass within the interior Atlantic Ocean. This process is found to be linked to the Southern Ocean sea-ice retreat at 125ka. In a transient simulation, the model of intermediate complexity iLOVECLIM also simulates a smaller ocean DIC content (360 PgC) at 125ka as compared to 115ka. More interestingly, it was able to reproduce spontaneous variations in bottom water δ13CDIC on magnitude and timescale (multi-centennial) found in the data reconstruction during the LIG. In the model, these isotopic variations arise due to the variable influence of depleted-δ13CDIC SSW and enriched- δ13CDIC Northern Sourced Water (NSW) — consistent with previously proposed interpretations of this proxy. The simulated watermass redistributions are associated to changes in AMOC strength suggesting that both the structure and strength of circulation could be varying under interglacial boundary conditions. Thus, our model is consistent with, and mechanistically underpins, previously proposed interpretations of this proxy as related to water mass shifts. Finally, two regions of deep convection have been identified and linked to these sudden large transitions: (1) the region between Norway and Svalbard, and (2) south of Greenland. In the model, (1) is subject to saline Atlantic water intrusion which activates the deep convection and the ocean circulation north of the Iceland-Scotland ridge. Within a few decades — through a series of processes involving sea-ice retreat, ocean heat release, and wind-stress anomalies — the polar surface waters retreat further north along the Greenland East coastline. Intrusion of high salinity Atlantic surface waters following polar water retreat triggers increased deep convection in region (2). As a result, the AMOC strength increases. The AMOC strength relaxes back to its lower state when the subsurface water south of Svalbard have cooled enough, which allows the sea-ice to grow, cutting off the heat released to the atmosphere and induced anomalies of temperature and wind-stress. In summary, the results presented in this study show that large changes in the interior ocean can be simulated owing to the differences in the early and late LIG boundary conditions. These differences, simulated in ocean carbon storage occur in concert with changes in (AMOC) strength and structure of circulation. Such rapid changes in circulation and ocean carbon cycling are of concern regarding the ongoing global warming as they are all affected by the changes in the sea-ice extent. More broadly the results emphasize that interglacial climate, circulation, and carbon cycling may all be abruptly perturbed. Under gradually changing climate conditions our model suggests AMOC state changes could even arise spontaneously due to coupled atmosphere-ocean-sea ice feedbacks.