Earth-logs

Now and again in the geological record, evidence turns up that suggests that the deep oceans were devoid of oxygen.  Ocean anoxia encourages burial of dead organic remains that gives rise to carbon-isotope “excursions”: signals of the anoxia itself.  A likely mechanism that starves the deep oceans of oxygen is the shut down of that part of the ocean “conveyor” driven by sinking of cold, dense brines, as happens today in the North Atlantic and around Antarctica.  Gases dissolve more efficiently in cold water than in warm.  Quite probably most oceanic anoxia events are related to global warming and increases in the “greenhouse” effect due to CO₂ rises in the atmosphere.  A group of US and British geoscientists have examined one such anoxia event in the Lower Jurassic (~183 Ma) of Denmark using both carbon isotopes and the density of pores (stomata) on fossil leaves (McElwain, J.C. et al. 2005.  Changes in carbon dioxide during an oceanic anoxia event linked to intrusion into Gondwana coals.  Nature, v. 435, p. 479-482).  Stomatal density is inversely related to the amount of CO₂ in the atmosphere, so is very useful in seeking evidence for an anoxia-climate link.

This particular anoxia event has been linked either to release of methane, which quickly causes warming and then oxidises to CO₂, from gas hydrate or to massive release of carbon dioxide itself.  McElwain et al. neatly show that the event first experienced drawdown of ”greenhouse” gas and cooling of around 2.5 °C, then sudden quadrupling of CO₂ and warming of around 6.5°C.  Such an odd pattern cannot be ascribed to methane release, but coincides with the formation of the Karroo-Ferrar continental flood-basalt igneous activity in southern Africa and Antarctica.  That involved massive intrusion into coal-bearing strata, whose thermal metamorphism would have released huge amounts of “greenhouse” gases.  Calculations of the amount of carbon mobilised to cause the shifts in CO₂ suggest between 2.5 and 4.4 trillion metric tons, vastly more than the probable amount of methane hydrate beneath the Jurassic sea floor.

enveloping glaciations during the Neoproterozoic Eon, that notion of “Snowball” conditions has received many severe knocks, charted by numerous items in EPN.  Geochemists and geologists from the Universities of Vienna and Witwatersrand realised that a good test of the hypothesis would be to concentrate on a rather obvious property of an ice-bound planet (Bodiselitsch, B. et al. 2005.  Estimating duration and intensity of Neoproterozoic Snowball glaciations from Ir anomalies.  Science, v. 308. P. 239-242).  Whatever falls on an ice sheet, whether it is cosmic dust from outside the Earth or ash from volcanoes, becomes trapped in the annual layers of ice.  When the ice melts, that accumulated content is transferred to the oceans very quickly.  With weathering in suspended animation during the glacial epoch, transport of many elements would have slowed to very low levels.  So, marine sediments deposited immediately after the diamictites that are allegedly glaciogenic ought to contain anomalously high levels of several elements.  The most important of these would be those which show very different abundance patterns in meteorites form those in terrestrial rocks.

Bodiselitsch et al. hit what seems to be “paydirt” in carbonates above a prominent diamictite in central Africa.  Their samples are impeccable, being from diamond-drill cores produced during evaluation of sediment-hosted mineralization in the famous Neoproterozoic Copper Belt of Zambia and Congo.  The core contains a prominent iridium anomaly at the very base of the carbonates, with a “signature” relative to other anomalous elements that points to a cosmic origin.  Normally such an anomaly would be ascribed to a meteorite impact, but in this case the coincidence would be too good to be true.  Instead, the authors use the magnitude of the anomaly to estimate how long cosmic dust had to accumulate to build up such a high level if it was released by rapid deglaciation.  Deep-ocean sediments from the last 80 Ma are a guide to the long-term accumulation rate of cosmic material.  If that rate is applied to the cap-carbonate anomaly, it gives a total time for accumulation in the hypothesised global ice cover of around 12 Ma.  Presumably this would have been from ice immediately overlying the area being studied.  An ice age that long defies any idea of more “normal”, astronomically forced glaciation, which would be expected to have cyclically formed and receded many times, thereby releasing the dust particles much more gradually.  Any anomalies would be expected in the diamictites themselves, yet there are none.  Although sample spacing is rather patchy through the entire succession, they are most dense around the anomaly itself.  Moreover, another suspected glaciogenic “package” higher in the sequence shows exactly the same iridium “spike”. 

Arguing against such support for the “Snowball Earth” hypothesis will be difficult, but other sequences require similar tests, most importantly those of Namibia, where Hoffman and colleagues developed their ideas, and the much more extensive deposits of Australia.  This diamictite sequence is reckoned to represent both postulated deep-freeze events of the Neoproterozoic, around 710 Ma (Sturtian) and 635 Ma (Marinoan). There is one nagging problem.  Data from one area are likely to record ice-retained cosmic dust only from ice in its immediate vicinity, and therefore do not represent the entire planet.  Much of the controversy is between supporters of a whole-Earth ice cover, and those who favour patchy glaciation (the “Slushball” model).  Unfortunately, Neoproterozoic stratigraphic correlation and radiometric age calibration is not sufficiently good to detect the same intervals elsewhere and look for anomalies there.  In fact, the stratigraphy is generally correlated from place to place by matching the diamictites themselves.  There is plenty of evidence that they may all coincide in time.

Tracking ocean circulation during the last glacial period

The use of various ocean-floor sediment proxies for climate change, such as the ups and downs of heavy ¹⁸O that chart waxing and waning continental ice cover, has progressively revealed the complexity of shifts during glacial and interglacial periods. Yet more emerged from finer-resolution time-series contained with Greenland and Antarctic ice cores.  The diversity of information that proxy for many different, climate-related processes has in the last decade enabled palaeoclimatologists to begin piecing together possible causative mechanisms, beyond the initial discovery of an astronomical signal in early oxygen-isotope records.  One of enormous significance is the possibility that sudden millennial-scale cooling and warming link to changes in ocean circulation, especially that performed by the Gulf Stream driven by thermohaline processes at high northern latitudes.  Shutting down that poleward transfer of heat, probably because freshwater made high-latitude surface water less dense, has been implicated in sudden cooling or “stadials”, and its restart linked to warming or “or interstadials”.  The last such sudden climate event, the Younger Dryas between about 12 and 11 thousand years ago, is widely believed to have resulted from a collapse of the Gulf Stream.  That has raised fears that current anthropogenic warming might achieve the same thing, thereby plunging Western Europe into a counterintuitive frigid period through loss of its maritime warming.

Ocean circulation has lacked a proxy that might help resolve such worrying scenarios, but it seems that one has arrived, because of improvements in mass spectrometry (Piotrowski, A.M. et al. 2005.  Temporal relationships of carbon cycling and ocean circulation at glacial boundaries.  Science, v. 307, p. 1933-1938).  Different bodies of ocean-surface water have subtly different chemical compositions, due to the varied geochemistry of surrounding landmasses.  Weathering of exposed rocks results in some elements entering solution in river water, and that mixes with surface water in the nearby ocean.  Among the most useful elements are those with an isotope to which radioactive decay of unstable isotopes of another element contributes.  A good example is ⁸⁷Sr that is formed when ⁸⁷Rb decays.  Where continents expose  large expanses of very ancient rocks they contribute more ⁸⁷Sr to seawater than do continents veneered with younger rocks.  Strontium isotopes have been used successfully for charting very-long term changes in the overall erosion of continental crust, in relation to climate shifts, but being related to calcium are taken up quickly by carbonate secreting organisms, such as foraminifera, at many different levels in the ocean as it circulates.  So they are not very useful for short-term studies.  A more useful isotopic system involving an daughter of slow radioactive decay is that of neodymium, because it does not get taken up in this way.  It does however enter the manganese minerals that slowly precipitate on the deep ocean floor.  Moreover, its isotopic composition varies greatly in different ocean-water masses.  Piotrowski et al. used neodymium isotopes from deep ocean cores to see if changes in this circulation proxy coincided with known climate proxies.  For interstadial, warming events there is a match, so a Gulf-stream control over millennial-scale climate shifts is indeed supported.  But for the start and end of the full glacial period control by ocean circulation did not happen.  Instead, changes in the neodymium record lag behind the climate proxies, suggesting climatic control of circulation, which then “kicked in” to boost changes that were well underway.

See also: Kerr, R.A. 2005.  Ocean flow amplified, not triggered, climate change.  Science, v. 307, p. 1854.

The competing periodicities of  the three astronomical “drivers” of climate – orbital eccentricity (~100 ka), axial obliquity (~40 ka) and axial precession (~20 ka) – lie behind several models for the climate changes of the last 0.7 Ma.  Taking in the theories that sway towards the influence of variables in the Earth system itself, around 30 models have some currency at present.  Since climate forecasters have to take account of which factors drive climate in the absence of human emissions, as well as piece together their own particular models, it is easy to see how critics of global warming get a wide hearing: compared with creationists, they have it easy!  Is there any way of resolving what is quite bluntly a theoretical mess?  It is a mess simply because the available data are so complex, and in the case of both main sources, ocean-floor sediments and ice cores, not only are their devils in the detail, but there are whopping contradictions, such as the mismatches in timing between the Greenland and Antarctic ice cores.  Add all the other sources, such as stalactites, tree rings etcetera, together with caveats like the difficulty in time calibration using ¹⁴C dating,  and the volume of diverse records become bewildering.  It is tempting that a reversion to a statistical approach, that includes more bells and whistles than hitherto (see Evolutionary rhythms below), can resolve matters.  Peter Huybers and Carl Wunch, of Woods Hole Oceanographic Institution and MIT, have tried that for pacing of the last 0.7 Ma of climate cycles (Huybers, P. * Wunsche, C. 2005.  Obliquity pacing of the late Pleistocene glacial terminations.  Nature, v. 434, p. 491-494).  Generally accepted “wisdom” holds that the last 7 glacial-interglacial cycles are paced by ~100 ka eccentricity forcing, even though it has the weakest effect on solar heating, by a very long way.  But there are smidgens of evidence for some interaction between that and the much stronger influence of changes in the Earth’s axial tilt or obliquity.  Huybers and Wunsch go for the Popperian rigor of first defining a null hypothesis, that obliquity has no effect, and then designing a test.   It isn’t easy to decide how the contrary hypothesis that it does can be evaluated though.  The clearest features in all climate records are the ends of glacial epochs or termination: they are sudden, sharp and generally look the same.  Most other features have some kind of pattern, but little consistent comparability.  Using the most advanced statistical techniques, which employ many iterations to test for stability in statistical models, they can show that the null hypothesis fails.  The positive result is that the time between terminations that are repeatedly modelled  falls into two envelopes, around 120 and 80 ka, which simple arithmetic shows are divisible by 40 ka.  But how can axial obliquity only have an effect every two of three of its cycles, while a single cycle does not appear in the time-series; is it nature skipping beats somehow.  One means that the authors suggest is that the underlying pace of eccentricity can effect the temperature at the base of ice sheets, depending on their thickness.  If they are thin, then the heating is insufficient to trigger ice-sheet collapse because the base is very cold, whereas if ice is thick the effects of thermal conductivity and heat flow makes the ice base warmer and more subject to perturbation beyond its failure limit.  It was at this point that I gave up, but wish the authors good luck in promoting their possibly unifying hypothesis for what finishes off glacial epochs…..

While I write this issue of EPN it is supposed to be early spring outside, and that is clearly what the ducks reckon as well – they are beginning to, er um, frolic.  But there has been two weeks of snow and frost.  Britain and the rest of Europe owe the frigid snap to cold air spilling westwards from northern Asia; the influence of the Siberian winter high-pressure area.  Although somewhat lost in the recent kerfuffles about whether or not global warming is a fact or a misreading of data, the inevitable build up of mid-continental cold dense air in winter might have interesting consequences, should climate warm.  Normally, areas far from the oceans remain dry as well as getting very cold through radiative heat loss in winter.  When spring comes, such snow as there is soon disappears and the extremes of cold are replaced by surprisingly high summer temperatures, as anyone who has visited Siberia or Northern Canada will know.  Should moist air find its way into such areas during winter, vastly more snow would fall.  Its melting would take longer, and more solar radiation would be reflected back to space in spring.  Such an albedo feedback could induce generalised cooling.  Now evidence has emerged that the earliest known growth of land ice in North America was linked to warming of the ocean from which winds blew over it (Haug G.H. et al. 2005.  North Pacific seasonality and the glaciation of North America 2.7 million years ago.  Nature, v. 433, p. 821-825).  In fact it is axiomatic that growth of continental ice sheets requires a supply of moisture and snow that exceeds the rate of summer melting and ablation, as well as cold winters.

Most theorising about the onset of Northern Hemisphere glaciation has centred on changes in North Atlantic circulation due to closures of the straits where the Isthmus of  Panama now links North and South America, and the start of southward deep-water circulation from the latitude of Iceland.  In fact both are known to have preceded the last Ice Age by a good 2 Ma.  The actual start around 2.7 Ma coincided with an increase in obliquity of the Earth’s orbit that would have led to periods with cold northern summers.  Without abundant mid-continent snowfall, that in itself would not have set ice sheets forming in earnest.  The multinational team of oceanographers studied sea-floor sediment cores from the sub-Arctic Pacific.  To their surprise, sea-surface temperatures provided by evidence from planktonic organisms show evidence at 2.7 Ma for on the one hand cooling of the sea surface (from foraminifer oxygen isotopes) yet considerable warming on the other (from organic chemicals secreted by coccolithophores).  Resolving this paradox requires a careful assessment of the ecological behaviour of the two groups of organisms.  The authors’ explanation involves the onset of density stratification in the North Pacific, so that the surface warmed quickly in summer, retaining warmth during autumn, and warmed slowly in spring from its minimum temperature.  Both result from the high thermal inertia of water.  The productivity of silica-secreting diatoms plummeted to a fifth of its earlier levels at 2.7 Ma as well, explained by ocean stratification reducing the supply of nutrients from deep water upwellings.  Intuitively, a warm sea upwind of the North American continental interior should have generated high snowfall in late autumn and winter.  Haug and colleagues modelled the contrasting effects of an ocean with water overturn and mixing with one that tends to become stratified, to simulate snowfall over the North American Arctic.  From a situation in the Pliocene with snowfall over Greenland and the Arctic islands, the scenario shifts to heavy snow over the whole Arctic in the earliest Pleistocene.  It seems that the trigger for the Great Ice Age was a hemisphere away from the “usual culprit”, the North Atlantic, although its vagaries, once glacial cycles were underway, probably controlled the details thereafter.