The planet slid from greenhouse to icehouse over the last hundred million years as carbon dioxide fell. The dinosaurs lived on a high-carbon-dioxide, hot world. We have long known that the poles were ice-free in dinosaur times. Early studies indicated that the equator then was not much hotter than today, but those early studies came with the warning that the main indicator used (isotopic composition of planktonic foraminifera) was subject to alteration after deposition that might have turned an indication of “hot” into an indication of “warm”. Recent studies, using other indicators and using very careful searches for unaltered foraminifera shells, are now indicating “hot” in the tropics during dinosaur times. The work is ongoing and a full consensus is not in, but tropical temperatures so hot that un-air-conditioned humans would have found it uncomfortable or even fatal to live on much of the planet now seem possible or even likely. Carbon dioxide remains the best explanation of the warmth, although current models, when given best-estimate carbon-dioxide loadings then, tend to simulate worlds a bit cooler than data indicate; whether this indicates shortcomings in data or models is unknown.
The planet saw widespread ice appearing at the poles about 35 million years ago, and generally carbon dioxide dropped and ice spread until recently. Details of that correlation remain unclear, with some central-estimate reconstructions indicating that some climate features are difficult to explain based on carbon-dioxide changes alone, but with the error bars including a carbon-dioxide explanation. (And the overall trend from greenhouse to icehouse is quite clearly a carbon-dioxide story. Furthermore, as more data have been collected, and better data, the mismatches between estimated carbon-dioxide level and estimated temperature have gotten smaller.)
Regionally, large and interesting changes occurred for reasons unrelated to carbon dioxide. The modern “conveyor belt” circulation in the Atlantic, for example, with surface flow directed northward from the Southern Ocean to near Norway, sinking, and return deep, does not seem to have existed more than a few million years ago when a seaway connected the Atlantic and Pacific Oceans across what is now Central America. (Now, the atmospheric transport of water vapor in the Trade Winds across Central America is not balanced by a return flow in the ocean beneath, so the Atlantic is saltier than the Pacific, and the “conveyor” circulation re-establishes the oceanic balance. With an open seaway across Panama, a much more direct route was possible. And, without the conveyor circulation, oceanic currents and coastal climates would have been quite different, although without a large globally averaged temperature change from the different currents.) In the ice-house world of the last few million years, Milankovitch cyclicity has driven ice-age cycling. The Earth’s orbit has many interesting features. These come from a few sources. First off, there are lots of planets out there, and some big ones. And all the planets run around the sun at different speeds. If you think of the solar system as a horse race, we keep passing Jupiter on the inside, and every time we pass, its gravity tugs on us a bit. The sum of all the tugs changes the Earth’s orbit a bit, giving the eccentricity changes described below. In addition, the rotation of the Earth causes the equator to bulge a bit. The planets, the sun and the moon (mostly the sun and moon) tug on this bulge, and that gives us the changes in obliquity and precession, just like a spinning top. As you might guess now, the important orbital features for this discussion are:
Think of an air-hockey table. Put the sun in the middle, nailed down, tie the Earth to it with a string, and hit the Earth. The Earth will zing around the sun. Put a little pin in the top of the Earth to be the North Pole. If you put the pin sticking straight up, you’re not there yet. The pin is inclined 21 degrees to 24 degrees from straight up, depending on when you look, going from 24 degrees to 21 and back over about 41,000 years. The larger the angle, the more the sun can shine on the North Pole (and on the South Pole, when the Earth is on the other side of the sun on the orbit!), and the less sun hitting the equator. This 41,000-year obliquity cycle moves some of the sun’s energy from equator to poles and back.
The air-hockey orbit in the previous section isn’t right; the orbit is eccentric (non-circular elliptical; think of a NASCAR track, although with a little curve even on the “straightaways”). A non-circular ellipse has two foci; think of two towers in the infield, both halfway between the straightaways, one a bit right and one a bit left when viewed from the main grandstand. The sun will sit at one of those tower positions (and the sun does not jump back and forth between the towers; it stays put). But, this is a weird NASCAR track; come back later, and the shape is changed a bit, going from almost circular to more squashed and back to almost circular over 100,000 years. (There actually is a 400,000-year modulation, so almost circular-slightly squashed-almost circular-more squashed-almost circular really squashed-almost circular-some squashed....) This change in eccentricity changes the total amount of sun reaching the planet a tiny bit; if you were in one of the towers, and the track were really squashed, the cars would spend a lot of time at the end far away from you where you had trouble seeing them, and only a little time at the near end, and if the cars are counting on being warmed by the “sun” from you, the extra time they spend far away reduces the total sun they receive. For the tiny changes in the Earth orbit, this is only a tenth of a percent or so in total sun received.
You may remember from the description of obliquity that the North Pole is inclined a bit. In addition to this angle changing, the North Pole also wobbles. Imagine putting your feet against a metal stake in the ground, grabbing the stake with your hands, leaning out until your arms are straight, and then having a friend push you in circles around the stake. Imagine a North Pole sticking up out of your head, extending your spine. The metal stake is “straight up”. If you bend your elbow and pull yourself toward the metal stake, your North Pole will point more nearly in the same direction as the metal stake, because you have changed your obliquity. But if you hold your obliquity the same (don’t bend your arm any more), and your friend pushes you around the metal stake, you are precessing.
Now, suppose you were doing this (metal stake, friend and all) on top of a NASCAR racer, with the sun in one of the towers in the infield. Your friend would have to push you really slowly the drivers would make about 10,000 laps before you got halfway around the metal stake! But notice that you would slowly switch from being on the infield-side of the metal stake when the car was at the end of the track closest to the sun tower (summertime for your North Pole, and wintertime for your South Pole), to being on the outside of the metal stake at that closest approach and on the near side of the metal stake at the farthest distance from the sun tower. This is precession. Notice that if you are close in northern summer, you are far in northern winter, giving a big difference between seasons in the north, but that close in northern summer is close in southern winter, and far in northern winter is close in southern summer, so when the winter-summer difference is large in the north, the winter-summer difference is small in the south, and when the winter-summer difference is small in the north, it is large in the south. Also, your friend is not pushing you with perfect consistency (and, bizarrely enough, the whole track is actually turning slowly, so that the straightaways switch slowly from being mostly north- south to being mostly east-west and on around to north-south again), so that rather than making a full circle of your metal stake every 20,000 laps or so, you typically make a full circle after either 19,000 laps or 23,000 laps. Also notice that, if the orbit/NASCAR track were a perfect circle, the two towers would be exactly in the center, the distance of the car from the tower sun would never change, so that this precession would not matter at all. Thus, precession matters a lot when the orbit is very eccentric, and precession matters little when the orbit is nearly round.
As scientists came to understand the Earth’s orbit and spin, calculation of the effects of these orbital features on the distribution of sunshine on the planet became possible. The most complete pre-computer treatment came from Milutin Milankovitch, so these are usually called Milankovitch cycles. Milankovitch predicted that, when ice-age cycles were understood, it would be found that the climate had varied with periods of 19,000 years, 23,000 years, 41,000 years and 100,000 years. Several decades later, when isotopic records of oceanic foraminifera were developed, these very periodicities were discovered — Milankovitch was right! And, because the different cycles affect north and south, and poles and equator, differently, it is possible to tell where the main controls reside.
The leading interpretation now is that poles are more important than equator, and north more important than south. When Canada and Eurasia received little summer sun, ice grew and the world cooled globally; when the sun increased in the high latitudes of the north, the world warmed and the ice shrank. The changes have been large — roughly 5 C to 6 C globally averaged — and switching from the modern level of about 10% of the land under ice (Greenland and Antarctica, primarily) to about 30% of the modern land area under ice (with glaciers over Erie and the Poconos in Pennsylvania, among many other places — note that when the ice spread, sea level fell, revealing land that is now under ocean, such that the total non-ice-covered land area was about the same then as now).
Oddly enough, northern sun has been more important than southern sun, with cooling in the south during some times when sunshine was increasing there. Many people have tried to explain this odd behavior in many ways, but so far the only successful explanations involve carbon dioxide. (The high albedo of the expanded ice contributed to the cooling, as did the sun-blocking effect of extra dust, plus shifts from trees to grasslands or tundra with higher albedo, but these together don’t explain the whole signal; the carbon dioxide, and a bit of methane and nitrous oxide change, were important.) Whenever the ice sheets have grown in the north in response to reduced sunshine there, carbon dioxide has dropped, and the carbon dioxide provides a successful explanation of the changes in the south. The path is: changing sunshine to changing things in the Earth system (temperature, ice volume, sea level, dust, etc.) to changing carbon dioxide to more changes in temperature in response, so the carbon dioxide is a positive feedback, not a cause.
The processes by which changing ice volume affects carbon dioxide are rather complex, involve many different pieces of the Earth system, and are a bit beyond our course. One, for example, is that ice-sheet growth in the north increases dust supply to the ocean (the glaciers grind up rocks, change winds, etc., increasing dust delivery especially in the north where there is a lot of land to make dust), which fertilizes plankton that turn carbon dioxide into plant, the plankton are eaten, the eaters poop, the poop sinks, and so carbon dioxide is moved into the deep ocean and away from the atmosphere, lowering atmospheric carbon dioxide. There exist many other mechanisms — covering 20% of the land with two-mile-thick ice sheets, lowering the sea level by several hundred feet, changing winds and currents, spreading sea ice in the cold, and other things constitute large perturbations to the Earth system, and it responded in a way that amplified those changes. The most important changes probably relate to shifts in southern winds — now, the winds howl around Antarctica, moving water to their left, hence north because of the Coriolis effect on our eastward-rotating Earth, and driving upwelling that brings CO2 back from the deep ocean, but during ice-age times the winds shifted up on South America and so left more CO2 in the deep ocean, lowering the atmospheric level.
The “skeptics” of climate change are fond of pointing out that temperature change probably started slightly before carbon dioxide change, and then concluding that carbon dioxide cannot be responsible for any of the warming. This is faulty logic, but of the sort that seems sensible to people who know nothing about the subject. (Suppose you run up a big debt on your credit card, and then you end up paying lots of interest on the debt until you go bankrupt. By the skeptic logic, you went into debt before you started paying interest, so the interest cannot have contributed further to your debt because the interest payments lag the debt in time. Wrong.)
A lot of very interesting questions are not fully answered with regard to the ice ages. But, the big picture is clear. The ultimate cause is tied to Milankovitch orbital features, which change the total amount of sunshine reaching a place during a season by 10-20% or even more (although with tiny globally-averaged effect). Many things happen in response to this cause, and carbon-dioxide response is especially important in the global signal. (Growing ice on Canada doesn’t directly make it much colder in Antarctica, but changing carbon dioxide does.) The changes have been large but slow. The 5 C to 6 C warming (10 F warming) from the last ice age, globally averaged, took over 10,000 years, or less than 0.1 F/century; the warming of the last century, tied especially to human activities, has been ten times faster, and the warming in the next century if we don't change our energy system is expected to be faster yet. Similarly, the carbon-dioxide changes of the ice ages were much, much slower than what humans are now doing. The ice ages provide further evidence of the warming effect of carbon dioxide, they allow us to test our models (which work pretty well), but they don’t provide any alternate explanation of recent temperature changes.