Reader Eric Worrall writes:
I was playing with Wood For Trees, looking at the relationship between Pacific Decadal Oscillation vs global temperature (Hadcrut 4), when the following graph appeared.
The interesting thing is PDO in this graph appears to have predictive skill for changes in global temperature – the changes in PDO appear to match changes in global temperature, once the graphs are normalised, but temperature lags PDO by around 5 years.
Source: http://goo.gl/hzOxW
Is it all just coincidence? Bad endpoint choice? Or does it in fact have some predictive value?
Readers are invited to weigh in.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
” Steven Mosher says: June 5, 2013 at 3:40 pm
PDO is temperature. predicting temperature from temperature tells you nothing. same with all the ocean patterns. OF COURSE they allow you to predict the global temps. they are part of the thing being predicted. Its like using CET to predict global temps. Of course you can.
Put another way. PDO is the effect”
The temperature is the temperature but it does lead the air temp. So a change in the ocean temperature results in a following change in air temperature over North America at least.
The PDO is definitely a leading indicator IMHO.
Post hoc ergo propter hoc. When an argument is a named logical fallacy, why make it?
If you prefer English, correlation (even lagged correlation) is not causality. If A and B have similar graphs (when rescaled, moved, etc) with or without A temporally preceding B, it leaves open:
* A causes B (desired/claimed conclusion)
* B causes A (yes, this is still possible, it is just less easy to explain. Think about that, the need for an explanation.)
* A and B are both caused by C (not shown), or C, D, E… (not shown). The lag might be due to one of these (not shown) causes — in the correct explanation.
* A really is B, so one isn’t really measuring two different things, one is measuring two different sides or aspects of the same thing. Steve Mosher points this possibility out above, although not really convincingly, given the size of the Pacific.
If the Pacific is almost half of the globe, and the PDO is a monotonic transform of the temperature of the Pacific as he seems to assert, why does global average temperature lag the average temperature of the Pacific? It is difficult to imagine an explanation for this lag if it is as simple as global temperatures are a direct average of the temperatures used to define/infer the PDO. Ordinarily if B (global temperature) is the average of A (Pacific temperatures) and D (everything else) and B is independently varying from A, then one does not expect B to systematically lag and follow A. In context, why is it only the Pacific that experiences global warming due to a global cause first, with everything else experiencing it later? So his argument appears inadequate to — without an explanation — contradict A causes B, although of course it is a possibility.
* A and B are are independently varying due to a variety of complex causes, and their lagged covariance is an accident of the particular values of their mutual parameters including a few that are shared. It will disappear in the future without a trace, just as it didn’t exist in the past (back before we were around with thermometers to look for it.
* A and B are randomly varying systems with shared characteristic timescales, and their lagged covariance is pure chance. Given only a few “transitions” of similar timescale, that isn’t as unlikely/impossible as one might think, and it is easy to find similar artifacts in any sufficiently long timeseries.
So, given all of these possibilities how can we choose between them? On the basis of explanations that make sense (including arithmetical/computational sense). One has to explain:
a) Why the PDO varies the way that it does. This is already a daunting task, as the PDO is not an independent phenomenon.
b) How the Pacific couples to the rest of the planet. This is of course part of the explanation for why the PDO varies the way it does.
c) How the PDO causes the rest of the planet to vary the way that it does. No, this isn’t circular reasoning, it is the observation that this is a coupled dynamic system so that separating out independent “causes” is all but impossible.
Here’s my “explanation”: The PDO varies the way it does because of the prior state of both the PDO and the rest of the world, including all of the major oscillations, the Sun, the state of GHGs and aerosols and volcanoes and the position of the Earth in its orbit and maybe other stuff we haven’t even realized is important yet, because of its current state (and the current state of a lot of those other independent or coupled variables) and a whole lot of physics. The rest of the world will vary in the future in a way that depends on the prior state of the rest of the world (including the PDO etc), on its own prior state, all those variables, as well as the current state, and possibly some unknown stuff.
What is the “cause” of the behavior of any part of this strongly coupled, nonlinear, chaotic system?
Everything. Nothing. CO_2 levels. Solar state. The particular angle of attack of some Pangean butterfly’s wings when they beat a hundred million years ago. Dark matter.
We cannot predict the climate. We cannot even predict the damn weather, not more than a week or so out. There are really good reasons we cannot predict the weather, and equally good reasons we cannot predict the climate. It is true that they aren’t quite the same problem, and sometimes one can predict the average behavior of a system in the long run (whatever that means) when one cannot predict its short time behavior at all reliably, but when I say we cannot predict the climate I mean that we cannot even understand the past behavior of the climate! We have no friggin’ idea why the MWP was warm, the LIA was cold, and why the world warmed (without CO_2 increase to drive it) since the Dalton minimum. We cannot predict the future state of one of the only important contributor of heat to the system, an enormously important cause whose effects on the Earth are complex and only beginning to be understood. To claim otherwise is an enormous act of intellectual hubris and scientific fraud — unless and until you can back up the claim with actual predictions, consistently validated. Or hey, I’d settle for a halfway decent hindcast or two, back to (say) 0 BCE or 16,000 BCE or 120,000 BCE or 50,000,000 BCE. The only thing we learn looking at the real climate record of the Earth is that it is always changing, that the changes are sometimes sudden and profound, and that we have no idea why they occurred or why they WERE either sudden, or gradual as the case may be, or gentle and moderate, or profound and catastrophic as the case may be.
Some of these things we are likely to never be able to properly prove or understand as the evidence is simply gone into the past. The Ordovician-Silurian transition — an ice age that began with 7000 ppm CO_2, and that peaked in glaciation a few million years later with CO_2 still at 4000 ppm. What’s up with that? Space aliens came and directed a freezing ray at the Earth, straight out of Buck Rogers? The Sun decided to turn off (partly) for a million years or so? A civilization consisting of highly evolved giant spiders had a nuclear war and triggered a nuclear winter a few million years long? Sure, we can propose more sensible alternatives, but honestly they will all still feel like science fiction, and in all probability none of them can either be verified/supported or falsified, at best they can be shown to be a consistent possibility.
Why is it so very difficult to say “we don’t know”?
rgb
Either a cause or an effect, it seems reasonable to see that none of the measures have a strong correlation with the CO2 concentration. Both PDO and AMO must be correlated with atmospheric temperature, after a normalisation, for instance made with the total area or volume of the effective thermal water body of the Pacific and Atlantic, to see how the graphic behaves. It must include the Indian ocean too. Like the Milankovitch cicles, the effect of a single parameter can be important but is the correlation of all parameters that matters. We must compare, for instance, peaks of phase and out of phase between these oceanic oscillations. The oceans are a lot more resilient to temperature change and should be the right normalised temperature proxy for the earth surface thermal behaviour.
Mosh, I guess you’ve never read Tisdale’s comments on the PDO. It is not temperature, it is an after effect of ENSO.
I’ve been pushing the PDO mode as the primary driver of global temperatures at Yahoo and other sites. The correlation is much better than CO2. I use slightly different dates but close enough. However, it still leaves open the question of what drives ENSO.
See:
Joe D’Aleo’s pdf on PDO+AMO at icecap.us, which shows, for example, the strong correlation between PDO+AMO and the US temperatures.
AMO is Atlantic Multi-Decadal Oscillation.
and his results are also presented at <a href="http://www.appinsys.com/GlobalWarming/PDO_AMO.htm"appinsys.com
appinsys.com also gives comparisons between temperature and PDO alone, AMO alone, ENSO alone, and AO-NAO (Arctic Oscillation and North Atlantic Oscillation). For ENSO, the “integrated Nino 3.4 SST index”–which is just a running summation of that index–correlates with the global SST anomaly; Bob Tisdale is the one who showed this, and of course he advertises his recent, inexpensive book on the subject of the ocean oscillations on his site.
This might be the answer to the question of what determines the ratio of El Nino’s to La Nina’s and so possibly world temperature.
Long-Term Lunar Atmospheric Tides in the Southern Hemisphere
Ian R. G. Wilson and Nikolay S. Sidorenkov
The Open Atmospheric Science Journal, 2013, 7, 51-76
http://www.benthamscience.com/open/toascj/articles/V007/TOASCJ130415001.pdf
Relevant part of .ABSTRACT
Finally, an N=4 standing wave-like pattern in the MSLP that circumnavigates the Southern Hemisphere every 18.6 years will naturally produce large extended regions of abnormal atmospheric pressure passing over the semi-permanent South Pacific subtropical high roughly once every ~ 4.5 years. These moving regions of higher/lower than normal atmospheric pressure
will increase/decrease the MSLP of this semi-permanent high pressure system, temporarily increasing/reducing the strength of the East-Pacific trade winds. This may led to conditions that preferentially favor the onset of La Nina/El Nino events.
If weather wasn’t variable, and previously known to be so, why would we have records extending back ~ 450 years ?
One of the features of the PDO has to do with atmospheric pressure. Since the jet streams are driven by differences in pressures the PDO could impact the jet streams and provide a mechanism to move the jets a little north or south as S. Wilde has mentioned many times. And, if this ties into solar/lunar tides as has also been mentioned previously, it could turn out to be the basic mechanism.
First off, thanks RGB for that comment.
The problem with any analysis of the data is that the data has been muddled with to the point where one has no idea if what one is looking at has any resemblance to reality or not. The only global temperature record I trust is the satellite record which unfortunately means only a snipet of time is available for earnest inquiry.
@richard M June 5, 2013 at 6:10 pm:
This is very tantalizing.
Total speculation: The Pacific being by far the largest non-land area on the planet, climate over it may have the capacity to organize in a way not possible elsewhere because of the complicating factors of irregular land forms. Kind of like a blank metal sheet where oscillations can form patterns in lycopodia (sp?) sprinkled on it.
This may be contributing of both ENSO and the PDO and allowing them to build up the power they do. And once they do, the patterns may achieve some stability/staying power.
Like I said, total speculation. But interesting to think about.
Steve Garcia
John West says:
June 5, 2013 at 6:17 pm
First off, thanks RGB for that comment.
The problem with any analysis of the data is that the data has been muddled with to the point where one has no idea if what one is looking at has any resemblance to reality or not. The only global temperature record I trust is the satellite record which unfortunately means only a snipet of time is available for earnest inquiry.
——————
Not so sure what the satellites are measuring.
Lots of adjustments.
Nuthin’ against the guys running them.
Eric Worrall
My theory is a multidecadal cycle oscillates relative to a secular trend. The recent warming (from 1973 to 2005) coincides with the warming phase of the multidecadal oscillation and its cooling phase should be from 2005 to 2037.
Here is a graph that shows the above pattern:
http://www.woodfortrees.org/plot/hadcrut4gl/mean:756/plot/hadcrut4gl/compress:12/plot/hadcrut4gl/from:1974/to:2004/trend/plot/esrl-co2/scale:0.005/offset:-1.62/detrend:-0.1/plot/esrl-co2/scale:0.005/offset:-1.35/detrend:-0.1/plot/esrl-co2/scale:0.005/offset:-1.89/detrend:-0.1/plot/hadcrut4gl/mean:756/offset:-0.27/plot/hadcrut4gl/mean:756/offset:0.27/plot/hadcrut3sh/scale:0.00001/offset:4/from:1870/plot/hadcrut4gl/from:1949/to:2005/trend/offset:0.025/plot/hadcrut4gl/from:1949/to:2005/trend/offset:0.01
Instead of IPCC’s “0.2 deg C/decade warming” the warming rate should drop to about 0 deg C/decade for the 30-years period from 2005 to 2037.
As a result, most of the sea ice lost in the warming phase from 1973 to 2005 is natural.
Steve Mosher,
I don’t usually disagree much with your comments, but in this case, I think you are mistaken. The PDO is NOT a temperature change per se, but an index which reflects the pattern of temperatures across the North Pacific, and trailing averages (eg 20 to 25 years trailing) of the PDO index do appear to correlate with changes in global average temperature. It is a bit hard to see how the PDO index from decades ago could have predictive power for present day average surface temperature if the PDO index is simply a measure of the global average surface temperature. See: http://rankexploits.com/musings/2013/pseudo-cyclical-contribution-of-the-pdo-to-earths-recent-temperature-history/
You get pretty much the same correlation by using sunspots instead of PDO. I think if the PDO had hands, it would just point up at the sun.
rgbatduke says:
June 5, 2013 at 5:12 pm
“Post hoc ergo propter hoc. When an argument is a named logical fallacy, why make it?…Why is it so very difficult to say ‘we don’t know’?”
Brilliant. Thanks.
Others have posted on the logic of correlation and causation. But those statements are mainly true for simple linear systems. Weather and climate are nonlinear dynamic systems (meaning they have at least one lag and one feedback).
In such systems, especially if complex (meaning many variables) even the language of causality gets very murky. Which says PDO or CO2 as ’causes’ of observed temperature change is probably simplistic. Whichnis my reply to the post’s question. But there is more.
Analogy. Which came first (a causation question), the chicken or the egg? From the egg’s perspective, a chicken is just another way to make another egg. The premise of the question is false, because they evolved together ( metaphore for earth’s climate, which plainlynrelies on the biosphere formthings like carbon sequestation, and atmospheric humidity from plant transpiration). To borrow from Willis’ important post on a single linear equation (actually his orbiter dictum), chickens and eggs are emergent properties of the system we call life. Anyway, you get the idea.
So the question about CO2s causative impact is almost like asking a chicken what impact the thickness of the carbonate eggshell will have on chickens. To which, if I were a chicken, I would cluck, if it gets a whole lot thicker real fast, fewer chicks will be able to hatch out. Not good. But check out this chicks beak. It can peck through a lot more than a wimpy shell twice as thick as now. And why do you think all eggshells are the same thickness? DDT thinned eggshells (cold analogy) and killed raptors (Rachel Carson Silent Spring, humans as raptors analogy). And how fast did you say this was going to happen? (Adaptation versus mitigation analogy).
Just some food for thought.
rgb
Because they can’t feel intellectually superior.
because they can’t full-fill their ideological objective.
because you sure as hell can’t get funding.
If you say we don’t know
Oops, I indented that last entry instead of outdenting it.
Richard M says:
June 5, 2013 at 5:49 pm
However, it still leaves open the question of what drives ENSO.
——————————————–
Some recently came up with the idea, that heat is currently sequestered into the deep ocean and will return and drive temperatures up in a couple of years or decades.
That will not happen because it would reduce entropy and violate the second law of termodynamics.
Temperature increases in the deep ocean are at least an order of magnitude smaller than at the surface, even if we believe the sudden jump while switching to the ARGO system.
That heat sequestered will not recombine when deep ocean water upwells to the surface and so will surface temperatures hardly be affected by the minusule increases in deep ocean temperatures during the last decades.
There will therefore also be no effect on ENSO.
The problem with any analysis of the data is that the data has been muddled with to the point where one has no idea if what one is looking at has any resemblance to reality or not. The only global temperature record I trust is the satellite record which unfortunately means only a snipet of time is available for earnest inquiry.
Me too. I have hopes of ARGO data in a decade or two, but at the moment there is way too much trouble with the data and way too few buoys.
Let me amend that — I don’t completely reject data from earlier measurements, but I pay it little attention when it is presented (as it always is) without error bars. Not just statistical estimates of e.g. s.d., but actual estimates of probable error (which will, of course, be strictly greater than s.d. to account for unknown systematic errors etc).
Of course if you include honest error bars, the curves are a lot less exciting, because most of the exciting behavior is variation on the order of the error bars over decades or more. Honestly, I rather wish that the satellite data was presented with an error estimate too.
rgb
Girma says:
June 5, 2013 at 6:34 pm
As a result, most of the sea ice lost in the warming phase from 1973 to 2005 is natural.
The sea ice lost in the Arctic correlates almost perfectly with the AMO lagged by a few years. While I think the PDO may drive global temperatures, the AMO has its own impacts.
That will not happen because it would reduce entropy and violate the second law of thermodynamics.
Ah, would you care to prove, justify, or otherwise support this assertion? I’m not defending the assertion that heat sequestration is occurring on a grand scale, but it is surely occurring all of the time on some scale, even if only the minor scale of sunlight warming the ocean by day and the heat being released at night. I cannot think of any good reason to think that this or any process akin to this would either reduce entropy or violate the second law.
Note well that even if cold water upwells in some location, if the cold water is warmer than the water that was there above in the first place (as, if you think about it, it is likely to be) it will have the effect of raising the air temperature above it, even if that air is warmer than the water. If you increase a boundary temperature in any heat flow problem, you will fairly generally raise temperatures everywhere as the heat flow adjusts, without violating the second law. This is not a static thermal equilibrium problem, it is a dynamical process in an open system.
rgb
I think if you redate lower temperature values by fifty years and deselect those you don’t like, your graph will clearly show that we are warming at a rate never seen before…I mean, what would Mike do?
feet2thefire says:
June 5, 2013 at 4:03 pm
@Theo Goodwin June 5, 2013 at 3:44 pm:
“You are extrapolating existing graphs into the future. That is not prediction, except in the trivial sense that you are asking “Will the future resemble the past?”
“???? You can’t be serious. ALL scientific predictions are extrapolations from existing evidence. All non-scientific predictions use crystal balls in place of existing evidence.”
Existing evidence consists of true observations of the environment. Observation statements imply only themselves. Science is a system of hypotheses that are taken as true because they have a long record of confirmation. They become confirmed when they are used to imply observation statements that are found to be true.
As an introduction to empirical hypotheses, read some internet account of Kepler’s Three Laws. Except for the radiation laws, climate science today exists entirely as empirical hypotheses.
As an introduction to theoretical hypotheses, read some internet account of Newton’s Theory of Gravitation, which implies Kepler’s work. The concept of gravity is theoretical.
We need a Kepler for the PDO. Then maybe that Kepler will be followed by a Newton for the PDO. They will discover and describe natural regularities that make up the PDO.
Short of a Kepler, we will know nothing of natural regularities that underly the PDO. The PDO might be one grand natural regularity.
Climate science is in its infancy.