Abrupt Climate Change During the Last Ice Age

Unlike the relatively stable climate Earth has experienced over the last 10,000 years, Earth's climate system underwent a series of abrupt oscillations and reorganizations during the last ice age between 18,000 and 80,000 years ago (Dansgaard 1984, Bond et al. 1997, 1999). These climate fluctuations were first discovered when scientists reconstructed past temperature variability over Greenland by analyzing tiny changes in the relative abundance of the oxygen-16 isotope versus the oxygen-18 isotope (noted as δ₁₈O and reported in parts per thousand) in ice cores recovered from Greenland glaciers. Each successively deeper ice layer represents a snapshot of Earth's climate history from the past, and together, the oxygen isotope record told a story of abrupt, millennial-scale climate shifts in air temperatures over Greenland between extremely cold stadial conditions and relatively mild interstadial periods during the last ice age (Figure 1) (Alley 2000, Alley et al. 2003). There are twenty-five of these distinct warming-cooling oscillations (Dansgaard 1984) which are now commonly referred to as Dansgaard-Oeschger cycles, or D-O cycles. One of the most surprising findings was that the shifts from cold stadials to the warm interstadial intervals occurred in a matter of decades, with air temperatures over Greenland rapidly warming 8 to 15°C (Huber et al. 2006). Furthermore, the cooling occurred much more gradually, giving these events a saw-tooth shape in climate records from most of the Northern Hemisphere (Figure 1).

Figure 1: Climate variability over the past 80,000 years.

Top panel shows the oxygen isotope record (δ¹⁸O_ice) from the Greenland Ice Sheet Project II (GISP II) ice core over the last 80,000 years (Stuiver & Grootes 2000). Colder air temperatures are indicated by more negative δ¹⁸O_ice values. Bottom panel shows changes in global sea level over the same time period (Waelbroeck et al. 2002), reflecting the waxing and waning of continental ice sheets during the last ice age. Note the relatively stable, small temperature changes reflected in the ice core δ¹⁸O values over the past 10,000 years when continental ice volume has been near modern levels, indicating the modern warm period known as the Holocene. In comparison, climate during the last ice age (between about 18,000 and 80,000 years ago) was much more variable. The abrupt warming and gradual cooling oscillations during this period of Earth’s history are know as Dansgaard-Oeschger cycles, or D-O cycles (Dansgaard 1984).

Although the D-O climate cycles have now been found in many other climate proxy records around the globe (Voelker 2002), the reason why Earth's climate was so much more variable during the last ice age is still unknown. Some theories suggest that changes in ocean circulation due to natural variations of North Atlantic surface water salinity were the trigger for the D-O events (the salt oscillator hypothesis) (Birchfield & Broecker 1990, Broecker et al. 1990a, Zaucker & Broecker 1992), while others argue changes in atmospheric circulation were the driver (the wind field oscillation hypothesis) (Clement & Cane 1999, Seager et al. 2002, Seager 2006, Seager & Battisti 2007).

In general, surface winds in the tropics, known as the trade winds, blow from east to west (Figure 2). As these powerful winds blow across the tropical Atlantic, they evaporate ocean water and transport the moisture across Central America where it rains out into the Pacific. This causes a net transport of freshwater out of the North Atlantic basin and makes the surface waters there very salty (Figure 2) (Broecker et al. 1990b, Zaucker & Broecker 1992). As these warm, salty waters of the tropical Atlantic circulate north to the sub-polar regions of the North Atlantic via the Gulf Stream, the ocean transfers heat to the atmosphere through evaporation and warms the North Atlantic region (Roemmich & Wunsch 1985). This evaporation process causes the surface waters to cool and become saltier (Conkright 2002). Because water density increases as temperature decreases and salinity increases, these waters eventually become dense enough to sink in the region south of Greenland and in the Norwegian Sea, forming North Atlantic Deep Water (NADW) (Figure 2). The NADW then flows southward along the bottom of the Atlantic Ocean carrying with it the salt that built up as the surface waters transited from the tropical to sub-polar regions. This transport of northward flowing surface waters and southward flowing deep waters is what oceanographers call Atlantic Meridional Overturning Circulation (AMOC) (Oppo & Curry, this Topic Room).

Figure 2: Sea surface salinity of the Atlantic and Pacific oceans.

The convergence of warm, moist air from the trade winds (purple arrows) leads a region of convection and cloudiness along the equator known as the Intertropical Convergence Zone (ITCZ) (grey line). The trade winds blowing across the tropical Atlantic cause a net transport of freshwater out of the North Atlantic basin and makes the waters there very salty (Broecker et al. 1990b, Zaucker & Broecker 1992). As these warm, salty waters of the tropical Atlantic circulate north to the sub-polar regions of the North Atlantic via the Gulf Stream (dark blue arrows), evaporation causes the surface waters to cool, and the remaining salt results in the formation of very cold, salty surface waters (Conkright 2002). These waters eventually become dense enough to sink in the region south of Greenland and in the Norwegian Sea, forming North Atlantic Deep Water (NADW). The NADW then flows southward along the bottom of the Atlantic Ocean (green arrows). This transport of northward flowing surface waters and southward flowing deep waters is what oceanographers call Atlantic Meridional Overturning Circulation (AMOC). (sea surface salinity map created using http://iridl.ldeo.columbia.edu/SOURCES/.NOAA/.NODC/.WOA09)

This theory assumes that NADW transports salt out of the North Atlantic when AMOC is strong, resulting in a gradual reduction in North Atlantic surface salinity over time. If surface waters at the sites of deep-water formation become too fresh, then AMOC would weaken or shut off because the surface waters would not be dense enough to sink. Furthermore, because the amount of tropical heat transferred to the high-latitude North Atlantic by ocean currents is reduced when AMOC weakens, ice sheets would be able to grow. As salt continued to accumulate in the North Atlantic during periods of reduced NADW formation, eventually surface waters at key sites of deep-water formation would become salty and dense enough again to sink, thus restarting AMOC and causing an abrupt warming in the high-latitude North Atlantic.

This proposed ‘salt oscillator hypothesis' was first put forth by Broecker et al. (1990a) as a mechanism to explain how salinity oscillations in the Atlantic Ocean could modulate the strength of AMOC and cause the D-O cycles of the last ice age. Since then, several studies have suggested that it only takes a small reduction in sea surface salinity to alter the rate of NADW formation, and that large inputs of fresh water from melting ice sheets to the North Atlantic could even cause a complete collapse of AMOC (Figure 3) (Clark et al. 1999, 2001, Stouffer et al. 2006). Furthermore, the presence of the ice sheets around the margins of the North Atlantic during the last ice age provided a freshwater source that could rapidly alter surface salinity at sites of deep-water formation. During warm interstadials, enhanced heat transport to the North Atlantic would cause melting along the margins of the ice sheets, gradually reducing surface salinity and weakening AMOC until stadial conditions returned (Figure 3).

Furthermore, computer models (Dahl et al. 2005, Lohmann 2003, Stouffer et al. 2006, Vellinga & Wood 2002, Zhang & Delworth 2005) of global ocean and atmosphere circulation show that a reduction in AMOC and the associated high-latitude cooling also causes a coincident cooling in the northern tropical Atlantic and a southward displacement of the tropical rain belt known as the Intertropical Convergence Zone (ITCZ). Indeed, proxy studies also show that the ITCZ migrated southward during stadials (Peterson et al. 2000, Wang et al. 2004). As a result of these changes, evaporation from the tropical North Atlantic is enhanced while freshwater input to the South Atlantic increases (Figure 3) (Krebs & Timmermann 2007a, 2007b, Lohmann & Lorenz 2000, Lohmann 2003, Stouffer et al. 2006, Vellinga & Wu 2004). Accordingly, changes in the tropical hydrologic cycle are predicted to have increased sea surface salinity in the tropical/subtropical North Atlantic during periods of reduced AMOC. Eventually, this positive salinity anomaly would be transported to the high-latitudes, increasing surface water density, and contributing to an abrupt shift to vigorous AMOC circulation and a return to mild interstadial conditions. Therefore, the tropical hydrologic cycle may act as a negative feedback response that increases North Atlantic salinity during periods of reduced AMOC.

Figure 3: Mechanism of the salt oscillator hypothesis.

During warm interstadials, when AMOC is stronger (a), enhanced northward oceanic heat transport results in warmer conditions in the North Atlantic. Warmer conditions in the North Atlantic cause the ice sheets around the North Atlantic to melt, gradually reducing surface water salinity. In addition, the more northerly position of the ITCZ during interstadials results in enhanced freshwater precipitation in the tropical North Atlantic, further reducing surface salinity. Eventually, surface salinity is reduced enough to weaken AMOC, shifting the climate into a cold stadial (b). During stadials, cooler conditions in the North Atlantic reduce meltwater input from the ice sheets, allowing surface salinity to increase. In addition, the ITCZ shifts southward during stadials, reducing the amount of freshwater input to the tropical North Atlantic. Both mechanisms result in an increase in North Atlantic salinity that eventually causes AMOC to strengthen, returning the climate system to an interstadial. (map created using www.planiglobe.com/omc_set.html)

An alternative hypothesis to explain the D-O cycles of the last glacial period has to do with large-scale changes in wind patterns over the North Atlantic (Romanova et al. 2006, Wunsch 2006). Mountain ranges are responsible for forming special atmospheric circulation features, known as stationary waves, which form as wind is pushed up and over topographic highs. Seager et al. (2002) showed that the configuration of Northern Hemisphere stationary wave patterns, which are strongly influenced by the height of the Rocky Mountains, plays an important role in determining the flow path of the modern subtropical jet stream. The jet stream is a fast flowing current of wind in the upper atmosphere responsible for steering storm systems in the mid-latitudes around the globe. As the jet stream flows over the Rocky Mountains, it is forced to dip southward over the central US and then turns sharply to the northeast to flow over the open North Atlantic and then northern Europe. Therefore, the presence of the Rocky Mountains in North America forces the jet stream flow to be less zonal (east-west oriented) and more meridional (north-south oriented) over the North Atlantic. Because the jet stream tends to separate colder, sub-polar air masses to the north from warmer, tropical air masses to the south, climate over northern Europe is warmer compared to climate at the same latitude on the western side of the Atlantic (e.g., Newfoundland).

During the last ice age, massive continental ice sheets up to five km high covered much of North America and northern Europe (the Laurentide and Fennoscandian ice sheets, respectively). Like mountain ranges, the ice sheets were tall enough to influence stationary waves and change the path of the jet stream. Indeed, computer modeling results suggest that regional elevation changes in the Laurentide ice sheet could have had a large effect on the position of the jet stream during the last glacial period (Figure 4) (Jackson 2000). Studies have suggested large changes in sea level (up to 35 m) across D-O cycles (Siddall et al. 2003) due to the melting of continental ice sheets. Therefore, abrupt topographic changes in Northern Hemisphere ice sheets could have forced large-scale atmospheric circulation reorganizations as continental glaciers constantly waxed and waned. According to this hypothesis, the path of the jet stream over the Northern Hemisphere oscillated between more meridional (Figure 4a) versus zonal flow (Figure 4b) as it interacted with changes in the elevation and overall size of the continental ice sheets, resulting in the saw-tooth pattern of D-O cycle temperature changes recorded in the Greenland ice core records (Wunsch 2006).

Figure 4: Mechanism of the wind field oscillation hypothesis.

The hypothesized position of the jet stream during warm interstadials (a) compared to cold stadials (b) during the last ice age. The presence of larger and thicker ice sheets during cold stadials forced the jet stream to shift farther south, resulting in a more zonal flow across the Atlantic Ocean. Note that the jet stream separates warm subtropical air to the south from cooler polar air masses to the north. Therefore, the more zonal flow of the stadial jet stream allowed cold, glacial conditions to extend farther south, resulting in a much colder climate in the North Atlantic and Europe.

Although climate scientists have worked hard to determine the ultimate trigger of abrupt climate change during the last ice age, it is likely that a combination of ocean and atmospheric circulation changes were involved. For example, a subtle shift in atmospheric circulation to a more meridional jet stream flow would encourage the transport of warm, salty water into the sub-polar North Atlantic, which in turn could lead to the reestablishment of strong AMOC and enhanced oceanic heat transport to the high-latitude North Atlantic. In this case, ocean circulation changes associated with AMOC may have amplified small changes initiated in the atmosphere on the transition into warm interstadials. Conversely, a sudden reduction in AMOC due to an influx of freshwater into the high-latitude North Atlantic region has the potential to trigger a regional cooling that can significantly alter tropical atmospheric circulation around the globe. Although we still do not know which happened first, interactions between both the ocean and the atmosphere must have played an important role in driving the dramatic climate oscillations of the last ice age. This paper only discusses two hypotheses to explain the abrupt climate shifts of the last ice age. To read about alternate hypotheses, see the comprehensive review paper by Clement & Peterson (2008).