Buy

Books
Click images for more details

Support

 

Twitter
Recent comments
Recent posts
Currently discussing
Links

A few sites I've stumbled across recently....

Powered by Squarespace
« Tribunal Dates | Main | Tabloid academics »
Thursday
Feb252016

Quote of the day, predictability edition

Even a fully deterministic system is fully unpredictable at climatic timescales when there is persistence.

From a Demetris Koutsoyiannis presentation.

PrintView Printer Friendly Version

Reader Comments (16)

Slide 3 has 'Common Sense' as the foundation for technology, which then vanishes from slides 4 to the end ;)

Feb 25, 2016 at 11:45 AM | Registered CommenterBreath of Fresh Air

Had to read the paper to understand what on earth he means by 'persistence'.


If I read it aright, its a term someone who doesn’t understand (the terminology of) chaos maths uses to describe a system with more than one attractor.

I.e. fully deterministic systems that are chaotic will suffer from sensitivity making their exact trajectory unpredictable, and if there is more than one attractor, they may average an 'orbit' round that for a long time before shooting off to another.

So the butterfly's wings in Brazil might cause a 7 year flood, or a seven year drought, whatever.

So 'persistence' means 'found an attractor to sort of orbit, for now'.

Feb 25, 2016 at 11:49 AM | Unregistered CommenterLeo Smith

Breath of Fresh Air said.. 'Common Sense' as the foundation for technology, which then vanishes from slides 4 to the end

Mmm. I have a deal of issue with the exact labelling on his hierarchy of knowledge, too, but full marks for actually having introduced the notion that knowledge is (best considered to be) hierarchical.

I think that what separates the Liberal Green Left mind from intelligence is the (lack of the ) concept that all facts are relative to a structure of belief, as well as to the world of 'whatever is the case'...

...but that's a bit off topic :-)

Feb 25, 2016 at 11:54 AM | Unregistered CommenterLeo Smith

Leo - excellent discussion here

http://www.climatedialogue.org/long-term-persistence-and-trend-significance/

Feb 25, 2016 at 12:34 PM | Unregistered CommenterPhil Clarke

I thought "persistence" was what was required by climate science model afficionados, in the absence of meaningful results.

Feb 25, 2016 at 12:47 PM | Unregistered Commentergolf charlie

Can anyone translate that phrase above into English for me..?

Feb 25, 2016 at 2:12 PM | Unregistered Commentersherlock1

One definition of "common sense" is that it is the set of prejudices that one has absorbed by age 18. I wonder what his definition is.

Feb 25, 2016 at 2:20 PM | Unregistered Commenterdearieme

@Phil Clarke :

That article you linked says the same - its is a non mathematicians ugly attempt to express what is called an attractor in chaos maths.

Feb 25, 2016 at 8:43 PM | Unregistered CommenterLeo Smith

" Even a fully deterministic system is fully unpredictable at climatic timescales when there is persistence. "

For practical purposes, this can rephrased as..
"Even a deterministic system is fully unpredictable if we don't know the details of that determinism"

And we don't. No need to refer to chaos math or philosophical logic. No need to argue about persistence.

Feb 25, 2016 at 8:48 PM | Unregistered Commentermothcatcher

@sherlock1

I will make a stab at a translation
Even a fully deterministic system
Even a system that is fully governed by (known) equations such that a given set of initial conditions will exactly determine a given result, and only one result...
is fully unpredictable
...still cannot have meaningful predictions made about it (because the sensitivity to initial conditions is too large and unmeasurable differences in starting conditions may have highly measurable effects on outputs)....
at climatic timescales
..over times scales large enough to allow those changes to propagate fully...
when there is persistence.
When the actual system is such that it it is chaotic and has more than one 'attractor'.

I think actually this youtube video showing three initial conditions that are very very close but eventually diverge into three totally different trajectories orbiting one of two attractors shows the principle.

https://www.youtube.com/watch?v=FYE4JKAXSfY

Feb 25, 2016 at 8:56 PM | Unregistered CommenterLeo Smith

A scientists' version of 'Special English' seems to be evolving whereby "sciencey-sounding" words (such as 'persistence') from a very limited vocabulary are forced into service in places where far more precise and appropriate words are available.

It may be because the writer feels that the word 'persistence' just sounds so much more *scientific* than 'attractor'. (After all, we want to sound like scientists - not debutantes!)

Furthermore, it doesn't pay to write clearly - because there is the risk that laymen might actually understand. And all of the professor's inacessible scholarly mystique would be lost!

Feb 25, 2016 at 11:31 PM | Unregistered CommenterJ Calvert N

Persistence in this context has a very specific meaning, albeit not as clear as it perhaps could be from the sound bite. By "persistence", Demetris is referring to "Long Term Persistence", aka fractal dynamics, aka Hurst-Kolmogorov dynamics. (Phil Clarke is exactly right with his link here, kudos for that).

Long term persistence describes a form of energy dissipation which is common in complex systems. It can apply to systems which are chaotic or not. Furthermore, some chaotic systems do not exhibit long term persistence. While chaos does indeed have relevance to weather and climate prediction, the statement made by Demetris is a separate issue to that of chaos theory.

Hurst-Kolmogorov dynamics is a stochastic (rather than chaotic) representation of energy dissipation in which the dissipation is approximately a constant power per octave, or a 1/f power spectral density relationship. This is based on a relationship first discovered analytically by Kolmogorov while analysing two dimensional turbulent fields, first empirically recognised in nature by Harold Edwin Hurst, formalised in a statistical sense by Benoit Mandelbrot, then taken by many others and applied to their own fields - including Demetris in the context of hydrology and climate.

Just to be clear on the relationship between chaos and long term persistence. It is not one-to-one; chaotic systems can exhibit long term persistence as an emergent property, or they may exhibit some other behaviour (short-term persistence or even anti-persistence). Similarly, non chaotic systems may exhibit long term persistence, likewise they may not. Long term persistence is fairly straightforward to identify in a system, given data across sufficient scales.

A youtube video was linked to an example chaotic system, the Lorenz attractor. Note the Lorenz equations contain just a single attractor. Also, the Lorenz attractor is an example of a chaotic system that does not exhibit long term persistence. That does not make it inherently predictable, but it is not an example of the class of problem that Demetris is discussing.

A better quality copy of the presentation can be found here (click through to "full text"):
Hydrology, society, change and uncertainty

Feb 26, 2016 at 12:20 AM | Unregistered CommenterSpence_UK

Even Stern has now realized that there is something wrong with the models:
http://www.nature.com/news/economics-current-climate-models-are-grossly-misleading-1.19416

Feb 26, 2016 at 8:46 AM | Unregistered CommenterPethefin

Phil Clarke, thank you for that link. Browsing through the comments section I found this from Demetris Koutsoyiannis:

"3. Signal vs. noise

In my view this is NOT a true dichotomy in geophysical sciences, while it is meaningful in electrical engineering and telecommunications.

Excepting observation errors, everything we see in climate is signal. The climate evolution is consistent with physical laws and is influenced by numerous factors, whether these are internal to what we call climate system or external forcings. To isolate one of them and call its effect “signal” may be misleading in view of the nonlinear chaotic behaviour of the system (see also dichotomy 5 below).

My reasoning as to why I regard this dichotomy as misleading has been exposed in earlier comments. To repeat it as briefly as possible: Let us assume that it is possible to remove the “signal” (the anthropogenic influence) from “noise” (whatever this is). Would the stochastic properties of the “noise” be different from those of the whole climate? As I wrote extensively earlier, the properties of the “noise” alone can be easily seen in older periods, those not affected by the “signal”. And it seems that stochastic properties remained unaltered. In contrast, climate models, which allegedly can distinguish “signal” and “noise” yield a “noise” which is fully inconsistent with past climate"

I believe this is also the stance of William Briggs, that there is no signal and no noise because you cannot distinguish "natural" from "man-made" since it is from the same origin. This is one of the main reasons why it is just so wrong to say that one can model the climate "as it should have been without manmade GHGs". That is just not possible.

Feb 26, 2016 at 8:52 AM | Unregistered CommenterAnders Valland

Thanks, Leo..!

I still think its b£llsh$t.....

Feb 26, 2016 at 2:53 PM | Unregistered Commentersherlock1

This paper is describing a number of issues that were addressed by the nuclear industry about 25 years ago, in dealing with models of reactor behavior. There are lots of phenomena that need to be considered, some of which are not well known or understood at all, but which have to be factored into the calculations. To deal with the uncertainty, one does a Monte Carlo simulation of the system, using a range of various parameters to represent the uncertain models. When you plot the results of many runs for the value that you are worried about (temperature, pressure, stress, etc), you end up with an "envelope" that gives you some idea of where you might end up. It is still possible to end up outside the envelope if the system encounters a "black swan" set of conditions, which is why it is important to properly parameterize the unknowns.

Engineers like this methodology because it gives them a design target for the variables that are the Monte Carlo inputs - you design for strength or temperature resistance that can withstand the "envelope". Then, you add a bit of additional margin to try to take care of the black swans, and, if you are really paranoid, you add additional margin or capabilities to take care of stuff you did cannot even think of.

I think that the climate modelers are trying to do something similar, but their use of this approach is NOT appropriate for a system like climate, which has so many unknowns and unknown unknowns, from biological reactions/feedback to cloud modeling . They think that they can just parameterize these unknowns and then use the limits of the envelope to scare people into a political solution that the climate modellers admire. As a public policy matter, this is similar to the use of "social science" results to drive legislation, and we know that most social science is not reproducible, not testable, not falsifiable, and has the same sort of unknowns that afflict climate science. It is an adjunct to the Precautionary Principle.

There are a bunch of papers in the nuclear modeling field that describe the technique that this author is talking about. It is good to see that someone is using these methods to critique climate science. The climate science community will complain that their peer-review process deals with it, but I don't think so - the Climate-Gate emails show that their peer-review process is completely broken.

In the end, it is all about generating pseudo-scientific numbers to impose a political system. It is not good science or good public policy.

Feb 26, 2016 at 5:57 PM | Unregistered Commenterrxc

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Post:
 
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>