Milankovitch Cycles
Posted on 22 July 2011 by Chris Colose
This post is intended to serve as a supplement to SteveBrown’s series on the Last Interglacial, beginning here.
Changes in the Earth's orbit brought about by astronomical variations have a strong impact on Earth’s climate. They serve as the pacemaker for the glacial-interglacial cycles over the Quaternary (roughly the last two and a half million years of Earth's history), and provide a strong framework for understanding the evolution of the climate even over the Holocene (the last 10,000 years, beginiing near the termination of the last glacial period). Milankovitch cycles are insufficient to explain the full range of Quaternary climate change, which also requires greenhouse gas and albedo variations, but they are a primary forcing that must be accounted for.
Orbital variations are also likely to be a generic feature of other planets, with strong implications for the fate of planetary atmospheres (for example, understanding the potential for habitability on other systems). This post will serve as a guide to what these so-called Milankovitch cycles are, how they work, and highlight some "to-be-done" work that remains.
Milankovitch cycles are classically divided into the precession, the obliquity, and the eccentricity cycles. These cycles modulate the solar insolation (i.e., the total energy the planet receives from the sun at the top of the atmosphere) or its geographic distribution. For example, figure 1 shows the solar insolation change at various latitudes in June over the last one million years.

Figure 1: June (daily averaged) insolation (W/m2) over the last 1,000,000 years (0=1950) at (blue= 90 N), (red = 60N), (green =30N), (purple=Equator), (light blue = 30S), (Orange=60S). Data from Berger A. and Loutre M.F., 1991, Insolation values for the climate of the last 10 million years, Quaternary Sciences Review, Vol. 10 No. 4, pp. 297-317
Each of the relevant Milankovitch cycles are described below:
Eccentricity: Eccentricity is a measure of how circular a curve is, with e=0 describing a circle, and e=1 describing a parabola. The orbital eccentricity therefore characterizes how circular or egg-shaped a planet’s orbit around the sun is (Fig. 2). The timescale of Earth’s eccentricity variation is ~400,000 years with a superimposed 100,000 year cycle. There is also an unimportant 2.1 million year cycle.

Figure 2: Circular and eccentric orbit.
Because of eccentricity, the distance of the Earth at perihelion (point closest to sun) is slightly different than the distance to aphelion (point farthest from sun).
Earth’s eccentricity is very moderate, never exceeding approximately 0.07 (almost a perfect circle). The modern day eccentricity is 0.016, and as a result, the solar insolation that hits Earth varies by ~6.4% over the course of a year. There are some more extreme examples: Pluto’s eccentricity is about 0.25, higher than any other planet in our solar system. HD 20782b, a newly discovered exo-planet almost 120 light years away has an eccentricity on the extreme end of ~0.97 (similar to Halley's Comet). Eccentricity can introduce very large "distance seasons" on a planet, although this also depends on the thermal inertia, which is large enough on a body with oceans (or a dense atmosphere) to moderate the changes between perihelion and aphelion. As we will see, Earth's seasonal variations are primarily deterimined by its axial tilt rather than its eccentricity.
Eccentricity is the only Milankovitch cycle that alters the annual-mean global solar insolation (i.e., the total energy the planet receives from the sun at the top of the atmosphere). For the mathematically inclined, the annually-averaged insolation changes in proportion to 1/(1-e2)0.5, so the solar insolation increases with higher eccentricity. This is a very small effect though, amounting to less than 0.2% change in solar insolation, equivalent to a radiative forcing of ~0.45 W/m2 (assuming present-day albedo). This is much less than the total anthropogenic forcing over the 20th century. However, eccentircity does modulate the precessional cycle, as we shall see.
Obliquity: Obliquity describes how tilted a planet’s axis is (Fig. 3 shows the obliquity of eight planets, plus Pluto which is labeled as a planet). The tilt of the Earth is ultimately what allows for the existence of seasons, since the Northern Hemisphere is pointed toward the sun in the boreal summer, and away from the sun in the boreal winter. The tilt of Earth’s axis (in other words, the angle between the spin–axis and a line perpendicular to the orbital plane) varies between about 22 to 25° (currently 23.5°) over a period of nearly 41,000 years, driving changes in the distribution of sunlight between the equator and high latitudes (with more tilt implying more sunlight at high latitudes, and less at lower latitudes; therefore more oblique orbits favor deglaciation on Earth).
Uranus is at the extreme end with a tilt of ~98 degrees; this would induce a very different structure of solar heating (where at certain times the North or South pole would be receiving most of the sunlight, and allow for a large migration of the solar “hotspot” over the course of one Uranian year); this should drive a different atmospheric circulation than on Earth. For highly oblique planets outside our solar system that have a surface, continents at the Polar Regions would be alternatively cooked and frozen, while the tropical latitudes would have two summers and two winters. At large obliquities (greater than about 54 degrees), the poles receive more annual-mean insolation than the planetary equator (Ward, 1974), and thus the annual-mean energy transport by the circulation would be equatorward.
Figure 3: Obliquity of the Planets, and the direction of spin (note that Venus rotates clockwise, in a retrograde fashion). Taken from www.solarviews.com. Click image to Enlarge
Precession: Precession does not describe how tilted the Earth’s axis is, but rather the direction of its axis. This changes what star is the “North Star” over time (today it is Polaris, but near the end of the last deglaciation it was Vega), and as described below, governs the timing of the seasons. This is illustrated in Figure 4 (a and b) below.


Figure 4: a. Schematic showing the influence of axial precession b. How precession changes the timing of the seasons (4b taken from Ruddiman: Earth's climate Past and Present)
There are two key precession effects: axial precession (see an animation here), in which the torque of the other planets exerted on the Earth's equatorial bulge forces the rotational axis to “wobble” like a spinning top; there is also an elliptical precession, in which the ellipse of the Earth itself rotates about one focus. This means that the elliptical shape of Earth's orbit rotates with the long and short axes of the ellipse rotating slowly in space. This moves the positions of aphelion and perihelion.

Figure 5: Change in the Earth's orbital plane. Even if the spin axis always pointed in the same direction (for example, on a perfectly spherical planet) it it would make a different angle with its orbital plane as the plane moved around
Precession also means that the solstices and equinoxes have changed positions, both with respect to the eccentric orbit, and with respect to the positions of perihelion and aphelion. Today, the positions of the solstices and aphelion/perihelion line up closely (Fig. 6), and the Northern Hemispheric summer occurs when the Earth is further from the sun, but this is not always the case.
The summer and winter solstices mark the longest and shortest days of the year, and we see the sun moving back and forth throughout the year between the Tropic of Cancer (23.5 N) and Capricorn (23.5 S), which is a reuslt of the revolution around the sun with a 23.5 degree tilt. Note also that the 90-23.5=66.5 degree mark defines the Arctic and Antarctic circles. At the shortest winter day (winter solstice), no sunlight reaches latitudes higher than this. The equinoxes mark the point where the length of day equals the length of night.

Figure 6: Positions of the Equinoxes, Solstices, Perihelion, and Aphelion. From Ruddimans Earths Climate: Past and Future
Today, the Northern Hemisphere winter occurs near Perihelion, and NH summer occurs when the Earth is farthest from the sun. At present, the Southern Hemisphere has a tendency toward hotter summers , and with a more moderate seasonal cycle in the North, although that simple idea is complicated by differences in land distribution and thermal inertia between hemispheres. However, about 13,000 years ago, the Northern Hemisphere summer would occur when the Earth is closest to the sun, and NH winter when it is furthest from the sun (Figure 4). This would enhance the strength of the seasonal cycle.
Precession varies on timescales of 19,000 and 23,000 years, and is thus important even over historical times. The precessional cycle is the key player behind the Holocene Climate Optimum, a time between ~7,000 and 5,000 years ago of particularly warm Northern Hemispheric extratropical summers, and colder tropical and extratropical winters.
It is important to note that under the precessional cycle, the change in solar radiation striking the Earth is opposite in each hemisphere, unlike the case of obliquity where a higher tilt will mean more intense radiation at both poles as the planet revolves around the sun (although, obviously at the local summer summer for both poles, and thus at different points in the orbit). Furthermore, eccentricity modulates the effect of precession. For zero eccentricity, the precession angle is irrelevant.
What Causes Milankovitch Cycles?
The changes in eccentricity of Earth’s orbit are due to alterations in the gravitational tugs induced by other planets. Jupiter has a very moderate eccentricity, but if it were larger, it would drive larger changes in Earth’s eccentricity. It therefore seems likely that exotic cases of highly eccentric orbits may be prominent in other solar systems, where various gaseous planets are known to exhibit large orbital fluctuations.
Obliquity and precession variations arise due to the torque exerted by gravity (i.e., a force that acts perpendicular to he spin axis of the top) which ultimately comes from the pull of the Sun and Moon on Earth’s equatorial bulge. Precession also varies due to the tilting of the Earth’s orbital plane, as shown above.
The periodicity of Milankovitch cycles is therefore subject to change over geologic time, as the length of day of Earth changes, and the moon becomes further separated from Earth. A shorter Earth-Moon distance would cause the precessional movement to have been larger and the precession and obliquity cycles would have been shorter, as would have occurred in geologically distant paleoclimates. For example, in the Upper Carboniferous (~300 million years ago), the ~41,000 yr obliquity cycle would have taken about 33,000 years (see e.g., here)
Milankovitch Cycles Beyond Earth
Milankovitch cycles are not unique to Earth, nor are the solar system’s orbital characteristics fixed in time. Even our own solar system may be unstable on timescales comparable to its age. In the inner Solar system, the planets' eccentricities exhibit chaos on billion-year timescales (Fig. 7). The lighter planets (Mercury and Mars) have the potential for large variations and in fact, it has been calculated that Mercury has ~1% chance of colliding with Venus or the Sun (or being ejected from the solar system) within the next five billion years (Laskar, 1994, Laskar & Gastineau, 2009).

Figure 7: Numerical Integration describing orbital paramters (10 Byr backward, note this is older than the age of these planets, and 15 Byr forwards). The larger planets behave more regularly. Based on J. Laskar, A&A 287, L9 (1994)
Mars has an obliquity that can vary chaotically between ~0-60°, which has severe implications for its climate evolution. Milankovitch cycles on Mars can actually play a role in redistributing ice on a global scale. In particular, it is thought that deposits of large amounts of water ice recently found in certain areas of the mid-latitudes of Mars (e.g., Holt et al., 2008) must have formed at a time when the climate was conducive to glaciation at middle latitudes, as there is no precipitation in these regions today. This probably requires a higher obliquity, a greater amount of sunlight at the poles, driving sublimation and vapor transport equatorward, where it can then be deposited at lower latitudes (Forget et al 2006). Earth can actually attribute its relatively mild variations in tilt to the stabilizing influence of the moon( Laskar and Robutel, 1993). It would also be possible to have higher obliquity variations if Jupiter were closer to Earth, even with a moon.
Figure 8, below, shows the relatively recent obliquity variations on Mars.

Figure 8: Recent obliquity variation on Mars (-20 Myr to 10 Myr). See Laskar et al (2004)
We don't need to think inside just this solar system though. The influence of exotic spin states or eccentricities is a rather hot topic in the planetary climate community right now (e.g., Spiegel et al., 2010), as new plants beyond our solar system continue to be discovered. Some open questions for example involve the ability to swing in and out of a Snowball planet (or a runaway greenhouse) at highly eccentric orbits. Can planets that undergo large variations in the axial tilt remain habitable? Can planets in binary (two-star) systems be stable? Some of these issues have been explored briefly, but with over 500 planets now discovered outside our solar system and many more expected to come, there's a good amount of work that needs to be done here. Earth, today, is stable in both its modern configuration or in a cold "snowball" configuration (i.e., if Earth were magically ice-covered, it would stay there, even keeping the solar constant and CO2 the same, due to the ice-albedo and water vapor feedback) (Figure 9).

Fig. 9: Bifurcation diagram of Temperature (purple curve) vs. Solar Insolation (blue curve). Because of the ice-albedo feedback, the equilibrium is stable at several points.
Figure 9 cuts into the heart of various planetary climate "extreme" problems. For a relatively circular orbit, the problem of determining where Earth falls into and out of a Snowball is challenging. What happens though if you make the planet slowly rotating? What if the eccentricity is very high, so it the planet swings in and out of the "habitable zone" over the course of one planetary year? Milankovich cycles have the potential to make this issue a lot more interesting, although it is not a solved problem.
Some Final Words
There are still a number of unresolved questions that remain in the astronomical theory of climate change, even during the more familiar Quaternary timeframe. For instance, while we know changes in the orbit pace ice ages, the precise way the three Milankovitch variations conspire to regulate the timing of glacial-interglacial cycles is not well known.
For example, about 800,000 years ago a shift of the dominant periodicity from a 41,000 yr to 100,000 yr signal in glacial oscillations occurred (called the Mid-Pleistocene Transition, see e.g., Clark et al., 2006), and while a lot of ideas exist for why this should be the case, there's no bulletproof answer to this. Explaining the 100,000 yr recurrence period of ice ages is difficult because although the 100,000 yr cycle dominates the ice-volume record, it is small in the insolation spectrum. Therefore, there's still a lot to be done here.
It seems that the Earth listens to the Northern Hemisphere when deciding to have an ice age. If the North and South are alternatively near and far from the Sun during summer, why has glaciation been globally synchronous? What connections are there between Northern insolation and Antarctic climate at the obliquity and precession timescales? What are the competitive roles between a further distance from the sun during summer and a longer summer, following Kepler's law? These quesrions are still not resolved (for a flavor of the discussion, see Huybers, 2009...see also Kawamura et al 2007; Huybers and Denton, 2008; Cheng et al 2009; Denton et al 2010 ). This problem also involves work at the interface of carbon cycle and ice sheet dynamics, processes that are in their infancy in terms of modeling.
Acknowledgments: I'd like to thank fellow SkS contributor "jg" for terrific work in piecing together various visuals used in this post.
Recommended Reading: I also recommend Tamino's multi-part series on Milankovitch cycles (the rest of the posts are linked at the bottom). Involves some math, but a good read.
Arguments






























I couldn't find the one he shows, but here's another from the Petit 1999 Nature paper:
The frequency spectrum of the temperature proxies (particularly delta-18 O - bottom left) shows peaks at the periods of the Milankovitch cycles. That makes the connection between the Milankovitch cycles and the glacial cycle, despite the fact the insolation looks nothing like the glacial cycle. The lack of similarity is the clue that the connection is more subtle.
The thing I'm having trouble with: from wikipedia: "But the mean solar irradiation for the planet changes only slightly for small eccentricity, due to Kepler's second law."
So, that's the mean solar irradiation averaged over the whole year that's staying the same? (Because the Earth will be faster when closer to the sun?) Meaning, the irradiance *is* affected by distance at any given moment?
There's two things here: the change in the solar radiation at different points in the orbit (e.g., between aphelion and perihelion) and the mean annual change.
With the former, the radiation changes as the inverse square of the distance. So, if you put Earth twice as far from the sun, it would receive four times less sunlight. For an eccentric orbit, the Earth will receive more sunlight when it is a bit closer to the sun than when it is a bit further, which is very intuitive.
For the annual-mean solar radiation however, the change that occurs between a circular or a more eccentric orbit is fairly small.
Adding to your answer to Dan's question, the changes in solar radiation occur between perihelion and aphelion, so that ~6.4% more solar insolation reaches Earth in January compared to July. When averaged over the course of the year, the change in solar insolation between high and low eccentricity is small.
Currently, the highest solar insolation reaches the Earth during the NH winter. In about 11,500, that will occur during the SH winter. While the amount of radiation reaching the Earth will not change, the amount absorbed will, due to the different albedo of the NH and SH. Many have speculation that this is the cause for the changes in temperature that occur between interglacials.
That star will be visible in NYC again around the year 10,000, at this time Deneb will be the 'pole star' in Cygnus. The March Equinox shifts through all the zodiacal constellations- now in Pisces- it will shift into Aquarius in about 600 years. In the year 3,000 Gamma Cephei will be the pole star.
My curiosities include:
- Intersections presumably represent energy balance?
- The unstable equilibrium is interesting. Increasing insolation there decreases temperature?
- Is there any way to tell from the diagram alone which intersections are stable and unstable?
- How do the curves vary over time? Clearly changes in insolation shift the straight line left and right. Then forced CO2 changes change the shape of the curved line? Over what timescales does the curve change?
- How does this figure relate to the glacial cycle?
... In a glacial, does the earth slip down to the unstable equilibrium, and then bounce back up because it is unstable (the temp change looks too big though)?
... Or do glacials correspond to the curve moving (seems more likely)?
... Or is the glacial/interglacial cycle just oscillation around the stable equilibrium? (And the increasing amplitude over the past million years represents a steepening of the curve?)
You are right that the intersections are energy balance points, where the solar flux equals the outgoing radiation. Having a temperature-dependent albedo is one way to get the structure in Fig. 9. I just said that the global albedo can range between 0.2 and 0.6 between two specified temperatures (lower albedo in the warm case, higher in a cold case) and parametrized the OLR to give a simple greenhouse effect, then made the plot. You can also make a plot of temperature vs. CO2 (instead of solar insolation).
Obviously this bifurcation structure in this post is a very "theoretical" and simplified one, so I wouldn't take it too seriously when trying to interpret glacial/interglacial changes, etc; in the real world there might be many stable equilibrium points, or it might take a sizable forcing to push away the climate from an unstable point.
In this plot, the limits of ~1300 W/m2 and ~2000 W/m2 are transition regions, so for example, if you start off in a warm climate and then gradually lower the solar constant, the climate cools smoothly; once you cross the ~1300 W/m2 mark, the snowball is initiated and you descend abruptly into the cold solution regime. Because of the high albedo, it now takes a higher solar constant (~2000 W/m2) than the original value to return back to the initial state. Actually getting out a snowball in the real world is still a pretty unresolved problem, but it probably takes a very large amount of CO2, as we see in the geological record for the Neoproterozoic glaciation. And, if it's too cold, excess greenhouse gases will just condense out on the surface, so it's not obvious that many planets at a distant orbit (or the outer edge of the "habitable zone") can even get out of a snowball, at least until the star continues to get brighter.
A lot of papers about snowball Earth discuss this, and there's a detailed treatment in Ray Pierrehumbert (and others) recent Neoproterozoic review paper that you can get from his web page (his textbook does as well, and so does Dennis Hartmann's in the ice-albedo feedback discussion). You can also talk about it in connection to a runaway greenhouse, or perhaps even smaller-scale phenomena like abrupt climate change, but there's debate as to whether this is an artifact of simpler models for many processes relevant to the real world. In this case, the key point is that the climate can equilibriate at multiple temperature solutions, and where it actually is depends on the history it took to get there.
The stability criteria is equivalent to stating that the slope of the absorbed solar curve is less than the OLR curve at the intersection point, but I would read these works cited above if you want a general overview of the mathematics or more detailed treatments.
I'll look into Ray's paper as soon as I can.
Does anyone have an answer to this question?
The match in climate changes on both hemispheres indicates that there are other forces in play as well.
I understand there are other forcings at work. But for the Milankovitch theory to hold up, the match between the hemmispheres should not be so.
I am working from memory at this point.
One thing that always stuck to me was also called the Stage 5 problem.
I will dig for the Northern/Southern thing. I thought this was common knowledge to anyone who has reserached the Milankovitch cycle.
"It seems that the Earth listens to the Northern Hemisphere when deciding to have an ice age. If the North and South are alternatively near and far from the Sun during summer, why has glaciation been globally synchronous?"
This question was asked in the article, so I won't dig for references as the article states this as well.
That is why I posed the question as there is really no known physical mechanism that would explain this phenomina which indiates we are missing something in the cause/effect relationship.
This is just a quick guess on my part. Admittedly I'd have to do some research to justify the claim.
As good a guess as any. I have read so many theories on this, with none of them proveable.
I know this article works on TSI basically because of the orbital changes. One thing this article does not mention, which I think at least bears thinking, is what effect does the magnetic field have? As we orbit with the Mil cycle, we are also doing a baycenter orbit.
Then there is the stage 5 question when the cycle does not match the timing.
I think the theory is interesting, but the lack of matching further back historically causes one to ponder the actual cause/effect question.
The good thing is it causes one to think.
On the 100k problem with milankovitch, I do agree that we have too many theories and not enough data to test them. Not much of relevance to current climate change however.
Eric... Same question.
The timing correlates fine. It is the rapid warming that has me puzzled. One would think that since the changes in eccentricity occur slowly, that the temperature would follow at a similar pace. Yet, that does not appear to be the case. At least, according to the ice core data.
This warming is enough to cause a slow (meaning a lot slower than what we're doing to the Arctic) advance or retreat of northern hemisphere snow cover. Because of the amount of land in the NH, this results in a substantial change in albedo, which of course drops temperatures, and advances/retreats snow cover further.
The drop in albedo further results in other feedbacks, primarily CO2, through things such as vegetation changes and ocean temperature changes. These, of course, evoke further feedbacks, as is well described by current climate science literature.
The fact that changes in TSI are so minimal, and yet the glacials/interglacials occur, is an important clue that climate sensitivity is high.
Ultimately, these effects all combine enough to cause the level of climate change required.
The main problem I've seen in the literature is in trying to identify the cause/mechanism behind what appears to be an abrupt release of CO2 (which is both detected in ice core measurements, and also required for the degree of climate change seen) early in the termination of a glacial period.
There's a lot of literature to be found just by searching for "CO2 glacial termination."
[DB] Fixed original text.
I have followed skepticalscience for a while without participating in the discussions, and my only contribution so far is the Norwegian translation of The Big Picture.
Regarding how the Milankovitch cycles impact the solar insolation globally, I found an interesting graph on page 18 in the paper Target CO2 by James Hansen & Co.
As you see, the difference between maximum and minimum global insolation during the last 400,000 years is only about 0.5 Watts/m2 with present day geographical and seasonal distribution of albedo. Assuming a slow-feedback climate sensitivity of 1.5 C / W/m2 gives a total climate change of only 0.75 C, much less than the 5 C difference between LGM and the Holocene. If the albedo was evenly distributed globally and not higher in the northern hemispere, the Milankovitch cycles wouldn’t be able to change the global average insolation at all.
The reason why northern hemisphere is controlling the glacial cycles seems obvious to me:
It’s not just because Arctic sea ice responds much faster than the Antarctic ice sheet, but because NH hemisphere has much larger areas that undergo positive albedo feedbacks during warming or cooling.
Just compare the huge changes in NH ice sheets between LGM (south to about 40 N in North America and 50 N in Eurasia) and today (Greenland) and the much smaller changes in the geographically isolated Antarctica.
Then we have the changes in the boreal forests known as the "taiga". During the ice ages the taiga wasn’t just pushed southward and replaced by ice sheets and tundra. The ice age climate was generally dryer than today, so much of the taiga (and also temperate forests) was replaced by open grassland (steppe, prairie etc) which has considerably higher albedo than the dark green coniferous forests that make up most of the taiga. The SH of course has nothing similar to this.
And finally we have the snow cover in temperate latitudes in winter. That snow cover would also expand southward and increase the NH albedo even more. Some may argue that this albedo increase wouldn’t matter much because it happens in winter when insolation is low, but at 40 N in late winter the solar elevation angle at noon is about the same as in northern Norway in June.
The areas in SH with a cold temperate climate (snow in winter) is negligible compared to NH and would expand far less during an ice age than in the NH.
So, the albedo feedback is much stronger in the north than in the south because of the distribution of the continents, and as Sfaerica mentioned above, this will have an effect on levels of CO2 and other GHG’s which will spread the warming (or cooling) to the southern hemispere.
The problem with this is that glaciation is mutual to both hemispheres. That fact shows the albedo etc links do not function.
This is one of the puzzling things about the cycle amongst many others.
[DB] The fallacy you fall into is the expectation that hemispherical glaciation is symmetrical, which it is not. Consider the hemispherical distribution of land vs water to gain a sense of the size of the mismatch (and the subsequent relative changes in albedo between glaciation and interglacial).
So your "fact" is completely NOT one.
As Dan and Dikran have pointed out... I don't see how your comment makes any sense. The two halves of the planet are quite obviously not separated by a giant, plexiglass divider. If a massive change in albedo in the northern hemisphere due to expanding or retreating year-round snow/ice cover reflects or absorbs a notably different amount of incoming solar radiation, then that is energy the planet loses/gains, period, year after year, for millenia.
Once that starts, the feedbacks start. That the change is initiated in one hemisphere only seems entirely inconsequential. The global temperature will rise/fall as a result.
And BTW, one of the Milankovitch cycles, the obliquity or axial tilt, will affect both hemispheres in the same way as both poles get more insolation and the tropics get less.
Starting at the bottom of a glacial.
Milankovitch triggers some warming.
Sea ice responds faster than land ice so more warming.
Oceans respond with CO2 outgassing.
Slowly the land starts to respond which is primarily northern hemisphere.
Moderate ice sheet retreat starts to produce more melting permafrost, more methane.
Later in the cycle, land ice sheets start to respond - it takes a fair old while to remove ice sheets kilometers thick. So albedo change and more warming, but only late in the cycle.
Also, as ice sheets retreat polewards, geography and trigonometry come into play - more change early, declining change later.
As ice sheets retreat they leave bare rock. Over 100's and 1000's of years this bare rock is slowly converted to soil and biomass. A carbon store.
Then Milankovitch starts to tip things the other way. A bit of cooling. First impact of cooling is to extend snow ranges towards the equator. A snowfall can have as much effect on albedo as a huge ice sheet, but it can happen much quicker. so albedo change can have a bigger impact earlier in the cooling phase.
Cooling triggers CO2 uptake by the oceans. More cooling.
However, all that new biomass on the colonised rock is still there. How long does it take to kill off those recent forests and return their CO2 to the oceans or the atmosphere. In the intervening period, will they hold CO2 levels up even as Milankovitch and Albedo are cooling things.
This surely makes for a nice complicated cycle over 1 glacial period.
And all of this is probably due to the assymetric configuration of the continents between morth & south. If the Earth's continents were arranged symetrically, we may not see ice ages.
And past arrangements of the continents probably had a huge impact on the Earth's predisposition to Ice Age behaviour or not.
Further, my understanding is that the sparse land masses north of 60 degrees South in the SH prevent the build up of continental ice sheets, and hence prevent the lapse into an ice age. On that basis I would suggest that if the SH had a similar land distribution to the NH, that would lock the Earth into permanent glacial conditions rather than prevent glacial conditions from forming.
This is a negative feedback (removing carbon), but the vegetative changes have a more direct effect of decreasing albedo especially with forest expanding into tundra, which is a quite rapid change in this context. And of course this change is much more pronounced in the NH cf to the south.
WRT early snow melt yes, that would occur during the early stage of the warming. But then the feedback available from albedo change would slow as the system waits for the ice sheets to recede.
In contrast, increased snow range during the cooling phase is a factor that would cut in early but then not suffer a later lag since snow can just keep extending without needing ice sheets to form first. So Albedo change due to snow/ice is likely to be a stronger early driver during the cooling phase but more of a lagging driver during the warming phase.
As for exposed rock & vegetation, the initial positive feedback is strong due to the rock being exposed. Later - perhaps 1000's of years, the albedo effect from this is somewhat reversed as vegetation replaces rock. Also at this point there is a negative feedback, opposing Oceanic CO2 outgassing as this new biomass soaks up CO2.
Then during the cooling phase, it is unlikely to be the exact reverse. Snow extends, killing off vegetation but without an intermediate bare rock phase. Also, die back of all that dying vegetation is then a carbon source, opposing the uptake of CO2 by cooling oceans.
So compare these patterns to the glacial cycles. Early warming - Milankovitch, CO2 (and perhaps Methane from melting permafrost) and early snow melt. Mid to late stage warming - ice retreat and exposed rock. Late stage warming, more ice retreat but diminishing impact as spherical geometry cuts in and slowdown of CO2 increase due to its logarithmic forcing and growth of vegetative biomass.
Then early cooling - Milankovitch, more snow. Mid cooling -even more snow, spherical geometry startiong to make this more effective, cooling oceans so CO2 drawdown but also compensating CO2 release from dying vegetation. Late stage cooling - Biomass has returned to previous levels, CO2 drawdown increases so CO2 starts to become a more potent late stage driver.
So simplistically, CO2 as larger early stage driver during warming with Albedo change cutting in more strongly at the end. And Albedo change is the stronger driver during early cooling with CO2 coming in more strongly at the end.
And this matches the Ice Cores. During warming CO2 is closely coupled to temps. During cooling, particularly the 1st half, much less so.
This would only be a general picture since there are a lot of other wrinkles - how and how fast ocean currents change for example. And no two glacial cycles from the cores are exactly the same. But within the limits of what we can quantify the ice cores certainly seem to behave substantially as we might expect.
Another minor point is that rock generally has a higher albedo than vegetation. The vegetation dissipates the energy differently in that about 10% of it is stored as chemical energy and then dissipated by the plant or animals later on, but the total energy absorbed by the surface is still greater.