Global Warming: Earth 10,000 AD

[Dinwar]

Well, considering volcanism is more or less random (as in, we don't have a good way to predict what will happen ,when it will happen, or where it will happen more than a few days/weeks in advance), I'd say this is a pretty substantial concern. I mean, if there's a Yellowstone-size eruption all the predictions from GCMs which use much smaller numbers more or less instantly become useless. If the results are much lower, they're also useless. We have a decent understanding of the average amount of volcanic activity over a significant period of time (a few hundred years or so), but an average is only a mathematical abstraction, not necessarily what happens in the real world.

You and I are looking at this from two different perspectives, however. I want to know more about the GCMs for purely entertainment reasons. I consider it fun to learn new things, particularly about something as important to political policy as these models. Therefore actionable predictions aren't a major consideration for me. I want to know how well GCMs deal with volcanic activity. Full stop. I may deal with practical aplications later. Or maybe not--I may decide that something else is more interesting after I learn more about this.

You, on the other hand, are (as far as I can tell) looking at this from a perspective of an engineer: "Don't tell me how the math works, just tell me what the equation is and what the answer is." (For the engineers in the room: I'm paraphrasing what my father, a civil engineer, used to say. We never really understood one another because I'm virtually incapable of using an equation I don't understand, while he always argued that understanding the equation isn't necessary once you know that it works.) This view is a very good thing--the whole reason these GCMs exist is because we're concerned with a problem and they can help us fix it or deal with it, so looking at them in terms of how to use the results is sort of the whole point.

If I've understood your possition correctly (and explained mine sufficiently), we're just coming at this from two different angles.

 

[CapelDodger]

Which makes how they deal with it even more interesting.

If you're that interested, go and find out about it. Don't expect to be spoon-fed it here. Let your inner scientist loose and do some research.

What I'd love to see is a range--models that have been run 1000 times or so (going off PAST's standard bootstrap/jack knife tests) using various levels of volcanic activity, ranging from none to a full-scale basalt trap cutting loose, and from the equatorial areas to the higher latitudes. That'd be the best way to demonstrate the effects of volcanism. We could probably cut down on the options using the most probable eruption locations, and eruption sizes.

Baring that I'd like to see 1000 papers on the topic, each covering a different section.

I know I won't GET it (supercomputers aren't cheap), but that's my wish list!

What we have is a relatively primitive climate model being fed the observed parameters of the Pinatubo eruption and modelling the climatic effects (which were short-term, as is usual for eruptions of any sort)perfectly adequately. Unless you want to put that down to luck you should give it up and move on, maybe look into the nuts-and-bolts of how it's done.

(Clouds are a favourite of many, by the way, and far more prevalent than climate-influencing eruptions. Anyone can find masses of uncertainty about cloud behaviour if they want to.)

 

[Dinwar]

Unless you want to put that down to luck you should give it up and move on, maybe look into the nuts-and-bolts of how it's done.

Give what up? Requesting someone reference the model they're talking about? Give up wondering how models will deal with the unknown of future volcanism? I *have* asked for the nuts-and-bolts; this wouldn't be the first time I've looked at a GCM from a nuts-and-bolts perspective. Nor is asking for 1000 reps unreasonable. Various models of mass extinctions do far more than a thousand reps, and it's standard for PAST's modeling (cladistics, mostly). Asking how many times the model has been run, and under what range of conditions, IS asking about the nuts-and-bolts of an inherently predictive model. One of my old professors got a pseudorandom number generator to generate Sepkoiski's Curve (with the exception of a few [not all] mass extinctions)--running a model once, to put it mildly, is not good enough to see if the results are real.

If you're that interested, go and find out about it. Don't expect to be spoon-fed it here.

I never did. I asked people who know more about the topic than I do (or at least know a different side than I do) where to start looking. When I had a question about alluvial fans I asked a sedimentologist for some references. It wasn't the only source I used, and I'd done some of my own research before asking him, so I wasn't asking to be spoon-fed information (this guy wouldn't do so anyway). However, when one realizes that others have more knowledge/experience in a certain topic I've never fealt it to be in bad form to ask those others for advice about where to direct one's studies.

For crying out loud, people, I asked for examples of the model because I was curious. I'm not saying that the models don't work, I'm not saying I don't believe anyone, I'm not trying to disguise my intentions in any way, I'm certainly not being lazy and not doing my own research. I'm at a complete loss as for why I've gotten the types of responses I have.

 

[CapelDodger]

Well, considering volcanism is more or less random (as in, we don't have a good way to predict what will happen ,when it will happen, or where it will happen more than a few days/weeks in advance), I'd say this is a pretty substantial concern. I mean, if there's a Yellowstone-size eruption all the predictions from GCMs which use much smaller numbers more or less instantly become useless. If the results are much lower, they're also useless. We have a decent understanding of the average amount of volcanic activity over a significant period of time (a few hundred years or so), but an average is only a mathematical abstraction, not necessarily what happens in the real world.

If Yellowstone goes off we're all screwed, so who cares if the models can tell us exactly how we get screwed? Models of such an event have, of course, been run and include a climate element.

You and I are looking at this from two different perspectives, however. I want to know more about the GCMs for purely entertainment reasons. I consider it fun to learn new things, particularly about something as important to political policy as these models.

You mix fun and politics? I keep them separate myself.

Therefore actionable predictions aren't a major consideration for me. I want to know how well GCMs deal with volcanic activity. Full stop. I may deal with practical aplications later. Or maybe not--I may decide that something else is more interesting after I learn more about this.

Go forth and learn, then, and do let us know how it works out for you.

You, on the other hand, are (as far as I can tell) looking at this from a perspective of an engineer: "Don't tell me how the math works, just tell me what the equation is and what the answer is." (For the engineers in the room: I'm paraphrasing what my father, a civil engineer, used to say. We never really understood one another because I'm virtually incapable of using an equation I don't understand, while he always argued that understanding the equation isn't necessary once you know that it works.) This view is a very good thing--the whole reason these GCMs exist is because we're concerned with a problem and they can help us fix it or deal with it, so looking at them in terms of how to use the results is sort of the whole point.

I thought the point was "fun"?

If I've understood your possition correctly (and explained mine sufficiently), we're just coming at this from two different angles.

You seem to be coming at it from at least two yourself - fun and serious stuff like politics.

Of course, everyday vulcanism only has a short-term effect and is of no bearing on what policies might be needed in light of AGW. Pinatubo faded into insignificance a long time ago.

 

[CapelDodger]

some geologists believe the term glacial period means that there are glaciers in existence, when they are gone, you have the interglacial period.

I very much doubt that. Interglacials occur between glaciations, not between "glacial periods".

 

[lomiller]

What I'd love to see is a range--models that have been run 1000 times or so (going off PAST's standard bootstrap/jack knife tests) using various levels of volcanic activity, ranging from none to a full-scale basalt trap cutting loose, and from the equatorial areas to the higher latitudes. That'd be the best way to demonstrate the effects of volcanism. We could probably cut down on the options using the most probable eruption locations, and eruption sizes.

Model runs ususaly already include typical volcanic activity. The scenarios you are asking for don't seem either typical or useful as they are very unlikely to occur. Climate models have been aimed at specific volcanic events like the lake Toba eruption.

http://www.sciencedirect.com/scienc...262c779b8153f23f5979c1edcc13d24b&searchtype=a

http://elements.geoscienceworld.org/cgi/content/abstract/4/1/41

 

[lomiller]

I mean, if there's a Yellowstone-size eruption all the predictions from GCMs which use much smaller numbers more or less instantly become useless.

And if an asteroid the size of the one 65M years ago hits it's all useless. Worrying about things that could, but almost certainly won't happen doesn't get us anywhere.

 

[CapelDodger]

Give what up? Requesting someone reference the model they're talking about? Give up wondering how models will deal with the unknown of future volcanism? I *have* asked for the nuts-and-bolts; this wouldn't be the first time I've looked at a GCM from a nuts-and-bolts perspective.

So you already know where to go to find about them. The same route you went previously.

Nor is asking for 1000 reps unreasonable. Various models of mass extinctions do far more than a thousand reps, and it's standard for PAST's modeling (cladistics, mostly).

Climate models are run as often as is practicable with given resources. Just as modellers of all sorts will make as many runs as they can. Why not?

Climate modelling is at least as well-resourced as your other examples, so I think we can rest easy that GCM's have been just as well exercised. I'm sure you can find the details out there on the internet.

Asking how many times the model has been run, and under what range of conditions, IS asking about the nuts-and-bolts of an inherently predictive model.

No, it isn't. "Nuts-and-bolts" refers to how it works, whereas the number of runs refers to measuring how well it works.

One of my old professors got a pseudorandom number generator to generate Sepkoiski's Curve (with the exception of a few [not all] mass extinctions)--running a model once, to put it mildly, is not good enough to see if the results are real.

Getting one out of one right (Pinatubo) with a primitive model, and at most a few dozen runs I should think, is a pretty powerful argument. It's also clear from historical records that the effects of Pinatubo were very like many similar cases. It's actually not a difficult problem to model.

For crying out loud, people, I asked for examples of the model because I was curious. I'm not saying that the models don't work, I'm not saying I don't believe anyone, I'm not trying to disguise my intentions in any way, I'm certainly not being lazy and not doing my own research. I'm at a complete loss as for why I've gotten the types of responses I have.

For a list of climate models see the IPCC reports. For details of how they work you'll have to look to the source institutions.

 

[Trakar]

Model runs ususaly already include typical volcanic activity. The scenarios you are asking for don't seem either typical or useful as they are very unlikely to occur. Climate models have been aimed at specific volcanic events like the lake Toba eruption.

http://www.sciencedirect.com/scienc...262c779b8153f23f5979c1edcc13d24b&searchtype=a

http://elements.geoscienceworld.org/cgi/content/abstract/4/1/41

And if an asteroid the size of the one 65M years ago hits it's all useless. Worrying about things that could, but almost certainly won't happen doesn't get us anywhere.

Well, and as you indicate, we can use these models to simulate the impact of these types of events on the climate, and these are occassionally run, but not as a part of ordinary climate research, as these are not technically climate events, but rather, rare and extraordinary geophysical events with a climate impact.

Interesting, no doubt, but not a foundational or primary focus of typical climate study.

 

[lomiller]

Climate models are run as often as is practicable with given resources. Just as modellers of all sorts will make as many runs as they can. Why not?

Climate modelling is at least as well-resourced as your other examples, so I think we can rest easy that GCM's have been just as well exercised. I'm sure you can find the details out there on the internet.

I've seen Gavin Schmidt generate model runs for blog postings they can't be that hard to get going (he was a co-author on a related paper later on so so the run probably doubled as preliminary research). 1000 runs, however, is way to much to ask for a relatively uninteresting topic that would probably never see publication

 

[BenBurch]

I've seen Gavin Schmidt generate model runs for blog postings they can't be that hard to get going (he was a co-author on a related paper later on so so the run probably doubled as preliminary research). 1000 runs, however, is way to much to ask for a relatively uninteresting topic that would probably never see publication

Depending on the model that can be a LOT of CPU cycles.

 

[Trakar]

Depending on the model that can be a LOT of CPU cycles.

Well, in addition to being expensive and resource intensive, it would simply be inefficient and inaccurate to throw in a lot of extremely rare or unlikely factors in the study of more general recent, current or near future climate as it would skew the average results for no reasonable purpose. I'm sure I've read studies in which the rates of vulcanism are varied among modelling runs, but if I recall correctly the range was fairly narrow in keeping with what was expected based upon what the rates have been over the last century or so. I don't want to get any more specific than this until I find the study(ies) I recall, but its been a while and an initial search has yet to prove fruitful. Catastrophism in the form of amplified or rampant volcanism is an interesting subject, but not clearly indicated to be a major factor in our planet's climate over the last, or next, 8,000 years.

 

[Trakar]

Well this is turning into a bigger task than I anticipated! I’m still looking for the specific volcanism forcing studies I remember. It’s a matter of recalling too few details and so having to detail skim through a large collection to find the specific bits and pieces I do recall to identify the proper papers, I'll keep at it over the next couple days,…meanwhile, for those interested I did run across some early climate modeling papers that are very useful in coming to terms with a lot of the basics:

“Modelling climate and the nature of climate models: A review”- International Journal of Climatology, Jan/Mar 1983 – unfortunately, I believe this one is still behind a paywall,

“Forty Years of Numerical Climate Modelling” – That’s another International Journal of Climatology article from 2001 with link - http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.117.8259&rep=rep1&type=pdf

“Global Climate Changes as Forecast by Goddard Institute for Space Studies Three-Dimensional Model” – Journal of Geophysical Research, 1988 with link - http://www.atmos.washington.edu/2008Q2/111/Readings/Hansen1988_climate_forecasts.pdf

“General Circulation Modelling of Holocene climate variability” Quaternary Science, 2004. Link- http://www.giss.nasa.gov/staff/gschmidt/qsr04.pdf

And here’s a few of the more important volcanic effect studies:

“On the surface and atmospheric temperature changes following the 1991 Pinatubo volcanic eruption: A GCM study” – Journal of Geophysical Research, 2002 with link - http://www.emc.ncep.noaa.gov/gmb/wx24fy/fyang/pdf/yang_jgr2002_volcano.pdf

“Volcanic Eruptions and Climate” - http://adrem.org.cn/Faculty/GongDY/class2008/VOLCANIC ERUPTIONS AND CLIMATE.pdf
(and to be honest, most of Alan Robock’s - http://climate.envsci.rutgers.edu/robock/robock_volpapers.html - body of work,…IMO)

But I suspect more of what you might be looking for is along the lines of this:

“A statistical volcanic forcing scenario generator for climate simulations” Journal of Geophysical Research, 2010 But it is still behind a paywall - http://www.agu.org/pubs/crossref/2010/2009JD012550.shtml

 

[Pixel42]

I'm at a complete loss as for why I've gotten the types of responses I have.

Having joined the forum fairly recently you don't know the history of AGW discussions here. Until the main thread was put on moderated status it was impossible to have an adult discussion on the topic. It was common for posters to arrive posing apparently innocent questions and then resort to sarcasm and abuse when they didn't get the answers they'd be led by denier blogs to expect. Long term posters on the subject are therefore understandably suspicious when they see newbies posing these sort of questions. Continue to post as sensibly as you have been and you'll soon earn their trust, but sadly on this particular issue trust does have to be earned, it isn't given until there is good reason to withdraw it as it would be on most other issues.

 

[Dinwar]

If Yellowstone goes off we're all screwed, so who cares if the models can tell us exactly how we get screwed? Models of such an event have, of course, been run and include a climate element.

Well, considering 1) it WILL happen again, and 2) I'm interested in this from a more academic angle (read, not necessarily just using these models to determine what to do about AGW), so how we're screwed is actually exactly what I'm looking for. If you're not that's fine by me--these models are very powerful, and can be used to answer a number of questions.

You mix fun and politics? I keep them separate myself.

What I actually said was that this topic (AGW, and GCMs) is very important to a lot of political decisions that are going on right now, and that's part of why I consider it fun to learn more about them. I didn't say that I mix politics with fun. This stuff is important, which is one reason why I find it interesting.

I thought the point was "fun"?

No, I said the point for me was fun. Your statements indicate that you're more interested in how GCMs can help us deal with AGW. I'm not as interested in that topic. I tend to look backwards, rather than forwards--I'm better trained at looking at fossils than predicting things that haven't happened yet.

You seem to be coming at it from at least two yourself - fun and serious stuff like politics.

Again, I mentioned that these models are important in a lot of political decisions being made. Because those are going to affect my life rather substantially, how these models work is of no small concern. However, that's hardly my only reason, or even the most important reason, for looking into them.

So you already know where to go to find about them. The same route you went previously.

Except that, as I've mentioned previously, the ones I've looked at don't deal with volcanism (or at least how volcanism is dealt with is seriously underreported).

No, it isn't. "Nuts-and-bolts" refers to how it works, whereas the number of runs refers to measuring how well it works.

In part. It's a useful tool in determining how messed up the models are (no slight against GCMs, but all models are messed up, it's only a question of how badly). And if something doesn't work well, that's a fairly good indication that the nuts-and-bolts side is screwy.

Getting one out of one right (Pinatubo) with a primitive model, and at most a few dozen runs I should think, is a pretty powerful argument. It's also clear from historical records that the effects of Pinatubo were very like many similar cases. It's actually not a difficult problem to model.

If it got it right once out of a few dozen runs that means that the model may or may not give accurate results the rest of the time. I'm not saying that the model is definitely inaccurate, but I've dealt with statistical approaches to problems too often to trust this sort of thing without looking into it (and that's what a GCM is under the hood--a statistical model). It's far too easy to see what you want to see, or to construct a model spaghetti-code style so that it works perfectly for the event you're ground-truthing it to.

Interesting, no doubt, but not a foundational or primary focus of typical climate study.

I never thought that this kind of thing would be, and I think people are focusing on one end-member of the spectrum of volcanic activity (or, to put it more accurately, they think that I'm focused on it). I fully acknowledge that a Yellowstone event is rare. A Chixilub event is even more rare (though a GCM of that event would hardly be without value, at least to paleontologists). What I meant was that there is a range of volcanic activity--some years there's more, some years there's less; some years the biggest eruption is small, some years it's large. Obviously high-amplitude events will be low-frequency. From my perspective (not an expert, but not completely ignorant either), if a model doesn't take into account this variation in volcanic activity it's not really very useful. It's missed a key component to our planet. And I'm not talking single events either--carbon dioxide emitted by deep-sea vents or low-amplitude, high-frequency eruptions (the mafic style of eruptions found in Hawaii, for example) will be just as important as Mount St. Helens, if not more so.

Yeah, I'm probably not going to find a single model that accounts for everything I think should be included. And some of what I think should be included may turn out to be irrelevant to the models--that's one reason I'm bothering to post about it, so that if my understanding is too far off base I can be corrected. But simply saying "If your most extreme end-member occurs we're all dead so we shouldn't worry about it" isn't a good critique of my statement.

 

[lomiller]

Well, considering 1) it WILL happen again, and 2) I'm interested in this from a more academic angle (read, not necessarily just using these models to determine what to do about AGW), so how we're screwed is actually exactly what I'm looking for. If you're not that's fine by me--these models are very powerful, and can be used to answer a number of questions.

Fine that you are interested but
1) it’s really not a climate change topic
2) it’s also a major goalpost shift from what you first asked
3) you have already been linked to modeling of an even larger eruption then Yellowstone

In part. It's a useful tool in determining how messed up the models are (no slight against GCMs, but all models are messed up, it's only a question of how badly). And if something doesn't work well, that's a fairly good indication that the nuts-and-bolts side is screwy.

Just a few statements about you said you said you are no longer asking about the models you have changed to asking about what they say about super-volcano eruptions. When given information you immediately move the goalposts and insist someone needs to model specific super-volcanoes.

Worse you insist they model hypothetical super-volcanoes, which isn’t a climate modeling problem at all. It’s a volcanology problem to figure out when the eruption could occur and what it may look like. Then climate models can probably predict the effect, but they need meaningful input data.

If it got it right once out of a few dozen runs that means that the model may or may not give accurate results the rest of the time.

Models of the day got the right results period. Full stop.

Again, I mentioned that these models are important in a lot of political decisions being made. Because those are going to affect my life rather substantially, how these models work is of no small concern. However, that's hardly my only reason, or even the most important reason, for looking into them.

But you are not asking about models at all you are asking about various “what if” scenarios. Climate Models are commonly used to investigate this type of problem, but the underlying scenario needs to be of interest for people to spend their time on that scenario.

 

[BenBurch]

...

But you are not asking about models at all you are asking about various “what if” scenarios. Climate Models are commonly used to investigate this type of problem, but the underlying scenario needs to be of interest for people to spend their time on that scenario.

Exactly, and the thing being asked for is like asking climate scientist to model an Asteroid impact of Dino-killer size as part of their predictions for human climate change. Might as well ask them to include the effects of unicorn flatulence in their models. Its a ridiculous request.

 

[Dinwar]

Just a few statements about you said you said you are no longer asking about the models you have changed to asking about what they say about super-volcano eruptions.

No. Please read what I've actually written--I said that it'd be interesting, and that because it's going to happen climate models which don't (can't) handle it are necessarily limited--BUT I have also acknowledged that such volcanism is rare, and that what I'm ACTUALLY after is seeing a range of volcanic activity accounted for, from some background level to a reasonable extreme.

YOU are focusing on the supervolcanoe eruptions, and cherry-picking quotes out of context like a Creationist. *I*, on the other hand, want to see how these models deal with the natural variability inherent in volcanism.

Worse you insist they model hypothetical super-volcanoes,

No. I said that my wish-list included a model which included, AS PART OF A SUITE OF RUNS ALONG A REASONABLE CONTINUUM OF VOLCANIC ACTIVITY, the effects on climate of a supervolcanoe eruption. I put that in caps so it's perfectly clear. I'm not saying I only want to see what will happen if one supervolcano cuts loose. I'm not saying the models need to deal with that possability. I'm saying that if you gave me a genie and I was restricted to wishing for GCM data that's what I'd wish for.

Models of the day got the right results period. Full stop.

No. NOT full stop. How did they do it? How much tweeking was involved? How close were they ("pretty close" tells me nothing--I'm used to a million years being a rounding error)? More importantly, to my mind, will this model work for other eruptions? Latitude has a significant effect on the climateological impacts of a volcanic eruption--a model that works perfectly fine for an eruption in the African rift vallies may fail misserably for a big eruption in Washington State, and neither may be able to deal with an eruption in the Arctic Circle.

But you are not asking about models at all you are asking about various “what if” scenarios. Climate Models are commonly used to investigate this type of problem, but the underlying scenario needs to be of interest for people to spend their time on that scenario.

The entire concept of projecting future climates is one gigantic game of "What If". I'm not saying it's without value--we're VERY good at the game, after all. But because we don't know the future (James Randi still has his million bucks) it's reasonable to expect people projecting future climate conditions to deal with the natural variability in the system. If the models are for volcanism as it stands right now and volcanism increases in a hundred years, the models either need to be able to adapt to that or are worthless.

But you should have noted something: I'm not asking for the results. I never have. What I'm asking for is how these models deal with a specific variable. This, believe it or not, has applications outside of predicting the climate in the future. Climatologists use paleoclimateological data to ground-truth their models--they sometimes run OIS 11 and see if their models yield what actually happened, for example. Before the closing of the Isthmus of Panama isn't really useful, because we were essentially a different planet back then, but GCMs have aplications in paleontology. Considering I'm working with Miocene to Pleistocene stuff anymore, this is somewhat interesting to me.

As I said before, this isn't the first time I've looked into GCMs. But the models I've seen in the past didn't discuss volcanism; they did discuss most other factors that I've seen discussed here. So I was curious. Understanding how GCMs handle volcanism will help me better understand these tools, both in terms of their predictive capability and their descriptive capability. Considering the variation in volcanic activity over the timescales I'm working in it's reasonable to ask how they deal with that variability--and pointing to a single eruption, while not worthless, hardly answers my question.

But hey, if you want to misrepresent my request as an attack (somehow; I'm still not clear how), and my indication of what I'd like to see as a demand that we focus on the most extreme examples I happen to mention, go ahead. Not like I haven't been misquoted or attacked for arguments I haven't even made before.

 

[Dinwar]

I think I need to re-state my request, if for no other reason than to clarify things:

I would like to know more about how GCMs deal with volcanism; therefore, I would appreciate it if those who know more than me can provide examples of GCMs which do so. What I would expect to find, if GCMs deal with volcanism adequately, is that these models are run multiple times, with a range of volcanic activity from some background (probably recent volcanic acticity) to high volcanic acticity (for example, the eruption of a cauldera). If I don't find that, I'll take what I do find.

I am not looking for this to know what will happen in the future, but rather to know how we determine it. It's an important issue in our time, and one that's certainly worth educating one's self on. Furthermore, because GCMs are sometimes used in my field it would benefit me to understand how they work (not necessarily what the results are), so that I can adequately evaluate those that I run across.

If this is not clear, please let me know which parts are the problem, so that I can clarify it. This is the perspective I've been coming at this from since I first posted my question, though it may not be clear.

 

[BenBurch]

OK, quite simply vulcanism is modeled as a temporary cooling effect. The CO2 and other GHGs emitted by a single eruption are very small in relation to the total in the atmosphere, but the aerosols are sometimes larger than the amount in the air at any time, and so have a significant effect. As they slowly rain out, the effect goes away.