Runaway climate change?

Wiki has an article on the subject, created by over enthusiastic folk. I tried to kill the witch on the grounds that it had no good definition; alas that didn't fly. My best effort at a defn that fitted reality was this but it didn't last. The wiki article will die a long slow painful death, but the term lives on elsewhere.

One of those places being Why we need a zero carbon world by All very nice people, no doubt, but over enthusiasts all. Why do we need a ZCW? Easy: We need a zero carbon world to drastically improve our chances of avoiding a runaway greenhouse effect that is irreversible. Um. But what *is* a RGE? Wiki thinks it's the same thing as RCC. I'm dubious. I think it might be the thing where our oceans boil away, in which case we can forget all about it as its not going to happen. ZCW seem to use RGE and RCC just about interchangably too, but I think its fairly clear they aren't doing a lot of thinking. Continuing, climate science indicates that runaway climate change is likely unless extremely rapid cuts in carbon emissions take place within the next few years. Well, since I don't know what RCC is, I suppose thats a little bit hard to argue with. But unless they mean Hansen, who is well out on a limb, its hard to see exactly what science they are referring to. And no article about RCC would be complete without fatuous nonsense like In 2001, it was projected that the sea ice on the Arctic Ocean would disappear in around 140 years. Now it is clear that it will be gone in a decade or so. Even the wild-eyed wackos don't say its clear it *will* be gone in a decade (unless they are hiding a very long tail nder "or so").

OK, enough ranting. can anyone point me at a credible defn of RCC or RGE?

[Update: thanks to Eli for the idea, and ABM for the H+S ref. I've now decided I know what RaGE is, and have dumped it into wiki: for the version I left, which might survive, and [[Runaway_greenhouse_effect]] for the current version, which could well be a disaster by the time you read this. Improvements (especially more refs) very welcome. Less impressed by comments attempting to define RCC -W]

More like this

Wow, that's a mess. There's Runaway greenhouse effect, Runaway climate change, and Abrupt climate change, only one of which is listed on the Glossary of climate change. I'm sure there's more, all taking about what certainly sound like the same thing - climate changing rapidly thanks to a total feedback gain above one.

Maybe the climate literature just hasn't put two and two together on this yet, but there's clearly something missing here... and William, I can't say I'm impressed by your current version of the text there - complete evaporation of a solid or liquid greenhouse gas is only one of several ways to bring nonlinearity into the system sufficient to bring the total feedback gain back under one again.

[Feel free to improve it, preferrably actually using real references rather than just some opinion as to what it might be supposed to mean :-) -W]

I don't know how other people define runaway change but I would define it as occurring when a previously stable equilibrium becomes unstable leading to rapid and potentially large changes. Or alternatively when the response to forcing becomes discontinuous. As will occur if feedback gain exceeds 1.

By James B. Shearer (not verified) on 15 Feb 2009 #permalink

This is a valid question.
However, I agree with James: Runaway Greenhouse Effect means that at some point anthropogenic climate change starts to feed off itself and positive feedback loops push earth climate into a new, much warmer equilibrum that's not just quantitatively, but qualitatively different from what went on during the last few millenia.

At least that's what most of the relevant literature seems to be about: Not a warmer version of the present, but a world climate that is fundamentally changed in many ways. That may well be possible, and I think the possible consequences are still widely underestimated.

The section "Positive feedback and runaway greenhouse effect" in "greenhouse effect" seems rather similar to the "Runaway climate change" article - is that a recent copy?

Is there a separate wikipedia article on climate feedbacks? That might be a suitable place for the real scientific definition - there *is* a definition of what amounts to runaway, which you can gather from the discussion in Bony et al (Journal of Climate 19:3445 (2006))

The current wikipedia article roughly gets to the scientific definition with this statement: "With radiation from the Earth increasing in proportion to the fourth power of temperature, in accordance with the Stefan-Boltzmann law, the feedback effect has to be very strong to cause a runaway effect."

From Bony et al that's essentially pointing to the Planck response, or as IPCC AR4 puts it, the "uniform temperature radiative cooling response". That response leads to a *reduction* in net forcing with a uniform temperature increase (an increase in net radiation outward) according to the distribution of temperatures around the planet. So any climate response that increases forcing (such as additional water vapor in the atmosphere) has to be added to this negative Planck response to get the net response to a change in temperature.

The final (linearized) temperature response is given by Bony et al equation A3. If the sum of the feedback gains (negative of response divided by Planck response) is greater than 1, i.e. the net response is positive, the linear assumption behind equation A3 breaks down and you get runaway.

As long as this net response is negative, climate is stable - the additional increments of forcing through these responses are successively smaller and converge. The closer the total response comes to zero, the larger the final temperature response will be to an initial forcing.

A second thing that should be coupled with such a definition is the question of timescale - water vapor is a quick response, ice albedo slower, CO2 and methane also slow. The rate at which the "runaway" happens is governed by the fastest "slow" response that, added to all the faster responses, first brings the total feedback gain over 1. There's probably evidence for this timescale from past episodes of rapid climate change...

William, what sort of change in forcing (increase in solar luminosity, plus GHG forcing) is required to cause RCC (i.e. oceans boiling to create a more Venuslike planet)? IIRC its been claimed that another half to billion years of solar evolution (is that 6% per GY, or is it a bit greater) was supposed to do the trick (but I think this stuff was very old). And could a short term increase in methane (by say all methane hydrates being released) potentially push the planet onto a RCC trajectory if it happened?

I should point out that the linearized "feedback gain" model I just described has absolutely no room for "tipping points" where the runaway starts - that would be a nonlinearity, where there may be some reason to believe feedbacks are increasing with increasing temperature/forcing. It seems unlikely to me that we have hit or are close to any nonlinearities yet, given the relatively small increment in temperatures so far (1 out of 33 C for the greenhouse effect as a whole, for instance).

That is, under the linearized "feedback gain" analysis, climate is either stable as a whole, or unstable, on the appropriate time-scale. Given the rapid climate changes seen during the ice ages, there is perhaps some reason to believe climate is fundamentally unstable on a multi-century timescale. On the other hand it seems to have been remarkably stable the last few thousands years...

Of course, if we don't define it, it can't exist. :-)

Geologic evidence for such events should be a reason to come up with a definition. The LPTM (aka PETM) is one example:…

The event seems to be a large methane release from methane hydrates, most of which happened in a period of time less than a hundred years. There are reasons to suspect that most of this warming was much quicker, less than a hundred years, and that the peak warming was larger than the average recorded over the whole period.

By Phil Hays (not verified) on 15 Feb 2009 #permalink

My operational definition of RCC is the point at which positive carbon cycle feedbacks overwhelm humanity's attempts to control atmospheric carbon. It has nothing to do with Venus, and everything to do with losing the last vestiges of (possibly delusional) hope of being able to "solve" the climate problem.

The way I've heard the term used, RCC means 'run out of control', i.e. the climate gets pushed into a state where the positive feedbacks become dominant compared to anthropogenic emissions, thus this is the point after which no attempt to cut emissions would be effective.

Hello William,

so it seems, that "runaway climate change" bandwagoon has grown up a little - Standord's Christopher Field has jumped in, obviously (citing, e.g. decreasing ocean CO2 airbourne fraction):…

Doom in, IPPC out :-)

[IPCC? -W]

BTW, prof. Barry Brook does not agree with You regarding the Viki Pope critics of "alarmism", see:…

[Vicky. No, BB doesn't agree, but his sources are thin - just a couple of newspaper quotes. No proper work -W]


Don't know anything about the science really, but in terms of why the idea should be catching on around now informally, isn't it a sort of delayed effect of denialism? Denialists always had a sideline in denying that Venus was so hot because of CO2 forcing. (The Sagan connection to nuclear winter helped there.) As more people realize that denialists are wrong about everything, perhaps it's tempting to assume that anything they've ever denied must be true.

The terms RCC and RGE just don't catch the flavor. How about using ELE (extinction-level event)?

The PT (PermianâTriassic) is the largest mass extinction in geologic history. The PT, and several other events share a lot of characteristics with the LPTM. Both the PT and the LPTM had a large volcanic source of CO2 gradually warming the climate, then sudden isotopic change in carbon, suggesting a methane hydrate meltdown. In the LPTM, some of the ocean became anoxic, in the PT and the Early Toarcian, most of it did.

The only thing we can say with confidence is that we haven't yet warmed the climate enough to produce a similar event. Yet.

By Phil Hays (not verified) on 16 Feb 2009 #permalink

It is fairly well defined by Venus, a situation in which none of the significant greenhouse gases can have a liquid/solid (or if you really want to stretch it, molecular, eg carbonates and hydrates) phase on the planet and everything gets boiled out into the atmosphere.

Here's an extract from Held & Soden 2000. Basically when the positive feedback gain exceeds unity then you get runaway situation. But this cannot be maintained in a finite system. In the case of Venus the greenhouse warming ended when the surface of Venus reached the melting point of sulphur.

What no one seems to understand is that when an abrupt climate change is happening, that is a case of a runaway situation. The sudden change in albedo due to the disappearance of sea ice, results in a short, but semi-permanent, climate change as the water vapour adjusts to the new temperature regime and cloud system.

In other words an abrupt climate change is just a small (water vapour) runaway event. But if you really want to unerstand what is going on, you have to be willing to make a pardigm shift and accept that runaway events can and have happened. That is difficult when the whole of Earth Science implicitly believes in uniformitariansism and rejects catastrophism.

Annual Review of Energy and the Environment
Vol. 25: 441-475 (Volume publication date November 2000)


Isaac M. Held and Brian J. Soden

The size of nondimensional ratio, βH2O, provides a measure of the strength of the water vapor feedback. If βH2O â 0.4, water vapor feedback increases the sensitivity of temperatures to CO2 by a factor of â 1.7, assuming that I and C are fixed.

If the value of βH2O were larger than unity, the result would be a runaway greenhouse. The outgoing infrared flux would decrease with increasing temperatures. It is, of course, self-evident that the Earth is not in a runaway configuration. But it is sobering to realize that it is only after detailed computations with a realistic model of radiative transfer that we obtain the estimate βH2O â 0.4 (for fixed relative humidity). There is no simple physical argument of which we are aware from which one could have concluded beforehand that βH2O was less than unity. The value of βH2O does, in fact, increase as the climate warms if the relative humidity is fixed. On this basis, one might expect runaway conditions to develop eventually if the climate warms sufficiently. Although it is difficult to be quantitative, primarily because of uncertainties in cloud prediction, it is clear that this point is only achieved for temperatures that are far warmer than any relevant for the global warming debate (22).

By Alastair B. McDonald (not verified) on 16 Feb 2009 #permalink

Fact is definitly : It get warmer and warmer, and in a few years there would be more water on this earth.
What could we do ? We must usw the safety brake...
And if we miss it? Then there would be alot of Problems for the humans and animals!

Lets wait...

>"A runaway greenhouse effect occurs when, on a planet with substantial reserves of greenhouse gases in liquid or solid form, some forcing occurs to begin to gasify them, leading via positive feedback to complete gasification of these reserves [1]."

Complete gasification certainly makes clear it is only an extreme event but is it true? Suppose there is some water and some methane hydrates and when both are present the gain is over 1 but once the methane hydrates run out the gain is less than one. I think this is a runaway effect and suggests your definition needs changing to allow just complete gassification of at least one of the greenhouse gasses.

However that creates another problem: If in addition there is some sea ice and this is expected to be completely melted with the gain expected to be less than 1 then this isn't a RGE (present circumstances?). However, if there is substantial quantity of sea ice and, when combined with gassifying greenhouse gasses, the gain is pushed up over one then I think it is a RGE. (If the albedo effect was greater than the GG effect this would be a runaway albedo effect rather than a RGE.)

This suggests that not only allowing a melting of an albedo changing solid to be allowed along with gassifying greenhouse gases but also a requirement for the gain to be over 1 until at least one of the constituents of the gain being over 1 is exhausted.

Or maybe we just have different views on how extreme and irreversable an effect has to be before it can be called a RGE and my opinion doesn't count for much.

(BTW I think some of the things I wrote are still in your revised piece from when a saw a need to distinguish feedback from runaway effects.)

There is a whole section 9.5 on "The runaway greenhouse" in the bible "Atmospheric Radiation 2nd edn." by Goody & Yung.

They point to Simpson's unsolved paradox where an atmosphere dominated by water vapour could not emit to space a flux greater than a certain value. If incident solar flux were to exceed this value the atmosphere could not respond. For Earth, Simpson concluded that the the critical flux is exceeded by solar flux in tropical regions.

What I am saying is that at present the tropical regions can export the extra heat to the polar regions, but if the Arctic sea ice disappears then in the summer without the high albedo from the ice, the solar flux will exceed the OLR from water vapour and disrupt the climate of the whole planet :-(

By Alastair B. McDonald (not verified) on 17 Feb 2009 #permalink

Alastair, I would be a bit careful about using G&Y. A lot has been learned about the water vapor continuum since then which could change the result.

U.S. pundit George Will clings to his disinformation policies:…

This fellow always gets my dander up, and I retreat to my old files gleaned from your older website, where I finally noticed your more up-to-date efforts. I just wanted to thank you for the work on the "'70s climate myth" reportage you began so long ago.

I have bookmarked your site and wish you the best. Thank you again. And for your similar efforts at Wikipedia.

>"If the [[clathrate gun hypothesis]] is correct, this would not be a runaway greenhouse effect in as much as the oceans would not boil away."

Umm., in this case the gg reserve is the clathrates not water vapour. Would these all be gassified - probably not but the available ones would be depleted. In what way does that fail to satisfy your definition of RGE?

[Its not a runaway unless all the GHG's vapourise. In my defn. Of course, I'm just winging it, you can invent your own if you like.

Oh, and of course the clathrates are weird - they are only non-vapour because they're sub surface. If brought to the surface, they would be vapour -W]

My previous comment seems to have been swallowed here..

crandles - precisely, I don't see that William's (Eli's hinted) new definition is coherent.

The clear criterion for "runaway" is whether the total feedback gain is greater than 1. Linearization gives just one value for the feedback gain, presumably less than 1 at present given our relatively stable climate. Nonlinearities in the system allow the feedback gain to change to either become greater than 1 (start a runaway or "abrupt climate change") or to become less than 1 again (to stop the runaway).

Running out of material by boiling all the oceans or evaporating all the methane clathrates is one dramatic way to induce nonlinearity in responses (to reduce feedback gain to less than 1 again), but only one of many, and it hardly seems deserving of special mention in this context, given that "runaway" seems to be used by reputable sources in many more ways than the Venus case.

[The clear criterion for "runaway" is whether the total feedback gain is
greater than 1.
says who? You could assign that to runaway climate change if you like. I don't see that William's (Eli's hinted) new definition is coherent. - your argument appears to be that is must be incoherent, because it disagrees with *your* pet defn. Thats illogical, captain -W]

"says who?" Says Soden and Held, who you referred to in your wiki update, thanks to Alastair! Let me quote the relevant part that Alastair cited above, again:

If the value of βH2O were larger than unity, the result would be a runaway greenhouse. The outgoing infrared flux would decrease with increasing temperatures.

Outgoing infrared flux decreasing with increasing temperatures happens if and only if feedback gain is greater than one, i.e. if total feedbacks on forcing are sufficiently positive to overwhelm the negative forcing effect of the (Planck - not quite Stefan-Boltzmann) uniform temperature response.

The meaning of "runaway" in this context is clearly the same as that of "Thermal runaway", which has its own wikipedia page and the definition there is "an increase in temperature changes the conditions in a way that causes a further increase in temperature".

An increase in Earth's temperatures could lead to "runaway" under that definition if and only if the increase in Earth's temperatures caused more heat to be retained, i.e. reduced outgoing infrared flux, precisely the condition Soden and Held cite for runaway. That is a coherent definition, and it boils down to:

runaway = total feedback gain greater than 1

And under that definition, even small "abrupt climate change" events may count as runaway as Alastair pointed out.

Do you have reputable sources using the terms runaway climate change or runaway greenhouse effect in a way that clearly limits it only to oceans boiling dry?

William has now clarified that his definition requires complete gasification of all greenhouse gases not just the reserve of greenhouse gases involved in the feedback. (Previously I wasn't sure if William was relying on the substantial for clathrates.) If that is what you want I think the definition should clarify this. It is a coherent definition but that doesn't mean it is right. If something like the melting point of sulphur being reached caused a situation where 75% of the oceans boiled away this wouldn't be RGE according to William. Coherent but I would beg to differ and claim this was a RGE event.

I could try to invent my own definition, but I would rather consult and see if there needs to be discussion of the alternate views. From Alistar's post "What no one seems to understand is that when an abrupt climate change is happening, that is a case of a runaway situation. ...
In other words an abrupt climate change is just a small (water vapour) runaway event. But if you really want to unerstand what is going on, you have to be willing to make a pardigm shift and accept that runaway events can and have happened. That is difficult when the whole of Earth Science implicitly believes in uniformitariansism and rejects catastrophism." (Is that in Held & Soden - it doesn't seem quite clear what is quote and what is Alistar's synthesis?)

It appears there could be a majority in favour of something more catastrophic like Williams definition while a minority believe that runaway situations can and have existed for short periods of abrupt climate change during Earths history. William claims such episodes are a 'constrained' version of RGE. Is there any evidence of this?

[This is a question of definitions. Definitions are very important, otherwise we talk rubbish. Like the newspapers, who really don't have a clue and don't care. I believe that it is valuable to have different words - runaway climate change, runaway greenhouse effect, abrupt climate change - mean different things *and to know what those things are* -W]

Just search runaway greenhouse on Google Scholar. The idea of a runaway greenhouse effect first surfaced wrt Venus and is well defined for planetary atmospheres. Here is the abstract from a 1969 paper by Ingersoll

Radiative-convective equilibrium models of planetary atmospheres are discussed for the case when the infrared opacity is due to a vapor in equilibrium with its liquid or solid phase. For a grey gas, or for a gas which absorbs at all infrared wavelengths, equilibrium is impossible when the solar constant exceeds a critical value. Equilibrium therefore requires that the condensed phase evaporates into the atmosphere.

Moist adiabatic and pseudoadiabatic atmospheres in which the condensing vapor is a major atmospheric constituent are considered. This situation would apply if the solar constant were supercritical with respect to an abundant substance such as water. It is shown that the condensing gas would be a major constituent at all levels in such an atmosphere. Photodissociation of water in the primordial Venus atmosphere is discussed in this context.

Section 3 discusses the conditions needed for a runaway greenhouse.

Crandles - I didn't fully explain why I thought William's definition was incoherent but you hit on a close example. Suppose you are in a situation where 75% of the oceans boiled away; ocean surface area declines precipitously and most of the planet's surface is dry land; relative humidity starts to drop. Under some conditions that would be sufficient nonlinearity to allow feedback gain to go below 1, and the runaway (my definition) stops. But by William's definition, since the oceans didn't completely boil away, there was no runaway?

But worse than that, H2O is far from the only GHG or potential GHG under warmer conditions - there are the planet's stores of CO2 in carbonate minerals, for example. Does William's definition require all those minerals also to be vaporized? And why are GHG's special in this regard - after all the "greenhouse effect" includes cloud warming effects (infrared absorption by condensed water vapor) - Venus has SO2 clouds, do we require all sulfate minerals to vaporize as well? Where does it stop? Or is "runaway greenhouse effect" somehow peculiar to a planet with oceans and water vapor only? But then, what if we find out Venus never actually had much water? It didn't have a runaway after all, despite the enormous present-day GHE there?

Eli - that's a fine early reference (and proves the term has been in the literature for a while) but it's a pretty limited discussion (one GHG, assuming all else constant or at least that planetary albedo is limited) of a more general issue... but the graph there of "Planck function" vs. optical depth is certainly a forerunner of the general condition.

Arthur, I think it actually was one of the first that attempted to relate the runaway condition to optical depth, but the term was pretty well accepted in the planetary atmosphere community before then wrt Venus.

And as I said in my original definition, once you get the condensed phases vaporized, the next step is to decompose the hydrates and the carbonates that are accessible on the surface. Sort of the reverse of what happened on earth.

In a sense this has become a discussion of whether a phenominological (however the hell you spell it) or a first principles definition is best, but it strikes me they are equivalent.


Do you have any more information on the water vapour continuum?

I agree with you that this has become a debate between boiling water and negative feedbacks exceeding 1, but surely Wikipedia should include both? The criteria for a WV runaway to start is not when the temperature reaches 100 C. It is at a lower temperature than that where a one degree increase will produce more water vapour to evaporate than is needed to raise the temperature by another one degree. (That is my best attempt at expressing a differential equation in words.) In other words, when the temperature gain from a one degree rise is greater than one. It is easy to see that, all other things being equal, this evaporation and boiling off of the oceans will only end when the basins become dry.

One can surmise on a dry Venus that the greenhouse effect from CO2 could cause the carbonates to disassociate, and a similar runaway to happen.

In the case of clathrates, the methane released would raise ocean temperatures, thus releasing more methane in a positive feedback role, until the reserves are exhausted. The oceans need not boil.

By Alastair McDonald (not verified) on 18 Feb 2009 #permalink

Grauniad informs, why arctic ice melting is "faster than predicted" by IPCC:

you guessed it: melting ponds!

also supported by this study (reported by Nude Socialist):…

Now, another positive feedback in Arctic found, informs Nature Geoscience:

you guessed it: N2O

not exactly runaway climate change, but not good news, either...


I can't help feeling that to any definition of 'runaway * effect' that is constrained by the end of the process is going to be high redefinition.

'Runaway * effect', I would suppose, derived originally from the idea of a runaway train - beyond the control of its drivers, hard or impossible to stop due to inertia, etc. A runaway train, though, is a runaway train regardless of how it and when eventually stops. It seems to me that most people are using 'runaway * effect' in a way consistent with this straightforward metaphor: when the process has reached the point when hauling on the brake has no effect, you've got yourself a 'runaway * effect'.

It does seems that there is a more constrained technical in play as well, at least for 'runaway greenhouse effect' wrt Venus and planetary atmospheres. But if most people are using 'runaway * effect' in the intuitive metaphorical sense way, it seems very silly to try to claim they're wrong just because people in a narrow discipline have evolved their own idiosyncratic definition. How people use a word is what it means... and if that means that different words do not always mean different things, well, that's language for ya.

I suggest that the term has two distinct meanings, and both should be recognized and defined.


I agree definitions are critical for scientific understanding although the fact that this is being done on wikip. leaves the door open for a future Connolley paper saying there was never a scientific consensus on RCC/RGE in the peer-reviewed literature.

I'm of the opinion that before you define runaway CC we need a clearer definition of climate change than the UN Framework Convention on Climate Change of 1992 where climate change was defined as:
âa change of climate which is attributed directly or indirectly to human activity that alters the composition of the global atmosphere and which is in addition to natural climate variability observed over comparable time periods.â

Under this definition even the rise in CO2 levels (without any change in temps) would constitute climate change.

Finally, would RGE be a natural phenomenon not an anthropogenic one (albeit one attributed indirectly)?

From Alastair McDonald above:

"What no one seems to understand is that when an abrupt climate change is happening, that is a case of a runaway situation."

[I saw that. I'm going to have to do another post on all this next week, but what I think of that is: what AMD hasn't understood is that this is a questions of definitions. Runaway is a *word*. It doesn't have a definition until you give it one. Now we could define it in such a way that ACC (like D-O events) count as runaway, but I think that would be a very bad idea, as it would muddle different things under the same banner. So I don't want to do that -W]

William (although it only struck me last night that you could also be a Billy C.) have you seen this report on "abrupt climate change"?…

So much comes down to words and not numbers IMHO.

[I'm sure I've bitched about it's defn before: causes substantial disruptions in human and natural systems. is crap. It means that anything pre-human cannot be ACC, by defn. Stoopid -W]

By Anonymous (not verified) on 20 Feb 2009 #permalink

Hi William - yes it comes down to definitions, but "runaway" has a pre-existing definition that I would have thought it was the duty of a wikipedia page to describe, not to come up with a new definition that somehow overrides previous understanding. "Thermal runaway" is well-defined and establishes sound reasonable criteria; audio feedback is a well-understood example of the positive-feedback criterion that's at the center of it.

Note that runaway climate change doesn't have to be warming - the "snowball earth" episodes are often referred to as coming from a runaway - due to ice-albedo feedback in that case. Instability is instability, on either side - if the trigger is a warming one, it will lead to runaway warming, if a cooling one, then runaway cooling.

Example citation on ice albedo runaway: "A Neoproterozoic Snowball Earth", Paul F. Hoffman et al, Science Vol. 281. no. 5381, pp. 1342 (1998) - "Kirschvink (17) proposed a snowball Earth, created by a runaway albedo feedback, in which the world ocean was virtually covered by sea ice but continental ice cover was thin and patchy because of the virtual elimination of the hydrologic cycle."

Now given that the term "greenhouse effect" doesn't usually include albedo terms, you could justify treating this as a distinct "climate change" rather than "greenhouse effect" runaway; nevertheless I expect there are lots of other examples in the literature (no, I'm not particularly familiar with it myself).

Strikes Eli that a useful way of defining RCC would be something along the lines of an irreversible change or a change that could not be reversed by random variation in the climate system over a period of 20-30 years.

Does it have to be called 'runaway climate change'?

I have always had a preference for 'catastrophic climate change'. Google reveals around 1.2 million hits for catostrophic, and 600k for runaway.

Of course in the layman mind and media land, 'catostrophic' tends to be more like really really bad and will happen any day. Although in science land I'm pretty sure it means a non linear situation with feedback greater than 1. And could also apply to the melting of the ice sheets over 100s or 1000s of years.

By Michael Hauber (not verified) on 21 Feb 2009 #permalink

Eli and Michael - yes, I think we're getting into definitions of more than runaway here. IPCC AR4 WG1 defined most of the relevant terms in the Glossary (Annex 1), starting on p. 941. Following are my suggestions for slightly more succinct definitions based on the IPCC versions:

* climate: The average weather. Specifically, the mean and variability of temperature, precipitation, wind, and other measurable attributes of a planetary atmosphere over a given period, typically 30 years.

* climate change: a change in the mean and/or variability of the properties that define the climate (when averaged over the climate definition period, typically 30 years).

* energy balance: the difference between total incoming and total outgoing energy. If climate is not changing, energy balance averaged over the climate definition period must be zero.

* global warming: climate change under conditions of a positive average energy balance

* global cooling: climate change under conditions of a negative average energy balance

* runaway climate change: climate change under which the response to the positive or negative energy balance situation acts to amplify, rather than reduce, the difference between incoming and outgoing energy, leading to self-amplifying change.

* abrupt climate change: climate change faster than the climate definition period, including an entry to and exit from a runaway state induced by nonlinearities in the climate system.

William, you seem to be heading down a top-down prescriptivist route on this definitions thing - which is certainly not how definitions are generally arrived at. You seem to be getting into a linguistic debate rather than simply looking for a clear way to describe a recognized term of art with a clear consensus on what it means - and in so doing, perhaps rather venturing out of your bailiwick? It might be worth discussing this with a linguist rather than with us folks here... Maybe try emailing one of the guys over at LanguageLog - Mark Liberman, maybe? Or Ben Zimmer - as the brother of your erstwhile SciBling Carl Zimmer, he may be open to discussing this.

[rather than simply looking for a clear way to describe a recognized term of art with a clear consensus on what it means - I don't think there is a clear consensus. Amongst the media there is a clear consensus to use muddy langauge and not really care. In science terms, the idea isn't really used much so doesn't get defined. I need to catch up on a large back log of comments; if I missed the place where you or someone else pointed me to this consensus, please do it again -W]

William, I wan't saying that there is a consensus - rather the opposite.

If the meaning of the term was well-established and you were simply trying to correct others' factual errors in their inexpert definitions, then it would be an appropriate application of your expertise (as it is when you correct others' factual errors when it comes to the science of AGW).

Identifying what a word or term means when there is not an obvious consensus, though, is not simply a matter of saying 'henceforth, this is what the terms shall mean' (unless you're doing that in a very limited way, as the IPCC reports do with 'likely', 'very likely', etc, where their definitions are explicitly restricted to the reports themselves).

I'm simply suggesting that if you really want to fight this battle it might be worth talking to a linguist about how best to go about finding definitions for these terms - and how to test whether the definitions you arrive at actually reflect how the words are used (through corpus searches and so on).

I know you want to constrain the terms in ways that would be useful, but if your definitions (no matter how potentially useful) are flatly at odds with how other people use/understand the terms you're going to be left with an unwinnable battle (and - dare I say? - one which may not be worth fighting).

'I donât know what you mean by "glory,"' Alice said.

Humpty Dumpty smiled contemptuously. 'Of course you donât â till I tell you. I meant "thereâs a nice knock-down argument for you!"'

'But "glory" doesnât mean "a nice knock¬down argument,"' Alice objected.

'When I use a word,' Humpty Dumpty said in rather a scornful tone, 'it means just what I choose it to mean â neither more nor less.'

'When I use a word,' Humpty Dumpty said in rather a scornful tone, 'it means just what I choose it to mean â neither more nor less.'

That is exactly what William thinks :-) He believes that runaways are something that can only happen on Venus. It is obvious to him that we on Earth are safe from them. And so he is looking for a definition which has that meaning - neither more nor less.

So he is searching through the common uses of the word to find one that corresponds to his ideas, but of course none does. Everyone has their own subjective idea of what runaway means - no two are the same. The only answer is to search for a technical term without applying preconceptions and the electrical engineering usage, based on simple mathematics, is obviously the best choice.

But it first requires understanding another term with a common usage - feedback. In common usage positive feedback is something good, but for an electrical engineer a circuit which suffers from positive feedback it is bad. This means that someone may think they understand what an engineer means by feedback and be completely wrong. So forgive me if I am telling you something that you already know.

When a signal changes in value, feedback is the amount the change is amplified by. Take a signal of 1 volt and increase it by 0.2 volts. The new signal is 1.2 volts. If the circuit has a positive feedback of 0.3 (30%) the output will be increased by the feed back by 0.3 * 0.2 volts i.e. 0.06 volts and the output signal will now be 1.26. But the circuit has a feedback of 0.3 so this additional 0.06 volts will be amplified by 0.3 i.e. 0.018 volts, which would also be amplified by 0.3 and so on.

Thus when a positive (or negative) feedback is applied the amplification is given by the geometrical series:
Sn = x*f + x*f&2 + x*f^3 ... x*f^n.

From schoolboy maths, the sum of that series is
Sn = x(1 - f^n)/(1 - f).

For n = infinity, and f less than 1, then f^n will equal 0.

So S = x/(1 - f).

If the feedback is steadily increased from zero, when it reaches 1 the denominator will be 0, and the sum will become infinite. In other words the voltage will runaway and the circuit will explode!

Saturated water vapour density increases exponentially with temperature. So the feedback it applies to global temperature will increase as global temperature increases. It will runaway when the temperature is high enough for a feedback greater than 1 to exist. That is what is surmised happened on Venus.

The big question is why has it not happened on Earth? I had my big eureka moment when I realised that it had!

Cheers, Alastair.

By Alastair McDonald (not verified) on 25 Feb 2009 #permalink

My sense of the meaning intended by lay climate activists when they talk of "runaway climate change" is more or less as Gareth and Frank define it: it occurs when positive feedbacks come into play on such a scale that they overwhelm mitigation efforts. But it seems a rather imprecise concept, depending as well on one's target and on the range of mitigation methods one allows oneself.

By mitchell porter (not verified) on 27 Feb 2009 #permalink

Suggestions/questions. (I am trying to understand this).

Is it possible to obtain a temperature spectrum defined (by me) as follows?

1. Allowed = consistent with energy balance
2. Forbidden = not allowed.

Example 1. : Simple Venus (see Houghton Global warming or Wikipedia above)
Was Venus unstable at all initial temperatures i.e. are there are no allowed values of
T ?
OR is there a critical initial temperature defined such that the spectrum of allowed temperatures is just
T is less than Tc ?

Example 2 : Earth with one dominant main feedback which can vary. Are all values of T allowed?
OR is there a critical value of T above which T is forbidden?

Example 3. Earth with several feedbacks of varying response times. e.g. lumps of ice, methane clathrates, oceans holding less CO2 etc.

Is it now possible that there are several forbidden temperature gaps which are removed from the spectrum? Global warming might then raise T until the bottom of such a gap. After that the warming would consist of the sum of two parts an unforced term and a forced term. (This may not be quite right for a non linear system). It would still be worth reducing anthropogenic CO2 in order to reduce the forced bit. The existence of these gaps would not in itself imply anything about the speed or magnitude of warming. At a higher temperature the top of the gap would be reached and more 'normal' global warming would resume with one less source of greenhouse gas to worry about.

Is this just more climate tripe? All I am doing is switching attention to the temperature variable.
Partial analogue to nuclear reactor. If the feedback parameter reaches 1 as a result of prompt neutrons there will be an explosion.
If delayed neutrons are required to get up to 1 you will have a controllable reactor.
Both involve runaway and in this case the first version might actually liberate less heat.
[This software has hysteresis. I clicked (when instructed ) to return to the previous page when it delivered me to a blank form after deleting my comment.]

By Geoff Wexler (not verified) on 27 Feb 2009 #permalink


Taking your nuclear reactor with its two states, both involve positive feedback, but only when the feedback getes greater than 1 is there a runaway and an explosion.

That's not MY definition. It is the standard engineering definition based on the simple mathematics I gave.

The thing gets more complicated when you come to the new expression "Tipping Point". What does that mean? [Tim Lenton, 2007] But using the terms feedback and runaway it is easy to restrict it to one of two states. It could be when the overall feedback in a dynamical system changes from negative to positive. Or it could be when the positive feedback reaches 1. (Ti

To see how there are two types of tipping point, let's consider the case of the Greenland ice sheet. Once the level of CO2 causes it to melt, the altitude of the upper surface becomes lower. The air temperature rises at lower altitudes so there is more melting (a positive feedback.) The following year, the CO2 will cause the same melting and and the decrease in height will also cause more melting. This will continue until the ice sheet has melted away. So until the CO2 had increased to a level to melt more of the ice sheet than is recovered each winter the ice sheet was stable. When the positive feedback from the lowering of the height of the surface started to be applied, and the overall feedback became positive, then there was no way back for the ice sheet. I would say that was when it passed its tipping point.

But other people (including Tim Lenton) say that the tipping point is when âa small change can make a big differenceâ in other words it has not been reached until the melting begins to runaway. In the case of Greenland, as the ice sheet becomes lower and lower, the surface temperature will become higher and higher. So the positive feedback aiding the melting will increase each year. One year it will be so great that the ice sheet will collapse. This is the other type of tipping point when the runaway happens.

So, switching to sea ice, I am saying we have already passed the tipping point for the Arctic sea ice, but it has not reached a runaway state yet. Perhaps this summer. Who knows?

William, thanks for letting me use your blog to sort out my thinking on runaway dynamical systems. Does it make sense?

Cheers, Alastair.

[I think all of this is rather more complicated than at first appears, which is why attempts to define runaway and tipping point usually come to grief. First off, the world has diurnal and seasonal cycles, so any defn of "tipping point" has to cope with the idea that Greenland, say, has a 40 oC difference between summer and winter (I made up the 40 oC, btw, but its approx correct). The huge seasonal cycle is a problem for the seaice tipping point folk. I prefer to reserve "runaway" for irreversible, and keep "abrupt" for fast. So Greenland melting isn't runaway. It might be a tipping point, if we knew what one of those was -W]

By Alastair McDonald (not verified) on 01 Mar 2009 #permalink

My comment/questions may have been a bit OT but it was not about terminology but about mechanisms and I have no answers so far. The bit about the nuclear reactor was just thrown in because it illustrates a possible distinction between speed and total effect (the reactor can probably produce more heat).

I don't think I agree with your first sentence but don't wish to labour the point. In that example the slow chain reaction has to run with a total feedback factor of 1 and might also be a runaway in William's sense (at the start of this thread) because it might be said to continue until the "oceans boil dry" (well.. approximately so remembering that there are control rods to pull out gently to keep the reaction going). As for irreversibility of the reactor that depends on the way you look at it. The fuel cannot be recovered but the fission rate can be raised and then lowered back to its previous value. The bomb is not reversible in any sense.

By Geoff Wexler (not verified) on 01 Mar 2009 #permalink