In the recent articles we mostly reviewed climate models’ successes or otherwise with simulating the last glacial inception.
Part Seven looked at some early GCM work – late 80′s to mid 90′s. Part Eight reviewed four different studies a decade or so ago. Part Nine was on one study which simulated the last 120 kyrs, and Part Ten reviewed one of the most recent studies of glacial inception 115 kyrs ago with a very latest climate model, CCSM4.
We will return to glacial inception, but in this article we will look at the end of the last ice age, partly because on another blog someone highlighted a particular paper which covered it and I spent some time trying to understand the paper.
The last 20 kyrs now have some excellent records from both polar regions. The EPICA project, initiated almost 20 years ago, has produced ice core data for Antarctica to match up with the Greenland NGRIP ice core data going back almost 800 kyrs. And from other research more proxy temperature data has become available from around the globe.
Shakun et al 2012
This paper is from Shakun et al 2012 (thanks to commenter BBD for highlighting it). As an aside, Bette Otto-Bliesner is one of the co-authors, also for Jochum et al (2012) that we reviewed in Part Ten. She is one of the lead authors of the IPCC AR5 for the section on Paleoclimate.
The Last Glacial Maximum (LGM) was around 22k-18 kyrs ago. Sea level was 120m lower than today as thick ice sheets covered parts of North America and Europe.
Why did it end? How did it end?
The paper really addresses the second question.
The top graph below shows Antarctic temperatures in red, CO2 in yellow dots and global temperatures in blue:
From Shakun et al 2012
Figure 1
The second graph shows us the histogram of leads and lags vs CO2 changes for both Antarctica and global temperature.
We can see clearly that the Antarctic temperatures started a sustained increase about 18 kyrs ago and led the global temperatures. We can see that CO2 is slightly preceded by, or in sync with, Antarctic temperatures. This indicates that the CO2 increases here are providing a positive feedback on an initial Antarctic temperature rise (knowing from basic physics that more CO2 increases radiative forcing in the troposphere – see note 1).
But what caused this initial rise in Antarctic temperatures? One possibility put forward is an earlier rise in temperature in the higher northern latitudes that can be seen in the second graph below:
From Shakun et al 2012
Figure 2
..An important exception is the onset of deglaciation, which features about 0.3ºC of global warming before the initial increase in CO2 ,17.5 kyr ago. This finding suggests that CO2 was not the cause of initial warming.
..Substantial temperature change at all latitudes (Fig. 5b), as well as a net global warming of about 0.3ºC (Fig. 2a), precedes the initial increase in CO2 concentration at 17.5 kyr ago, suggesting that CO2 did not initiate deglacial warming. This early global warming occurs in two phases: a gradual increase between 21.5 and 19 kyr ago followed by a somewhat steeper increase between 19 and 17.5 kyr ago (Fig. 2a). The first increase is associated with mean warming of the northern mid to high latitudes, most prominently in Greenland, as there is little change occurring elsewhere at this time (Fig. 5 and Supplementary Fig. 20). The second increase occurs during a pronounced interhemispheric seesaw event (Fig. 5), presumably related to a reduction in AMOC strength, as seen in the Pa/Th record and our modelling (Fig. 4f, g).
..In any event, we suggest that these spatiotemporal patterns of temperature change are consistent with warming at northern mid to high latitudes, leading to a reduction in the AMOC at ~19 kyr ago, being the trigger for the global deglacial warming that followed, although more records will be required to confirm the extent and magnitude of early warming at such latitudes.
The interhemispheric seesaw referred to is important to understand and refers to the relationship between two large scale ocean currents – between the tropics and the high northern latitudes and between the tropics and Antartica. (A good paper to start is Asynchrony of Antarctic and Greenland climate change during the last glacial period, Blunier et al 1998). Perhaps a subject for a later article.
Then a “plausible scenario” is presented for the initial NH warming:
A possible forcing model to explain this sequence of events starts with rising boreal summer insolation driving northern warming. This leads to the observed retreat of Northern Hemisphere ice sheets and the increase in sea level commencing, 19 kyr ago (Fig. 3a, b), with the attendant freshwater forcing causing a reduction in the AMOC that warms the Southern Hemisphere through the bipolar seesaw.
This is a poor section of the paper. I find it strange that someone could write this and not at least point out the obvious flaws in it. Before explaining, two points are worth noting:
- That this is described a “possible forcing model” and it’s really not the paper’s subject or apparently supported by any evidence in the paper
- Their model runs, fig 4c, don’t support this hypothesis – they show NH temperatures trending down over this critical period. Compare 4b and 4c (b is proxy data, c is the model). However, they don’t break out the higher latitudes so perhaps their model did show this result.
From Shakun et al 2012
Figure 3
The obvious criticism of this hypothesis is that insolation (summer, 65ºN) has been a lot higher during earlier periods:
Figure 4
We saw this in Ghosts of Climates Past – Pop Quiz: End of An Ice Age.
And also earlier periods of significant temperature rises in the high northern latitudes have been recorded during the last glacial period. Why were none of these able to initiate this same sequence of events and initiate an Antarctic temperature rise?
At the time of the LGM, the ice sheets were at their furthest extent, with the consequent positive feedback of the higher albedo. If a small increase in summer insolation in high northern latitudes could initiate a deglaciation, surely the much higher summer insolation at 100 kyrs BP or 82 kyrs BP would have initiated a deglaciation given the additional benefit of the lower albedo at the time.
As I was completing this section of the article I went back to the Nature website to see if there was any supplemental information (Nature papers are short and online material that doesn’t appear in the pdf paper can be very useful).
There was a link to a News & Views article on this paper by Eric Wolff. Eric Wolff is one of the key EPICA contributors, a lead author and co-author of many EPICA papers, so I was quite encouraged to read his perspective on the paper.
Many people seem convinced the Milankovitch theory is without question and not accepting it is absurd, see for example the blog discussion I referred to earlier, so it’s worth quoting extensively from Wolff’s short article:
Between about 19,000 and 10,000 years ago, Earth emerged from the last glacial period. The whole globe warmed, ice sheets retreated from Northern Hemisphere continents and atmospheric composition changed significantly. Many theories try to explain what triggered and sustained this transformation (known as the glacial termination), but crucial evidence to validate them is lacking.
On page 49 of this issue, Shakun et al use a global reconstruction of temperature to show that the transition from the glacial period to the current interglacial consisted of an antiphased temperature response of Earth’s two hemispheres, superimposed on a globally coherent warming. Ocean-circulation changes, controlling the contrasting response in each hemisphere, seem to have been crucial to the glacial termination.
Once again, a key climate scientist notes that we don’t know why the last ice age ended. As we saw in Part Six – “Hypotheses Abound” – the title explains the content..
..Some studies have proposed that changes in ocean heat transport are an essential part of glacial termination. Shakun et al. combine their data with simulations based on an ocean–atmosphere general circulation model to present a plausible sequence of events from about 19,000 years ago onwards. They propose that a reduction in the AMOC (induced in the model by introducing fresh water into the North Atlantic) led to Southern Hemisphere warming, and a net cooling in the Northern Hemisphere. Carbon dioxide concentration began to rise soon afterwards, probably owing to degassing from the deep Southern Ocean; although quite well documented, the exact combination of mechanisms for this rise remains a subject of debate. Both hemispheres then warmed together, largely in response to the rise in carbon dioxide, but with further oscillations in the hemispheric contrast as the strength of the AMOC varied. The model reproduces well both the magnitude and the pattern of global and hemispheric change, with carbon dioxide and changing AMOC as crucial components.
The success of the model used by Shakun and colleagues in reproducing the data is encouraging. But one caveat is that the magnitude of fresh water injected into the Atlantic Ocean in the model was tuned to produce the inferred strength of the AMOC and the magnitude of interhemispheric climate response; the result does not imply that the ocean circulation in the model has the correct sensitivity to the volume of freshwater input.
Shakun and colleagues’ work does provide a firm data-driven basis for a plausible chain of events for most of the last termination. But what drove the reduction in the AMOC 19,000 years ago? The authors point out that there was a significant rise in temperature between 21,500 and 19,000 years ago in the northernmost latitude band (60–90° N). They propose that this may have resulted from a rise in summer insolation (incoming solar energy) at high northern latitudes, driven by well-known cycles in Earth’s orbit around the Sun. They argue that this rise could have caused an initial ice-sheet melt that drove the subsequent reduction in the AMOC.
However, this proposal needs to be treated with caution. First, there are few temperature records in this latitude band: the warming is seen clearly only in Greenland ice cores. Second, there is at least one comparable rise in temperature in the Greenland records, between about 62,000 and 60,000 years ago, which did not result in a termination. Finally, although it is true that northern summer insolation increased from 21,500 to 19,000 years ago, its absolute magnitude remained lower than at any time between 65,000 and 30,000 years ago. It is not clear why an increase in insolation from a low value initiated termination whereas a continuous period of higher insolation did not.
In short, another ingredient is needed to explain the link between insolation and termination, and the triggers for the series of events described so well in Shakun and colleagues’ paper. The see-saw of temperature between north and south throughout the glacial period, most clearly observed in rapid Greenland warmings (Dansgaard–Oeschger events), is often taken as a sign that numerous changes in AMOC strength occurred. However, the AMOC weakening that started 19,000 years ago lasted for much longer than previous ones, allowing a much more substantial rise in southern temperature and in carbon dioxide concentration. Why was it so hard, at that time, to reinvigorate the AMOC and end this weakening?
And what is the missing ingredient that turned the rise in northern insolation around 20,000 years ago into the starting gun for deglaciation, when higher insolation at earlier times failed to do so? It has been proposed that terminations occur only when northern ice-sheet extent is particularly large. If this is indeed the extra ingredient, then the next step in unwinding the causal chain must be to understand what aspect of a large ice sheet controls the onset and persistence of changes in the AMOC that seem to have been key to the last deglaciation.
[Emphasis added].
Thanks, Eric Wolff. My summary on Shakun et al, overall – on its main subject – it’s a very good paper with solid new data, good explanations and graphs.
However, this field is still in flux..
Parrenin et al 2013
Understanding the role of atmospheric CO2 during past climate changes requires clear knowledge of how it varies in time relative to temperature. Antarctic ice cores preserve highly resolved records of atmospheric CO2 and Antarctic temperature for the past 800,000 years.
Here we propose a revised relative age scale for the concentration of atmospheric CO2 and Antarctic temperature for the last deglacial warming, using data from five Antarctic ice cores. We infer the phasing between CO2 concentration and Antarctic temperature at four times when their trends change abruptly.
We find no significant asynchrony between them, indicating that Antarctic temperature did not begin to rise hundreds of years before the concentration of atmospheric CO2, as has been suggested by earlier studies.
[Emphasis added].
Ouch. In a later article we will delve into the complex world of dating ice cores and the air trapped in the ice cores.
WAIS Divide Project Members (2013)
The cause of warming in the Southern Hemisphere during the most recent deglaciation remains a matter of debate.
Hypotheses for a Northern Hemisphere trigger, through oceanic redistributions of heat, are based in part on the abrupt onset of warming seen in East Antarctic ice cores and dated to 18,000 years ago, which is several thousand years after high-latitude Northern Hemisphere summer insolation intensity began increasing from its minimum, approximately 24,000 years ago.
An alternative explanation is that local solar insolation changes cause the Southern Hemisphere to warm independently. Here we present results from a new, annually resolved ice-core record from West Antarctica that reconciles these two views. The records show that 18,000 years ago snow accumulation in West Antarctica began increasing, coincident with increasing carbon dioxide concentrations, warming in East Antarctica and cooling in the Northern Hemisphere associated with an abrupt decrease in Atlantic meridional overturning circulation. However, significant warming in West Antarctica began at least 2,000 years earlier.
Circum-Antarctic sea-ice decline, driven by increasing local insolation, is the likely cause of this warming. The marine-influenced West Antarctic records suggest a more active role for the Southern Ocean in the onset of deglaciation than is inferred from ice cores in the East Antarctic interior, which are largely isolated from sea-ice changes.
[Emphasis added].
We see that “rising solar insolation” in any part of the world from any value can be presented as a hypothesis for ice age termination. Here “local solar insolation” means the solar insolation in the Antarctic region, compare with Shakun et al, where rising insolation (from a very low value) in the high northern latitudes was presented as a hypothesis for northern warming which then initiated a southern warming.
That said, this is a very interesting paper with new data from Antarctica, the West Antarctic Ice Sheet (WAIS) Divide ice core (WDC), where drilling was completed in 2011:
Because the climate of West Antarctica is distinct from that of interior East Antarctica, the exclusion of West Antarctic records may result in an incomplete picture of past Antarctic and Southern Ocean climate change. Interior West Antarctica is lower in elevation and more subject to the influence of marine air masses than interior East Antarctica, which is surrounded by a steep topographic slope. Marine-influenced locations are important because they more directly reflect atmospheric conditions resulting from changes in ocean circulation and sea ice. However, ice-core records from coastal sites are often difficult to interpret because of complicated ice-flow and elevation histories.
The West Antarctic Ice Sheet (WAIS) Divide ice core (WDC), in central West Antarctica, is unique in coming from a location that has experienced minimal elevation change, is strongly influenced by marine conditions and has a relatively high snow-accumulation rate, making it possible to obtain an accurately dated record with high temporal resolution.
WDC paints a slightly different picture from other Antarctic ice cores:
..and significant warming at WDC began by 20 kyr ago, at least 2,000 yr before significant warming at EDML and EDC.
..Both the WDC and the lower-resolution Byrd ice-core records show that warming in West Antarctica began before the decrease in AMOC that has been invoked to explain Southern Hemisphere warming [the references include Shakun et al 2012]. The most significant early warming at WDC occurred between 20 and 18.8 kyr ago, although a period of significant warming also occurred between 22 and 21.5 kyr ago. The magnitude of the warming at WDC before 18 kyr ago is much greater than at EDML or EDC..
From WAIS Divide Project (2013)
Figure 5
We will look at this paper in more detail in a later article.
Conclusion
The termination of the last ice age is a fascinating topic that tests our ability to understand climate change.
One criticism made of climate science on many blogs is that climate scientists are obsessed with running GCMs, instead of doing “real science”, “running real experiments” and “gathering real data”. I can’t say where the balance really is, but in my own journey through climate science I find that there is a welcome and healthy obsession with gathering data, finding new sources of data, analyzing data, comparing datasets and running real experiments. The Greenland and Antarctic ice core projects, like NGRIP, EPICA and WAIS Divide Project are great examples.
On other climate blogs, writers and commenters seem very happy that climate scientists have written a paper that “supports the orbital hypothesis” without any critical examination of what the paper actually supports with evidence.
Returning to the question at hand, explaining the termination of the last ice age – the problem at the moment is less that there is no theory, and more that the wealth of data has not yet settled onto a clear chain of cause and effect. This is obviously essential to come up with a decent theory.
And any theory that explains the termination of the last ice age will need to explain why it didn’t happen earlier. Invoking “rising insolation” seems like lazy journalism to me. Luckily Eric Wolff, at least, agrees with me.
Articles in the Series
Part One – An introduction
Part Two – Lorenz – one point of view from the exceptional E.N. Lorenz
Part Three – Hays, Imbrie & Shackleton – how everyone got onto the Milankovitch theory
Part Four – Understanding Orbits, Seasons and Stuff – how the wobbles and movements of the earth’s orbit affect incoming solar radiation
Part Five – Obliquity & Precession Changes – and in a bit more detail
Part Six – “Hypotheses Abound” – lots of different theories that confusingly go by the same name
Part Seven – GCM I – early work with climate models to try and get “perennial snow cover” at high latitudes to start an ice age around 116,000 years ago
Part Seven and a Half – Mindmap – my mind map at that time, with many of the papers I have been reviewing and categorizing plus key extracts from those papers
Part Eight – GCM II – more recent work from the “noughties” – GCM results plus EMIC (earth models of intermediate complexity) again trying to produce perennial snow cover
Part Nine – GCM III – very recent work from 2012, a full GCM, with reduced spatial resolution and speeding up external forcings by a factors of 10, modeling the last 120 kyrs
Part Ten – GCM IV – very recent work from 2012, a high resolution GCM called CCSM4, producing glacial inception at 115 kyrs
Pop Quiz: End of An Ice Age – a chance for people to test their ideas about whether solar insolation is the factor that ended the last ice age
Twelve – GCM V – Ice Age Termination – very recent work from He et al 2013, using a high resolution GCM (CCSM3) to analyze the end of the last ice age and the complex link between Antarctic and Greenland
Thirteen – Terminator II – looking at the date of Termination II, the end of the penultimate ice age – and implications for the cause of Termination II
Fourteen – Concepts & HD Data – getting a conceptual feel for the impacts of obliquity and precession, and some ice age datasets in high resolution
Fifteen – Roe vs Huybers – reviewing In Defence of Milankovitch, by Gerard Roe
Sixteen – Roe vs Huybers II – remapping a deep ocean core dataset and updating the previous article
Seventeen – Proxies under Water I – explaining the isotopic proxies and what they actually measure
Eighteen – “Probably Nonlinearity” of Unknown Origin – what is believed and what is put forward as evidence for the theory that ice age terminations were caused by orbital changes
Nineteen – Ice Sheet Models I – looking at the state of ice sheet models
References
Global warming preceded by increasing carbon dioxide concentrations during the last deglaciation, Shakun, Clark, He, Marcott, Mix, Liu, Otto-Bliesner, Schmittner & Bard, Nature (2012) – free paper
Climate change: A tale of two hemispheres, Eric W. Wolff, Nature (2012)
Synchronous Change of Atmospheric CO2 and Antarctic Temperature During the Last Deglacial Warming, Parrenin, Masson-Delmotte, Köhler, Raynaud, Paillard, Schwander, Barbante, Landais, Wegner & Jouzel, Science (2013) – free paper, in submission form (rather than published form)
For interest Valérie Masson-Delmotte is one of the two co-ordinating lead authors for AR5 for the Paleoclimate section, Frédéric Parrenin is a contributing author.
Onset of deglacial warming in West Antarctica driven by local orbital forcing, WAIS Divide Project Members, Nature (2013) – free paper
Notes
Note 1 – see for example, CO2 – An Insignificant Trace Gas? – an 8-part series on CO2, Atmospheric Radiation and the “Greenhouse” Effect – a 12-part series on how radiation interacts with the atmosphere, Visualizing Atmospheric Radiation – a 13-part series to help us get a better grasp of how this works including “does water vapor overwhelm CO2″, “is CO2 saturated” and many other questions.
The Science of Doom 
SoD
Hello again, and thanks for the mention in the main article. Just a few quick thoughts for now:
1/ From Eric Wolff’s News & Views response to Shakun et al. (2012):
This may be suggestive that the extent of the NH ice sheet is the key missing variable: if the large ice sheets at the LGM are inherently unstable they may respond to an increased insolation from a low level by disintegrating. Smaller ice sheets earlier during the last glacial may have been more resistant to even sustained insolation increases, especially as they did not extend to the lower latitudes seen by the time of the LGM. If the process of large ice sheet disintegration is self-propelling once initiated and produces large and sustained NH freshwater fluxes, strong inhibition of NADW and so AMOC will follow.
2/ From Parrenin et al. (2013):
Is it not the case that Shakun 12 proposes Southern Ocean warming as the trigger for CO2 release and consequent Antarctic/global warming? I’m not sure if your “ouch” is fully merited here?
3/ It would be useful if you can show insolation curves for a span of high SH latitudes alongside the WAIS Divide core comparisons (your fig 5).
BBD,
1. It may be so. Hypotheses abound. This is one of them.
2. The ouch point is that up until Parrenin et al 2013 it has appeared clear that CO2 lagged Antarctic temperatures. This is the stated position in many papers and also in the IPCC reports. It isn’t “ouch, they made a mistake”, it’s “ouch, the sequence of events is still uncertain”.
The Shakun et al 2013 paper has some possible ambiguity between Southern Hemisphere records and Antarctic records in their wording, but in any event clearly states in some parts that Antarctic temperatures lead CO2:
And then the possible ambiguity:
And Wolff, in his commentary on the paper says:
If I have misunderstood the exact position of Shakun et al, it is in any case clear that Parrenin et al are overturning the consensus position.
3. Good idea, I will generate a few.
Thanks for the response. To be clear, I do not dispute that there are considerable mysteries here, but I await an alternative physical mechanism replacing orbital pacing of deglaciations if doubt is to be cast over this relationship. Until something more plausible that “stochastic” appears, the orbital trigger remains firmly on the table. I share with Tom Curtis the view that over-much is made of the gaps in knowledge at the expense of the very clear relationship between deglaciation and (in particular) the obliquity cycle.
Very briefly, wrt Parrenin, one paper does not “overturn the consensus”. Parrenin might be wrong 😉
Also, I was not very clear about *which* orbital parameters would be helpful – specifically, is it possible to examine SH obliquity? Precession too, if it is simple to generate the curves. If not, please do not spend time on this unless you wish to.
The Figure 2b of the Shakun et al paper shows that SH temperature leads CO2 by 620±660 years. The modus is at 0, but most of the distribution on the leading side, with a long tail to an even larger lead.
Pekka,
It’s the mechanism of dating that is under question – that is, the values in the graph are being questioned.
The question is the age of the air, t2, trapped in the ice at depth d vs the age of the ice, t1, at depth d. The ice is porous for Δt years after being deposited as snow until eventually the ice is compacted enough so that air can no longer be circulated out of that depth d.
We have a very confident calculation of time t1 when we can count annual cores. We have a pretty confident calculation of time t1 when the ice gets a bit older as it’s constrained by some basic physics relating to flow & pressure.
But what is Δt? To calculate that you need to know or assume quite a bit more.
SoD,
I have understood that dating is a problem. It’s also a problem that can be studied and papers can be written on that. There may be other as or more severe problems that are hiding in the total lack of data on them. They may manifest themselves in differences between climate models, but they may avoid even that form of detection.
What I find really intriguing is the shape of the 800 kyear delta 18O curve. Even if the timing is significantly wrong the shape is not severely distorted. History seems to tell that the highest and lowest ice volumes cannot be maintained long in relative terms and that a sharp phase shift ends those periods. What’s the nature of that phase shift? Why the volume of ice seems to go down to very low values after a maximum?
And, why is this kind of behavior quite “recent”?
BBD,
I already have the calculated values of TOA insolation for every day of the year for the last 150 kyrs at every 1′ of latitude.
I have already written a couple of routines that produce:
a) the insolation curve at latitude, x and day of year, d for the last 150 kyrs, e.g.,
b) a set of comparison contour plots (as seen in the Pop Quiz article) for any pairs of dates, e.g.:
It’s the work of moments to produce these graphs for different values.
What would you like to see? Do you have latitudes and dates that you would like to see the curves for?
I also have the data for the last 500 kyrs in a different file for every 5′ of latitude and on the first of every month.
If you specifically want just obliquity and precession then many people have already calculated these, they are values that are not dependent on latitude or season, just on year.
Well, let’s think about it (this is not the same as an actual request for curves!) What would be useful?
My sense is that we need to look at the progressive change in insolation from about 40N to 65N from 22ka to 11ka. First day of month (MAM JJA) at 5 degree intervals every 500y. Perhaps that might yield a more informative picture of how the maximum extent LGM NH ice sheet experienced orbital forcing.
What do you think?
Then the same for the SH but for austral spring and summer months.
* * *
Are you disputing the dating of the d180O reconstruction btw? I got the sense that you dispute the reliability of astronomically tuned chronologies in our previous discussion. Have you ever discussed your doubts about all of this with an actual paleoclimatologist? If so, how did they react?
Sorry d180O >>> d18O
BBD,
By the way, it’s fairly easy to use Greek letters in HTML posts. You just enclose the letter name between an ampersand and a semicolon. If you capitalize the letter name, you get the capital letter. For example delta gives you δ and Delta gives you Δ.
For anyone interested, check out Comments and Moderation for how to get nice formatting & Greek letters.
BBD,
We can’t use orbitally-tuned proxy dates to show that temperatures & ice sheet volumes are in sync with the orbital hypothesis. Or to show how well they match. We are assuming the hypothesis we want to test.
We just had a short discussion on this in Part Nine, where I show the equation used in one proxy reconstruction.
That doesn’t mean it isn’t the best way to date proxies when we don’t have any independent dating method. Because there is clearly a precession and obliquity signal in the proxy data. What is of interest is what proportion of power is within the precession and obliquity bands.
Re your earlier comment of January 9, 2014 at 1:18 pm, there is a 100 kyr signal – approximately – in the proxy data. There is no 100 kyr signal in the insolation curves. Here’s one I produced earlier:
[…] shown definitively that they are indeed linked. Science of Doom has an interesting recent post that discusses this […]
If temperature doesn’t lead CO2, that raises a new problem, you now need a non thermal mechanism for increased CO2. Therefore, I’m highly skeptical of the validity of the new dating of Parrenin, et. al.
And as I said in another thread, the real problem isn’t why interglacial periods end, it’s why they begin at all.
When ice sheets are several kilometers thick and 10-20 degC colder at the top than at ground level from the lapse rate, you might also wonder why they melt. (As I commented on another post, continents sink into the mantle under the weight of all that ice (and the ocean floor rises. Perhaps terminations don’t occur until sinking reaches a critical point.)
DeWitt,
I hadn’t looked into it until recently, but I have come to understand that there is a problem with the “thermal mechanism” for increased CO2. As in, explaining the increase in CO2 quantitatively.
Parrenin et al may be wrong, that is very possible. But the authors include some very experienced researchers so I’d bet their reservations were also very strong. Jean Jouzel is a geochemist with 40+ years of research, here are his papers. Take a look at Frederic Parrenin’s papers.
Not suggesting the argument from authority, otherwise Science of Doom would not exist. Just suggesting that it’s worth taking seriously.
SoD,
Have you had a look at the papers of Lourantou et al (2010) and Fischer et al (2010) (references 2 and 23 of Parrenin et al). They discuss mechanisms that have influenced changes in atmospheric CO2 concentration. Google Scholar finds free copies of these papers as final (Fischer) or as manuscript (Lourantou).
SoD,
I agree that the temperature change alone is insufficient to produce the CO2 concentration change on a purely solubility basis, i.e. Henry’s Law. If that were the case, then interglacial periods would likely be stable rather than fairly quickly beginning to slide back to glacial conditions. There needs to be some sort of reservoir that accumulates CO2 that eventually becomes unstable. Then when it releases the CO2, it overshoots the equilibrium value so that CO2 peaks and then declines. But a temperature change still seems like the most logical trigger for the beginning of the CO2 evolution.
I started to read the Fischer et al review article (the author list is impressive: Fischer, Schmitt, Lüthi, Stocker, Tschumi, Parekh, Joos, Köhler, Völker, Gersonde, Barbante, Le Floch, Raynaud, Wolff).
Some excerpts from the intro:
and conclusions
Obviously there’s still much to do.
Pekka,
I should read the review article myself, but since you’re reading it, I’ll ask anyway. Does the review article imply that in the previous interglacials that once past the peak CO2 concentration, the draw down commences automatically?
DeWitt,
As far as have understood the paper it doesn’t discuss the onset of deglacification at all. Rather the paper looks at the role of Southern Ocean in making possible the difference of about 100 ppm between interglacial and glacial maximum. The paper concludes that changes in circulation are most important in that but that other processes have also an important role.
A discussion of the changes in circulation patterns is an important part of the paper, but the onset or timing of these changes is not discussed at all.
Pekka,
That was my reading too. The correlations between δD and CO2 didn’t include any possible phase shift either.
Sea ice in the SO is currently advancing, if that means anything other than the polar seesaw.
BBD,
How about this approach to displaying the data – Hovmöller plot?
The two Hovmoller plots are insolation anomalies (with respect to today) averaged (area-weighted) between 65’N and 80’N (top) and 65’S and 80’S (bottom):
Click to expand
This set is the exact same data but less pretty, with contour lines displayed:
Click to expand
Now I have written the routing to display I can churn out all kinds of variants very easily.
And here is the last 30 kyrs only:
Click to expand
DeWitt Payne
I’ll try a *semi*colon next time, as you originally suggested…
😉
These are fascinating, SoD. Thanks for posting them up. I did wonder whether Hovmoller plots might be a good way of visualising the insolation change data. But they only convince me more that we should be looking at Δ insolation at lower latitudes (see previous comment).
If – and this is of course purely an ‘if’ – the key to this is NH ice sheet extent – then we need to look down, not up. Quick summary of thinking so far:
– Deep insolation minima = maximum extension of NH ice sheets (~40N eg LGM for MIS3)
– At ~40N, ice sheets are fundamentally more vulnerable to Δ insolation than at 65N
– Dynamical collapse of ice sheets once initiated is self-propelling and engages strong positive ice albedo feedback
– The above provides a hypothetical basis for NH deglaciation under only moderate Δ insolation
* * *
DeWitt Payne
Thanks for the formatting tip. I’m trying it out in the above comment.
BBD,
Here’s more Hovmoller plots in lower latitudes.
50’N – 65’N and 50’S – 65’S over the last 150 kyrs
Click to expand
40’N – 50’N and 40’S – 50’S over 150 kyrs
Click to expand
50’N – 65’N and 50’S – 65’S over 30 kyrs
Click to expand
40’N – 50’N and 40’S – 50’S over 150 kyrs
Click to expand
BBD,
For formatting – &[greek letter written in text];
last character needs to be the semicolon, you had a colon
(I fixed up the 3 I noticed).
.. a minute later I see you already figured it out..
SoD
Well, that didn’t really dispel any mysteries. I have to admit that I’m more confused than enlightened by all this, but perhaps the same applies to you? Thanks for the discussion, the plots, and the insight that we are missing something paradoxically both subtle and vast.
Also thanks for correcting the formatting bork in my previous.
BBD,
Definitely.
re: the quiz plots (got this wrong also), Does the correct answer G have the steadiest values over the southern oceans? It would make some sense that the heat goes into ocean and after a while starts to chew on the ice sheet edges much like seen currently. Thus the ‘start-Eemian’ plot would be nice to see. There are also many Dansgaard-Oescher events during last glacial that could be of interest, could these be astronomically driven (likely not, the periodic collapse of some (was it the WAIS extension over the southern ocean??) ice shelf sounds like a convincing reason) ?
jyyh,
Did the Hovmoller plots above help to answer your question?
Thanks for the plots, never seen these before..rather just reading what’s available and forming various hypotheses. Tried to overlay the NGRIP core data on these plots… if they answered the question, well kind of, Dansgaard-Oescher events cannot be orbitally driven since orbital variations are too slow. Still I’m of the opinion it’s a matter of taste whether it’s spring insolation in the south or spring in the north (65-80) what’s driving the glacial-interglacial variation.There are still discrepancies, f.e. why 97kybp-90kybp didn’t start an interglacial… This thing is not settled, orbitally driven yes, specific area which initiates these not so much. So I’m in the Connolley camp here :-). I mostly trust the radioactive dating of the cores, having done one experiment with radioactive nucleotides that clearly showed half-life exists, but in these core issues I’m not too sure. Could the isotopes be of different permeability f.e? (RE:BBD d18O) Also, who’s to say there has been snow accumulation every year over the areas the cores have been bored?
You seem a bit unhappy with “Many people seem convinced the Milankovitch theory is without question and not accepting it is absurd”. But “Milankovitch theory” isn’t clearly defined. If its just that orbital factors pace the ice ages, then yes I’m convinced. If its the central importance of summer insolation at 65 N, then no I’m not.
William,
I like a concrete hypothesis that can be falsified. Right now the hypothesis can’t be falsified because I don’t know what it is.
Which is another way of saying there is no theory (in a scientific sense). It is more like a banner or a party.
As you comment, the “Milankovitch theory” is a number of different theories. I gave some examples of them in Part Six – Hypotheses Abound.
How do orbital factors pace the ice ages and what precisely are you claiming by “pacing”?
What is the evidence for this hypothesis?
And how could this hypothesis be falsified?
That’s a somewhat puzzling question. You must be aware of ays imbrie shackleton (http://www.mantleplumes.org/WebDocuments/Hays1976.pdf) and its successors.
William,
Puzzling it may be, but I find asking people specifically what they believe rather than accepting a vague statement as “probably could be right because lots of people also accept the vague statement” has a lot of merit.
Yes, I am familiar with Hays, Imbrie and Shackleton, we had a look at them in Part Three of this series.
Here is one main thrust of their paper:
I.e., the paper finds frequencies in the proxy data that match frequencies in the orbital data. That is an interesting start but is not a complete theory.
Here is the second main thrust of their paper:
That is, temperature and ice core data show variation at frequencies that match precession and obliquity but they also show their main variation at the 100,000 year time period that doesn’t show up in the insolation spectrum.
They suggest:
It’s a great paper for 1976. It doesn’t explain why ice ages start and end. It doesn’t explain the 100,000 year time period.
How do orbital factors pace the ice ages and what precisely are you claiming by “pacing”?
Hopefully I am not still puzzling you with requesting an explanation.
You asked “How do orbital factors pace the ice ages” and I thought that a bit odd, because the well-known answer is the Fourier analysis in HIT and succeeding papers. But perhaps all we’re establishing is that we’re looking at the same evidence. Or indeed that others in this conversation are doing the same. So, yes, the main evidence is that the periods match.
> their main variation at the 100,000 year time period that doesn’t show up in the insolation spectrum.
That’s wrong (unless I’m really mis-remembering). 100 kyr is there, its just not as powerful as you’d expect from the effects. But that’s hardly a fatal problem: the idea that the “amplification” isn’t the same at all frequencies is hardly surprising. This is just the well-known https://en.wikipedia.org/wiki/100,000-year_problem
I hope you can forgive me saying that I’m finding your response puzzling again: you know about the 100 kyr problem (at least, I’ll be very surprised if you don’t) and part of that is that 100 kyr is there, but at reduced power, so how can you say that 100 kyr doesn’t show up in the insolation spectrum?
FWIW, the various (incorrect) attempts to get the timing, and the explanations, for the ice age cycles in order is (IMHO, and I’ve written this down somewhere but can’t find it now) the most-got-wrong theory in history.
William,
I’m attempting to state your specific hypothesis for you, because you haven’t clearly stated one. You refer to Hays et al 1976 and mention the ‘well known Fourier analysis’ so I quote from Hays et al 1976:
Is this your hypothesis?
A) We believe the 100 kyr ice age cycle is a non-linear response to the 100 kyr eccentricity cycle.
B) We don’t have a physics explanation for A).
Yes / No / In-between?
And no, the 100 kyr does not show up in the insolation spectrum. Fig. 4 in the iconic Hays 1976 paper demonstrates this. I have produced one as well – you can see it earlier in the comment thread.
You appear to be using the argument for incredulity. ‘I can’t believe you don’t understand something so basic and simple, why am I even bothering to explain it?’
Perhaps you don’t intend to do that. Perhaps I’m not very bright. Perhaps you can indulge my desire for specifics.
If your hypothesis is ‘the insolation spectrum does produce a 100 kyr period‘ then:
A) Produce this evidence (the spectrum)
B) Explain a physics link between the insolation spectrum and the climate response
Showing that the largest climate spectral response in the last 800kyrs is of the same frequency as the tiniest orbital forcing that barely shows up in the insolation spectrum and just saying ‘read Hays et al 1976’ is not the kind of approach we take on this blog.
> That’s wrong (unless I’m really mis-remembering). 100 kyr is there
OK, I’m wrong. Its not in HIS’s pix.
There are nice pix in ftp://ftp.elic.ucl.ac.be/loutre/paper/Dom/Berger_paleoc_05.pdf but still not the one I want.
The eccentricity variations at 100 kyr inevitably modulate insolation, though. Do you disagree? (Your earlier “There is no 100 kyr signal in the insolation curves. Here’s one I produced earlier” does suggest that you think there is no 100 kyr signal at all). OTOH “the tiniest orbital forcing that barely shows up in the insolation spectrum” suggests otherwise.
> Is this your hypothesis?
I don’t really have my own hypothesis.
> A) We believe the 100 kyr ice age cycle is a non-linear response to the 100 kyr eccentricity cycle.
Yes. Muller has a different theory, but I’m not sure anyone else goes for it (http://www.pnas.org/content/94/16/8329.full).
> B) We don’t have a physics explanation for A).
I’m pretty sure people have proposed a variety of physics explanations. None are definitive or without flaws.
William,
I did the calculation in Part Four:
That is, the value is too small to be seen in the insolation curves. So there is a value, but it’s at noise level.
In conceptual terms the impact of perihelion and aphelion is accentuated by larger eccentricity. So when aphelion is in NH summer, a higher eccentricity will make that NH summer colder.
The value is very low and changes very slowly, which is why no one can find a way to link the impact of 10 mW/m2 change per kyr due to eccentricity variation through any kind of believable physics to the inception/termination of ice ages.
However, I have been thinking about this some more and do have a slightly random suggestion to quantify the impact of eccentricity on insolation in a brute force way, which won’t take a whole lot of time.
I have a Matlab program to calculate the insolation at any latitude on any day of the year for the last million years. I’ve already precalculated the last 150 kyrs (in 1kyr increments) for every day and every 1′ which is about 10 million values. There are some memory limits so I’m not sure what can be done in terms of granularity.
1. Perhaps calculate the monthly average for every 1′ latitude for the last 800 kyrs – that’s 1.7M values.
2. For each month, each latitude, produce the fft of the 800 kyrs. Normalize so the sum of each fft =1. Calculate the amount of spectral energy between fa & fb ( where these values should represent the 100 k signal we are looking for).
3. Show the stats on max, mean, and which lat/month has the highest eccentricity signal. Or anything else anyone wants to see.
My questions:
a) what are fa and fb? Or a family of values to test against
b) what is the noise threshold we identify – in advance – that the normalized energy between fa & fb should be above
I expect that this exercise is going to demonstrate that there is no 100 kyr signal above noise at any latitude in any month.
> annual solar insolation
But of course there’s no need to restrict yourself to annual solar insolation, as you go on to discuss. I’d guess that 5 degrees would be good enough. And you don’t need to store all latitudes, you only want the answers.
SoD,
Just for curiosity. Which factors are taken into account in your application?
Based on ignorance of quantitative facts I would have expected that many body interactions within the solar system might make a calculation that extends one million years to the past quite complicated.
Pekka,
It’s Jonathon Levine’s Matlab program, based on a paper:
function radiation = insolationjl(t,month,day,latitude)
%radiation = insolationjl(t,month,day,latitude);
%
%This program computes the insolation at times in argument t,
%on (month,day) at latitude, for any part of the last
%three million years, using the data of Quinn, Tremaine, and Duncan (1991).
%
%REF: Quinn, T.R. et al. “A Three Million Year Integration of the Earth’s
%Orbit” The Astronomical Journal 101 pp. 2287-2305 (June 1991).
%
%Example: rad = insolationjl([0:1:1000],5,30,40) gives the insolation
%in W/m^2 on May 30 at 40ÁN latitude,
The integration of Quinn et al seems to take everything essential into account. The more recent standard reference Laskar et al (2004) comments on that:
As far as I understand they consider their new 2004 integration essential for periods much longer than 1 billion years. For the periods you have considered there’s probably no difference.
The statement in Hays et al.
is not true because if the spectra shows ANY precession signal at all then it is a kind of proxy for eccentricity. The two are related. Eccentricity modulates the strength of the precession signal. If the earth’s orbit was circular then precession would have no effect whatsoever on insolation.
The eccentricity signal is very weak in the Fourier power spectrum of insolation, but it modulated very strongly the precession signal which is strong.
It’s misleading to look only at the power spectrum over long intervals, because Fourier decomposition is insensitive to that particular type of strong influence on summer insolation at high latitudes.
The paper Lin and Wang: EMD analysis of solar insolation Meteorol Atmos Phys 93, 123–128 (2006), DOI 10.1007/s00703-005-0138-7 is related to this. Their EMD method is non-stationary and can therefore isolate also the modulating eccentricity variations, not only stationary variations from eccentricity that are killed in standard Fourier decomposition by the faster precession cycle.
SoD,
You might look at the power spectrum of the square of insolation (or square of the deviation from the mean value). The nonlinearity of that should reveal a rather strong eccentricity signal in the spectrum.
As Clive noted eccentricity has an important influence on insolation, but the linearity of Fourier decomposition makes the signal very weak in normal frequency spectrum as the positive and negative phases of the higher frequency precession cycle cancel each other.
Fourier analysis is insensitive to many nonlinear effects like modulation of a higher frequency signal in this case.
Another paper related to the issue discussed in my (and Clive’s) above comments is Berger, Loutre, and Mélice: Equatorial insolation: from precession harmonics to eccentricity frequencies.
The figure linked to below shows the maximum insolation -Smax (at midsummer’s day) for both the north and south poles – solid and dashed curves. These show an out of phase precession signal at both poles. The top curve is the earth’s eccentricity. Note how the eccentricity determines the amplitude of the signal and how it diminishes with eccentricity. The zero line is for current values with the south pole receiving 20 W/m2 in summer more than the north pole.
If now we look at the difference between the two – that is the asymmetry between maximum insolation poles we then get the red dashed curve. This isolates the 41,000 year cycle in obliquity. If you like this is the N-S gradient caused by obliquity. The solid red curve is the total insolation received each summer at the North Pole in Mwatts/m2 – (scaled down by a factor 10^6).
click here for figure
Pekka,
Thanks, Pekka. I’ll take a look.
My signal analysis is quite rudimentary. I’m not sure whether I’ve just forgotten a lot, or never really appreciated the subject.
I had the view that even though eccentricity is modulated via precession we would either see it in the frequency spectrum or we would find it via the values of insolation when compared with the period of interest (in this case, time of deglaciation).
This might be naive and I’m keen to understand how to analyze these problems in a professional way.
If you have some other recommendations for brushing up my knowledge in this area, it would be appreciated.
I have read this paper, but will reread.
Clive,
This point I understand.
Whether I know how to analyze the data in the correct manner is another question. So if you have some suggestions as to how to analyze the data I would be interested to hear it.
Generally I have looked at the actual values of insolation with respect to the key time points of glacial inception and glacial termination. Glacial inception appears to have some promise. Glacial termination – at least the last two – appear to kill any general theory.
How should we analyze the data?
SoD,
Fourier analysis just identifies if any harmonics present in a signal. I think it is better instead to make a least squares fit to the data with a sum of Milankovitch harmonic terms. The fitted variables then give you the amplitudes and relative phases. There is no physics involved at all. When I did that I found the following results for the LRO4 stack.
last 900,000y
last 5.3My
When you then compare the fit with the calculated Milankovitch cycles the match is then almost perfect. The main signal throughout the entire period is the 41,000y obliquity signal and it aligns perfectly with the astronomic calculations. For the last 800,000y the alignment is perfect for the eccentricity. What is not explained however is the physics behind why this happens. It also doesn’t describe well the saw tooth shape of the last glacial period nor the one 5 cycles earlier. Both had weak values of eccentricity.
SoD,
Here we see just one more example of technical problems we face, when linear analysis does not work.
Linear analysis would be fine in handling eccentricity, if all important mechanisms would have time constants much longer than precession. Then only the totals over full precession cycles would count and the influence of eccentricity would be as small as Fourier analysis tells. That’s however hardly the case, but switching from linear analysis to non-linear can be done in infinitely many ways.
The articles of Lin&Wang and Berger et al describe two possibilities out of this infinity. What they tell clearly – and what we can see by eye immediately looking at temporal development of insolation – is that eccentricity is a major factor in spite of the fact that linear Fourier analysis makes the signal very weak.
With the small number of glacial cycles of approx 100 kyr length no formal mathematical analysis is likely to tell more than naked eye can see from the curves. Any choice of the method is affected by what we know about the phenomena already. Therefore they do not give essentially more objective answers than direct comparison. It’s not possible to give well justified confidence limits for any conclusions on that basis. The research has an exploratory nature, not that of statistical inference.
The main conclusion that I draw from the recent discussion is that there’s no real basis to talk about 100,000,year problem. Having a problem in particular in that is an artifact of limited methodology of analysis, rather than lack of a strong 100 kyr cycle.
Looking at the details of last deglaciation is an obvious case for the first problem to study, similarly could we pick as subject of research the details of glaciation phase, or conditions of switching from one phase to another. Frequency analysis may be an almost useless tool in that. Looking directly at temporal development may well be a much better choice. It’s probably better to compare also precession related signals directly on time axis because Fourier analysis cannot take into account the modulation by eccentricity. The main processes of ice ages seem to be so essentially non-linear that the fully linear Fourier analysis is not a good choice.
Clive,
The 41 kyr obliquity signal is the most important tuning signal in the creation of the Lisiecki and Raymo age scale. Therefore the strength of that signal cannot be considered similarly with signals that are not used in tuning.
The 23 kyr cycle is also used in tuning, but is evidently not as important as they write:
Even so it’s a bit surprising that you don’t find much 23 kyr signal in the data.
Agreement with the eccentricity cycle is relevant as it’s not used in tuning, but the very precise fit on the timing of most deglaciations can be considered a forced feature created by the tuning of the 41 kyr cycle rather than an empirical result.
Really interesting. Thanks Clivebest and Pekka Pirilä.
Pekka,
I’ll repeat.
You can’t tune to a signal that isn’t there. Sure the level of confidence goes down because you’ve used degrees of freedom for the tuning, but that doesn’t mean you can ignore the results. And there are other constraints, such as deposition rate for sediment cores, on the tuning.
This tuning reminds me very much of the use of a lock-in amplifier to retrieve a sinusoidal signal from noise. Under the right conditions one can recover a signal when the signal to noise ratio is -60dB. The signal to noise ratio for the sediment cores is a lot higher than that.
To put it another way, you have a time series that has increasing accuracy uncertainty or bias on the time axis as the age increases rather than the signal axis. That uncertainty produces phase shifts in the apparent signal compared to the presumed excitation signal. There is little or no reason to believe that those phase shifts are real rather than artifacts of the sample dating. Thus tuning the time axis to remove the phase shifts. Similar uncertainties affect the estimates of the timing of the inception of the glacial to interglacial transition.
DeWitt,
I agree that the signal is almost certainly a real one and that tuning can be performed. I tried to formulate my comment in a way that’s fully consistent with that. From accepting that the signal is a real one we cannot conclude, how accurately tuning is expected to produce the correct timing.
It’s natural to expect that the phase of the precession cycle affects also the timing by shifting the changes in insolation by several thousands of years up to half length of the precession cycle or about 10 kyr, from what obliquity and eccentricity would imply. 10 kyr is not essential in this comparison, but not insignificant either. There are certainly also other potential causes for some error in timing. Tuned data cannot be used as an empirical test of details that are directly dependent on the accuracy of tuning.
Errors in the original data affect also tuning leading potentially to unrealistically good agreement in comparison to the quality of the data.
Pekka,
This is a very helpful comment.
I can see clearly now (from the various comments and the Lin & Wang paper) that Fourier analysis doesn’t help much with these kind of problems, perhaps with much of climate science – at least, Fourier analysis might obscure more than reveal given the nonlinear processes involved.
The EMD process isn’t clear to me, but I expect it will take some time and a textbook to become conceptually clear.
However, the key point seems to be – if I understand your comment above – to review the time series, perhaps in the way we have done in this article with the Hovmoller plots ??
My primary interest at the moment is to understand the best dating of glaciations and deglaciations and identify these against the time progression of solar insolation.
SoD,
Some further observation. I use the 900 kyr history shown by Clive as reference for this. (Sod: You could edit the link in his comment; it contains the same link twice in succession and that makes it fail.)
It seems to tell something about the timing in spite of the observation that the preciseness of the agreement in timing may be due to tuning, but it fails to tell some essential points.
– Glacial cycles are asymmetric, deglaciation is much faster than glaciation, but the curve drawn is symmetric by nature while individual peaks have asymmetric structure. (The failure is clear in two out of the six latest glacial cycles.)
– The curve has still the nature of the linear solution although we don’t know any linear contribution from eccentricity cycle.
Your posts seem to indicate similar thinking to what I prefer:
1) The glacial cycles are not amplified orbital signals but consist of phases of glaciation (longer) and deglaciation (faster) separated by state shifts.
2) There are no lasting states of glacial maximum or interglacial, but both are essentially turning points between the more persistent states of glaciation and deglaciation.
3) What we should understand include:
i) What’s the nature of the state of deglaciation? The paper of Fisher et al is on this problem.
ii) What’s the nature of the state of glaciation?
iii) Why are the situations of extremes in glaciation (interglacials and glacial maximums) so short lived?
iv) Why does the deglacification end usually only at a very low level of ice, i.e. the process of deglaciation is seldom interrupted?
v) Why does the phase of glacification lead mostly to a high maximum, but this is not as regular as (iv)? (The explanation may be simply in the duration of the phase.)
vi) Finally: What determines the timing of the state shifts?
To me the case seems that the state shift is triggered easily with very little or very much ice, but requires stronger changes in insolation to occur during the state of ongoing glaciation. My question (iii) may thus be very tightly linked to timing when combined with the rates of change (questions (i) and (ii)).
I can’t help thinking that something basic is still missing. Milankovitch theory can only just be part of the story. So I am going to stick my neck out with a (crazy) new proposal:
The 100,000 and 400,000 year cycles in the ellipticity of the Earth’s orbit are caused by regular gravitational effects of the other planets as they orbit the sun, particularly those of Jupiter and Saturn. Every 100,000 years the orbits of Jupiter and Saturn align themselves so that their net gravity perturbs the Earth’s orbit causing it to elongate and become more elliptical. This cycle reaches a maximum every 400,000 years in regular fashion.
The moon is also effected by the same regular (Milankovitz) induced variation in its orbit around the sun. This also causes an increased elliptical orbit of the moon around the Earth. Tidal forces vary as 1/r^3 so small changes in distance can have large effects on tides.
Read more here:
Could the moon be the trigger for interglacials
Feel free to criticise !
Pekka (re: January 10, 2014 at 3:10 pm),
That’s an interesting paper. It reminded me that the whole series of papers published in Quaternary Science Reviews in that issue (Vol 29, 2010) is really good. It is basically the EPICA research team publishing on all their topics.
The introductory paper is Barbante, Climate of the last million years: new insights from EPICA and other records.
I’m not sure how many are freely available, but here is the list for anyone interested:
Thank you for this very well written piece! This is the first of your blog posts that I have read and found it to be both educational and acutely insightful. This subject is peripherally related to my research as a PhD student in geology studying depositional sequences that were controlled in large part by past sea-level cycles. (Tectonics and sedimentation rate are key factors as well.) However, I’m still very much a non-expert in the nitty gritty of the climate forcing mechanisms, such as Milankovitch cycles. I feel that I’ve learned quite a bit from this blog already so thanks for taking the time to write this!
One last thing:
(Apologies in advance if this is not the appropriate space for this comment. Feel free to snip by all means.)
As someone curious about climate science but beyond fatigued with what I consider the disingenuous tactics from each “side,” which seem primarily concerned with political point-scoring and/or posturing rather than with honest scientific discussion, “Science of Doom” has catapulted to the very top of my list of favorite climate blogs. I can’t tell you how refreshing it was to read a blog post about climate science that discussed both strengths and weaknesses of a study without once treading anywhere near the trappings of partisan rhetoric.
TronCarter,
Thanks very much. I annoy or upset lots of people so it’s great to hear your comments.
SoD,
You write in a comment above
That’s obviously true for those orbital periods that are used in calibration, but using only some of the periods in tuning and searching for others in the tuned timeseries makes perfect sense.
A relatively recent paper where tuning is used is Dreyfus et al (2007) where only the precession periods of 23 kyr and 19 kyr are used in tuning the timing of δ18O signal from EDC.
[…] 2014/01/09: TSoD: Ghosts of Climates Past – Eleven – End of the Last Ice age […]
Frank asked on January 18, 2014 at 9:53 am (replying here so it’s not buried in the threads):
I don’t know much about ice sheet dynamics.
A recent paper, Insolation-driven 100,000-year glacial cycles and hysteresis of ice-sheet volume, Ayako Abe-Ouchi et al, Nature (2013) did a GCM model with a parameterized ice sheet model (pre-calculated for different conditions from a more complex model).
Here are a few quotes from that paper and I comment a little at the end:
I have lots of questions about this paper. I just emailed the lead author to ask about the simulation itself – which is over 400 kyr.
In previous articles we saw Smith & Gregory 2012, with a 120 kyr simulation that could only happen by having lower resolution (x10 improvement in speed) and all forcings speeded up by 10 (x10 improvement); and a second paper, Jochum et al said:
So I’ve asked how they managed to produce a 400 kyr simulation – amazing supercomputers? or some other magic? It’s important to understand any assumptions and simplifications in a model before getting any further.
On the paper’s results, the slow rebound of the earth’s crust is a critical part of the effect that they find.
I don’t know what the actual values of isostatic depression are for the LGM, probably can be found in a paper somewhere with some searching.
I have lots of other questions to ask on this paper, like ‘did the model find that Antarctica led the N. Polar regions’?
and ‘did the model find that the time period of terminations matched the real time, or did it coincide with peak insolation’?
SOD: Thanks for the reply. I was wondering if the NH warming lagged SH warming because so much energy was consumed in melting the ice. For a 100 m SLR in 10 centuries (1 cm/yr over 70% of the earth, 0.7 cm/yr, 7 kg water melted/m2/yr), I calculate a power input of 0.074 W/m2 from every square meter of the planet. If only the energy arriving in the NH was used to melt NH ice caps (and little melting occurred in the SH), 0.15 W/m2 of power would be needed, a fairly small number compared with the forcings driving current climate change. The buildup of ice sheets is several times slower than melting and so represents even less power. From this “power input” perspective, ice ages and interglacials are pretty trivial affairs.
When I see how far south the Greenland ice cap reaches, I always wonder why it is still there. The only answer that makes sense to me is that the top stays below freezing because of its altitude and the lapse rate, so accumulation on the top can compensate for melting at lower elevations near the coast. But if were that simple, the altitude of the Laurentide Ice Sheet would have preserved it too.
Until you produced this paper, I thought that a role for “glacial isostatic subsistence” (to coin a name for the opposite of glacial isostatic rebound) in terminations was wild speculation. According to Wikipedia, rates of glacial isostatic rebound are about 1 cm/yr in some places today, 10-20,000 years after the ice sheet melted. If a rate of “subsistence” 10-fold higher existed near the end of an ice age, the elevation change associated with subsistence would be 1 km/10 millennia. That makes it possible that terminations can be triggered (perhaps by orbital changes) once enough compression has occurred. Many ice covered areas could be below sea level, even though that sea level was 100m below today’s. This map of Antarctic shows just how different things might have been – the usual map apparently shows what Antarctica would look like after the ice cap melts and rebound occurred.
Maybe the bigger continental ice sheets subsided more than the Greenland Ice Sheet and that is why the latter survived.
Nitpick: The opposite of rebound would be subsidence, not subsistence.
[…] In Eleven – End of the Last Ice age we saw the sequence of events that led to the termination of the most recent ice age – Termination I. […]
FYI, I used a Hovmoeller plot in here: http://erimaassa.blogspot.fi/2014/01/brutality.html
I for one cannot understand how the NH (NGRIP) and SH(EPICA) can be so much offset from each other so the picture is just an artistic interpretation of things long gone… most adjustments had to be done on the EPICA record with a couple shorter ones to the NGRIP… for the lack of better explanation I could say Antarctica could have been so cold and dry during the ice age there was no snow accumulation on the EPICA site. This and couple other explanations still leave two adjustments unanswered and the match to the 40-50S insolation isn’t too good. Anyway, from the plot, 143ky ago might be a good place for the start of previous termination.
jyyh,
It’s important to understand the assumptions behind the data.
First of all though, when you say “offset” do you mean amplitude or time?
On amplitude, it’s clear that Greenland has recorded much greater swings in temperature than Antarctica.
On time, this field is a work in progress.
I think there are a few papers that really help in getting across this technical discipline.
1. One-to-one coupling of glacial climate variability in Greenland and Antarctica, EPICA Community Members (2006) – free paper
This lines up the DML core (high resolution, but last 150 kyrs only) with NGRIP:
However, the timescales are not properly aligned. The subject of timescales for Antarctic ice cores is complicated. A bit of background from a few papers is not essential, but really helps:
2. Extending the Vostok ice-core record of palaeoclimate to the penultimate glacial period, Jouzel et al (1993) – free paper
3. Climate and atmospheric history of the past 420,000 years from the Vostok ice core, Antarctica, Petit et al (1999) – free paper
4. Eight glacial cycles from an Antarctic ice core, EPICA community members (2004)
Then the EPICA group worked on new timescales. The 2004 paper put the ice core data on the “EDC2” chronology. Later they developed the EDC3 chronology:
5. The EDC3 chronology for the EPICA Dome C ice core, Parrenin et al (2007)
Then, realization that more work needed to be done:
6. Consistent dating for Antarctic and Greenland ice cores, Lemieux-Dudon et al (2010):
I don’t want to overwhelm you with technical details and papers that cover lots of ground, but the field is in transition and ensuring you have the right dataset – and you understand the assumptions built into the dataset – is something of a journey (a journey that I am on).
Note – A lot of papers in this field are not publicly accessible because they are published in Nature or Science (I can email you a few specific ones if you need them).
thanks TSoD, I meant time of course, it’s entirely possible that I have a version of EPICA that’s not been reworked (2004), hence it’s possible the EPICA2006 paper is likely to do the same way better than I would have been able. I was just eyeballing and it seemed that the colder it is on Antarctica, the more the dates on the dataset I have might be skewed. The spikes on Antarctic data look like developing more like one would assume by the slow orbital forcings with regular inclines and falls (normal distribution) so the starts of the warmings within the glacial would almost all be SH driven if the times can be matched between NH and SH. Since I’m not a working scientist, I’m not expecting to find anything notable or new. ‘Sometimes a blind chicken…’ and so on. And there’s nothing wrong in repeating work done by others, if just to make things more clear to myself or the few blog readers. I’ve also stated in the blog (sometime in 2008) that the responsibility is the readers so I’m not overtly concerned of the scientific accuracy of my blog posts ;-). There’s quite a bit of reading.
jyyh ,
I will try and write an article summarizing the ice cores when I am on top of the relevant work.
Thank you for a very interesting blog. I think it is a more reliable blog than most other. It gives som answers, and a lot of questions. I learn a lot of reading about the basics, but think most of the assumptions in predicting climate are of less value. Are the forcings a huge guesswork? If it is true that there is a heating in the deep oceans, how can that come about? It`s like a mysterious force, Like a hand of God (or Satan), dragging heat down into the deep. Can there be a heating without the work of gases, directly into water? Can there be a heating of deep ocean over thousand of years, melting ice ages? Can there be an imbalance of radiance at TOA lasting for thousand of years, providing the heat down in the ocean. Can the icecap be a kind of lid so that the ocean is not cooling, but building up heat in the depths. So many questions, and so few answers.
It’s a real joy to come across thoughtful discussions on this topic. There seems to be entirely too much hype.
I know most laypersons consider that we’re not in an Ice Age, but technically we’ve been in one for roughly 2.6 million years. The Holocene is merely the current interglacial of the current Ice Age. So, talking about the last Ice Age ending 10,000-19,000 years ago is incorrect, because interglacials still have ice at the poles and covering continental material.
The paper by W.S. Broecker, “The Present Interglacial: How and When?” provides an interesting discussion of mechanics and timing of interglacial endings. Though no one can know when the current interglacial will end, it should sober anyone to discover that the Holocene is already running 600 years over the average length of an interglacial in the current Ice Age.
For all we know, the current interglacial could start ending later today. Though many interglacials have taken many hundreds of years to end, some have ended in as little as 50 years–mirroring the abrupt start and end of the Younger Dryas.
It would seem prudent to explore methods for ending the current Ice Age, or at a minimum, preparing for the next glacial period. Severely decreased evaporation during a glacial period will result in desertification and failed crops for something like 90,000 years.
What would it take to end the current Ice Age? What methods could be implemented using current technologies? If we spent as much effort on this as we’ve seen put toward bashing “warmth,” perhaps we could avoid the more terrifying cold. With global warming, people might be inconvenienced. With global cooling, billions might die.
I’m curious what your thoughts are on this.
There is a new proposal by Ralf Ellis as to what triggers an interglacial.
Does Gaia end Ice Ages?
As ice sheets expand and ocean temperatures fall, so CO2 levels eventually fall dangerously below 200ppm, threatening the survival of plants based on C3 photosynthesis. Boreal forests and temperate grasslands will begin to die back exposing soil to weathering. Desertification ensues and strong winds will transport dust storms over the ice sheets. With very little annual snowfall during a glacial maximum, this dust layer will build up and thereby reduce the ice albedo. The next ‘Grand Summer’ precession cycle at maximum eccentricity will now be able to rapidly melt back the ice sheets, because finally lower albedo ensures more heat is absorbed by the ice sheets.
Interesting.
If you believe Ruddiman, we’ve already solved, or at least substantially delayed, the problem of a new glacial period by a combination of land use/land cover changes associated with the invention of agriculture and fossil fuel combustion to return long lost carbon to the atmosphere. Otherwise the planet would already be well on its way to a new glacial period.
We are just really lucky to be living in an era of low orbital eccentricity. Otherwise the next ice age would have already started because northern summers are weak because they coincide with earth-sun aphelion. The effect is much less than normal because maximum eccentricity is just 0.016 during this cycle. A similar glaciation occurred 400,000 years ago and that interglacial lasted about 25,000 years. A much larger drop in summer insolation will occur in 15k years time which should end this interglacial. Any extra help we can get from CO2 emissions might then be thought beneficial !
http://clivebest.com/blog/?p=7046
Clivebest,
The MIS 11 (marine isotope stage 11) interglacial did indeed last for more than a full precession cycle. Dating is uncertain, but from what I have read (Ruddiman, for instance) it seems likely that the minimum July insolation at 65N during that interglacial was 439 W/m^2 at 417 ka (thousand years ago). At present, July insolation at 65N is 427 W/m^2, significantly less. Some people have dated the MIS interglacial so that it spans the minimum at 397 ka, which was 415 W/m^2. I think the 417 ka date is much more plausible since, unlike the 397 ka date, it provides an explanation for why that interglacial was so long. Note that the insolation at any given age is not in question, the uncertainty is in the dating of the interglacial.
You may be interested in this paper:
Critical insolation–CO2 relation for diagnosing past and future glacial inception
A. Ganopolski, R. Winkelmann & H. J. Schellnhuber
Nature 529, 200–203 (14 January 2016) doi:10.1038/nature16494
The past rapid growth of Northern Hemisphere continental ice sheets, which terminated warm and stable climate periods, is generally attributed to reduced summer insolation in boreal latitudes1, 2, 3. Yet such summer insolation is near to its minimum at present4, and there are no signs of a new ice age5. This challenges our understanding of the mechanisms driving glacial cycles and our ability to predict the next glacial inception6. Here we propose a critical functional relationship between boreal summer insolation and global carbon dioxide (CO2) concentration, which explains the beginning of the past eight glacial cycles and might anticipate future periods of glacial inception. Using an ensemble of simulations generated by an Earth system model of intermediate complexity constrained by palaeoclimatic data, we suggest that glacial inception was narrowly missed before the beginning of the Industrial Revolution. The missed inception can be accounted for by the combined effect of relatively high late-Holocene CO2 concentrations and the low orbital eccentricity of the Earth7. Additionally, our analysis suggests that even in the absence of human perturbations no substantial build-up of ice sheets would occur within the next several thousand years and that the current interglacial would probably last for another 50,000 years. However, moderate anthropogenic cumulative CO2 emissions of 1,000 to 1,500 gigatonnes of carbon will postpone the next glacial inception by at least 100,000 years8, 9. Our simulations demonstrate that under natural conditions alone the Earth system would be expected to remain in the present delicately balanced interglacial climate state, steering clear of both large-scale glaciation of the Northern Hemisphere and its complete deglaciation, for an unusually long time.
http://www.nature.com/nature/journal/v529/n7585/full/nature16494.html
Frank,
Looks interesting indeed, but I can’t access it for a year since I don’t have a subscription and won’t pay $50 or whatever it is they want. In a year I will be able to get it through my university library if I have not forgotten about it. That seems to be a new policy implemented by the Nature “family” of journals.
Mike M. I could not access the full paper either, but the major conclusions appear to be derived from Figure 3, which can be opened in full page view.
The problem of EMICs (earth models of intermediate complexity) is they replace physics with parameters, and run lots of simulations – choosing the result that matches reality.
Ok, big oversimplification but basically how it works. Unless something has changed recently.
I realize that a similar criticism can be leveled at GCMs but it’s not quite the same. At least, the tuning/matching is less obvious and not based on the method of many runs with different parameters.
The method in EMICs – run lots of simulations with different parameters and pick the one that works. Is it different from curve fitting? Not unless the normal process to avoid over-fitting is taken.
Normal process to avoid over-fitting = run 1/2 or 2/3 of the time period with the process, then verify that the simulation result in the last 1/2 or 1/3 matches the actual measurements.
Caveat – above description is from memory based on a number of papers. I will review some in the next few days and comment further.
So, the abstract should read “one of many possible EMICs predicts ….”. One might ask how about the ECS for this particular EMIC. If ECS is 10 K for 2XCO2, then a modest change in CO2 plus polar amplification might enable modest changes in CO2 to modulate development of an ice age.
When calculating the social cost of carbon, the benefit of avoiding an ice age probably should out weigh all of the harms. 😊
I wrote a brief article on this topic – Ghosts of Climates Past – Twenty – Note on EMICs.
Sometimes a Great Notion can be gleaned from formerly disproved theories such as the Hamaker/Weaver hypothesis of the 1980’s which spoke to the carbon depletion and de-mineralization of soils (loss of top soil and worldwide forest fires). It de-emphasized the increasing CO2, but admitted that the holy grail of what actually could lead to a virtual lock of our jet stream that would enable both the building of an ice sheet over North America and Northern Europe.
That could all remain in place by the differential heating and cooling of the globe with the northern latitudes becoming cooler from volcanic eruptions around the northern ring of fire, while tropical regions continue to warm up! This differential warming and cooling gradient produces more weather patterns that can produce the needed snowfall over the right areas while kicking in the proper tropical heat engine to produce the greatest results!
A kind of Pineapple Express overdrive you could say. Volcanic ash does not move from the northern hemisphere to the south since winds do not circulate between hemisphere’s. Only Equatorial volcanoes can put ash into either hemisphere. Never doubt how quickly a high latitude ash eruption can cool down the arctic. As you know the angle of incidence of sunlight is greater nearer to the poles. That means that light not only hits the ground with a lower intensity at a greater angle but ALSO MUST TRAVEL THROUGH MORE OF THE ATMOSPHERE BEFORE IT HITS THE SURFACE OF THE EARTH, SO IT IS MORE PRONE TO BEING INTERCEPTED BY THE PRESENCE OF VOLCANIC ASH WHICH ALSO CAN HAVE THE IMPACT OF SEEDING CLOUDS PRODUCING EVEN MORE SNOWFALL ETC.
Our planet, just like our bodies, operates from the interaction of several different kinds of feedback loops which can facilitate different results depending on the various inputs involved, climatology must try to factor in a tremendously large number of highly interacting (with each other) variables that must be just right to produce the extended snowfall to transfer that much water to create a continental ice sheet. Keep in mind once it is there it may no longer receive the same weather that created it. There are far more ways to reduce or intercept light. Only sunspot cycles can increase the overall total radiation we get from the sun. Melting ice comes much more easily. Some consideration of large catastrophic events could also deliver the large transfer of the water required as well!