A recent compilation of instrumental and proxy indicators suggests that, following more than a millennium of relative stability, the Atlantic Meridional Overturning Circulation (AMOC) began to weaken in the mid-1800s and then again around the 1960s [Caesar et al., 2012]. The linkage between many of the paleoceanographic proxies and AMOC strength relies on the strong relationship between observed and simulated patterns of North Atlantic surface and subsurface temperature variation over the past ~150 yr - which in the models is associated with changes in AMOC strength arising from simulated variation of open-ocean convection in the Labrador Sea [Caesar et al., 2018; Rhamstorf et al., 2015; Thornalley et al., 2018]. However, the connection between convection in the Labrador Sea and AMOC strength is now challenged by an extensive network of hydrographic observations suggesting that the meridional mass transports that define AMOC strength are driven largely by surface to deep water conversion in and around the Nordic Sea Basin - at least for the period of comprehensive observation since 2014 [Lozier et al., 2019]. Furthermore, the resolution of purported AMOC proxy records is typically such that they must be subjected to multi-decadal smoothing before they can been meaningfully inter-compared and/or related to the instrumental record.

Here, we provide an annually- to sub-annually- resolved marine sediment record of near-surface hydrography in the eastern branch of warm, salty Atlantic water inflow to the Nordic Seas for the period ~ AD 1750 to ~AD 1992 (core top age), with an estimated absolute chronological uncertainty of 1-15 years (1 sigma, 1-6 years after AD 1870) based on 210Pb and Cs dating, chemical identification of historic tephra and wiggle-match 14C dating. Planktonic d180 results indicate stable near-surface hydrographic conditions from ~AD 1770 to ~AD 1900, at which time we detect low amplitude variability toward higher d18O values (lower calcification temperatures) followed by the abrupt onset of much larger transient dO18 increases in AD 1950 (±1-2 yr) that appear to persist until to end of the record ~AD 1990 (see Figure). Correlation of the isotopic time series to the instrumental record of Sea Surface Temperature (SST) in the North Atlantic reveals the same spatial pattern as the so-called subpolar North Atlantic “warming hole” that (in large part) defines the AMOC SST Index [Rhamstorf et al., 2015; Caesar et al., 2018; Osman et al., 2019]. The most recent part of isotopic record is also well correlated with the temperature of the densest Nordic Sea overflow waters leaving the basin through Denmark Strait (available since AD 1950), indicating that the near-surface inflow characteristics recorded at the study site are imparted to deep overflow waters by recirculation and deep convection within the basin in just a few years, consistent with regional hydrographic analyses [Eldevik et al., 2009]. Overall, our analysis suggests that large d180 anomalies recorded at the core site are associated with i) cooling, freshening and expansion of the North Atlantic Sub-Polar Gyre (SPG), ii) increased transport of cold, fresh SPG waters through bathymetrically-steered currents into the eastern branch of the Atlantic water entering the Nordic Seas where they may iii) influence surface buoyancy and deep water formation within the Nordic Sea Basin [Hatun et al., 2005]. Although not conclusive, this is consistent with the possibility that changes in temperature and salinity of the SPG and Labrador Sea may influence AMOC via a downstream response in and around the Nordic Sea Basin rather than via open ocean convection in the Labrador Sea itself.

Although the proximate forcing of the SPG is not entirely clear, the downstream response to changes in the SPG appears to have increased substantially around AD 1950. We briefly examine the roles of atmospheric blocking events, associated changes in sea ice and freshwater transport, and of increased Greenland Ice Sheet run off in contributing to the sudden change in response at that time.