Reconstructing Sea Ice Conditions in the Arctic and Sub-Arctic Prior to Human Observations

Sea ice is a sensitive parameter characterized by a high variability in space and time that can be reconstructed from paleoclimatological archives. The most direct indication of past sea ice cover is found in marine sediments, which contain various tracers or proxies ofenvironments characterized by sea ice. They include sedimentary tracers of particles entrained and dispersed by sea ice, biogenic remains associated with production under/within sea ice or with ice-free conditions, in addition to geochemical and isotopic tracers of brine formation linked to sea ice growth. Reconstructing the extent of past sea ice is, however, difficult because proxies are only indirectly related to sea ice and require the use of transfer functions having inherent uncertainties. In particular, we have to assume a correspondence between sea ice cover values from modern observations and the sea ice proxies from surface sediment samples, which is a source of bias since the time intervals represented by modern observations (here 1954-2000) and surface sediments (10 0 ―10 3 years) are not equivalent. Moreover, suitable sedimentary sequences for reconstructing sea ice are rare, making the spatial resolution of reconstructions very patchy. Nevertheless, although fragmentary in time and space and despite uncertainties, available reconstructions reveal very large amplitude changes of sea ice in response to natural forcing during the recent geological past. For example, during the early Holocene, about 8000 years ago, data from dinocyst assemblages suggest reduced sea ice cover as compared to present in some subarctic basins (Labrador Sea, Baffin Bay, and Hudson Bay), whereas enhanced sea ice cover is reconstructed along the eastern Greenland margin and in the western Arctic, showing a pattern not unlike the dipole anomaly that was observed during the 20th century.