we are done.

the only variability you can predict is the variability of your models. That variability may or may not have any connection with reality.

I did have that caveat in my previous post. Obviously, when using a model, you are limited by the extent to which the model has the same sort of features as the real system. As I noted, the models do seem to at least roughly capture the sort of variability seen in the real world.

In fact, the very mention of “initial condition sensitivity” is an admission of that lack. You are saying that, even if you plug in what you now know, your calculations yield wildly varying results.

Again, not knowing everything doesn’t mean we know nothing. Yes, the sensitivity to initial conditions means that we should not make predictions that depend on these initial conditions. This means not believing the individual jigs and jags up and down in any given run of the model. However, things like the trend over 100 years or the magnitude of the variability (i.e., the characteristic size of those ups and downs) turns out to be independent of the initial conditions.

Are you now agreeing there are natural cycles in climate that can swamp puny CO2 effects?

What I am saying is what happens is exactly the same as what happens for the seasonal cycle: There, we see that the day-to-day weather variations dominate over the seasonal cycle…so for trends on the scale of days to a few weeks, we are likely to see trends that are opposite to what the seasonal cycle predicts (e.g., we can see cooling trends in the spring and warming trends in the fall). However, as we take the trend over longer periods (say, a month or two or three), the trends due to the seasonal cycle become much more resilient even though the weather noise is still there.

I doubt you would want to claim that the seasonal cycle is a “puny” effect in Rochester even though the noise can dominate over it over short periods.

Even if you aren’t, a constantly increasing forcing factor should have a constantly increasing effect, if only on the average. A 10 year hiatus means that preceding and following periods must show an increased effect for the average to be realized. You now need to explain this variability.

Actually, the several year period up to the late 90s did have a greater than average trend, I believe. Obviously, we will have to see what the future holds in store.

The entire business of bringing them up is a scare tactic. It’s like claiming we live on a plateau with a western cliff so all progress toward the west should be discouraged immediately cease. Never mind that we haven’t the faintest idea whether it’s even there, or how far if it is.

It is not a scare tactic to acknowledge very real dangers even if we are still hazy on the details on some of them. It is important to quantify what we know in each case, but simply to completely ignore the possibilities because we are uncertain about them is foolhardy. (By the way, for internal consistency, I assume that with the attitudes you express here, you must have been one of those who were on the front lines of the protests against our invasion of Iraq?)

“Give me all your money or life as you know it is over, citizen. Yeah and that means that the polar bear gets it too…”

Sorry, but you can do nothing of the sort. For one, the average is the prediction! Secondly, the only variability you can predict is the variability of your models. That variability may or may not have any connection with reality. Since no single model is good enough as tested through predictive accuracy (the reason for averaging them in the first place), you can’t use them to for probability statements — even tacit ones.

I used a dice roll because it is a synonym for random (meaning unpredictable) event. You can only speak of probable outcomes in dice rolls because the number of possible outcomes is completely known. If you know little about dice, you haven’t a leg to stand on. The climate is far more complex than a pair of dice with a vast, likely infinite, number of possible outcomes. You are going to have a hard time showing you know enough of them to support predictions of them. In fact, the very mention of “initial condition sensitivity” is an admission of that lack. You are saying that, even if you plug in what you now know, your calculations yield wildly varying results.

think of the seasonal cycle. Certainly, we can make a more robust prediction regarding the average temperature for the first half of April relative to the average temperature for the first half of January than we can regarding, say, the average temperature

Are you now agreeing there are natural cycles in climate that can swamp puny CO2 effects? Even if you aren’t, a constantly increasing forcing factor should have a constantly increasing effect, if only on the average. A 10 year hiatus means that preceding and following periods must show an increased effect for the average to be realized. You now need to explain this variability.

I don’t find this fact particularly reassuring as we apply a significant forcing to the climate system.

The entire business of bringing them up is a scare tactic. It’s like claiming we live on a plateau with a western cliff so all progress toward the west should be discouraged immediately cease. Never mind that we haven’t the faintest idea whether it’s even there, or how far if it is.

With so many different outcomes to choose from how can they be predictive?

…

The only thing useful (in terms of prediction) from so many possible outcomes IS the average.

I would say sort of, but not quite. What one can predict is the average and also some measure of the amount of variability that you can expect to get about that average. So, in the case of future climate change, we can predict the average trend that we expect and how much variability one might get around that trend if we look over, say, a 10 year time span. (This assumes that the models do a good job capturing such variability, which I admit I am not up on the details of, although my impression is that they do at least a pretty good, although certainly not perfect, job of capturing such variability.)

Getting back to the seasonal cycle analogy (sorry to keep returning to this, but the more I think about, the more useful I realize it is): One thing that is definitely useful is to predict how the average temperature varies over the year. However, there are other things that might be useful too. For example, let’s say that we have data for February here in Rochester that shows that there has actually been a negative trend over the month for temperatures. One might want to ask, “How compatible is this negative trend with the seasonal cycle theory (since it predicts that the average February should have a positive temperature trend)?” With an appropriate climate model, one would presumably be able to answer this sort of question.

Hmmm, yasss! “Sensitive to initial conditions” is code-speak for “there’s a lot we don’t understand”. Dice rolls are “sensitive to initial conditions”. It’s a bit of hubris to claim knowledge of the effects of a small component on the outcome when the major drivers can’t be predicted, isn’t it?

Well, yes, dice rolls are “sensitive to initial conditions” and yet I think you would agree that if I took a die and altered it by painting the side with one dot to have six dots then you could make some predictions as to how this die will behave on average relative to a regular die even though you can’t predict what the next role would be. Likewise, in regards to the seasonal cycle analogy, even if we can’t predict the day-to-day weather for more than about a week in advance, I think you will agree that we can make some general predictions regarding what the temperature will be in July relative to what it will be in January. And, in fact, the same climate models that show extreme sensitivity to initial conditions still show that under a given forcing such as an increase in CO2, the basic change in the global climate is predictable even while the individual jiggles up-and-down from year-to-year are not (because they depend so delicately on the initial conditions).

10 years represents 1/10 of a century and 1/3 of the period that seems to be the major “proof” of said forcing. One would think steady forcing would cause steady change but a flat period implies it happens in spurts. A 10 year hiatus implies those spurts happen under no more than 2/3 of an interval that is supposedly sufficient to use as a climate indicator.

Again, think of the seasonal cycle. Certainly, we can make a more robust prediction regarding the average temperature for the first half of April relative to the average temperature for the first half of January than we can regarding, say, the average temperature for the first half of February relative to the average temperature for the first half of January. And, it doesn’t mean that the seasonal cycle happens in spurts; it just means that there is considerable weather noise in the system and thus the longer period you look at the more likely you can see the underlying trend above the noise.

However, I find it hard to believe we know enough to know the true (vs. hypothesized) cause(s) let alone being able to predict when one is nigh.

I haven’t claimed that exactly where the tipping points are is particularly easy to predict. However, unlike you, I don’t find this fact particularly reassuring as we apply a significant forcing to the climate system.

With so many different outcomes to choose from how can they be predictive? A bit like predicting the Super Bowl with wild point spread predictions isn’t it? Somebody will win and we’ll know which predictor was right (er, closest) as soon as the game is over, no? Like a common weather forecast in my parts during the summer: sunny with a possibility of thunderstorms. IOW: you may or may need an umbrella. Doesn’t leave much to chance — and is conveniently hard to pin down to a wrong, eh? nudge nudge.

The only thing useful (in terms of prediction) from so many possible outcomes IS the average.

… El Nino and La Nina that no one claims to be able to predict accurately because they are very sensitive to initial conditions

Hmmm, yasss! “Sensitive to initial conditions” is code-speak for “there’s a lot we don’t understand”. Dice rolls are “sensitive to initial conditions”. It’s a bit of hubris to claim knowledge of the effects of a small component on the outcome when the major drivers can’t be predicted, isn’t it?

This one: Is a period of, say, 10 years over which the temperature is flat something that the climate models predict should never happen in a world of steadily increasing greenhouse gas forcing?

10 years represents 1/10 of a century and 1/3 of the period that seems to be the major “proof” of said forcing. One would think steady forcing would cause steady change but a flat period implies it happens in spurts. A 10 year hiatus implies those spurts happen under no more than 2/3 of an interval that is supposedly sufficient to use as a climate indicator.

a tipping point would be where the place where a small perturbation could cause the climate to change by a large amount

Every reversal with differing slopes ultimately has one then — it’s just a matter of degree. OK. However, I find it hard to believe we know enough to know the true (vs. hypothesized) cause(s) let alone being able to predict when one is nigh. I seem to recall you mentioning some difficulty due to sensitivity to initial conditions.

The only legitimate reason for model averaging is if uncertainty in model selection makes any individual model untrustworthy.

No. It is also useful in averaging over the unforced aspects of climate in order to see the general trend due to the forced aspects. Take our analogy to the seasonal cycle: It is nice to produce a pretty plot that shows the average temperature in Rochester as a function of the day of the year and clearly demonstrates the smooth cyclic seasonal cycle forcing that would be much noisier if we just looked at a single year’s temperatures.

However, my point was not to get in a discussion of whether it is good to average models but simply to point out that the correct thing to look at is what the individual runs of the models predict, not the averages predict, in order to get an idea of the different possible courses of the climate.

Even if I did buy it, it looks as if each model output consists of nothing BUT short-term spikes and no single one agrees with the record over the previous decade although the graphic at RC makes it hard to tell.

Noone is claiming that they do. Getting those individual spikes right would entail being over to predict short-term oscillation in climate such as El Nino and La Nina that noone claims to be able to predict accurately because they are very sensitive to initial conditions. (I.e., even the exact same model run with just a slight perturbation to its initial conditions produces a different pattern of spikes and dips.) However, what one can do is look at those individual runs to see how much, say, the trend over a 10 year period can vary. Remind what the original question was: Is a period of, say, 10 years over which the temperature is flat something that the climate models predict should never happen in a world of steadily increasing greenhouse gas forcing, or is it something that should not be that uncommon?

Hmmm… repeat again, please, your definition of “tipping point”.

In my view, a tipping point would be where the place where a small perturbation could cause the climate to change by a large amount. Think of being at the top of a hill between two valleys where a push one way or the other will send you into one valley or another. In the Younger-Dryas example, there seems to be a point at which there is a dramatic change in ocean currents that leads to a very different climate state.

The only legitimate reason for model averaging is if uncertainty in model selection makes any individual model untrustworthy. IOW: when you can’t decide which model to use. It’s like polling and getting wildly differing opinions. No one opinion can be labelled as “correct” and the best estimate is the average of all of them. The test for predictive power is always made using the average. If you CAN point to one of the outputs and say “THIS is the right one!” it begs the question of why you didn’t use it alone in the first place. So I don’t buy your explanation.

Even if I did buy it, it looks as if each model output consists of nothing BUT short-term spikes and no single one agrees with the record over the previous decade although the graphic at RC makes it hard to tell.

I didn’t say that they were. I don’t know why Smokey linked to that graph in the first place except that he always likes to link to these deceptive graphs from totally ridiculous sources

Yeah that particular link was completely unwarranted. Nonetheless, I note that you avoided the gist of what preceded it, which was a good point. Namely,

Before discussing “tipping points” as if they are actually proven to exist, first, try to falsify the hypothesis that the climate is acting normally.

Younger-Dryas events

Hmmm… repeat again, please, your definition of “tipping point”.

Such as…?

Younger-Dryas events ( http://en.wikipedia.org/wiki/Younger_Dryas ).

Please define “not very long”. The graphs at RC, particulary this one only show negative excursions lasting say 1-3 years with no prominent flattening.

No…Look at the distribution of trends. What you are apparently noticing is that the thick black line, representing the average over all the different climate model runs behaves that way. However, that is not the correct thing to look at: The average over all different runs is akin to averaging over all the weather and short-term climate variations (such as the ENSO events). It is useful for seeing the overall climate trend (and is, as expected, pretty steadily upward except when there are volcanic eruptions) but it is not a representation of what one expects the earth’s climate system to actually follow because in the earth’s climate system we do not average over such weather and climate variations. What the earth is actually following is akin to a single run of a climate model, with one particular realization of the weather and climate.

One could make an analogy with seasonal cycles here. If I showed you a plot of the AVERAGE daily temperature here in Rochester over the course of a year, you would see a smooth oscillation with temperatures rising steadily in the spring and falling steadily in the fall. However, if we actually plotted the temperature over any given year, you would not see such smooth steady trends, but would instead see lots of noisy ups and down superimposed on this general cycle.

Lucia’s opinions on Monckton graphs are hardly evidence for “tipping points.”

I didn’t say that they were. I don’t know why Smokey linked to that graph in the first place except that he always likes to link to these deceptive graphs from totally ridiculous sources like Monckton or IceCap that purport to show that AGW is wrong. So, I commented on it.

FYI: Technically speaking, mathematics is deductive while science is abductive.
http://en.wikipedia.org/wiki/Abductive

Second of all, I don’t think the evidence of tipping points comes from climate models at all. It comes from the past paleoclimate history showing some dramatic shifts in climate or greenhouse gas levels or the like, along with an understanding of how highly nonlinear interactive systems tend to behave.

Such as…?

Except that large variations in the temperature trends, including negative trends, is exactly what the models do predict over time periods that are not very long: http://www.realclimate.org/index.php/archives/2008/05/what-the-ipcc-models-really-say/langswitch_lang/in

Please define “not very long”. The graphs at RC, particulary this one only show negative excursions lasting say 1-3 years with no prominent flattening.

Even Lucia, no friend of the IPCC and its predictions, finds the Monckton graphs to be by her euphism “artful”: http://rankexploits.com/musings/2009/more-artful-graphs-from-monckton/ A real skeptic might show a little more skepticism towards his sources.

Lucia’s opinions on Monckton graphs are hardly evidence for “tipping points.”

An exact word for word, puctuation for punctuation duplicate of Joel Shore (18:43:30) : minus the blockquote.

How odd!

We know enough about the drivers of climate to make reasonable suppositions about the past. For example, the warming period following a glacial period is surmised as a combination of a small change in the orbit of the planet combined with the effects of the release of greenhouse gases such as CO2.

“…it must be shown that the death was not only unnatural and not explainable by previous causes (even if unknown) but that it also has human origin.”

AGW has ruled out natural causes such as solar variation for especially the past 30-odd years of warming. We know that the 35 per cent rise in CO2 is due to human causes; we know the heat-retaining properties of CO2; we know the atmosphere is warming.

In other words, on the crime analogy, the usual suspects have been ruled out, and we have the murder weapon, a motive and the corpse. Them humans done it, m’lud!

The case for anthropogenic CO2 explaining the end of the last century has not been made. This case MUST be made before discussing anything arising as a consequence.

Well, in your opinion, the case has not been made. However, in the opinion of the scientific community as a whole, as expressed through the normal organs of that community, such as the U.S. National Academy of Sciences, it has.

REPLY: I can’t tell if you are really interested in discussion of if you are simply trying to blast enough comments here to promote your gaming website, ht83 dot com linked in your name. If you are interested in commentary, please lose the link. As I don’t allow those types of links here. – Anthony

There is no proof that a “tipping point” is right around the corner, or that such a creature even exists. By repeatedly opining on where these “tipping points” will be, you assume that such a thing as a “tipping point” exists outside of computer models or James Hansen’s fevered imagination.

First of all, there is no “proof” of anything in science. Science is inductive. If you want proof, go to mathematics which is inductive. Science simply deals with evidence.

Second of all, I don’t think the evidence of tipping points comes from climate models at all. It comes from the past paleoclimate history showing some dramatic shifts in climate or greenhouse gas levels or the like, along with an understanding of how highly nonlinear interactive systems tend to behave.

Since the past decade’s flat to cooling climate behavior shows us that models can not predict, then you cannot credibly argue that such a thing as a future “tipping point” even exists.

Except that large variations in the temperature trends, including negative trends, is exactly what the models do predict over time periods that are not very long: http://www.realclimate.org/index.php/archives/2008/05/what-the-ipcc-models-really-say/langswitch_lang/in

Before discussing “tipping points” as if they are actually proven to exist, first, try to falsify the hypothesis that the climate is acting normally. If you can do that, you’re a better man than all of the IPCC’s political appointees put together: click

Even Lucia, no friend of the IPCC and its predictions, finds the Monckton graphs to be by her euphism “artful”: http://rankexploits.com/musings/2009/more-artful-graphs-from-monckton/ A real skeptic might show a little more skepticism towards his sources.