Novel Holocene temperature reconstruction of Greenland summit from GISP2 nitrogen isotope data reveals similarities with North Atlantic Circulation and temperature proxies

Ice core data offers a popular tool to reconstruct high latitude paleo-climate and are often compared to reconstructions from other paleo-archives (e.g. marine sediment cores, pollen records, etc.). For instance, oxygen-stable-isotope (δ¹⁸O) of the ice-core water samples is commonly used to reconstruct past site temperatures. However, many high latitude marine temperature reconstructions show low accordance with the spatially inhomogeneous ice core δ¹⁸O-based reconstructions over the Holocene on multi-centennial to multi-millennial timescales. For example, many of the δ¹⁸O-based Greenland temperature reconstructions do not show a consistent Holocene Thermal Maximum (mid to early Holocene warm period, e.g. for Greenland from 5.4 ± 1.4 ka to 8.6 ± 1.6 ka b2k) found in many high-latitude climate records. The isotopic composition of the water samples provides a rather robust proxy for reconstructing paleo-temperatures for times where large temperature variations occur. In the Holocene where temperature variations are comparatively small, changes in seasonal distribution of precipitation as well as of evaporation conditions at the source region may dominate water-isotope-data variations. In addition, the change of elevation of the Greenland ice sheet over the Holocene leads to an additional temperature signal which is able to mask multi- millennial temperature trends. The use of nitrogen stable isotopes of ancient air trapped in the ice cores provides an alternative for ice core based site temperature reconstructions. This method uses the stability of isotopic compositions of nitrogen in the atmosphere at orbital timescales as well as the fact that changes are only driven by processes in polar firn. Thus gas-isotope-based reconstructions are independent from changes in precipitation seasonality or source signature. Here we present a high-resolution Holocene temperature record from Greenland summit, reconstructed based on nitrogen stable isotopes data (δ¹⁵N) from the GISP2 ice core. The reconstruction was conducted by exploiting a Monte Carlo based firn model inversion technique on GISP2 inert gas isotope data, which leads to robust uncertainty estimations. The most robust temperature estimate (T(δ¹⁵N)) was compared to a variety of North Atlantic sea surface temperature and terrestrial temperature proxies, showing comparable signatures for multi-centennial to multi-millennial signals. Our record reveals that the warmest period of the Holocene (Holocene Thermal Maximum, HTM) at Greenland summit occurred from 5.4 to 9.2 ka b2k. The HTM was composed by three distinct warm-phases interrupted by several centennial-scale cooling events. We find evidence for a rapid cooling beginning at about 5.4 ka b2k connecting the HTM with the Neoglaciation (long-term cooling trend, 0.4 to 5.1 ka b2k) and for a late Holocene warm-phase 1.3 to 2.2 ka b2k. Furthermore, Greenland warm-phases occurred mostly in times of low solar activity and are synchronous to three Bond-events (ice rafted debris depositions). We find evidence for a coherence of AMOC variability and Greenland summit temperature during the Holocene and conclude therefore that the latter is most likely mainly driven by changes in North Atlantic circulation patterns and AMOC variability for multi-centennial to multi-millennial variability.