Many people from high latitudes have heard about ice ages, but don’t necessarily know much. Relics of the last one glitter menacingly in high mountains and around the poles. Beetling crags, great mountain bowls from which mists boil and ice-carved superhighways sweeping down, U-shaped, from highland areas bear unmistakeable witness to recent glacial action. For European and many North American geoscientists it leaves an enduring impression from their earliest high-school field trips. Evidence and processes deduced from it leap out from the landscape. More than that, glacial erosion lays bare the bones of underlying rock architecture better than anywhere else.
Ice-age glaciers were not confined to mountains, but spread as vast sheets across surrounding plains. The Antarctic ice sheet gives a hint of their former magnitude at high northern latitudes. Armed with rocky debris lifted from their sources, powered by their accumulated mass and gravitational potential, few obstacles were able to divert them. Inexorable rasping and plucking filled the ice with debris of every possible grain size. Dumped at the ice sheet’s front, where melting and evaporation balanced continual supply, or at its base, this glacial muck built the glutinous boulder clay or till on which much of North America’s wheat crop now grows. There is no mistaking till with its jumbled mix of clay to house-sized fragments. Where it stops marks the furthest advance of any of the ice sheets. Dusseldorf, Watford and St Louis are redeemed by sitting close by signs of such glacial exhaustion. Bare land beyond the ice-sheets felt their influence too. When glacial ice melts it does so in a fearsome way. There is so much meltwater that, whether an area is climatically wet or dry, land beyond the ice front is inundated. Powerful meltwater streams form sheet floods across flat ground to lay down a superabundance of newly liberated sediment. This creates a bewildering variety of landforms and blankets of rapidly dumped gravels, sands and silts. The bulk of the Netherlands and its unrelenting flatness would not be there but for melt streams that followed the present Rhine and Meuse valleys from their source in an Alpine ice cap.
Seasonally frigid ice-front plains repeatedly froze and thawed to open ice-wedged cracks, sometimes clear today in road cuttings through till and glacio-fluvial gravels. Ice-marginal land has little vegetation to grip the newly settled sediment. The remaining large ice masses build up dense, cold air above them, which spills off as high winds. The world clenched in glaciation was a very dusty place, and much of the finest rock flour was whipped across mid latitudes in the northern hemisphere by such katabatic winds. Dust storms raged across northern Eurasia, North and South America to build the thick yellow, loess soils of western China, the US Great Plains and the Eastern foothills of the Andes. Signs that the tropics were both drier and windier than today lie in larger tracts of now vegetated sand dunes than are found beyond modern deserts. The great tropical lakes of East Africa dried up almost completely at the time when northern ice sheets reached their maximum extent.
Once there were thought to have been just four ice ages in Europe, deduced from the deposits that they left behind. Sedimentary terraces laid down by glacial melting in the Eastern Alps showed four separate pulses in tributaries of the Danube, which gave their names to each glacial advance. Any Earth scientist who studied in the 1960s is unlikely to have forgotten Gunz, Mindel, Riss and Würm. Land evidence is, however, not trustworthy for divining intricacies of climates past. Glaciers are adept at removing signs of their predecessors; they simply ride over and snatch the bulk of them away. Evidence of earlier ice ages on land comes in isolated yet tantalizing snippets. Terraces of modern rivers yield large-mammal fossils, in some cases hairy elephants and rhinos from glacial epochs, in others are great surprises for high latitudes of semi-tropical faunas, but always fragmentary and rarely going back far in time. Peats built from plant debris in ponds contain layered sequences with blown-in pollen grains that record changing vegetation. Beetles in peat are even more useful. Experts know intimately their current climatic ranges and habits, and most are very sensitive to environmental change. Beetles and pollen help fine tune ideas about changing ecosystems as well as climate.
Turbid rivers sourced in ice caps dump sediment at continental margins, which collapses oceanward as unstable slopes build up. Icebergs calved from floating ice shelves that extend from continental ice sheets drift far over the deep oceans eventually to broadcast their debris load to the sea-floor. Alien pebbles then punch into the thin ocean-floor ooze of fine clays and dead plankton. Such dropstones are just one of many keys to climate shifts held in the almost complete time record preserved in ocean-floor sediment. Different groups of plankton, particularly foraminifera or forams, inhabit different latitudes and various levels in the ocean-water column, some of which are sensitive to water temperature. Their fluctuations in cores through the sediment column are direct but qualitative indicators of changes in ocean water that lay above and ocean drilling site. Since foram hard parts grow from sea water and the calcium and CO2 dissolved in it, their fossil shells lock in aspects of the changing chemistry of the oceans. The opportunity to use fossils’ chemistry as a quantitative guide to past climate was the brainchild of the American geochemist Harold Urey – yes, the same man who encouraged his graduate student Stanley Miller to attempt a synthesis of life’s building blocks (Chapter 13). In 1946 Urey predicted that shelly faunas’ extraction of bicarbonate ions from sea-water at different temperatures to lay down their calcium carbonate shells should differ for oxygen’s two main isotopes, 16O and 18O, because of their minute contrasts in chemical properties (Chapter 17). He reckoned that the lower the water temperature the greater the likelihood that the heavier of the two should be taken up in shells. Experimenting with captive molluscs he showed that this indeed happened in a measurable way; for each fall of 1°C, δ18O increased by 0.5‰.
Urey’s breakthrough did not work quite as he originally hoped. As well as varying with temperature, δ18O in a fossil shell depends to a much greater degree on the isotopic composition of the sea water itself, and that does not stay the same, largely because of the waxing and waning of glacial ice (Chapter 17). Sea-surface temperature varies with latitude and because of shifting currents, but bottom water is likely to remain stable and only a few degrees above freezing point whatever the global climate. Water is a mixture of two isotopic forms, H216O and H218O, which have slightly different properties. ‘Heavier’ water has a slightly higher latent heat of vaporisation than the ‘lighter’ form – to vaporise H218O needs more heat than does H216O. So, when water evaporates the resulting vapour contains more ‘light’ H216O than does liquid water. If all water vapour finds its way back to the ocean as rain and river flow nothing changes. But if some remains as snow on land to become long-lived glacial ice, 16O is gradually extracted from the oceans. As a result δ18O increases in sea water while the volume of water lessens because some is locked in ice caps.
The change in oxygen isotope composition occurs everywhere at the ocean surface and, given time, mixes to the deepest levels as the oceans circulate. This tendency swamps the effects of ocean temperature change on oxygen isotopes. Any shell-secreting marine organism inherits an indirect measure or ‘proxy’ for continental ice volume and the global climate during the course of its life. Surface dwelling, or planktonic, forams mix this effect with that from the local sea-surface temperature, but only the ice-volume signal enters shells of benthic forams living at constant temperature in bottom water. Data from fossil planktonic forams can be adjusted for ice volume using δ18O in benthic ones that occur with them in sediment cores, to yield a proxy for sea-surface temperature. Starting in the 1970s, measurements of δ18O in foram shells from ocean-bed cores built up a detailed, continuous record of how global climate has changed that now extends back 30 Ma. It is very different from the conclusions based on the sequence of glacially-derived sediments on land.
Figure 21.1 shows a 2.5 Ma record of δ18O compiled from an ocean-floor sediment core. The highest values (shown as troughs in Figure 21.1) correspond to periods with a maximum volume of ice locked on the continents, low values (peaks) showing when glaciation was at a minimum (interglacials). A similar pattern occurs in all cores, irrespective of latitude and the varying rates at which sediment has built up. Consistency of the signal is a sure indication that the fluctuations reflect global changes. You saw in Chapter 17 that glaciation in the Northern Hemisphere began at about 2.5 Ma ago, at the start of the Pleistocene Epoch. Since then the δ18O record shows 40 to 50 major cycles of glacial and interglacial conditions – roughly one cycle every 40 ka up to 700 ka ago followed by 100 ka episodes.
Figure 21.1 Record of the variation of δ18O in benthic forams for the last 2.5 Ma from a Pacific Ocean floor sediment core. Note: the δ18O axis is inverted by convention. The higher the value the more ice was locked on continents, and the colder global climate. Different variations of signal power with frequency occur in three distinct sections.
An important issue raised in Part I was James Croll’s prediction, later evaluated by Milutin Milankovic, that astronomical forces affecting the Earth’s celestial motion modulate its solar heating. Spurred by the repetitions in the ice-volume record that emerged in the 1970s mathematicians took the oxygen-isotope time series apart as a test of the Croll-Milankovic hypothesis. They converted the δ18O signal to measures of power over a range of periodicities, using a process termed Fourier analysis that treats such time series as mixtures of sine waves of every conceivable frequency. An analogy is tuning a radio signal to see which wavebands contain broadcasts. Figure 21.1 shows that Pleistocene climatic ‘broadcasts’ were made in only 4 bands, with peaks at 100, 41, 23 and 19 ka, exactly as Milankovic predicted for variations in eccentricity of the Earth orbit (100 ka period plus a smaller one every 413 ka ), in the tilt of its axis of rotation (41 ka) and in precession of that axis (23 and 19 ka) (Figure 4.3). But the powers for these periods shown in Figure 21.1 do not stay the same. For the earliest part of the record the 41 ka signal dominates, then the 23 and 100 ka peaks pop up, and for the last 700 ka, the 100 ka signal dominates. Reality, in this case, is more odd than theory.
We can be sure that gravitational effects of the Giant Planets on the Earth’s motion have stayed much the same since planetary orbits in the Solar System stabilised ~3.8 Ga ago. So why did the relative influence of each process change during the Pleistocene Epoch of ice ages? By far the weakest of the three influences on solar heating, the eccentricity variation, now has the greatest effect. And there are deeper questions. To shift climate from maximum to minimum glaciation needs a change in global average temperature of at least 5°C, a shift of between 2 and 8% in solar heat input. All the astronomical processes acting together at the same time cannot generate more than one tenth of this difference. Stranger still, the ice-volume signal tracks the variation in solar warming predicted for high latitudes in the northern hemisphere, not that for the Earth as a whole. The three astronomical processes give much the same heat-input curve in both hemispheres, but the pattern for the south is slightly shifted in time relative to that for the north. This results in a warming trend in one hemisphere occasionally being matched by a cooling one in the other.
Clearly, our planet’s climate is not entirely in the ghostly clutches of the giant gas planets. For most of its history it has gone its own sweet way, although a faint Milankovic signal can be detected when intricate sedimentary layers of many ages are analysed. About 2.5 Ma ago global climate lurched into an oscillation that it had not shown since the long-lived glacial episode of the Carboniferous and Permian Periods. The astronomical timing for the Plaistocene is spot-on, but more in the sense of the tiny electronic encouragement that a pacemaker gives to an arrhythmic heart. The full climate system involves far more than fluctuations in the ‘deposit’ side of the solar energy budget. How much is temporarily banked depends partly on Earth’s albedo. That fluctuates with changing proportions of the surface covered by water, bare soil, vegetation, ice and clouds; measurable now but only guessable for the past. It also changes with probably random peaks and troughs of volcanic emissions into the stratosphere of high-albedo sulfuric acid aerosols. Then there is the ‘greenhouse’ effect of atmospheric water vapour, carbon dioxide and methane. Atmospheric water is double edged, delaying heat loss when a gas, but reflecting it away when tiny droplets form clouds. Carbon dioxide and methane are intricately bound with living processes and those of oceanographic and geological kinds in the carbon cycle. A drop of 100 parts per million in atmospheric CO2 gives a 1.4°C global temperature fall, but there simply hasn’t been enough CO2 to be withdrawn to give the 5°C decrease needed to shunt climate into a full ice age. That requires a fall of about 350 ppm, but the highest CO2 concentration in the last 3 Ma has been a mere 300 ppm, before modern industry forced it to rise. Heat is shifted around the Earth by circulating air and ocean water, complicated by Coriolis Force, density changes in ocean water because of evaporation and formation of sea-ice, and the feedback influence of ice-controlled rises and falls in sea level that subtly change the routes taken by flow.
Hopping from one to another factor in a reductionist quest for causes is clearly futile. We are not dealing with some arcane motor on an engineer’s test bench, but just about everything there is in the Earth System! Grappling mathematically with the conspiracy of climatological reinforcements and cancelling outs is still not achievable, except in a grossly simplified way. Neither detailed information about all the variables (and ones of tiny magnitude can be very important), nor the computing power are available today. In fact there is no comprehensive hypothesis to test! Each month brings in evidence for some newly found connection between climate and one factor or another. Part of continuing research involves refining measurements, mainly the closeness of their spacing in time to reveal perhaps crucial intricacies – a problem long familiar to geologists probing the stratigraphic column. So close to our own time, spanning the conditions of humanity’s evolution and covering a topic central to our survival, matters of fine time-detail and matching between different indicators present plenty of frustrations. But a basic, albeit qualitative, framework for the ice ages is emerging. Most of the clues come from climatic proxies closest in time to our own era; those obtained by drilling into the upper layers of marine sediment and ice caps that still remain.
Evidence for the climate engine
Looking at the last interglacial-glacial cycle’s record (from ~120 to 20 ka) in cores from the North Atlantic floor at higher latitudes than 35°N, Hartmut Heinrich, a German oceanographer, showed in 1988 that six levels in drill cores had unusually large amounts of coarse debris – ‘coarse’ in this context means particles larger than 0.1 millimetres, compared with less than 0.01 mm for deep-sea sediments. He concluded that the ice sheets flanking and floating on the North Atlantic periodically broke up to launch an iceberg ‘armada’ equatorwards. From the varying thickness of the layers of ice-rafted debris (IRD) and by matching rock fragments to their bedrock sources Heinrich showed that most of the icebergs set out from Canada and Greenland, but were accompanied by lesser flotillas from Scandinavia. So, the calving simultaneously affected all land ice around the shores of the Arctic and North Atlantic Oceans, and was not due to local instabilities. It was a response to regional climatic changes. The fossils in the IRD layers confirmed a global change, as you will see. Among them were abundant shells of a foram species that lives today only beneath the Arctic winter sea ice. They turned up in cores as far south as the latitude of Lisbon. The ice-loving forams disappear from cores at the same time as do the sandy IRD fragments. Sea ice pushed south with the icebergs and then periodically retreated. The oceanic polar front extended south of 40°S at the maximum of the last ice age, about 20 to 18 ka ago (Figure 21.2). Apparently, there was no Gulf Stream to carry warm tropical water to high northern latitudes. And the same occurred, to a lesser extent, during each of the earlier Heinrich events.
Figure 21.2 Estimated positions of the North Atlantic ocean polar front (the boundary between cold and warm surface water) at different times since the maximum of the last ice age.
While oceanographers plumbed the depths of oceanic ooze, glaciologists drilled into the Greenland and Antarctic ice caps. Such masses continually flow outwards, so do not possess an indefinite record. In Greenland, drilling can reach only a little further back in time than the end of the second-last ice age. But the pickings are rich, because ice traps both the dust that continually falls on its surface and air that is enveloped by snowfall. The ice itself records air temperature over the ice caps calculated from variations in the δ18O and ‘heavy’ hydrogen or deuterium (δD) content of glacial ice itself. Calibrating depth with time from the annual layers that can be counted back 8600 years in the upper parts of the Greenland ice cores and allowing for compaction, scientists can match data changes with time from the ice cores with those in marine sediments. Far from a constant decline in temperature to the last glacial maximum around 20 ka ago, they found around 25 episodes since about 85 ka that begin with rapid warming followed by slow cooling. The cold parts in five of these Dansgaard-Oeschger events – named after their discoverers – coincide with Heinrich’s iceberg armadas (Figure 21.3). Interestingly, perhaps crucially, they appear clearly in the Greenlandic ice record but not in the far longer one from Antarctica.
Climate seems to have a whole series of pacings, some more regular than others. For the Dansgaard-Oeschger-Heinrich cycles the periods are shorter and more irregular than any for which astronomical factors could be held accountable. Compared with the record at lower latitudes, best shown in pollen samples from sediments deposited in swamps and lake beds, each cooling pulse in the ice and marine record matches evidence for retreating vegetation and falling lake levels in the tropics. The sudden warmings link with increased atmospheric methane in ice-core gas bubbles, possibly as a result of short-lived boosts to global plant life brought by increased humidity in warmer episodes. The simplest, cheapest and most finely resolving means of analysing ice-core records is to measure the ice’s electrical conductivity, which varies with the content of acids in the ice. Volcanoes continually emit sulfurous gas that quickly falls out in rain or snow. With this reasonable assumption, ice throughout a core ought to show good conductivity, albeit with ‘spikes’ due to major eruptions. It does not. Warmer parts of the Dansgaard-Oeschger cycles show clear evidence for acidity, but ice from the cold episodes conducts virtually no current. This can be explained by volcanic acid that must have fallen with snow being neutralised during cold episodes by fine carbonates carried in dust. This matches well with measurements of varying dust content in the ice, and indicates that cold episodes were windy, and dry surfaces bare of vegetation were widespread. The cheap and cheerful conductivity measurements are much more closely spaced than any others, and the changes from warm-wet to cold-dry are abrupt. Though the conductivity evidence is not calibrated accurately to time, glaciologists estimate that the transition from one extreme state to the other may have taken about ten years, perhaps even less! What triggered the changes must therefore have been equally brusque.
Two important features emerged as time resolution in climate analysis improved. First, the exact sequence of changes in 19 of the Dansgaard-Oeschger cycles popped up in cores from a small, deep marine basin at 34°N, just offshore of Santa Barbara in California. Sedimentation there has been ten times faster than is the case far out in the Pacific Ocean, and so rapid changes are easier to spot. The fluctuation measured in the Santa Barbara Basin is shown by varying thicknesses of sediments that reflect changing oxygen content of water that flows into the basin. Today the basin has bottom water devoid of oxygen and sediments with fine bands build up, for there are no bottom dwelling creatures able to snuffle around in the mud for food falling from above. When oxygen was present in the bottom water, burrowers stirred up the sediment to give it a homogeneous character, and it is also less rich in organic material and so lighter in colour. The pattern of alternations between stirred and unstirred muds in the cores match with the Greenland ice-core temperature data. Laminated muds correlate with warm parts of the cycles, and stirred mud with cold parts. The explanation is complicated, but the observation serves to show that whatever happened in detail in the North Atlantic affected the Pacific too; even the most intricate climatic events gripped the whole northern hemisphere.
Figure 21.3 Detailed records from 20 to 65 ka before present for the North Atlantic region. (a) Greenland ice-core oxygen isotope proxy for air temperature above the ice cap; (b) Sea-surface temperature variations estimated from oxygen isotopes in planktonic forams from a sediment core at 60°N SW of Iceland. Pale-blue bands represent Heinrich events; red lines are Dansgaard-Oeschger events.
A second detailed and highly revealing study of marine cores is from between northern Scotland and the Faroe Islands. It involves a close look at variations with time in the abundance of many kinds of foram. Each of the Dansgaard-Oeschger cycles resolves itself in the ice core δ18O data into a period of 2 to 5 thousand years of cooling to full glacial cold and then a sudden emergence to much warmer conditions, within 100 years at most (Figure 21.3a). Roughly the same pattern emerges from estimates of sea-surface temperature using oxygen isotopes in planktonic foram shells (Figure 21.3b). Then the process more or less repeats itself. As you will see, this sort of detail helps flesh out the processes underlying at least the shorter term fluctuations of climate.
For the last 11700 years global climate has been warm and stable compared with events going back more than 100 ka. During the Holocene Epoch the rise of agriculture then technology, explosive population growth and the ‘Age of Reason’ have had it easy climatically. But the Earth’s emergence from the last ice age was by no means simple. From about 15000 years ago warming took temperatures to almost present levels by 13000 years ago. Glaciers melted rapidly, vegetation invaded formerly sterile areas and humans followed herds into the newly reclaimed lands. A time to shed furry clothes, a time of plenty, but a breathing space that came to a very sudden end. Until only three decades ago, the clearest sign of a change at this time was in northern peat deposits, in which the pollen of a small alpine plant, the mountain avens (Dryas octopetala) suddenly overwhelms all other types at 12900 years. Such preponderance had occurred once before on the way out of the depths of the preceding glaciation. Every record of climate analysed since then in the Northern Hemisphere, whether marine, ice core or tropical, shows that what happened was not just the explosion of a rather pretty little white flower. Within a few years global temperature plunged almost to full glacial levels. The ice was back! Much of the majesty of Britain’s upland areas was carved by this last re-advance of glaciers. But it was not a time to linger and enjoy the spectacle. As well as gripping cold, dust storms whipped across northern lands once again and grazing largely vanished from large areas of polar desert and steppe tundra, especially around the North Atlantic. Deserts also spread in the tropics and central Asia. This Younger Dryas event lasted over 1200 years, until temperatures rose to reach the present warm, stable state at about 11700 years ago, again with alarming pace – over a mere decade or so. The sheer suddenness of the Younger Dryas is absent from the Antarctic records, possibly because that continent was already in a frigid spell that had begun a thousand years earlier. Interestingly, the Antarctic ice record extends back 800 ka, with a chance of reaching a million years. More of this later.
The proxy records of ice volume, temperature, humidity and windiness gives us a fairly clear picture of the course of climatic events, at least in the northern hemisphere and probably globally. Simply accounting for them by back and forth movement of ice sheets and the North Atlantic polar front is circular and explains nothing. Although the Sun’s heating fluctuates differently in the northern and southern hemispheres, because of astronomical factors, the signals in the ice-core record from Antarctica broadly match those from Greenland. The astronomical influence in both hemispheres is entirely that affecting the northern hemisphere. Getting to the roots of climate change means looking to the north and to Earth processes there, and how they might have gripped the course of world events through other processes. Such linkages must be capable of inducing huge energy shifts, and doing that extremely suddenly.
Until the last 200 years there was never enough carbon dioxide available in the atmosphere to account for climate shifts into full glacial conditions even if it was removed entirely. However, being a ‘greenhouse’ gas it must have played some role. Atmospheric mixing ensures that changes in its composition spread globally in less than a year. Gas bubbles in ice cores preserve a unique record of atmospheric composition as well as those of air temperature, acid rain and dust gleaned from the ice itself. Figure 21.4 shows the last part of the Antarctic ice record, in which CO2 tracks temperature, rising to peaks in warm episodes and falling as climate gets cooler. While school-book physics says that more gas dissolves in water as things cool down, that can only account for a tiny part of the changes. Explaining the atmospheric CO2 pattern requires looking to sea life, its death and burial on the sea floor. That biological-geological cycle can pump the gas, or at least the carbon formerly in it, from the atmosphere to store it in marine sediments. The close correlation between the two records in Figure 21.4, suggests that cold conditions were associated with explosive growth in the biological productivity of the oceans and a rain of dead matter to the sea floor. That needed nutrients. Today’s surface ocean waters far from land contain a good blend of nutrients familiar to gardeners – potassium, nitrogen and phosphate – yet the remote oceans do not bloom with marine plant life. One crucial nutrient is currently lacking; once more, dissolved ferrous iron (Fe2+) which is a micronutrient that is essential for phytoplankton. Glaciers deliver ground-up rock flour in huge volumes, so fine that it can circulate in currents far out from land. The same applies to the dust picked up during the dry, windy periods of cold. Mineral dust contains iron, so glaciation may have fertilised the oceans to pump one greenhouse gas out of air by biological means. This would reinforce cooling, but it would not explain it. The ‘greenhouse’ effect can be struck out as the supreme engine for ice-age ups and downs.
Figure 21.4 Carbon dioxide fluctuations in the Antarctic ice cap matched to changing polar air temperature relative to that now.
We could examine each of the many factors involved in the climate system to try and tag the ultimate culprit, but the same chicken and egg riddle rears up in each case. Reductionism is not the way forward, but trying to grasp the whole system. Climate for the last 2.5 Ma has been unstable; a monumental truism. Here’s another; the Earth system itself provides a great many contributory factors with some degree of linkage to climate (and in many cases among themselves). Were the Earth a completely inert pool ball the main energy input from the Sun would willy-nilly vary in a complex way governed by the Milankovic effect. Like the astronomical ‘drivers’ the multiplicity of Earth-system factors are not in perfect lockstep, and some have opposing influences. A metaphor might serve to clarify things somewhat.
For the moment substitute water flows into and from a bucket for all the combined influences on climate. There are drops, drizzles and floods of influence from each, some positive and others negative at any one time. Suppose that the bucket represents the overall state of climate. Not an ordinary bucket with a handle, but one pivoted so that it can rotate (Figure 21.5). At the beginning of one state a variety of climatic influences start filling the bucket. For a while it doesn’t matter which influence is the greatest, how many there are, nor whether or not each fluctuates in its own right. While the centre of gravity of the system lies beneath the pivot the system is stable, even though the state changes. Once the flow begins to push the centre of gravity above the pivot the state suddenly becomes unstable at some critical level. One drop more from whatever influence tips the bucket over and spills its contents. The system enters a different state, to be filled to the cusp of sudden transformation again. The higher the pivot is placed on the bucket the longer it takes to become unstable, and the larger the final influence to overturn it. Those ‘causes’ that are part of the Earth system, like the greenhouse effect or ice sheets, become transformed into ‘effects’, indeed they oscillate between the two. Those that participate from far beyond, the astronomical factors, pulse away regardless. Despite their miniscule influence on solar warming, given a sufficiently unstable system their very regularity makes them the most likely candidates for the final catastrophic push: the tiny conductors that control the tempo, signal the crescendos and diminuendos, and cue in or out the instruments in a mighty orchestral symphony.
Figure 21.5 The bucket analogue for the climate system.
The broad Cenozoic trend towards global cooling was undoubtedly Earth’s own geological doing (Chapter 20). Eventually a limit was passed when climate became extraordinarily sensitive; i.e. the metaphorical bucket’s pivot descended lower. Many geologists reckon that the critical shift was the closure of the straits between the Americas (Figure 17.7) that stopped efficient exchange of water and the energy that it contained by tropical flow between the Pacific and Atlantic oceans. That occurred around 3 Ma ago. Warm water could then drift north and south from the tropical Atlantic Ocean to increase humidity and winter snowfall at high latitudes. This would not have much effect on the long-established, but thermally isolated Antarctic ice cap, glaciers from which must eventually flow to the sea, float as ice shelves and break into icebergs. Although holding 90% of the world’s ice it cannot draw more water into itself from the oceans than it had achieved when it first formed to roughly its present, almost complete extent. It would simply get thicker and higher, spread faster and supply more icebergs to remelt. Ice caps in the northern hemisphere can spread on continents surrounding the Arctic Ocean as far as conditions permit. They can both draw down and resupply ocean water. In changing their extent, they change the albedo of the northern hemisphere and thus part of solar heating. Antarctica cannot do this.
Glacial ice does not form just because the climate is cold. Today there are vast areas of low plains at high northern latitudes, such as the North American Prairies and Russian Steppe, where in winter any undue exertion may freeze our lungs and car tyres become brittle. They have no glaciers today because they are deprived of the necessary moist air. An influx of maritime air over such areas could make ice sheets form, for instance from a change in Pacific Ocean currents along the west side of North America. The weak changes in solar heating brought about by fluctuating axial tilt and precession can shift the limit for potential glacier formation by as much as 600 km north or south. Given moisture supply to the repeated accumulation centres of continental ice sheets in Canada and Scandinavia, the 41, 23 and 19 ka cycles in land-ice volume before 700 ka seem simply explained, but not the 100 ka pacing seen thereafter which matches the much weaker eccentricity influence. It is useful to look more closely at huge ice sheets.
Building to as much as 5 km thick, continental ice creates its own topography. Continental crust is forced down 270 m for every kilometre of ice sitting on it. Around the ice some uplift results from displacement of the weak asthenosphere towards ice-free crust. Therefore, as an ice cap forms moist air is forced higher to shed more snow. When it begins to melt, as solar heating increases in the astronomical cycle, bedrock topography does not spring back immediately to its pre-glacial state. The asthenosphere is sufficiently viscous for rebound to be sluggish – ten thousand years after their ice caps vanished, Canada, Scandinavia and northern Britain are still rising. Bowed-down crust stays depressed while the ice melts. The more ice melts, the lower its remaining surface and the faster melting becomes. Starting at low latitudes the melting increases the surface gradient on the ice cap, so that unmelted northerly ice flows more rapidly into the depression, reinforcing the rate of melting. As solar heat input slackens once more, depressed land has had insufficient time to reach the former elevation where ice first got a grip.
Since ice ages began 2.5 Ma ago, glaciers have returned many times, each with a prodigious capacity for planing off the surface of the crust and dumping the debris elsewhere, as often as not in the sea. The crust has thinned and during each interglacial warm period it bobs back to lower and lower elevations in the ice-cap heartlands. Each ice age would then need more climatic encouragement to get a grip. Yet back they came with a 41 ka pace until 700 ka ago. Maybe the general cooling of the preceding 60 Ma continued as a background to counter the erosional effect. Perhaps the flip to the tiny influence of the eccentricity and 100 ka pacing marks a threshold when only the combined effect of 100, 41 and 23 ka signals gave sufficient cooling to encourage glaciation in ground-down areas. This sort of procedure must go on to some extent, but it is clearly not the whole story. It does not explain the short frequency Dansgaard-Oeschger events that are such a dramatic part of the last ice age and probably earlier ones.
Ocean water circulates globally, presently taking cold deep water from the North Atlantic and the Antarctic on a meandering journey across both hemispheres and from Atlantic to Pacific (Figure 2.3). Both wind and its guidance by Coriolis Force, and the drag to replace sinking near-polar brines move surface waters in swirling current streams. Ocean circulation, with a pace measured in hundreds of years, moves far more solar energy annually than the more rapid agency of winds because of water’s far greater thermal capacity. Flows of deep water involve two means of increasing salinity and thereby density; freezing out of pure water in sea ice and evaporation at low latitudes (Chapter 2). Cold brines that sink and flow equatorwards at deep levels have only two sources, in the North Atlantic and in those embayments in Antarctica where sea ice can form. There is barely any such flow from the Arctic Ocean to the Pacific where the shallows of the Bering Straits hamper any deep escape from the Arctic Ocean.
The North Atlantic Deep Water draws a balancing flow polewards, in the form of the Gulf Stream and North Atlantic Drift that keep high latitudes warmer and wetter than they might otherwise be. But there is more than that process alone. Intermediate depths in the North Atlantic are occupied by warm dense water that sinks after becoming saltier through evaporation in the tropics. It too moves polewards to break surface south of Iceland when winter storms there push surface waters aside. High winds evaporate from this warm upwelling, cooling it and making it more briny and dense still. This sinks together with brines formed by freezing of Arctic Ocean surface water, further ‘pulling’ the surface flow. Between them these two processes release energy at high latitudes that is equivalent to a quarter of the average solar heating there. All this stems from past history, closure of the Isthmus of Panama and opening of the Atlantic before that; an air, ocean and tectonic conspiracy of influences.
Suppose this exquisite interconnection somehow switched off because the Gulf Stream was no longer ‘pulled’ towards the Arctic Ocean; what then? Land surrounding the North Atlantic must cool, and since its previous warming was mainly by air flow from warm ocean water of the Gulf Stream, the cooling would be almost instant. The polar front would spread southwards, as it so manifestly did during the Dansgaard-Oeschger-Heinrich events and during the Younger Dryas (Figure 21.2). There are plausible ways of throwing the switch. The most obvious is the most contrary. Given enough warming by the Gulf Stream, Arctic sea ice would stop forming, thereby reducing the source of the deep equatorward flow. Global warming itself precipitates glaciation? A sobering notion for us today. A surface layer somehow made less salty will also do the trick, because the cold sinking brine needs an unusually salty water supply to begin the process.
Calculations suggest that a decrease of only one or two parts of salt in every thousand is sufficient. Today the Atlantic is one part per thousand more salty than the Pacific due to ‘export’ of evaporated moisture by tropical easterly winds across Central America, enough to sustain the deep water flow. Less evaporation during a period of cooling might decrease Atlantic salinity, but that would be a slow process. The third trigger is influx of fresh water into the Arctic cold-brine factory. Having a low density it would float on ocean water mixing slowly with salt water to dilute it. That would still freeze but the residual water would not become salty enough to sink. But where would vast amounts of fresh water come from? Increased rain is a possibility, but in the throes of an ice age the obvious source is from melting of the ice sheets themselves. The Dansgaard-Oeschger (D-O) cycles involve sudden warming followed by steady cooling; i.e. retreat of the ice sheets followed by their steady re-advance. Heinrich’s armadas of icebergs, whose melting must freshen surface waters in the Atlantic, may have been crucial. Three of them took place during the coldest part of D-O cycles and immediately precede rapid warming, another coincides with the rapid warming and the youngest shown on Figure 21.3 is within the last glacial maximum. Is this a case of ‘binge and purge’, the great ice caps periodically reaching such a thickness that they collapsed?
Study of forams in ocean-sediment layers spanning the short-period climate changes of D-O events details the changes in ocean flow that accompanied them. Different foram species chart the back and forth flows of different water layers. During the coldest part of each D-O cycle the corresponding ocean-floor sediment layers are full of Arctic forams that thrive beneath sea ice. But a stagnant deep circulation results in a build-up of more saline water formed in the tropics unable to be shifted north because the Gulf Stream was shut down. Instead, it must spread slowly polewards at intermediate depth beneath the cold but less-saline surface layer. Its assembly of warm-water forams enters the North Atlantic cores shortly before the rapid warming above the Greenland ice cap. Eventually this water must well to the surface, perhaps destabilising extensive glacial ice shelves whose break up may result in overthickened glaciers on land surging seawards; hence Heinrich’s armadas. This is the start of the short-term Dansgaard-Oeschger cycles, many of which follow rising sea-surface temperature in open water (Figure 21.3b). The arrival and cooling of saltier intermediate water once more produces dense brine to sink and restore poleward circulation, dragging in warm surface water to give a brief temperature rise until temporarily renewed melting overwhelms the system with fresh melt water at the surface. Once more the circulation is stopped and the glaciers expand.
The biggest of the climatic shocks, the Younger Dryas that took place just while the last ice age waned, had its greatest effects around the North Atlantic Ocean. However, its abrupt signal emerges from climate records around the northern Pacific and in the Andes of South America. Yet in Antarctic ice records from the same time period shows warming; a period called the Antarctic Cold Reversal. The evidence for what triggered the Younger Dryas comes from two sources: indications of changes in salinity in the Gulf of Mexico from its planktonic forams, and the history of a vast lake that formed in central Canada at the retreating ice front. Lake Agassiz, named after the first scientist to postulate ice ages, still exists as a small relic in the form of Lake Winnipeg. It formed a staging post for the meltwater from the North American ice sheet. At first lobes of ice blocked its escape directly to the North Atlantic and the filling lake spilled through the Mississippi Valley to the Gulf of Mexico. Plankton there show signs of low saltiness until 12900 years ago, when normal salinity returned. The sediments of Lake Agassiz, dominating the wheat fields of Manitoba, show that the lake level dropped drastically at that time. It had drained somehow, but not through the Mississippi. A glance at the map of North America shows the most likely outlet, through the Great Lakes system and the St Lawrence valley to the North Atlantic. The Younger Dryas matches with a simple event that would have shut down Atlantic circulation, a flood of fresh meltwater floating out across the North Atlantic.
The Antarctic Cold Reversal – its warming at the time of the Younger Dryas – has been explained by the shutdown of North Atlantic deep water formation. Instead of northward transfer, warm Atlantic surface water at low latitudes would have ‘backed-up’ and spread southwards to increase temperatures at higher southern latitudes and over Antarctica. Close comparison between the Greenland and Antarctic ice records through the last 120 ka reveals similar mismatches between short-lived events. Periods of warming in Antarctica lag behind the rapid warming events in Greenland by roughly 220 year. This climatic bipolar see-saw also seems to have resulted from oceanic heat exchange across the Equator that fluctuated with the Dansgaard-Oeschger events.
The extreme climatic events so briefly covered in this chapter formed the backdrop for human evolution and migration, for cultural change in the broadest conceivable sense, and for our future survival. We now have an inkling of how climate change links in with Earth’s foibles as well as the astronomical pacemaker. Humans are now major players in the weaving of climatic events, but before ten thousand years ago we were at its whim.