skeptic2 said:
I believe this happened once. The whole Earth was covered in ice and the oceans were frozen down to 1 km deep in some places. It is believed that it was reversed by the release of CO2 and other greenhouse gasses from volcanoes. With little or no vegetation to use the CO2, it slowly built up in the atmosphere until the planet warmed up.
Quite so! Great example. This is colloquially called "snowball earth", and the most recent possible occurrence is some 635 million years ago. To put that in context, the so-called Cambrian explosion of animal life was about 530 million years ago.
The snowball model is not universally accepted; but my understanding is that it is becoming more and more a recognized feature of Earth's past. The conventional explanation for how it occurs is indeed a runaway feedback effect.
Sometime I'll try to put together a post with a highly simplified model of an idealized planet which has some of the features described so far in the thread. That is, of runaway albedo effect leading to total ice cover in some cases (the snowball hypothesis) and with other cases where a sub-critical positive feedback merely amplifies the effects of any perturbation (the current situation).
As a bit of a foretaste, here is a description of the effects.
Earth currently intercepts about 342 W/m^2 from the Sun. The albedo is 0.31, which means 0.3*342 = 106 W/m^2 is reflected, and 236 is absorbed. Our surface temperature is, averaged out, about 15C. We radiate the energy from the Sun as thermal infrared radiation, most of which arises high in the atmosphere, where temperatures are much cooler. The characteristic temperature of our thermal radiation is -18C, which corresponds to high in the troposphere.
Now the albedo of ice and snow is very high. It reflects a lot of light. For example, over the Antarctic the albedo is about 0.8.
Imagine what would happen if a mad scientist managed to snap freeze the entire Earth, overnight, even right through the tropics. That would be bad... But we'd hope that that the Sun might melt all the ice, starting with the equator and back to current conditions.
Alas, not so. With the whole Earth having an average albedo of 0.8, the amount of absorbed energy would be 0.2*342, or about 65 W/m^2. The characteristic temperature would be way down below -85C. Even allowing for a significant warming effect from the atmosphere, this would still be well below freezing at the surface, even up to the equator.
Put another way; given the current sunlight levels, there are two stable equilibria for ice cover. There's what we have now. And there's a snowball.
Don't panic. A feature of this kind of bistable situation is that the "gain" of the feedback (presently stable, obviously) increases with ice cover. If you can just push the ice cover past the point where the gain goes critical, you set up the runaway effect. It turns out that for the Earth and our present atmosphere, this edge of stablility is way way colder than the ice ages or anything else for hundreds of million of years. So we're safe from the snowball.
Conversely, once in a snowball state,
something has to push past the edge in the other direction. Basically, you just need to heat things up enough to start melting ice in the tropics. That suddenly starts a reverse runaway, which melts all the ice right up to the poles; a huge and sudden shift that dwarfs anything else we can see in the records of paleoclimate. This is indeed what is thought to happen at the end of the snowball periods. It ends abruptly, with the Earth suddenly converted to very warm throughout.
In control theory, this is called hysteresis.
Cheers -- Sylas
