Making Sense of the Mid-Pleistocene Transition (MPT)

I recently read a fascinating study looking to understand why the “Mid-Pleistocene Transition” (MPT) happened about 1 million years ago. The MPT refers to the change where the glacial-interglacial cycles of the Pleistocene (the last 2.6 million years) changed from cycles of ~41,000 years to ~100,000 years. Both periodicities are synced to Milankovitch Cycles. During the early Pleistocene, the periodicity matches cycles in the earth’s obliquity, but after the MPT the glacial-interglacial cycles began to track more closely with variations in the earth’s orbital eccentricity. So the big question has been why the change?

A recent paper by Willeit published in Science Advances[1] examined this by producing a model that was able to reproduce the MPT. In fact it accurately reproduced the maximum extent of the ice sheets as well as extent of sea level rise. It made use of the carbon cycle, volcanic activity, and changes in regolith cover. The model results indicate that continents had built up regolith in Northern Hemisphere for millions of years. During the early Pleistocene the successive advances of glaciers cleared out the regolith on the continents of the Northern Hemisphere, and this was the most significant factor in causing the MPT.

Willeit’s study concludes that the earth’s system is sensitive to relatively small variations in CO2 (the model uses a climate sensitivity of around 3°C),[2] and the progressive removal of regolith exposed bedrock, which impeded movement of glaciers. And the pulverized bedrock produced dust that eventually settled on the glaciers, reducing their albedo and increasing melt rates. The gradual increase in exposed bedrock made for more stable glaciers, and consequently glacial cycles shifted to the eccentricity cycles about 1 million years ago. “Our transient modeling results demonstrate that both previously proposed mechanisms—regolith removal and gradual lowering of CO2—are essential to reproduce the realistic evolution of climate variability during the Quaternary, and their combination controls the timing of regime changes of climate variability. Note that a gradual change of the regolith cover causes a rather rapid (few hundred thousand years) transition from the 41- to 100-ka world, in good agreement with observational data.”

The ice core record only goes back about 800,000 years, but scientists have reconstructed the variations in global average temperatures throughout the glacial and interglacial cycles. Friedrich et al 2016[3] examined the last 784,000 years and came up with two independent reconstructions of surface air temperature (SAT) variations during the glacial-interglacial cycles in good agreement with each other in timing and amplitude of SATs.

Friedrich et al 2016 also estimates the forcings due to orbital variability and sea level as well as CO2, dust-albedo and ice-albedo feedbacks. The authors were then able to calculate sensitivity (S), recognizing that sensitivity is different for warm and cold climates. “The resulting mean of S for cold climates (Scold) amounts to 0.48 K W−1m2, which corresponds to 1.78 K per CO2 doubling. For warm climates, the value (Swarm)is more than two times larger, attaining 1.32 K W−1m2 or 4.88 K per CO2 doubling. The average of S over the entire 784-ka range can be calculated from a linear regression of the SAT/radiative forcing dataset. It amounts to 3.22K per CO2 doubling.” Since warmer climates have higher sensitivity than colder climates, the long-term average for sensitivity would underestimate projections for future warming. Nevertheless, estimates for climate sensitivity continue to average around 3 k per CO2 doubling.

When I initially read that Swarm is greater than Scold, I immediately began to wonder, if sensitivity increases with temperature, would that lead to a runaway feedback loop? The paper never answers that question. But my thinking on this is that there has to be an endpoint to this if climate is going to stabilize at some equilibrium. The most likely candidate for this is is albedo. As a thought experiment, let's start with a snowball earth - ice covering the entire planet to the equator. If a warming signal occurs, the albedo feedback should remain essentially at 0 until the planet warms enough for open water to appear at the equator. At that point, more heat is absorbed and less reflected, so sensitivity should increase. As the planet continues to warm, ice retreats and sensitivity should increase with it until most glacial ice disappears. As the surface area of ice approaches 0, the albedo feedback from ice must also approach 0, and sensitivity should decrease. There may also be affects from the amount of dust that can land on the surface of ice (reducing its albedo) that could also affect sensitivity. So my assessment on this is that while Scold is smaller than Swarm, Swarm should be much larger than Shot (where Shot is sensitivity with no ice sheets).

Since climate change discussions rarely deal with the effects of AGW significantly beyond 2100, we often refer to ECS as a constant, and I suspect that entirely right to do, given the time scales involved (decades to hundreds of years). But on time scales of millions of years, it may be more valuable to discuss variability in sensitivity. I've seen other studies investigating this, and in a future post I may address the studies I've found for further consideration.


References:

[1] M. Willeit, A. Ganopolski, R. Calov, and V. Brovkin, "Mid-Pleistocene transition in glacial cycles explained by declining CO2 and regolith removal", Science Advances, vol. 5, pp. eaav7337, 2019. http://dx.doi.org/10.1126/sciadv.aav7337
http://advances.sciencemag.org/content/5/4/eaav7337

[2] M. Willeit, “First successful model simulation of the past 3 million years of climate change” Realclimate, 2 April 2019. http://www.realclimate.org/index.php/archives/2019/04/first-successful-model-simulation-of-the-past-3-million-years-of-climate-change/

[3] Friedrich et al, "Nonlinear climate sensitivity and its implications for future greenhouse warming," Sci. Adv. 2.11 (2016): e1501923.
https://www.researchgate.net/publication/309791338_Nonlinear_climate_sensitivity_and_its_implications_for_future_greenhouse_warming

[4] Sun, Y., Yin, Q., Crucifix, M. et al. Diverse manifestations of the mid-Pleistocene climate transition. Nat Commun 10, 352 (2019). https://doi.org/10.1038/s41467-018-08257-9
https://www.nature.com/articles/s41467-018-08257-9