Towards an understanding of Ice Ages

An Ice model driven by NH insolation but adjusted for dust albedo does a pretty good job at reproducing the last 8 glacial cycles. Details are described below.

Simulated Ice Volume model in blue compared to measured LR04 Ice Volume data.

Prior to the Mid Pleistocene Transition (MPT) roughly one million years ago,  glacial cycles followed a simple 41Ky obliquity cycle. This  can be understood if Northern and Southern Hemisphere ice sheets waxed and wained independently, because the average of the NH summer and SH summer peak insolation follows obliquity.  Raymo et al. [1] suggested that the East Antarctic Ice sheet melted back to bare land prior to 1M years ago. The ebb and flow of the southern ice sheets (Patagonia included) with SH precession simply offset  that in the Arctic because the precession component is always out of synch between North-South summers. The average of both just leaves the in-synch obliquity component which is why it dominates the Global Ice volume data before MPT.

Hubers (2) also pointed out that the integrated summer insolation which he calls total “caloric summer” also follows obliquity. The reason for this is due to Kepler’s law. Conservation of angular momentum during the earth’s elliptical orbit ensures that each hemisphere’s precession enhanced maximum insolation  has a shorter summer by several days. So when calculating the total  energy received at 65N or the poles, the  precession term simply cancels out. Both the above effects are summarised in figure 1., which also highlights the role played by eccentricity in modulating peak insolation at each pole.

Figure 1. Maximum and total solar insolation calculated at the poles during last 600,000 years. Both the total annual insolation and the N-S asymmetry clearly show the underlying effect of the 41,000 obliquity signal. STOT is equivalent to total calorific summer and (Smax NP – Smax SP) leaves just  the obliquity signal which affects both hemispheres equally.

Raymo proposed that once Antarctica became permanently covered in ice after MPT the symmetry was broken, and NH insolation thereafter became the dominant driver of  glacial cycles. So why didn’t ice ages simply follow the NH Peak 65N insolation? This is the unsolved mystery of ice ages as to why they  apparently switched instead to a 100Ky “eccentricity” cycle. Eccentricity modulates the strength of the precession term and each interglacial does indeed coincide with eccentricity maxima. However,  no-one has properly explained why other larger NH  insolation peaks fail to have much effect, while far smaller peaks eventually succeed in melting back the ice sheets (figure 2).

Figure 2. The last 800,000 years of glacial cycles. Ice Volume (LR04) in grey, Eccentricity in black, EPICA (Antarctic) temp (red), dust(purple), CO2(yellow). The current interglacial is most similar to that 400,000 years ago which also coincided with low eccentricity. Click to expand

In 2006 Gerard Roe published a paper “In defence of Milankovitz” [3] where he showed a good qualitative agreement between the rate of change of the SPECMAP[4] Ice volume fit with Arctic summer melting, as measured by maximum insolation at 65N. The agreement though was too good, partly because the SPECMAP fit itself had used the 65N insolation for dating the stack.  The LR04 stack[5] however, contains many more geographically spaced cores which are independently timed. When I checked the LRO4 derived DV/DT against N-pole summer insolation I found that there was indeed a correlation but the agreement was not nearly as good as Roe’s. However what Roe did demonstrate is that NH peak insolation is a major driver in the ablation and growth of the ice sheets, but cannot on its own explain the ice volume data. So what is the missing link needed to explain the dynamics of recent ice ages?

Ellis and Palmer[6] proposed that the missing link was the dust albedo effect. As glaciations deepen and CO2 levels fall below 200ppm. The arid conditions cause less snow which combined with CO2 starvation cause die back of boreal forests and increased dust deposits over the ice sheets lasting thousands of years. You can see this effect in figure 2. They propose that this decreases the ice surface albedo sufficiently such that the next NH peak summer is sufficient to rapidly melt back the ice sheets caused by a chain reaction  ice albedo feedback. One difficulty with this explanation is that Greenland dust data is only available for the last 200,000 years and so the assumption needs to be  made that Antarctic dust is also a proxy for NH glacial deposits. This seems a reasonable assumption at least for for the last glacial period (figure 3).

Figure 3: Comparison of Greenland GRIP data with EPICA Antarctic data.

Willeit and Ganopolski [7] have also recently highlighted the importance of dust albedo and snow ageing during the last glacial cycle.

“The surface energy and mass balance of ice sheets strongly depends on the amount of solar radiation absorbed at the surface, which is mainly controlled by the albedo of snow and ice.

Both the snow ageing effect and the effect of dust deposition on snow albedo play a fundamental role in reducing surface albedo, particularly in the ablation areas”

They used an “Earth system model” – CLIMER-2 and were able to reasonably well model the last glacial cycle. Yet the details still remain obscure because such complex climate models are black boxes. Surely there must be a simpler physical explanation for the slow growth and sudden collapse of ice ages.

Does dust enhanced albedo play the critical role in recent glacial cycles?  To answer this question, I decided to try to develop a simple dust model based loosely around the original Imrie and Imrie Ice model [8]. First I tried a very simple equation.

-\frac{DV}{DT} \propto (1 \pm b) \times S(1 -\alpha)

where b is a nonlinearity constant between ice growth and decay, S is NH insolation, \alpha is the ice albedo which depends on dust levels.

I have tried two versions for \alpha so far.

a)   \alpha(t) = 0.5 - fac \times \int{dust.dt}


b)   \alpha(t) = 0.5 - fac \times (dust(t-15))

The first assumes that a critical accumulation of dust over one glaciation is the prime driver once NH induced ablation starts, whereas the second assumes that a critical threshold 15,000 years ago is required. Scale factors are needed to match insolation in W/m with Ice Volume in Benthic \delta_{18}O . I used b = 0.03. The model is written in Python.


First we look at the integral dust results a) as shown in Figure 4. The dust albedo model improves the agreement between Insolation and -DV/DT

Figure 4. Shown below are the model results in blue  compared to -DV/DT data in red. The upper plot shows again -DV/DT in red compared to the 65N Insolation. Below this is the LRO4 ice volume data and dust data scaled for comparison.

The agreement of the albedo weighted NH insolation with  -DV/DT is now much better. But the real test of the model though is to reproduce the Ice Volume results by integrating forwards in time based on the start value 800,000 years ago. The only input for this is the new NH insolation and the recorded EPICA dust data. Figure 5 shows the result.

Figure 5. A comparison of the Model A simulated Ice volume by integrating -DV/DT forward in time  This is compared to the  the actual LRO4 Ice volume data. Overlaid in red is -DV/DT proving that the integration is sound.

In general the agreement is remarkably good considering the time scale. However we see two problems caused by overcompensation for dust albedo in the early cycles. Perhaps Antarctic dust was not fully coupled to that in NH.

The results from model b) which includes a 15,000 year time lag in dust contributions are shown in figures 6 and 7.

Figure 6. Overview for the delayed dust albedo model. The comparison of -DV/DT for ice volume and Dust induced albedo is shown below.

Figure 7: Simulated Ice Volume results by integrating the delayed dust albedo model B forward in time. This is compared to the actual LR04 ice volume data in black/red. as in Figure 5.

The agreement is better, but we see potentially another problem in that the holocene is now not so well modelled due to latent effects of dust from LGM. However, the timings and basic shape of glacial cycles are reasonably well represented. So I am fairly hopeful that a solution to the mystery of ice ages can be found.

This is still a work in progress but you can download the current software and all relevant data here


  1. M. E. Raymo et al. Plio-Pleistocene Ice Volume, Antarctic Climate, and the Global ?18O Record. Science  28 Jul 2006: Vol. 313, Issue 5786, pp. 492-495
  2. Peter Huybers. Early Pleistocene Glacial Cycles and the Integrated Summer Insolation Forcing,  Science 313, 508 (2006)
  3. Gerard Roe, In defense of Milankovitch, GEOPHY RES LETTS, VOL. 33, L2470
  4. Imbrie, J., Hays, J.D., Martinson, D.G., McIntyre, A., Mix, A.C., Morley, J.J., Pisias, N.G., Prell, W.L., and Shackleton, N.J., 1984. The orbital theory of Pleistocene climate: Support from a revised chronology of the marine ? 18O record. In Berger, A., et al., (eds), Milankovitch and Climate. Hingham, MA: D. Reidel, pp. 269–305.
  5. E. Lisiecki &Maureen E. Raymo, A Pliocene-Pleistocene stack of 57 globally distributed benthic D18O records, PALEOCEANOGRAPHY, VOL. 20, PA1003, 2005
  6. Ellis, Ralph and Michael Palmer (2016) “Modulation of Ice Ages via precession and dust-albedo feedbacks” Geoscience Frontiers
  7. Matteo Willeit and Andrey Ganopolski, The importance of snow albedo for ice sheet evolution over the last glacial cycle,  Clim. Past, 14, 697–707, 2018
  8. John Imbrie and John Z. Imbrie,  Modeling the Climatic Response to Orbital Variations, Science, Vol. 207, No. 4434 (Feb. 29, 1980), pp. 943-953


About Clive Best

PhD High Energy Physics Worked at CERN, Rutherford Lab, JET, JRC, OSVision
This entry was posted in Climate Change, Ice Ages and tagged . Bookmark the permalink.

31 Responses to Towards an understanding of Ice Ages

  1. Lance Wallace says:

    4K (your second sentence) should be 40 or 41 K.

    “I have tried various two models” Perhaps delete “various”?

  2. J Martin says:

    Inyeresting. I will have to read it again.

  3. oldbrew says:

    This might be of interest…

    Glaciers Created a Huge ‘Flour’ Dust Storm in Greenland
    By Rafi Letzter, Staff Writer | November 2, 2018

    Researchers have written and speculated about glacier-flour dust storms in Greenland for a long time, according to NASA. But it took until this September for investigators to spot such a massive plume of the elusive dust forming and drifting 80 miles (130 kilometers) northwest of the far-northern village of Ittoqqortoormiit. Glacier flour is a fine dust created when glaciers pulverize rocks, NASA wrote. While satellites had occasionally spotted smaller storms of the stuff, this one was “by far the largest.”

    • ralfellis says:

      Yes, Ganopolski has made several papers based upon that same glaciogenic argument. However, the dust in the Greenland ice sheet has been identified by isotopic analysis, as being from the Gobi desert region. So none of Ganopolski’s arguments make much sense.

      The dust originated from the Gobi because CO2 levels got so low, that the whole of the Gobi region became a shifting-sand CO2 desert. And the great dust layers on the Loess Plateau in China demonstrate how much dust the Gobi was producing.


  4. Javier says:

    However, no-one has properly explained why other larger NH insolation peaks fail to have much effect, while far smaller peaks eventually succeed in melting back the ice sheets (figure 2).

    Of course it has been explained, by Peter Huybers, by Chronis Tzedakis, and by me. It just takes a long time to sink, perhaps a generation.

    NH peak summer insolation is not decisive to trigger an interglacial unless is above 549 W/m2 at a time when obliquity is >23°. Otherwise it is a third order factor that can help produce an interglacial when synchronized with obliquity.

    Second order factor is global ice volume. This was observed by Parrenin in 1998. The more global ice volume the better. If LR04 d18O > 4.90 per mil and obliquity > 23° you get an interglacial, no matter the value of NH insolation. High global ice volume values strongly promote interglacials at high obliquity.

    And first order factor is obliquity. Every interglacial takes place at an obliquity window (>23°), and every interglacial ends ~ 6 kyr after obliquity drops below 23°. No obliquity window no interglacial. That’s a hard rule. Interglacials are still taking place at a 41 kyr cycle, they are just skipping beats.

    The 100-kyr cycle is not an interglacial cycle. There are 11 interglacials in the past 800,000 years. Do the math. Slightly less than 82 kyr because twice in that period two interglacials took place separated by 41 kyr. Don’t believe me? You have already seen this, yet you are still unconvinced.

    Milankovitch got it right when he defined caloric half summer as the half of the year when every day receives more insolation that any of the other half. It was tough to calculate without computers, so he did it for one year every 10,000. After Berger 1978 Milankovitch theory was corrupted with the idea that NH 21st June insolation was good enough, and easier to incorporate into the first model by Kutzbach in 1981. Every single problem of Milankovitch theory can be tracked to that corruption of the theory. Despite NH peak summer insolation being unable to explain interglacials, everybody, including you, holds to it even if they are told it is wrong. Why?

    Milankovitch theory has absolutely no problem and explains perfectly well interglacials, including their anomalies, like MIS 11c and MIS 7e. Your dear dust is a consequence, not a cause, exactly the same as CO2, and together with albedo and other factors (increased volcanic activity, rising sea level), are just part of the positive feedback pack towards quick melting associated to a very high ice volume.

    Follow this flow chart:

    Interglacial flow chart. A simple flow chart incorporates the criteria deduced from the past 1 million years of glacial cycle. Interglacials do not take place when obliquity is below 23°. If obliquity is above 23°, interglacials require an ice volume higher than the equivalent to 4.55‰ benthic ?18O or a very high 65°N July summer insolation, above 549 W/m2. If insolation is lower, interglacials require an insolation above 521 W/m2, or very high ice volume, around 4.90‰ ?18O. There are two cases why interglacials fail to take place. The most common is case 1, when ice volume is not sufficiently high (8 cases). Case 2 is when there is sufficient ice but insolation is too low, associated to very low eccentricity (3 cases).

    We do really need to rely less in models and more on the evidence. Models are built on assumptions that are generally wrong.

    • ralfellis says:

      So why would a high ice volume trigger an ice sheet collapse, rather than a retreat back to a more stable volume? It makes no sense. If the ice sheets were happy at 80% of maximum volume, why not just retreat back to the stable 80% level.


      Also the ice age cycle is NOT 100 ky, and NOR is it 82 ky. It is a variable cycle that started off as about 90 ky, but has increased to about 115 ky. And these cycle lengths are the same as the equivalent clusters of four or five precessional cycles. Quite clearly the controlling factor in recent ice ages is precession, not obliquity.

      On the Image below SGY stands for the Seasonal Great Year.
      (A title for the precessional cycle, which is quite variable.)

    • ralfellis says:

      And Javier….

      Please remember that this is an article about an ice-age model based upon dust-albedo feedbacks. It is not an opportunity for you to post vast diatribes about your obliquity theory.

      So what do you think is wrong with including dust-ice-albedo as a terrestrial feedback system, that assists in the initiation of interglacials?


  5. Clive Best says:

    Thanks Javier for a clear explanation. I agree that total caloric summer for both poles depends only on obliquity. Likewise the asymmetry between NH and SH also follows obliquity. I like the plot of Epica temperatures aligned with obliquity delayed by 6,500 years. It is impressive.

    So your rule works, but it doesn’t say why it works. For example why should there be a threshold in ice volume needed for an interglacial? If you believe ice albedo feedback then each glaciation would end up as a snowball earth. Likewise falling CO2 is a feedback. So I need a physical explanation as for why the rule works.

    I am not wedded to the dust albedo factor but it does offer a possible physical explanation for why the rule works. Dust peaks whenever dry arid conditions ensue which is normally at the end of a glacial period. My simple ice model kind of works although it needs some fine tuning.

    How do you explain the Rho result that DV/DT ~ Insolation65N , or is it just because timings are artificially chosen to align?

    • Javier says:

      So your rule works, but it doesn’t say why it works.

      I think it is pretty obvious. We go back 2 Myr and obliquity alone was enough to melt the ice and produce an interglacial. Then the planet continued cooling and ice build-up was too much for obliquity to overcome, as it has a 6 kyr delay and only 20.5 kyr of rising phase. On practical terms about 10 kyr to get the job done and so much ice melts slowly. So obliquity became dependent on high latitude North summer insolation. Since precession period is about half of obliquity’s, it means that half of the time they are properly aligned and cooperate to get the ice melted. In MIS 3, for example they were not properly aligned so no interglacial was possible. But when eccentricity is low and no interglacial is produced the planet accumulates a huge amount of ice, so the workload doubles the next time. Around 1 Ma the amount of ice was so much from one interglacial to the next that a new factor entered the scene. Otherwise interglacials might have stopped occurring or would have rarely taken place. Parrenin observed this factor that he called ice-sheet instability. High ice volumes strongly promote interglacials. This is an observation.

      The explanation is not albedo. Albedo is a puny beast. The great majority of albedo is atmospheric, and the increased albedo from ice can be easily compensated by a small reduction in cloud cover, and we know glacial periods are arid. Ice sheets become unstable when they grow too large. We know this because of Heinrich events that produce armadas of icebergs at ~ 6000 years intervals. Younger Dryas is a Heinrich event. Ice shelves become prone to collapse, and when ice volume is very high it means that a great part of the ice is lying below current sea level, on continental platforms. The factor that is most effective at melting ice is not insolation, and not dust. It is water. At the same temperature water melts the ice several times faster than air. And high obliquity delivers the increased solar energy right at the water at high latitudes. As the ice melts the sea level rises and more ice melts, so the sea level rises more. The rest of the factors (CO2, dust, volcanic activity, albedo) also contribute but are much slower. Melting by rising sea levels at the base of glaciers and ice sheets probably accelerates their speed and collapse.

      This new powerful negative feedback by ice volume has as a consequence that interglacials become warmer the higher ice volume is prior to their termination. Check it. Thus temperature oscillations have become wider in the last 800 kyr and the climate is more extreme.

      For example why should there be a threshold in ice volume needed for an interglacial?

      Because the time allowed to melt the ice and cause an interglacial before obliquity starts falling is limited by the obliquity cycle. The speed of the melting is directly proportional to the amount of ice. More ice, faster melting. Ice lower than the threshold and the melting is too slow to do it before the obliquity window closes. Higher than threshold and the melting will take place fast enough to produce an interglacial before obliquity starts falling.

      If you believe ice albedo feedback then each glaciation would end up as a snowball earth. Likewise falling CO2 is a feedback. So I need a physical explanation as for why the rule works.

      I believe ice albedo is hugely overestimated. The Arctic melting from 1995 to 2007 was nothing short of spectacular. 30% of summer sea ice has been lost. The ice albedo feedback must be several times higher now, yet sea ice has not melted in 10 years. The planet is about atmospheric albedo, not ice albedo.

      I am not wedded to the dust albedo factor but it does offer a possible physical explanation for why the rule works. Dust peaks whenever dry arid conditions ensue which is normally at the end of a glacial period. My simple ice model kind of works although it needs some fine tuning.

      Possible explanations greatly outnumber correct explanations, that’s why hypotheses abound. One has to be extra-careful about causality and correlations. Dust is a consequence of glacial periods, so the correlation can not support that dust is a cause, as it is already explained. Feedback factors are defined as consequences that modify the gain of the response. By that definition dust is a feedback.

      How do you explain the Rho result that DV/DT ~ Insolation65N , or is it just because timings are artificially chosen to align?

      The result is not very impressive. The peaks that clearly align are the peaks when an interglacial was produced, and in most cases that means that obliquity, 65N insolation and ice volume were working together.

      How do you explain the lack of effect that 65N insolation had on ice volume at 375 ka? or at 450 ka? or at 175 ka?

    • ralfellis says:

      The calorific summer (total energy for the whole summer) is not the full answer, because of the low angle of the spring and autumn Sun, as per my longer post below. Low elevation insolation cannot warm high albedo ice sheets, and so most of that ‘all summer’ obliquity insolation you are relying upon is reflected away.

      Soot on Snow experiment: bidirectional reflectance measurements of contaminated snow:

      In addition, the full calorific summer tends to mask the effects of precession, because spring and autumn precession are in opposition – while obliquity can act for the whole summer throughout the entire precessional cycle.

      This makes the effects of precession look weaker. But if it is only high elevation mid-day mid-summer Sun insolation that can be absorbed by high albedo ice sheets, then it is the effects of midsummer insolation that are paramount in initiating interglacials.

      It is precession that can deliver that high intensity midsummer insolation, not obliquity. And precession is strongest when eccentricity is high, which is why we get the strong correlation with the ~100 ky eccentricity cycle.


  6. Clive Best says:

    I like your argument, but I am not so sure that obliquity did become dependent on NH high latitude (precession) insolation, certainly not in a systematic fashion (see plot below)

    Ice sheets become unstable when they grow too large. We know this because of Heinrich events that produce armadas of icebergs at ~ 6000 years intervals. Younger Dryas is a Heinrich event. Ice shelves become prone to collapse, and when ice volume is very high it means that a great part of the ice is lying below current sea level, on continental platforms. The factor that is most effective at melting ice is not insolation, and not dust. It is water.

    I have thought about this before – see The straw that broke the camel’s back. High ice volume is the same as low sea levels. Once the land bridge forms between Siberia and North America the only access to the Arctic Ocean is through the North Atlantic. Sea levels at LGM were 120m lower than today.

    High obliquity moves the tidal flows further north because the moon’s orbit spans a higher latitude range, likewise the sun. A lunar standstill produces the maximum tides at high latitudes every 18.6y. The increased ebb and flow of tides coincident with obliquity enhanced summer insolation is sufficient to accelerate ablation of the ice sheets. The sea rise feedback you describe then helps complete the job.

    Following your comment I decided to plot the same thing compared also to eccentricity modulated NH insolation at 65N.

    The deepest glaciations always coincide with low eccentricity. Sea levels at LGM were 120m lower than today and Ice volume second only to the Anglian. However the next glaciation will be the most severe the world has ever seen because eccentricity reaches the lowest levels since 2.4 million years ago.

    • ralfellis says:

      As can be seen, interglacials do not follow obliquity on many occasions, like 660, 450, 370, 170, and 50 ky ago. It is clear form these missing cycles, that whether one champions obliquity or precession as the dominant orbital cycle, that a terrestrial feedback system is also required to initiate an interglacial.

      And since all of the interglacials appear to be associated with NH precessional-Milankovitch maxima, it seems clear that this feedback system has to operate only in the NH. (Otherwise, some interglacials would be triggered by a SH precessional maximum.) So it is unlikely that a global feedback like CO2 can be involved with interglacial initiation.

      What feedback systems are there, that only operate in the NH? The most obvious and likely is dust-ice-albedo. The NH contains all the major continents, and therefore all the major ice sheets, and so the albedo of the NH will change much more than the SH during the ice age cycle.

      Not only did Antarctica never lose its ice sheets during interglacials (as can be seen today in the Holocene), but it could not be darkened much by dust either. All the major dust sources, like the Gobi, are in the NH – while the SH had to depend upon the puny Atacoma desert for its dust. It might be interesting to graph the albedo changes of the NH and SH during the ice age cycle, but they will certainly be greater in the north than the south.


  7. ralfellis says:

    One possible reason for the change from obliquity interglacials before the MPT, to eccentricity-precessional interglacials afterwards, is the differential insolation absorption characteristics of bare earth and ice.

    Clive mentions here Huyber’s obliquity inspired ‘calorific summer’ (the total of energy for the whole summer) as being a significant metric. But much of this early spring and late autumn energy is delivered at a very low angle of solar elevation. This is fine for bare earth, which can absorm that low angle insolation, but ice and snow will reflect snd reject that low angle insolation. (The Sun being low on the horizon during a high-latitude spring and autumn.)

    So obliquity can warm a low albedo world, but it has great trouble warming an ice-sheet dominated high albedo world. (Even when the ice sheets are covered in dust, because the dust will sink down in the ice, and hide itself from low elevation insolation.)

    So what is required, to warm an ice-sheet dominated world, is high intensity, high angle of elevation, precessional insolation. The king of intense insolation that is delivered in midsummer, by orecession. And THAT, I believe, is why the world changed from obliquity interglacials to precessional interglacials, at the MPT boundary, some 800 ky ago. The difference was the increasing size of the northern (mainly northern) ice sheets, which passed a certain size-threshold.


  8. ralfellis says:

    The reason why the 15 ky year time-lag for dust is required in this model, is because of surface concentration of dust. What an interglacial requires, to run to completion, is for dust to continually appear on the surface, where it is concentrated and lowers the albedo of the ice even further. (Experiments have demonstrated that dust tends to stay on the surface, rather than being easily washed away.)

    In which case, we require at least 15 ky years of dust layers within the ice sheets, that can ‘rise’ to the surface as the interglacial proceeds. The length of the dust deposition era is limited by the precessional cycle. Any more than 15 ky, and we are back into the previous precessional cycle, which would have already intiated an interglacial. (The theory assumes that both dust and a high-precession Milankovitch ‘Great Summer’ are required, to trigger an interglacial).

    The only exception to the 15 ky limit, is when eccentricity, and therefore precession, are so low, that there is no strong Milankovitch ‘Great Summer’ and no strong precessional insolation, to trigger an interglacial. In which case the climate has to wait another ~22 ky, for the next Milankovitch maximum.

    See the position 270 ky ago in fig 2. Plenty of dust had already being deposited, and the conditions were right for an interglacial. But the next Milankovitch ‘Great Summer’ was so weak, than nothing happened. The world had to wait patiently for another 22 ky, when a much stronger precessional maximum triggered the interglacial.



    Note: I use these terms interchangeably…. Milankovitch maximum; Precessional maximum; Great Summer. (Although the Milankovitch-precessional maxima does include obliquity.)

    I like the term Great Summer because it is very descriptive. When the NH has its Great Summer, the SH is having its Great Winter – so the Great Summer is like an Annual Summer, except it lasts for about 22 ky.

    And this descriptive view challenges the idea that a global feedback like CO2 can greatly effect interglacial initiation. Like an annual winter, the ice age ‘winter’ is terminated by effects in the NH, rather than by global effects. (The NH is dominant in ice age modulation, because it contains the great continents and therefore the great ice sheets.)

    So what melts the annual winter snows in Canada – the summer insolation in Canada, or the ambient temperature in Argentina? Logic dictates that it is NH insolation and NH feedbacks that control interglacial initiation, and not a global feedback like CO2.


  9. Hi Clive thx for your website which seems to be the only source of understanding re global warming at the moment.

    I have a comment. Both the media and scientists focus on ice melting in the Arctic Ocean and claim that the decreased albedo will heat the earth.

    This conclusion is not at all obvious to me.

    There is a line somewhere between the equator and each pole that joins all the points on the Earth’s surface that are at the average temp of the Earth. N of this line the Earth is losing heat, S of it the Earth is gaining heat. (Focussing for a moment on the Northern Hemisphere) huge quantities of warm air and sea water are crossing this line from South to North, and huge quantities of cold air and water are moving in the opposite direction.

    Albedo effects both absorption and emission. Since emission dominates in the North, doesn’t less ice mean MORE cooling?

    Putting it another way, if you take a blanket off the (relatively warm) Arctic Ocean, doesn’t it cool the Earth?

    Of course it’s complicated. For one thing, the ocean is always frozen anyway in the winter, down to roughly Iceland I think. (The calculations depend a lot on how much water is moving around.)

    Thx, Patrick

    • Clive Best says:

      It’s true that the earth’s weather systems are driven by heat flowing from the tropics to the poles. In winter the poles receive no sunlight so albedo is irrelevant, and you are right that ice then acts more like a blanket stopping excessive radiative heat loss. In summer however the opposite is true, so less ice should mean more absorbed solar energy.

      The other factor would be a change in latitude of convective cells. This could increase/decrease meridional heat flow towards the poles. This happens naturally over 41ky with changes in Obliquity (earth’s tilt).

  10. Thx so much for your reply, but am I missing something fundamental?

    > In winter the poles receive no sunlight so albedo is irrelevant,

    Surely albedo affects both absorption and emission (equally, when you consider any particular wavelength)? Isn’t that the point of the Leslie Cube?

    All faces of the cube are at the same temp, but some absorb and emit much more.

    Whether any IR radiation from say the Arctic Ocean gets all the way out into space I don’t know. But some seems to go pretty far because on a clear (but damp) night the roof of my car gets ice on it even though the car park is at 4C. There seem to be gaps that allow IR from my car to get away – at least to higher altitudes if not all the way to space.

    > In summer however the opposite is true, so less ice should mean more absorbed solar energy.

    Yes, but we know that polar regions on the whole LOSE heat.

    Clearly this is complicated. I’m very interested in all comments because I am a biologist not a physicist.

    • Clive Best says:

      In the sense that high albedo = low IR emission then yes. However if incoming insolation = 0 then the surface will still cool but just slower when covered in Ice and snow. IR from the Arctic/Antarctica does escape to space. During the polar winter the lapse rate collapses and there is little convection, so the main way the surface cools is via IR.

      In the extreme case of Antarctica the tropopause disappears and the Stratosphere essentially touches the surface. T < -50C

  11. PS I did check that snow has a very high albedo in the IR part of the spectrum as well as the visible part.


    > Surely albedo affects both absorption and emission (equally, when you consider any particular wavelength)?

    If that were not true you could make a perpetual motion machine from a Leslie Cube (I think).

    • Frank says:

      Patrick: If you check again, I believe you will find that Ice and snow have a high emissivity for thermal IR. Therefore their absorptivity of DLR will also be high.

      The term albedo may apply only to the reflection (non-absorption) of SWR, which includes near IR. So it is possible using the term albedo may lead to some confusion. Emissivity generally means emissivity at room temperature – in other words the emission of thermal IR. So emissivity usually refers to the IR wavelengths of interest to climate science.

  12. Alan Lowey says:

    Hi Clive, I’m convinced the Moon was captured around 1mya and altered the tilting of the planet. New physics tidal forcing due to 5 exotic moonlets which orbit around Earth’s equatorial plane can account for climate cycles of 5.7yr Douglass/ENSO, 11.1yr sunspot, 88yr Gleissberg, ~210yr Suess-DeVries & ~1500+-500yr millennial cycle imo. It’s the strong gravitational interaction between the moonlet cores when on the equatorial plane which can increase solid body Earth tides. This pushes warm equatorial waters to higher latitudes, increasing precipitation which falls as snow in polar regions. This article on small changes to tilt causing termination was an inspiration

    • Alan Lowey says:

      If the 405,000-year climate cycle evident in rocks going back hundreds of millions of years is confirmed, then it suggests tidal forcing due to the Earth’s 100,000yr inclination cycle, due to exotic core interaction with the Sun.

      It’s this combination of tidal forcing, the 100kyr and separate 41kyr cycles, which gives the controversial glacial data imo.

      • Clive Best says:

        For sure there is a 400,000 year cycle. There also appears to be a 1.2 million year cycle. The inclination cycle may perhaps modulate the moons orbit around the earth. So an interesting. Someone needs to calculate if the JPL ephemeris shows a strong effect? Tidal forces are more difficult to relate directly to climate forcing.

        • Alan Lowey says:

          Thank you for the reply. The 1.2 million year cycle relates to Earth’s changing angle of tilt, so also applies to this hypothesis. There’s also a ~27 million cycle of mass extinction events related to asteroid impacts which correlates with the solar system traversing the galactic plane. This also lends itself to a strong gravitational interaction between exotic matter cores along a plane with the galactic centre being the source which solves the galaxy rotation anomaly.

          This is new physics beyond Einstein. Mercury’s anomalous precession has an alternative explanation due to an exotic matter core which has a strong gravitational interaction with the Sun’s core when the planet crosses the solar plane during it’s orbital inclination cycle.

        • Alan Lowey says:

          I find increasing tidal effects to be very easy to apply to climate change. It’s the increase in high tide and low tide which is being observed in recent times. This is attributed to sea level rise but an increase in solid body Earth tides hasn’t been considered. A satellite study could measure these to rule out the possibility I’m suggesting. It’s the difference between the UN insisting on going to carbon neutral unnecessarily and detrimentally impacting nation economies. School children needn’t be terrified of runaway greenhouse gas global warming due to Arctic methane release. The list goes on..

    • Alan Lowey says:

      Update: the captured Moon would have come with it’s own moonlets because only the 88yr and ~210yr cycles are robust in the data 98mya. It suggests Ex-V (millenial) and Ex-II (~11.1) came with the Moon and were captured by the Earth’s exotic equatorial plane. Ex-1 (5.7yr) could be missed in the data or came with the Moon and settled more quickly into the exotic equatorial plane capture.

      There’s also two main strengths to the strong gravitational interaction between exotic cores. The Earth’s inclination 100kyr cycle would be just a strong forcing due to it’s ~23.5° tilt whilst the moonlet forcing could be ultra strong due to the alignment of their core axes.

      I would guess the 88yr and ~210yr moonlet tidal forcings, assuming very low axial tilt relative to the equatorial plane, have a coefficient at least twice that of the Earth’s inclination effect.

      So prior to the Mid-Pleistocene Revolution the Earth’s inclination orbital forcing would have been moderate, giving a regular glacial cycle with thin ice sheets, and then after the capture, the Ex-V millennial cycle would have been very impactful, causing increased and irregular glaciation effects.

      • Alan Lowey says:

        Prior to the MPR, the 88yr and ~210yr moonlets would have had very low axial tilt and therefore ultra strong gravitational forcing, giving rise to the dominant 41,000yr tilt cycle. After MPR and Moon capture these two exotic moonlets would have been disrupted, reducing their forcing, giving rise to the dominant 100kyr inclination and new millennial forcing.

        • Alan Lowey says:

          A simple test of the hypothesis is to see whether the 88yr and ~220yr tidal forcing cycles exist in the oxygen isotope data prior to 1 million years ago:

          • Alan Lowey says:

            “We have run spectral analyses of sediments from the late Cretaceous, which show significant 88 and 210 year periodicities. The debate is what is the nature of these periodicities – particularly as this is in the Cretaceous (98Mya) and not recent past!”

Leave a Reply