Ice age triggers

Why can a slight wobble in Earth's orbit swing the whole planet into an ice age?

The Earth's orbit is not perfectly steady. Its shape stretches and relaxes, its axis tilts a little more and a little less, and the direction that axis points swings slowly round. These are small adjustments — the total sunlight the planet receives in a year barely changes. Yet the marine sediment record carries the same rhythms all through the last few million years, with kilometres of ice arriving and leaving in time with them. How does a nudge that alters the annual energy budget by a fraction of a percent move that much ice?

Start by being precise about what the orbit changes. Eccentricity, the stretch of the orbit, varies on a roughly 100,000-year beat. Obliquity, the tilt of the axis, swings between about 22.1 and 24.5 degrees every 41,000 years. Precession, the wobble of the axis direction, sets which season the Earth is closest to the sun in, on a roughly 23,000-year beat.

Now notice something about that list. Only obliquity changes the annual mean sunlight reaching high latitudes by much. Eccentricity's effect on the yearly total is tiny, of order a tenth of a percent, and precession changes the annual total essentially not at all — it moves energy between the seasons and the hemispheres, then takes it back.

So the quantity that matters cannot be the annual total. What else is on offer? Milutin Milankovitch's insight was to ask not how much energy arrives but when: specifically, how much summer sunlight falls near 65 degrees north, where the continents that can hold ice sheets sit. Winter snow falls almost regardless; what decides whether an ice sheet grows is whether the summer is cool enough to leave some of it behind. That is a seasonal, hemispheric quantity, and precession and obliquity move it a great deal even when they barely touch the annual total.

That converts a small forcing into a larger one, but not a large enough one. So what amplifies it? Two feedbacks do most of the work. Snow and ice are bright: where they persist they reflect sunlight that bare ground or sea would have absorbed, so the region cools further and more snow survives the next summer. That ice-albedo feedback turns a marginal survival into a growing sheet. Second, carbon dioxide falls as the world cools, from roughly 280 parts per million in warm intervals to about 180-190 at the depth of a glaciation — colder, more stratified oceans hold more carbon, though which mechanism dominates is still argued. Lower carbon dioxide cools the whole planet, including the southern hemisphere, which the orbital forcing by itself would have pushed the other way.

Now the honest part, because the theory has real trouble. Across the last 800,000 years the dominant rhythm is the roughly 100,000-year one — and eccentricity, which carries that period, is by a wide margin the weakest of the three forcings. The strongest signal in the data is paced by the feeblest push. That is the 100,000-year problem, and it is unsolved. Worse, the rhythm has not been constant: before about one million years ago the record cycled at 41,000 years, matching obliquity, then switched. The orbit did nothing to warrant that switch, and this mid-Pleistocene transition remains unexplained, though the leading suggestions involve the ice sheets themselves changing — thicker sheets once glacial erosion stripped soft sediment from their beds, or a system that skips one or two obliquity beats before collapsing.

The refinement that connects everything is the word pacemaker, worth taking literally — it is the title of the 1976 paper by Hays, Imbrie and Shackleton that found the orbital frequencies sitting in deep-sea cores and made the case stick. Orbital changes do not cause ice ages the way a match causes a fire. They set the timing of a system with its own large internal responses, and those responses supply most of the amplitude.

That reframing also disposes of a common objection. In Antarctic ice cores, carbon dioxide rises a few hundred years after temperature begins rising at the end of a glaciation, which is sometimes read as showing it cannot matter. Read it the other way: orbital change starts the warming, the warming releases carbon dioxide, and the carbon dioxide drives the bulk of the global change that follows. That is what a feedback amplifier looks like in a record — it must lag the trigger to be doing its job.

Two limits, stated plainly. The theory explains the pacing well and the magnitude only with the feedbacks doing most of the work, and their size is still an active question. And the two big anomalies — the dominance of the weakest forcing, and the switch from 41,000 to 100,000 years — are open. A theory that predicts the beat but not the strength or the mode switch is a real and useful theory with real and unfixed holes in it.

• Why the southern hemisphere warms in step with the north, when the orbital forcing pushes them in opposite directions.
• What actually happened at the mid-Pleistocene transition, and how the regolith and ice-thickness hypotheses are being tested.

Every term the collection defines is gathered in the glossary.