What On Earth? Is The Gulf Stream Broken?

[Rapid Descent]

Oh FFS, how many times does this ridiculous garbage come up?

The current cold spell (which is affecting the UK, and some parts of Scandinavia) is caused by a local weather system over the area. It has nothing to do with a slowing, or anything else, of the gulf stream. Indeed, other parts of Europe (further east) have had above average temperature for the time of year.

The gulf stream is a wind driven gyre. The winds are driven by the coriolis effect as the earth rotates. Ocean salinity has little or no effect on it. The only way to stop the gulf stream would be (1) to stop the earth rotating or (2) rearrange the continents. Global warming is not going to cause either of these things.

The North Atlantic Drift, a thermohaline circulation that loops up and around the north atlantic (hence the name). This is dependent on salinity. The North Atlantic Drift is dependent on the gulf stream, but the converse is not true. The ocean circulation patterns probably add around 2-3 degrees C to our winter temperatures at best, and have no effect whatsoever on the summer temperatures.

[South Lorne]

Oh FFS, how many times does this ridiculous garbage come up?

The current cold spell (which is affecting the UK, and some parts of Scandinavia) is caused by a local weather system over the area. It has nothing to do with a slowing, or anything else, of the gulf stream. Indeed, other parts of Europe (further east) have had above average temperature for the time of year.

The gulf stream is a wind driven gyre. The winds are driven by the coriolis effect as the earth rotates. Ocean salinity has little or no effect on it. The only way to stop the gulf stream would be (1) to stop the earth rotating or (2) rearrange the continents. Global warming is not going to cause either of these things.

The North Atlantic Drift, a thermohaline circulation that loops up and around the north atlantic (hence the name). This is dependent on salinity. The North Atlantic Drift is dependent on the gulf stream, but the converse is not true. The ocean circulation patterns probably add around 2-3 degrees C to our winter temperatures at best, and have no effect whatsoever on the summer temperatures.

...Geography teacher or Scientist...?.....

[Rapid Descent]

...Geography teacher or Scientist...?.....

Geography teacher???

That's fighting talk where I come from.

[Kurt Barlow]

You've obviously not seen the forecast have you?

Apparently there is a snow storm coming to Saudi Arabia, so I'd batten down the hatches, no seriously.

Excellent - that will be a novelty. I have heating and the house is well insulated. I expect it will be rain here as we are effectively on a spit with the sea one side and shallow lagoon the other.

Its 23 today and cloudy which is unusual.

[Kurt Barlow]

That's like winter here isn't it!

I know a guy who works in the 'Kingdom' and he suffers from the cold when he comes back to England.

Aha - so am I right in guessing you too are in the Magical kingdom / neighbouring state? Whereabouts?

[Kurt Barlow]

Apologies that's not what I meant, I should have said that's like winter there i.e. it doesn't get much cooler.

no I'm on the British South Coast freezing my b@lls off!

It gets a lot colder in land. Im at Ras Tanura which is on a coastal spit and means Cape Oven in Arabic

[Rapid Descent]

Well, I'm still waiting for fluffy666 to provide supporting evidence of his claim that an LTP model for background variability has no physical basis. I provided 3 peer reviewed references, two of which demonstrated the physical basis for the model, and have not heard anything back yet. Still waiting, fluffy. While I'm here, let's just take apart this little diatribe:

Skeptics will never, as an article of faith, admit that global warming is in that the strongest theory explaining current observations by a very long way, and they will also never admit that there could be any negative consequences.

Firstly, science doesn't work by assessing the "strongest theory", it works by falsification, and there is much that the current theory simply gets wrong (as I have outlined in previous threads). Science *always* holds "we don't know" as a valid answer, and fluffy fails to recognise this.

Skeptics will rarely if ever criticize one another even when their respective sects directly contradict one another (It's not warming! It's warming because of the Sun!).

True of some sceptics, not of others. Hasty generalisation. I have criticised many people on the sceptic side as well as on the warmist side for getting their facts wrong.

They are perfectly happy to lie, conspire, threaten and bully in service of 'the cause' whilst being hypersensitive to even the slightest hint of such behavior from 'the other side' - for example, plagiarism, stealing emails, breaking into labs and issuing death threats are fine, but getting annoyed with people who routinely lie and misrepresent isn't.

Yawn. Yes, people lie, threaten, bully - it is human nature. Sceptics do it, warmists do it. Death threats are a routine part of being a publicly known sceptic as well - just that most sceptics recognise it as such and don't play the victim card.

Plagiarism? Are we referring to John Mashey's conspiracy-theory laden "report" into the Wegman congressional report? Firstly, the plagiarism charge is not proven, but is under investigation. Secondly, as Steve McIntyre showed, whilst the Wegman report's referencing is sloppy in places, the practices are entirely consistent in the field, with Wegman's referencing about on a par with Bradley's text book. So here we have a case where, once again, "bad behaviour" exists on both sides of the fence, probably amounts to little more than sloppy referencing, but fluffy with his rose tinted specs can't tell the difference.

Whenever the latest denialist hypothesis comes up - be it the sun, some sort of natural cycle, internally generated changes, cosmic rays or whatnot - you can be sure that it will be hyped as The One True Explanation by Good Honest Scientists. Critical examination of the hypothesis will be lacking, even if said hypothesis violates basic physical law or relies on a highly dubious chain of causality.

This is once again a hasty generalisation on the same grounds as above, but I'm still waiting for you to tell me which physical law the LTP model that I proposed violates. Fluffy, you asserting that this is the case does not make it so.

On the other hand, any paper by 'the enemy' will be minutely combed through

What, such as "going to town" on a peer review? Such as an 88-page review of a 10-page paper, as we recently saw? Once again, fluffy's blinkers make him accuse the sceptics of behaviour which is well evidenced on his side of the debate as well.

- in the case of the original MBH 98 paper, for over a decade - and the slightest error or inconsistency used to dismiss it in its entirety.

ROFLMAO. The "slightest error"? MBH98 is a dreck, and is riddled with errors, from improper modification of PCA, to questionable sampling criteria, to statistical overfitting, to incorrect statistical significance calculations, to farcical confidence intervals which even today nobody has figured out exactly how they are calculated (although we do know that they are wrong, since correctly calculated CIs are infinite for some parts of the reconstruction)

As far as fanaticism goes, I can simply point out that it isn't me starting these threads. It's the skeptic faithful who routinely pick up the latest meme from the denialosphere to post here to spread the word.

As far as fanaticism goes, you can't see that the behaviours you describe exist on both sides of the debate, and the most important thing is to rise above them - once again though, I'm not holding my breath.

[Kurt Barlow]

So what are you doing for Christmas?

Is Syd still available?

Maybe - maybe not Either way I would abstain for medical reasons.

Xmas here - with colleagues and Medical staff. I think my task is to make yorshire puddings and chicken liver pate!

[fluffy666]

Well, I'm still waiting for fluffy666 to provide supporting evidence of his claim that an LTP model for background variability has no physical basis. I provided 3 peer reviewed references, two of which demonstrated the physical basis for the model, and have not heard anything back yet. Still waiting, fluffy.

You may have to provide a link if you want an answer. Some of us actually work for a living, and hence cannot log on every day.

I snipped the whining and rambling from the rest of your post, couldn't be bothered answering. Seriously, you have made a huge raft of assertions with zero evidence to back them up, this being the 'Gish Gallop' approach. I'm not rising to that, I don't have time.

[Kurt Barlow]

Apologies that's not what I meant, I should have said that's like winter there i.e. it doesn't get much cooler.

no I'm on the British South Coast freezing my b@lls off!

I have when in the Military, been in Saudi (Riyadh) in the winter.

Nice American complex, great.

First Shamal of the season today.

That made driving around Khobar fun

[Rapid Descent]

You may have to provide a link if you want an answer. Some of us actually work for a living, and hence cannot log on every day.

LOL, you were the one who insisted that LTP was not physically sound without any kind of link or argument - just assertion. When asked to justify your position, you are suddenly too busy. I do understand that you may be busy - I often am as well, and don't post. I'm open to challenges on my thinking, but I'd prefer scientific challenges rather than hand waving. I had to interpret what I thought you were saying (which was largely a failure to understand basic physics, I think) and gave you three papers (four but one is a two-parter) to consider.

You claimed to have some expertise in LTP so the papers should be familiar to you. But a quick reminder:

"The Hurst Phenomenon - A Puzzle?" by Vit Klemes, 1974, the Czech hydrologist, provides a simple circular reservoir model of the hydrological cycle that exhibits LTP. Link here (paywalled, unaware of a free copy on the web, but your tech library / uni should get you access)

The following is a 2-part paper showing how LTP is expected in the hydrological cycle from the principal of maximum entropy:

"Uncertainty, entropy, scaling and hydrological stochastics, 1, Marginal distributional properties of hydrological processes and state scaling" by Demetris Koutsoyiannis, Hydrological Sciences Journal, 50 (3), 381–404, 2005, link here

"Uncertainty, entropy, scaling and hydrological stochastics, 2, Time dependence of hydrological processes and time scaling", by Demetris Koutsoyiannis, Hydrological Sciences Journal, 50 (3), 405–426, 2005, link here

None of this breaches any law of physics that I am aware of; of course the LTP in the hydrological cycle imprints itself on the global temperature through clouds, water vapour, vegetation, etc. Variability in the hydrological cycle will swamp CO2 as both water vapour and clouds have a much greater effect on atmospheric temperatures. As I'm sure you know, LTP is pervasive at lower frequencies; it will always (at some point) dominate over STP / Markovian dependencies. The final question is whether evidence supports the idea of two bands (STP at high frequencies, LTP at low) or just consistent LTP-dominated. This question is addressed by the following paper:

"Global dependence in geophysical records", Mandelbrot and Wallis, Water Resources Research 5, 321-340. 1969. Link here (paywalled)

Feel free to explain to me how any of these papers are unphysical or break any known laws of physics.

I snipped the whining and rambling from the rest of your post, couldn't be bothered answering.

My whining and rambling was a direct response to your whining and rambling. If you don't like it, can I suggest you don't do it in the first place?

Seriously, you have made a huge raft of assertions with zero evidence to back them up

Your original claims had zero evidence to back them up. So I was responding in kind.

this being the 'Gish Gallop' approach. I'm not rising to that, I don't have time.

If the cap fits.

[jonb]

I remember watching a documentary a few years back, it said warming at the ice caps causes more icebergs to fall away from the Arctic ice and drift south and also more cold water to sink below the Arctic and travel south before resurfacing near the gulf stream. It was posited that the combined effect of these two actions could turn off the gulf stream.

Thinking about it , this could be part of a self regulating system on earth that keeps the planets temperature stable overall, as the rising temperatures at the pole(s) causes a shift in things like the gulf stream that causes thing to cool down again.

Having said all that, I think I've seen snow and ice before during the winter months, and if I remember correctly, every time it's happened before in my life, the gulf stream hasn't switched off and we haven't entered a new ice age

If the gulf stream stops, it will mean colder temperatures in Europe, but also presumably warmer temperatures wherever the heat is not coming from. That could mean more desert as well as more frozen arctic regions and that is not a good thing.

[jonb]

All very dramatic until one realises that the Artic ice if melted makes no difference to the volume of water in the oceans. ,

Greenland ice does though, as does Antarctic ice.

[fluffy666]

None of this breaches any law of physics that I am aware of; of course the LTP in the hydrological cycle imprints itself on the global temperature through clouds, water vapour, vegetation, etc.

Errm, no. 'Of course' will not do. You have to actually demonstrate it.

This is where the big leap of faith (on your part) comes in.

Variability in the hydrological cycle will swamp CO2 as both water vapour and clouds have a much greater effect on atmospheric temperatures.

Only if a variation in the hydrological cycle - for instance a change in the distribution of rainfall - changes atmospheric water vapour content. Unproven and physically unlikely, given the residence time of water vapour. Indeed, the classic example of such a process, the PDO, has no influence on the temperature trend. Likewise ENSO.

I'd also point out that you seem extremely certain as to what cloud feedback sign and magnitude is. Which is somewhat odd, because it's generally regarded as a significant source of uncertainty.

Further, I'd have to point out that if feedbacks just happened to be arranged such that any change in temperature was self-sustaining - i.e. a random hike of 1K lead to feedbacks which made 1K-over-previous the new normal - the there would be nothing to stop the climate system randomly walking off into Snowball or Venus modes. The fact that this patently has not happened fairly convincingly refutes the hypothesis.

Feel free to explain to me how any of these papers are unphysical or break any known laws of physics.

They don't break any laws of physics.

However, they are describing variations in local or regional rainfall patterns. Not global temperatures. The issue is with the sudden leap from one to another, which simply cannot be justified. LTP in global temperatures - without an extremely good physical explanation - would indeed violate basic energy conservation.

And you have not given a physical explanation. None of the referenced papers give such an explanation; they don't address the issue. Furthermore, the whole idea of internally generated variation in the climate system has been looked at by the climate science community, and found wanting; there is no way to make the energy budgets work. If you can find a way to do this, then go ahead, fame and fortune await.

My whining and rambling was a direct response to your whining and rambling. If you don't like it, can I suggest you don't do it in the first place? Your original claims had zero evidence to back them up. So I was responding in kind.

Actually, I asked who started (practically) all the global warming threads; that's a simple question of numbers and anyone can look at the evidence. Stolen emails.. you may have noticed the odd bit of attention paid to these in skeptic circles. I'm not aware of climate scientists doing the same, ditto for the lab break-ins. Even the mention of MBH 98 got the pavlovian response.. So the idea that I was ranting without evidence was false. However, your accusations that climate scientists do the same does appear to be without evidence. Simply saying 'same to you' does not cut it.

[Rapid Descent]

Errm, no. 'Of course' will not do. You have to actually demonstrate it.

This is where the big leap of faith (on your part) comes in.

Oh, this is just hilarious.

In the last thread, I provided a slew of evidence that demonstrates that LTP is present in virtually all temperature-related geophysical records. Both instrumental and proxy. You could not counter this, so you turn round and insist that the theory is no good because there is no physical basis for it.

I provide a physical justification which breaks no laws of physics, and you go quiet. Eventually, after a few weeks, you resurface on another thread, and finally admit this:

They don't break any laws of physics.

So, we've established that your last claim is simply wrong. But you now quickly switch your arguments to something completely different, without a pause for breath to release how wrong your previous statements were.

You then come out with this:

Only if a variation in the hydrological cycle - for instance a change in the distribution of rainfall - changes atmospheric water vapour content.

Typical realclimate bull****. The residence time of water vapour has nothing to do with it whatsoever. The concentration of water vapour is a nonlinear function of a wide range of parameters, including, but absolutely not limited to, temperature, land use, ocean circulation, wind speed, recent precipitation, the list goes on. These things vary over a wide range of time scales. Your myopic belief of ergodicity driven by residence time is astonishingly naive.

You clearly demonstrate that you do not understand the hydrological cycle, nor the factors that drive it. If you really understood the papers I linked in the previous post, you would understand that the hydrological cycle is very prone to LTP, we expect to find it, and we do find it when we measure it. This is the dominant driver of most geophysical processes on the earth today, which is why we find LTP throughout these processes, as demonstrated by Mandelbrot in the paper above.

Professor Koutsoyiannis expresses it better than I do in his address to the EGU:

Can we prove statistically that His significantly different from 0.5?

• Definitely, yes, in all of the cases, in which the null hypothesis involves pure randomness ("roulette" behaviour); see Cohn & Lin (2005).

• In fact this is indirectly done in all abundant publications that detect "trends" in hydroclimatic time series (usually locating alternating trends in different time periods or different sites of a region; see Hamed, 2008).

• Just replace the fallacious alternative hypothesis "deterministic trend" with "H> 0.5".

Unproven and physically unlikely, given the residence time of water vapour. Indeed, the classic example of such a process, the PDO, has no influence on the temperature trend. Likewise ENSO.

This is funny too. ENSO and PDO are essentially related, patterns of ocean circulation, across different time scales. ENSO on the monthly to annual scale, PDO on the decadal scale. What you don't realise is that there is also a PCO, and a PMO, which are the same things on a centennial and millenial scale. And these WILL create trends on the decadal scale, but not at longer scales (where 10kyr-100kyr variations create trends, and so on). Yet ENSO and PDO have eluded deterministic modelling. Stochastic LTP modelling is the only approach that successfully describes the behaviour of the system. This is what these hydrologists are trying to explain to you, but you are not able to understand.

So, in fact, ENSO and PDO are PERFECT examples of how powerful LTP is as a tool, and how much deterministic modelling has failed. Thanks for that.

And you have not given a physical explanation. None of the referenced papers give such an explanation; they don't address the issue.

Yes, they do. I can lead you to the papers, but I can't make you understand. Or do you not think entropy a meaningful part of science? Funny, many important theories are based on that principle.

Furthermore, the whole idea of internally generated variation in the climate system has been looked at by the climate science community, and found wanting; there is no way to make the energy budgets work. If you can find a way to do this, then go ahead, fame and fortune await.

It's already being done, through the peer reviewed literature as I speak. I've been linking the papers for you. The fact that you and many climate scientists are in denial about it won't change that fact.

Actually, I asked who started (practically) all the global warming threads; that's a simple question of numbers and anyone can look at the evidence.

Anyone can look at the evidence of LTP. But you demanded I present it. Yet when it comes to you presenting evidence, other people are expected to go find it. Funny how many double standards we find in climate science and the advocates of such...

[Rapid Descent]

More to add to this, to demonstrate the entirely unreasonable position that fluffy666 adopts on this point. This is a rounder description of the quote from Prof. Koutsoyiannis in the post above, which is short but technical.

The choice here is not "LTP or nothing". I'm sure fluffy accepts there is an unpredictable component of the system. The consensus position terms this "weather", and uses a Short Term Persistence ("STP") noise model. So the question is really, which model is most appropriate? We can add other models, such as independent, identically distributed ("IID") models into the mix as well.

We can reject IID models by checking for serial correlation. Time series that exhibit time dependency cannot be "IID" by definition.

We are left with STP and LTP as two possible options. (There are other possibilities, but until we have a model to base it on, other hypothetical possibilities are impossible to analyse). By rejecting LTP, fluffy is asserting STP. But we can test and reject STP: at longer time scales, STP series will have a Hurst exponent H tending to 0.5. All measures of global temperature exhibit H > 0.9, clearly showing LTP. In a similar way that we can reject IID, we can reject STP.

These are all different models of natural variability. Despite fluffy's protestation, there is no more physical basis to presume STP over LTP. We don't have the full equations of climate, so we cannot determine which the most appropriate model is from this route. Our best bet is to estimate from the data. And the data all points one way: LTP every single time.

[snowflux]

GEOPHYSICAL RESEARCH LETTERS, VOL. 33, L06718, 4 PP., 2006

doi:10.1029/2005GL025591

Long-term persistence in climate and the detection problem

Diego Rybski, Institut für Theoretische Physik III, Universität Giessen, Giessen, Germany

Armin Bunde, Institut für Theoretische Physik III, Universität Giessen, Giessen, Germany

Shlomo Havlin, Minerva Center and Department of Physics, Bar-Ilan University, Ramat-Gan, Israel

Hans von Storch, Institute for Coastal Research, GKSS Research Centre, Geesthacht, Germany

We have analyzed six recently reconstructed records (Jones et al., 1998; Mann et al., 1999; Briffa, 2000; Esper et al., 2002; McIntyre and McKitrick, 2003; and Moberg et al., 2005) of the Northern Hemisphere temperatures and found that all are governed by long-term persistence. Due to the long-term persistence, the mean temperature variations σ(m, L) between L years, obtained from moving averages over m years, are considerably larger than for uncorrelated or short-term correlated records. We compare the values for σ(m, L) with the most recent temperature changes ΔT i (m, L) in the corresponding instrumental record and determine the year i c where ΔT i (m, L)/σ(m, L) exceeds a certain threshold and the first year i d when this could be detected. We find, for example, that for the climatologically relevant parameters m = 30, L = 100, and the threshold 2.5, the values (i c , i d ) range, for all records, between (1976, 1990) for Mann et al. (1999) and (1988, 2002) for Jones et al. (1998). Accordingly, the hypothesis that at least part of the recent warming cannot be solely related to natural factors, may be accepted with a very low risk, independently of the database used.

Summary: Northern Hemisphere temperatures do indeed exhibit long-term persistence, but it is very unlikely that the current warming can attributed solely to natural factors.

[Rapid Descent]

Good stuff snowflux! I do appreciate your evidence-based approach, which we can actually discuss and get our teeth into (even if we ultimately disagree), to fluffy's assertions.

Rybski et al is actually one in a chain of papers which have included important discussions on the consequence of LTP in the climate system. I'll get on to that in a moment. I'd like to highlight one aspect of the paper you cite in particular:

Northern Hemisphere temperatures do indeed exhibit long-term persistence

Indeed! And I am not aware of any scientist who has studied this issue and come to the opposite conclusion in the peer-reviewed literature.

In addition, I would like to remind everyone that I do not doubt the greenhouse effect is a real, physical effect. However I am contrasting two models:

1. STP climate, with a time constant between 1 year and 30 years, coupled with a high climate sensitivity to forcing to achieve the interglacials

2. LTP climate, without the need for a high sensitivity to forcing.

Of course, there are other models but I think model "1" represents the IPCC view, and "2" represents my view (and that of others). Model "2" does not preclude the possibility of anthropogenic factors, but does suggest there is really no evidence for the high sensitivity. So it is possible we will find an anthropogenic effect. But that effect is likely to be of less concern.

Getting back to the Rybski et al paper. It is important to understand this is part of an (ongoing) back-and-forth between scientists, and I consider it to be an interesting and valuable contribution to the debate. The papers that form this discussion are:

Cohn and Lins, "Naturally Trendy" (2005), in which Cohn and Lins demonstrate that using instrumental records and an assumption of LTP, the warming in the record fails statistical significance tests of anomalous warming.

Rybski et al, the paper you link above, largely agree with Cohn and Lins about being unable to detect a trend in instrumental data, but instead of using instrumental data, use paleoclimate reconstructions and a different test, arguing that in fact you can detect a difference in these reconstructions. The problem with this is that you have to believe the paleoclimate reconstructions are accurate enough to do this with. Which leads us to...

Koutsoyiannis and Montanari 2006, link here, is a response to Rybski et al, and highlights a number of issues with the Rybski paper. In particular, Rybski et al clearly underestimate the uncertainty not only in the paleoclimate reconstructions, but also in their own parameter estimations, leading to an (incorrectly) high degree of confidence. In fact, when accounting for these issues, the significance once again disappears, the problem being in the quality of the reconstructions as much as anything else.

The most recent paper is Halley 2009, who argues once again that "unnatural" influence can be detected in the paleoclimate reconstructions. Halley is also a good paper, but is frustrating that important caveats in the document were left out of the abstract. In particular, Halley finds that the parameters he estimates from paleoclimate reconstructions have non-overlapping confidence intervals. He also observes non-overlapping confidence intervals between paleoclimate reconstructions and instrumental records, and argues that this is an indication of unnatural agencies influencing the temperatures.

Unfortunately, this makes no sense: if reconstructions that are supposedly measuring the same thing do not have overlapping confidence intervals, it means your confidence intervals are wrong. Full stop. Indeed, the fault most likely lies in the paleoclimate reconstructions rather than Halley's analysis. This leads us on to the recent discussion in the Annals of Applied Statistics (link here) on paleoclimate reconstructions, with the rejoinder (link) and many commentators agreeing that confidence intervals are seriously underestimated in these reconstructions. This must be fully addressed (as well as several other caveats in Halley's paper) before we can have any confidence in his findings.

As stated in Halley's paper, more work is required before we can draw much in the way of conclusions. But all do agree that the LTP model is more appropriate, and this does call directly into question our current understanding of climate change, in particular the claimed need for high sensitivity to forcing.

[snowflux]

I'm no expert on this, but I don't see how LTP can be a model of anything. I thought it was purely a statistical phenomenon, of relevance only in determining whether or not the currently observed temperature variations are statistically significant. I don't think the existence of LTP in itself has anything to say regarding the physical causes of the temperature changes observed in the past, and I don't see anything in the papers you cite that indicates otherwise.

I can't help thinking you've got the wrong end of the stick here.

[Rapid Descent]

I'm no expert on this, but I don't see how LTP can be a model of anything. I thought it was purely a statistical phenomenon, of relevance only in determining whether or not the currently observed temperature variations are statistically significant. I don't think the existence of LTP in itself has anything to say regarding the physical causes of the temperature changes observed in the past, and I don't see anything in the papers you cite that indicates otherwise.

I can't help thinking you've got the wrong end of the stick here.

LTP is a stochastic model, in the same way that STP and IID are. They represent models of variability from all sorts of mechanisms and sources. They are "models" because they are not the real thing itself, but a mathematical representation of that thing.

Lets take a more down to earth example, from physics. Electronic circuits provide a nice example from physics because relatively trivial circuits can be used to generate all three types.

Consider a communications receiver, say a radio receiver operating at a few GHz. The electromagnetic radiation from a nearby transmitter causes voltage fluctuations in the receiving element, which we would term the "signal". Lets just say we mix down that signal to baseband, filter it, and sample it at the Nyquist frequency of the filter. Now that receiver not only has "signal", but also noise. The noise comes from random fluctuations of electrons in the receiving aerial (this occurs in all conductors). The random fluctuations are a function of the temperature of the conductor which receives the signals. Because we sampled at Nyquist, these randomly fluctuating electrons will behave, to a very good approximation, as an IID result. Because the electron movements are random, they cannot be deterministically predicted. However, we know from Boltzmann's work, that the magnitude of the thermal voltage effect will be kBT, where k is Boltzmann's constant, B is the bandwidth of the filter and T is the temperature of the aerial.

So we do not have a deterministic model which will tell us exactly what fluctuations to expect; but we have an entirely physical stochastic model with which we can fully describe the properties of this noise. This is a perfectly good model of what is happening to the circuit. We can add to this model the signal (since the system is linear, such an addition is valid).

Lets imagine the same circuit, but sampled more often than the Nyquist frequency. Because of this, the samples will no longer be independent, and assuming a simple filter, the noise in the circuit will be Markovian, or STP. Once again, we cannot define a model that deterministically predicts this noise. But we can define a stochastic model that fully characterises it.

Finally, an LTP model. I'm going to break from the example above, and consider instead the voltage fluctuations on a carbon composite resistor with a high current flowing through it. This leads to excess noise, which is a type of LTP noise. Once again, the fluctuations are not deterministically predictable - so only a stochastic model of the noise is viable. There are many models for different component types that describe the characteristics of the noise.

So, for all of these different examples, are the noise models (IID, STP, LTP) real models or not? To me, they characterise very effectively a behaviour of the real world. Some of them have fully understood reasons behind why these characteristics are what they are, some are more empirically derived, but they are still, by definition, models.

[snowflux]

LTP is a stochastic model, in the same way that STP and IID are. They represent models of variability from all sorts of mechanisms and sources. They are "models" because they are not the real thing itself, but a mathematical representation of that thing.

I know what a model is; I wrote one for my PhD. LTP is not a model. LTP may be incorporated into a model, or may be evident from a model, but it cannot be a model.

[Rapid Descent]

I know what a model is; I wrote one for my PhD. LTP is not a model. LTP may be incorporated into a model, or may be evident from a model, but it cannot be a model.

I suspect we are at cross purposes here. I think you are using a very narrow definition of the word "model" and I am using a broader definition.

A model, at the heart of it, is merely a characterisation of a system. It may be something as explicit as an equation, or it may merely be a statement about the behaviour of the system. Anything that represents the system, but is not the system itself, is a "model".

For example. Say I have a black box, which generates a stream of numbers, and I am given a task of producing a model of that data stream. I may observe the numbers coming out of the box and draw some graphs; I might note that the data looks to be normally distributed, from a histogram. I might note the numbers appear to be random. I might even do some tests to verify this; perhaps a runs test to provide evidence that the numbers are appear independently random, or perhaps a more sophisticated test such as the Durbin-Watson test. I might perform standard tests for normality, such as the Lilliefors test or the Bera-Jacques test.

Assuming these tests pass, I might make a declaration, that I have characterised the series of numbers (with supporting evidence) as IID, normally distributed, with an estimated mean of (mu) and a standard deviation of (sigma). Now this statement is just a characterisation of the data coming out; it is not the true data stream, which may in fact not be truly normal, just that I have not carried out enough samples or tests to find this out. My statement about the data is just a characterisation; this statement is, in itself, a model of the data.

I may now chose to write a simulation of the data stream, so my colleagues can simulate output which captures these characteristics. I might put together a linear congruential generator to produce a stream of random numbers, and process this stream through a Box-Muller transform. Now I have an actual set of equations which characterise the system. I think you are (narrowly) referring to this as "the model".

But I would be a little critical of such a narrow view. Because this is just one implementation, and a rather imperfect one at that. My linear congruential generator only generates a limited set of statistically valid random numbers. The Box-Muller transform also does not produce a true normal distribution (although it is close enough for most purposes). In fact, this model is merely a characterisation of my original model (which was merely a statement that the data are normally distributed with defined first and second moments).

So, in this example, I actually have two distinct models. Both are simply characterisations of the true system, and therefore provide models of that system; the first being a descriptive model of the characteristics, the second itself being an imperfect model of the first model, but an actual implementable set of equations.

I understand your desire to narrowly limit the term model (or rather, "the model") as the set of equations that you use at the end. But I don't really like this use of the term: to model something, whether it be an Airfix kit, or a mathematical description of a data set, is to create a caricature of the real system. And a simple estimate of system properties (whether it be normally distributed, or IID / STP / LTP) is a form of caricature of the system, and therefore a model.

Very philosophical this I suspect we will have to agree to disagree over our respective definitions of the word. But if at least we understand the different definitions we use, I don't think it should be a barrier to communication!

[fluffy666]

Oh, this is just hilarious.

I'd prefer 'annoying bone headed denialism', but there you go.

In the last thread, I provided a slew of evidence that demonstrates that LTP is present in virtually all temperature-related geophysical records. Both instrumental and proxy. You could not counter this, so you turn round and insist that the theory is no good because there is no physical basis for it.

No. Hydrological records. Not 'temperature related', and no slew of evidence. No matter how many times you declare this, it won't make it true. Shifting rainfall from one region to another does not break any laws of physics. Changing the temperature of an object for no reason DOES break the laws of physics. You are the one trying to twist the argument here.

Typical realclimate bull****. The residence time of water vapour has nothing to do with it whatsoever. The concentration of water vapour is a nonlinear function of a wide range of parameters, including, but absolutely not limited to, temperature, land use, ocean circulation, wind speed, recent precipitation, the list goes on. These things vary over a wide range of time scales. Your myopic belief of ergodicity driven by residence time is astonishingly naive.

No, it's reality based. You can come up with as much as you like here, but that does not make it true. You have to actually provide evidence, and you have not. Water vapour has a short residence time and therefore cannot stay out of temperature based equilibrium for any significant period of time.

I'd note as well that you forgot to address one of the more obvious problems with your hypothesis, notably the absence of any mechanism for the climate wandering off to the Snowball or Venus states. We must all just be really, really lucky.

And since you mentioned RealClimate, you (or other people reading this, more likely) may wish to see what they have to say on the subject:

http://www.realclimate.org/index.php/archives/2008/08/hypothesis-testing-and-long-term-memory/

(PDO, ENSO, etc).. And these WILL create trends on the decadal scale, ..

That's strange, because this has been looked for and not found. You cannot keep declaring things to be true when they are not; this may be a useful debating technique (come up with numerous plausible sounding claims that take a lot of research to debunk) but it is not a way of establishing a scientific hypothesis.

It's already being done, through the peer reviewed literature as I speak. I've been linking the papers for you. The fact that you and many climate scientists are in denial about it won't change that fact.

Hmmm. Yes, papers from the 1960s and 1970s.. not a problem per se, but not exactly indicative of a hot research area.

And generally, you announce scientific breakthroughs after the paper has been published and, ideally, reacted to. Not before the papers are published, otherwise you can get embarrassed - think Mr Watts with his surface stations paper that been 'about to be published' for over a year now.

Anyone can look at the evidence of LTP. But you demanded I present it. Yet when it comes to you presenting evidence, other people are expected to go find it. Funny how many double standards we find in climate science and the advocates of such...

No. I ask for a physical model of how LTP in temperature records works (note the word 'temperature'). You then link to papers about hydrology and give me a load of hand waving waffle about water vapour with absolutely ZERO evidential backing (but plenty of abuse for asking). All you have done is run a statistical technique, known for generating false positive results, against a few single-location records, and then claimed that this overturns pretty much all of climatology. But of course, climatologists are 'in denial' about that. What with Gallelio and all. Oh, and climate scientists are all full of bull****, naturally. A professor of statistics looks at the evidence on LTP and points out the flaws, and that is 'bull****'.

Here's a challenge for you, then: Assuming that climate sensitivity is extremely low as you claim, explain the current temperature trend. You cannot just say 'LTP'; there has to be a physical reason of some sort that it not GHG-related (and a large, obvious one at that). We know that the oceans are not getting dramatically colder - which could be a source of heat to the surface; we know that ice sheets are not forming and we have a pretty good handle on things like surface albedo. So I've very curious as to what your explanation is. Where is the energy coming from?

[fluffy666]

I suspect we are at cross purposes here. I think you are using a very narrow definition of the word "model" and I am using a broader definition.

A model, at the heart of it, is merely a characterisation of a system. It may be something as explicit as an equation, or it may merely be a statement about the behaviour of the system. Anything that represents the system, but is not the system itself, is a "model".

The difference is that between purely descriptive/empirical models, and physical models.

Imagine dropping objects from the tower of Pisa. You could model this in two ways.

First, you could just measure the mass and surface area of the objects being dropped, time how long they took to fall, and hence arrive at a derived relationship that would give you the time to fall from the tower for any object. This is an empirical relationship.

Second, you could derive a equation based on gravity and air resistance (itself a function of pressure, temperature,etc), using already-established laws*. Then you could predict the time to fall. This would be a physical model. Importantly, a physical model is independent of any measurements already made, at least for a specific system.

The interesting thing is that the empirical relationship would, for the example given, almost certainly give better answers, especially if you stayed within the bounds used to establish it. It would certainly give better results than using the standard equations of gravity alone. But you would not use such a relationship to disprove gravity, of course.

*Yes, technically gravity itself is empirical. But a very widely-applicable empirical relationship.

[Tin Foil Hat]

Total waste of energy - all that fuss and we get one poxy shower with maybe a cm or 2 of snow and that's it. Not even enough for a snowman for my little boy.

FFS.

Did anyone get decent volumes of snow today or was this a universal wet fart of a weather warning?