Evaluating the impact of Mediterranean overflow on the large-scale Atlantic Ocean circulation using neodymium isotopic composition

Palaeo-oceanographical archives in the Mediterranean Sea indicate the occurrence of Sapropel depositions, which are characteristic of anoxic bottom conditions and of a stratified sea. The impact of such drastic changes in the thermohaline circulation of the Mediterranean Sea on the large-scale circulations in the North Atlantic basin is not fully understood. Here we evaluate the impact of a direct perturbation of the Mediterranean Sea circulation through different idealized simulations with freshwater release rates of 20, 50, 100, and 200 mSv (1 mSv = 10(3) m(3)/s), using a the IPSL-CM5A-LR global coupled atmosphere-ocean model, with simulations up to 1000 years long. The simulations are compared to available data for the Mediterranean Sea and the North Atlantic. The Neodymium isotopic composition (epsilon(Nd)) anomalies in the North Atlantic are clearly dependent on the rate of fresh water release in the Mediterranean Sea, and subsequent modification of Atlantic meridional overturning circulation (AMOC) intensity and subpolar gyre (SPG) extension. Based on comparison with seawater epsilon(Nd) records from Rockall Trough, we suggest that a limited release of freshwater around 50 mSv may be more representative of hydrological conditions of the Mediterranean Sea during the Sapropel S1 deposition. Our results indicate that a decrease in Mediterranean overflow was clearly a factor in the intensification of upper AMOC following the Early Holocene period, when a large eastward expansion of the subpolar gyre is simulated. Also, eastward expansion of the subpolar gyre due to freshwater release in the Mediterranean (e.g. Sapropel-like events) and its impact on the epsilon(Nd) signature highlight the fact that the modification of ocean circulation in the North Atlantic basin might be more complex than a straightforward change in the AMOC, since there may also be a large impact on barotropic circulation. The simulations also produce a significant decrease in the epsilon(Nd) values of the Iceland-Scotland Overflow Water (between -3.5 and -1.3 epsilon(Nd) unit), which is consistent with changes observed during the S1 event in the reconstructions. Thus the 3-dimensional changes in baroclinic and barotropic circulations of the ocean, as reconstructed in some epsilon(Nd) records, are broadly consistent with the sensitivity simulations performed, and might thus be partly related to changes in Mediterranean overflow, which would have contributed to modification of the Atlantic Ocean circulation following a complex 3-dimensional pattern that is detailed in the paper.