Past and future ice age initiation: the role of an intrinsic deep-ocean millennial oscillation

Abstract. This paper offers three interdependent contributions to studies of climate variation: (1) the recognition and analysis of an intrinsic millennial oceanic oscillation that affects both Northern and Southern high latitude climates, (2) The recognition of an oceanographic switch to ice-free seas west of Greenland that explains the initiation of the Last Ice Age, and (3) an analysis of the effect of increasing salinity in the seas east of Greenland that suggests the possibility of the initiation of an ice age threshold climate in the near future. In the first contribution the millennial oscillation in the flow of the North Atlantic Drift reported by Bond et al. (1997) is proposed to be part of a 1500 yr intrinsic deep ocean oscillation. This oscillation involves the exchange of North Atlantic intermediate-level deep water (NADW) formed in the seas east of Greenland with Antarctic Bottom Water formed in a shallow-water zone at the edge of the Antarctic continent. The concept of NADW formation is already well known, with details of the sinking water flowing out of the Greenland Sea observed by Smethie et al. (2000) using chlorofluorocarbon tracers. The concept of Antarctic Bottom Water formation is also already well established. However, its modulation by the changing fraction of NADW in the Southern Ocean, which I infer from the analysis of Weyl (1968), has not been previously discussed. The modulated lower-salinity Antarctic Bottom Water that reaches the northern North Atlantic then provides negative feedback for the cyclic variation of NADW formation as proposed here. This causes the 1500 yr bipolar oscillation. The feedback suggests the possible sinusoidal character of the proposed oscillation model. The model is consistent with the cooling of the Little Ice Age (Lamb, 1972, 1995), and it also correctly predicts NASA's observation of today's record maximum area of winter sea ice on the Southern Ocean and the present observed record low rate of Antarctic Bottom Water production cited by Broecker (2000). The sinusoidal form of this conceptual model is therefore reinforced by both old and new data, and provides insights into world-wide climate change.

The second contribution of this paper is a hypothesis for the initiation of Pleistocene ice ages, typified by the Last Ice Age that began 120 000 yr BP. Instead of the classical Northern high-latitude summer cooling caused by orbital precession and changes in Earth's axis inclination, this hypothesis proposes the sudden onset of year-round ice-free seas west of Greenland, with greatly increased precipitation in the ice sheet nucleation regions of Baffin Island, northern Quebec, and Labrador. Devon Island ice-core studies by Koerner at al. (1988) and deep-sea sediment data reported by Fillon (1985) support the concept of ice-free seas west of Greenland and imply the initial meteorological conditions that are proposed here. These conditions are consistent with the heavy precipitation inferred by Adkins et al. (1997) from deep-sea sediment data. The changes in northeastern Canada were accompanied by quite cold conditions in northern Europe, inferred by Field et al. (1994) from tree pollen data. The European cooling was probably caused by loss of the recurring Iceland low-pressure system due to the dominant effect of a frequent stronger low-pressure system over the Labrador Sea, as postulated in this paper. The key to ice-free seas west of Greenland is the loss of the near-surface stratification that normally enables sea ice to freeze. Using the high-resolution European Space Agency's ENVISAT system, I have monitored the flows through the Nares Strait and found that the dominant southward flow of lower density polar water into Baffin Bay correlated with the growing area of seasonal sea ice forming early in the winter in the Bay near the southern end of the Strait. This implies that low-salinity polar water was the cause of the stratification. A search for the cause of the stratification loss then became a search for the cause of the loss of the southward flow of polar water. The loss could have occurred if denser and more saline Atlantic water replaced the polar water in-flow. Medieval historical records suggest that an analogous partial replacement probably did occur during the early medieval climatic optimum, with some warmer Atlantic water removing the thick perennial sea ice along Greenland's north coast. The NADW formation rate and the Spitsbergen-Atlantic Current (SAC) flow were then near maximum values. I hypothesize that enough of the thick perennial sea ice along Greenland's north coast was removed by the penetration of the SAC flow into the polar ocean to enable a medieval voyage eastward along the coast in AD 1118. This voyage is implied by an old map record showing Greenland realistically as an island. An even stronger SAC flow associated with a stronger maximum in the 1500 yr intrinsic oscillation of the oceanic system was the likely trigger for the initial conditions of ice-sheet growth when the Last Ice Age began.

The third contribution of this paper is the hypothesis that modern society's activities might cause a repetition of the transition to an ice age threshold climate within one or two decades from 2013. This possibility depends on a continuing increase of salinity in the seas east of Greenland, with a corresponding increase of NADW formation and the SAC flow. The increase is currently being driven by the increasing rate of the saline Mediterranean outflow that contributes to the North Atlantic Drift. The rate increase is a consequence of the increasing salinity of the Mediterranean Sea as reported by European oceanographers (Science, 279, 483–484, 1998). The rising salinity of the Mediterranean and its increasing outflow is attributed to the diversion of nearly all the in-flowing rivers for irrigation. A further substantial salinity increase should occur with the loss of all perennial polar sea ice possibly within one or two decades from 2013 if the present trend of loss continues. The trend is displayed on the University of Illinois internet site: http://arctic.atmos.uiuc.edu/cryosphere/. The increasing salinity of the Greenland Sea is now reflected in an increasing northward winter penetration by the SAC flow. According to Lamb (1972), during the early 20th century at the time of maximum extension of sea ice in April, open water normally extended only as far north as the southern cape of Spitsbergen at about 76.6° N. But in Aprils of 2013 and 2014, open water extended 380 km farther northward to the north coast of Spitsbergen. When the SAC was running strongly to replace sinking NADW in February of 2014, I observed open water extending about 730 km north from the cape into the polar ocean to latitude 83° N, where the penetration of the SAC flow was beginning to obstruct the southward flow of polar water. Even greater seasonal extensions of the SAC flow are expected with an additional Greenland Sea salinity increase after the loss of all perennial polar sea ice. This could cut off southward movement of polar water through the Fram Strait during much of the winter, and send annual pulses of the denser Atlantic water of the SAC flow into the sea north of Greenland. If these annual pulses begin to occur and allow enough denser Atlantic water to flow southward through the Nares Strait, the Baffin Bay stratification would be lost and a switch to an ice age threshold would occur. The severity of the resulting cold regional climate might have a disruptive effect on higher-latitude societies.