Guest Post by Willis Eschenbach
I must admit to being greatly bemused by Michael Mann’s new (and sadly, paywalled) opus magnum about the Atlantic Multidecadal Oscillation (AMO), sometimes called Atlantic Multidecadal Variability (AMV). Here are a couple of quotes from our boy on the subject, emphasis mine:
“The AMO, defined as a 40-60 year timescale oscillation originating in coupled North Atlantic ocean-atmosphere processes, is almost certainly real“
and
“This is a key finding of Knight et al (2005) (of which I was a co-author) as well as Delworth and Mann (2000) [the origin of the term ‘Atlantic Multidecadal Oscillation’ (AMO) which I coined in a 2000 interview about Delworth and Mann w/ Dick Kerr of Science].”
followed by Mann 2021
“Two decades ago, in an interview with science journalist Richard Kerr for the journal Science, I coined the term the “Atlantic Multidecadal Oscillation” (AMO) to describe an internal oscillation in the climate system resulting from interactions between North Atlantic ocean currents and wind patterns. … Today, in a research article published in the same journal Science, my colleagues and I have provided what we consider to be the most definitive evidence yet that the AMO doesn’t actually exist.”
I do enjoy Mann’s implication that he was the discoverer of the AMO phenomenon, when in fact it had been described in detail in 1994 by Schlesinger and Ramankutty, six years before the publication of Delworth and Mann. Also, in the linked Kerr article in Science that Mann refers to above, despite discussing the name “Atlantic Multidecadal Oscillation” in detail, Kerr never says that Mann named the phenomenon … but I digress.
First, what is the AMO? It is a slow temperature swing of the Atlantic, most visible in the North Atlantic. Here’s a graphic of the oscillation.
Figure 1. Long AMO, from NOAA. This shows a period of about 65 years. There are various instrumental versions of the AMO data. This is the longest instrumental version of the AMO held by NOAA, starting in 1856.
Since the first description of the AMO in 1994, the phenomenon has been extensively studied by any number of scientists. A search on Google Scholar shows 31,300 web pages discussing the AMO. So why does Michael Mann now claim it’s not a natural variation of the Atlantic?
Because “state-of-the-art” climate models say so … his study starts like this:
An analysis of state-of-the-art climate model simulations spanning the past millennium provides no evidence for an internally generated, multidecadal oscillatory Atlantic Multidecadal Oscillation (AMO) signal in the climate system and instead suggests the presence of a 50- to 70-year “AMO-like” signal driven by episodes of high-amplitude explosive volcanism with multidecadal pacing
(Protip—any time someone starts out by talking about “state-of-the-art climate models” you can safely ignore their claims … but again I digress.)
Mann’s claim in his new paper, “Multidecadal climate oscillations during the past millennium driven by volcanic forcing“ (paywalled), is that in preindustrial times what people have been calling the “AMO” was actually a stable Atlantic that was being forced by sporadic volcanic eruptions that just happen to have the same frequency as the AMO. But then that volcanic forcing has died out in modern times, and just in the nick of time volcanic forcing has been replaced by anthropogenic forcing … funny how that works. In M. Mann’s world, it’s always the humans who are to blame.
In any case, I thought I’d see what I could learn from the data in both the instrumental and proxy AMO records, along with the volcanic records discussed by Mann. To start with, here’s the Amman et al. dataset that Mann et al. used of 61 tropical eruptions that they say drove the AMO before modern times. I’ve shown the eruptions as vertical lines. On top of these volcano lines, I’ve overlaid several of the empirical modes of a Complete Ensemble Empirical Mode Decomposition (CEEMD) analysis of the eruptions, showing the various longer-term cycles in the data.
Figure 2. Tropical volcanic eruptions, and various CEEMD modes.
Here’s the thing about signals. As the brilliant mathematician Joe Fourier showed way back a couple of centuries ago, any signal can be decomposed as the sum of underlying signals of various periods. CEEMD is like Fourier analysis, except it doesn’t break a signal down into regular sine waves. It breaks a signal down into underlying signals that can change over time, as you can see above.
Now, is there a cycle in the eruption data similar to the ~ 65 year period of the AMO? Well … kinda. But since each and every signal can be broken down into underlying signals, it may just be by chance. The underlying signals have to have some period, and it might just be fifty to sixty years, as in the volcanos.
So that’s the volcanos. How about the proxy records of the AMO? The main one that is discussed by Mann is the Wang et al study, “Internal and external forcing of multidecadal Atlantic climate variability over the past 1,200 years“. The data is available here. It’s based on “a network of annually resolved terrestrial proxy records from the circum-North Atlantic region.” In that study, Wang et al. distinguished between what they called “AMV”, Atlantic Multidecadal Variability”, and the AMO. They said that something like 30% of the variability of the AMV was from volcanoes, and when that’s taken out we’re left with the AMO. Me, I doubt that, because modern volcanoes show little effect on the AMO. I also wanted to see how well the eruptions matched up with their data, so I’ve used their raw “AMV” data.
First I looked at how well the Wang proxy records matched the instrumental records shown in Figure 1. I’ve also added in the 50-60 year empirical mode of the CEEMD analysis of the Amman eruption records shown above in Figure 2.
Figure 3. Two AMO records and one eruption record, 1856 to present.
We see a couple of things in Figure 3. First, the Wang paleo proxy AMV (red) is very close to the modern instrumental AMO (blue).
However, the Amman eruption data is a quite poor match to the modern AMO data. This is no surprise. Look at Figure 1. If you don’t know which year the huge Pinatubo eruption occurred, you couldn’t tell it from Figure 1.
Next, I looked at the longer term view of that same data. Figure 4 shows that result.
Figure 4. Two AMO records and one eruption record, 800 to present.
Again, some interesting things in Figure 4. First, the average length of the cycles in the Wang paleo AMV is 65 years, which matches the modern data.
However, as in the modern period, there’s a very poor fit between the Amman eruption data and Wang paleo data. Among other things, the period of the eruption data averages 55 years, not the 65 years of either the Wang paleo data or the modern instrumental data. So although at times it matches up with the Wang data, it goes into and out of sync with both the instrumental AMO and the Wang AMV data.
So … how did Mann et al. come to their conclusions? As mentioned above, computer models …
The CMIP5 Last Millennium multimodel experiments provide a pseudo-ensemble of N = 16 simulations driven with estimated natural forcing (volcanic and solar, with minor additional contributions from astronomical, greenhouse gases, and land-use change) over the preindustrial period (the interval 1000 to 1835 CE is common to all simulations). We estimate the forced-only component of temperature variation by averaging over the ensemble, based on the principle that independent noise realizations cancel in an ensemble mean.
(In passing, let me note that it is certainly not always true that averaging a number of model outputs means that the “noise realizations cancel”. But again I digress …)
I rather did like the idea of a “pseudo-ensemble”, however … is that a bunch of random computer models hanging out on a street corner smoking cigarettes and pretending to be an ensemble? But I digress …
And what were their conclusions (emphasis mine)?
The collective available evidence from instrumental and proxy observations and control and forced historical and Last Millennium climate model simulations points toward the existence of externally forced multidecadal oscillations that are a consequence of competing anthropogenic forcings during the historical era and the coincidental multidecadal pacing of explosive tropical volcanic activity in past centuries. There is no compelling evidence for a purely internal multidecadal AMO-like cycle.
His claim is that for about eleven centuries, “explosive tropical volcanic activity” made it look like there is an AMO. And coincidentally, just when the volcanic forces left off, a competition between CO2 and sulfate forcings caused the AMO swings.
You’ll forgive me if, given what I see in the Figures above, I don’t find that argument even slightly compelling.
Finally, this is what I love about studying the climate. The science is far from settled, and that gives me the opportunity to learn something new from every paper that comes out.
Here on our dry northern California coastal hillside, rain is forecast starting tomorrow morning and lasting two days. However, around here, rain forecasts even twelve hours out are sometimes way wrong, and it’s generally true for rain forecasts three or four days out. Funny thing about chaotic systems. They tend to be … well … chaotic.
[NOTE: It’s now “tomorrow morning” when the rain was supposed to start … bright sunlight and not a cloud in the sky. Gotta love chaotic systems.]
Seems like out here in the real world, the modelers don’t have that whole “noise realizations cancel” deal completely worked out … but I digress.
My best regards to all, skeptics and mainstream folks alike,
w.
PS—I sign everything I write with my initial, “w.”, and for the same reason I choose my words very carefully—because I wrote them, I take ownership of them, and I know that it is always possible I will be called upon to defend them. However, I can’t defend your interpretation of my words. So when you comment, please quote the exact words that you are discussing. This avoids endless misunderstandings.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
So far in his career Mann has denied the existence of the MWP, minimised the LIA and now seeks to demonstrate that the AMO is not a quasi-periodic internal variation but an artefact of external forcings.
He “proves” this by referencing a climate model – because the climate model doesn’t produce the AMO, except by inducing it by applying external (prior) forcings based on the combination of aerosol forcing curves and well mixed GHG forcing curves, ergo the model is true and the explanation arrived at. So by a circular argument Mann has arrived at the point where he can argue his sophisticated climate model explains anything he wants.
The problem with this argument (other than the circular reasoning, of course) is that it depends entirely on the input priors to the climate models. Without the input priors (forcing curves) climate models just generate random noise.
By the same logic it is now a simple step to argue that any observation can be explained by adding something suitable into the prior forcing. What’s next – El Nino’s inserted into the priors? Then the models will be almost perfect.
Climate models are basically driven via a reductive decomposition . By creating a series of input prior forcings, the model is driven to give the required output ie match the temperature series. It is self-delusion on a grand scale. The proof this is true is that if we average many models the random behaviour largely cancels and we are left with a temperature response over time. The average temperature response over time can be almost perfectly reconstructed from the input prior forcings, ergo the GCM’s are just lipstick on the pig.
What’s even more incredible is that people believe this crap.
“In M. Mann’s world, it’s always the humans who are to blame.”
His manntasy is: Humanns save Gaia from humans.
From a comment posted by me a few days ago to the RealClimate blog:
“Once again we see the old, tired “explanation” for the mid-20th century cooling as the masking of an underlying warming trend by industrial aerosol emissions. . .
When we examine temperature records for regions with little or no industrial activity, lo and behold: the expected warming trend fails to appear. Now I have no idea whether or not the “Atlantic Multidecadal Oscillation” is a real phenomenon or an artifact. But using industrial aerosols to explain the very real cooling trend we see from ca. 1940 through ca. 1979 (40 years!) would appear to be a serious error, produced by the failure to invoke the most fundamental scientific controls.
For details, I refer the reader to the following blog post (http://amoleintheground.blogspot.com/2021/03/thoughts-on-climate-change-part-10.html#more ), where I present relevant temperature data for the Arctic, the Antarctic, Africa, Madagascar, Siberia, Afghanistan, Burundi, Haiti, Kyrgistan and New Caledonia. In all cases, the data for the period 1940-1979 fails to reveal any trace of the assumed warming trend, despite the absence of much in the way of industrial activity.”
Predictably, my comment was consigned to their “bore hole” in a crude attempt to divert attention from embarrassing evidence.
We had slow refreeze this winter in the Arctic and I wonder if this could have meant currents flowing from the Arctic increased the volume entering the North Atlantic. The years since 2007 have been fast freezing up till now as I recall.
Whatever is causing the ups and downs in the Earth’s temperatures, the AMO temperature profile is repeated all over the world. The AMO profile shows it was just as warm in the Early Twentieth Century as is is today.
Here is the US surface temperature chart (Hansen 1999) which closely resembles the temperatue profile of the AMO.
Other regional surface temperature charts from all over the world show the same temperature profile as the AMO and the US Hansen 1999 chart.
Hansen 1999:
Weekly_rise March 9, 2021 5:26 am
You’ve left out some words, and in doing so you are conflating modelworld with the real world. Let me fix it for you.
Like Shania Twain said, “That don’t impress me much“. All you’ve told us is that climate models are unable to replicate AMO variability unless they’re fed some specially selected modeled forcings. Guess what? They’re also not able to replicate the PDO, the El Nino/La Nina phenomenon, tropical cyclones, the AMO, and a host of other natural phenomena.
Not only that, but the specially selected modeled forcings are modeled eruptions up until modern times, when somehow they miraculously die out and are replaced just in the nick of time by modeled anthropogenic CO2 and sulfate forcings … yeah, that’s totally legit.
All that means is that the models are not up to the task … it does not mean the AMO is not the result of internal variability.
w
Willis, surely you see why your comment is a non-response? Maybe there is something wrong with the models here, and that’s the explanation, but you have not shown that. Why do you believe, specifically, the model control runs don’t show AMO while forced runs do? Can you provide evidence to substantiate your position?
Weekly, I have no idea why the model control runs wouldn’t show a modeled AMO, while when they are forced with some specially selected model forcings they do show a modeled AMO. I also have no idea why model runs don’t show a modeled El Nino, or a modeled AAO, or many natural phenomena.
The curiosity to me is why you would take any of that as evidence about the real-world AMO.
In addition, some of the model control runs DO show a modeled AMO. Here’s Mann et al. on the subject.
You do seem confused, however, about the abilities of models. Here’s the 411.
I’ve been programming computers about as long as anyone on the planet. I wrote my first program in 1963, nearly sixty years ago now, on Hollerith punch cards for a computer that took up a small room. And based on a lifetime’s experience in the field, I can assure you of a few things.
1) A computer model is nothing more than a physical realization of the beliefs, understandings, wrong ideas, and misunderstandings of whoever wrote the model. Therefore, the results it produces are going to support, bear out, and instantiate the programmer’s beliefs, understandings, wrong ideas, and misunderstandings. All that the computer does is make those misunderstandings look official and reasonable. Oh, and do it really fast. Been there, done that.
2) “Iterative” computer models, where the output of one timestep is fed back into the computer as the input of the next timestep, are notoriously cranky, unstable, and prone to internal oscillations and to falling off the perch. Generally, they need to be “fenced in” in some sense to keep them from spiraling out of control.
3) As anyone who has ever tried to model say the stock market can tell you, a model which can reproduce the past absolutely flawlessly may, and in fact very likely will, give totally incorrect predictions of the future. Been there, done that too.
4) This means that the fact that a model can hindcast flawlessly does NOT mean that it is an accurate representation of reality.
5) There is an entire branch of computer science called “V&V”, which stands for validation and verification. It’s how you can be assured that your software is up to the task it was designed for. Your average elevator software has been subjected to more V&V than any of the computer models.
6) Computer modelers are all subject to a nearly irresistible desire to mistake Modelworld for the real world. Unless some computer model’s software has been subjected to extensive and rigorous V&V. the fact that the model says that something happens in modelworld is NOT evidence that it happens in the real world.
7) The more tunable parameters a model has, the less likely it is to accurately represent reality. Climate models have dozens.
8) The climate is arguably the most complex system that humans have tried to model. It has no less than six major subsystems, viz, the ocean, atmosphere, lithosphere, cryosphere, biosphere, and electrosphere. None of these subsystems is well understood on its own, and we have only spotty, gap-filled rough measurements of each of them. Each of them has its own internal cycles, mechanisms, resonances, and feedbacks, and each one of the subsystems interacts with every one of the others. Finally, there are both internal and external forcings of unknown extent and effect. For exmple, how does the solar wind affect the biosphere? Not only that, but we’ve only been at the project for a couple of decades. Our models are … well … Tinkertoy representations of real-world complexity.
9) Many runs of climate models end up on the cutting room floor because they don’t agree with the aforesaid programmer’s beliefs, understandings, wrong ideas, and misunderstandings. They will only show us the ones that they agree with, not the ones where the model went off the rails.
As a result of all of these considerations, anyone who thinks that the climate models can “prove” or “establish” or “verify” something that happened five hundred years ago or a hundred years from now is living in a fool’s paradise. They are in no way up to that task. They may offer us insights, or make us consider new ideas, but they can only “prove” things about what happens in modelworld, not the real world.
Finally, be clear that having written dozens of models myself, I’m not against models. However, there are models, and then there are models. Some models have been tested and subjected to extensive V&V and their output compared to the real world and found to be very accurate. So we use them to navigate interplanetary probes and design new aircraft wings and the like.
Climate models, sadly, are not in that class of models. Heck, if they were, we’d only need one of them, instead of the dozens that exist today and that all give us different answers … leading to the ultimate in modeler hubris, the idea that averaging those dozens of models will get rid of the “noise” and leave only solid results behind.
I hope this explains my position on M. Mann’s claims about models and the AMO.
w.
Willis, this answer is worth a separate post.
(More especially the part starting with ‘I’ve been programming’)
Thanks, Wim, good idea. I’ll do that. I’m tired of folks accusing me of “not liking models” and that kind of thing.
Appreciated,
w.
Willis:
You are blowing smoke.
“Models” cannot model El Ninos or La Ninas because they are caused by random volcanic eruptions, and increasing or decreasing levels of Anthropogenic SO2 aerosol emissions,
Further, there is zero evidence that CO2 has any climatic effect, and any model incorporating CO2 will necessarily fail.
You are doing everyone a great disservice by your failure to recognize the overwhelming effect of atmospheric SO2 levels.
Oh, piss off. I don’t stand for anyone telling me that I’m “blowing smoke”. I tell the truth as best I know how. Go whine at someone else. Not interested.
w.
Willis:
I was actually responding to your point #8.
You are normally right on, but in this instance you completely ignored the essentially perfect correlation between changes in atmospheric SO2 aerosol levels, and temperature changes.(of which I have made you aware), and instead described Earth’s climate as being an exceedling complex.system.
It is, in fact, an very simple system, with no cycles of any kind. Add SO2 aerosols, and it cools down. Decrease SO2 aerosols, and it warms up.
Burl, you accused me of “blowing smoke”. Go blow your smoke at someone else. Not interested. We’ve been through this before, and I found nothing of note in your theory.
w.
Willis:
You keep calling my analyses a “theory”
The actual theory is that SO2 aerosols are the Control Knob for Earth’s Climate.
The graphical data which I present is PROOF of the theory–every significant increase or decrease in average anomalous global temperatures can be shown to be, or attributable to, changing levels of SO2 aerosols in the atmosphere.
And you are incapable of finding anything of note in my theory!! It is the only proven climate theory in existence.
Was something unclear in my last comment?
w.
“Weekly, I have no idea why the model control runs wouldn’t show a modeled AMO, while when they are forced with some specially selected model forcings they do show a modeled AMO. I also have no idea why model runs don’t show a modeled El Nino, or a modeled AAO, or many natural phenomena.
The curiosity to me is why you would take any of that as evidence about the real-world AMO.”
Willis, I propose simply that this is a thing worth exploring. Mann has developed a hypothesis about why the control runs don’t show an AMO and forced runs do, and he provides a lot of solid evidence supporting it. Simply saying, “I bet the models are wrong, though” does not let us off the hook. If they’re wrong we need to understand why they are wrong and to understand what that tells us about the system. This is what I mean when I say the responses I’ve seen have been inadequate and not compelling.
Your general points about models are well taken, if overly cynical in my view. I’m a strong proponent of the aphorism that all models are wrong, but some are useful.
Weekly, you say:
Sorry, but that is simply nonsense. He has not provided one scrap of evidence that anything is happening. He has provided computer control runs and computer forced runs. If the results of computer runs actually were evidence I’d be a rich man, and so would a lot of other people. But they’re not. They are just as likely to be wrong as any other human calculation. Evidence is facts, observations, logic, math. Solid. Testable.
And in fact, regarding your claim that he has a “hypothesis about why the control runs don’t show an AMO and forced runs do”, according to his study the AMO appears in some unforced runs and not in others. Here you go:
Next, you say
It tells us NOTHING about “the system”. It only tells us something about the models. Again you are conflating modelworld with real world.
Consider again the fact that some control runs produce an AMO and some don’t. Does that tell us anything about “the system”? Not a thing. Those runs that reproduce an AMO phenomenon may be right for the right reasons.
Or they may be right, but for the wrong reasons.
Or they may be wrong for the right reasons.
Or they may be wrong, but for the wrong reasons.
But that tells us NOTHING about “the system”, only about the models.
Finally, as a lifelong computer programmer, I couldn’t disagree more with the claim that “All models are wrong but some are useful.” Consider the CFD models that the Boeing engineers use to design wings on jumbo jets. Are you going to tell me with a straight face that those models are wrong? If you truly believed that, you’d never fly again.
Not only that, but while models that are right are absolutely useful, for a model that is wrong to be “useful”, we need to understand WHY it is wrong. But with the complex iterative climate models with dozens of parameters required, where the output of one cycle is used as the input to the next cycle, determining where it went off the track is nearly impossible. Was it an error in the parameter that specifies the ice temperature at 10,000 feet elevation? Was it an error in the parameter that limits the formation of melt ponds on sea ice to only certain months? No way to tell.
In addition, the gridsize of the computer models are far larger than important climate phenomena like thunderstorms, dust devils, and tornados. If the climate model is wrong, is it because it doesn’t contain those phenomena?
Heck, we don’t even know if the Navier-Stokes equations as they are used in the climate models converge to the right answer, and near as I can tell, there’s no way to determine that.
Finally, as I mentioned above, a computer model is nothing more than my ideas made solid. If I think CO2 is the secret control knob for the global temperature, the output of any model I create will reflect that.
But if I think (as I do) that the temperature is kept within narrow bounds by emergent phenomena, then the output of my model will reflect that.
Are the outputs of either of my models “evidence”?
Not on this planet.
So I fear Mr. Mann has not provided us with evidence of anything, and the record of historical eruptions that he claims are the secret cause of the AMO (but only up to the 1800s or so) simply do not support his claims.
Best regards,
w.
“And in fact, regarding your claim that he has a “hypothesis about why the control runs don’t show an AMO and forced runs do”, according to his study the AMO appears in some unforced runs and not in others. Here you go:”
The fact that these two models turned out to be outliers is rather one of the major factors motivating the study, is it not? A primary source of evidence for the AMO as an internal oscillation over paleo timescales came from the fact that extended (1000+ year) control run simulations with these two models show AMO-like oscillations. Mann et al.’s 2020 study shows that control runs with the CMIP5 ensemble do not. This latest paper builds on this finding to show that the apparent AMO oscillations in the paleo record coincide with volcanic forcing.
The CFD models used by Boeing are almost certainly wrong, since a model by definition is a representation of a real system that at some level is too complex to capture in a model. Being wrong doesn’t mean you can’t fly a jumbo jet or land a rover on Mars. Agreed that we have to try to understand why wrong models are wrong – that’s how we improve our understanding. I also agree that it’s extremely hard to figure out why a given model is wrong, which is why so many really smart people spend such an awfully large amount of time trying to do it.
Conjecture possibly explaining the mechanism of the AMO?
1) High tide in the North Atlantic occurs over numerous lines of longitude affecting diverse straits to Arctic ocean (from North Atlantic) for a relatively protracted daily time period.
This sea water influx to the Arctic ocean is comparatively deep and cold mixture compared to the outflow of warmer surface water at the Bering Strait.
The Bering Strait is 85 km wide at it’s narrowest and only 90 metres deep in places.
Hydraulic tidal pump – Colder water of mixed depth IN to Arctic Ocean from North Atlantic. Warmer surface water OUT to Pacific.
Duration – Relatively LONG time period
2) High tide of extremely short daily duration at Pacific region passing through the narrow Bering Strait from the Pacific into the Arctic Ocean.
Hydraulic tidal pump – Small volume of warmer surface water IN from Pacific to Arctic Ocean. Small amounts of mixed depth water OUT to North Atlantic.
Duration – Relatively SHORT time period.
The overall effect of the above processes would be to eventually create a water temperature imbalance that possibly goes some way to explain the periodic AMO?
The correlation between AMO and UK Sunshine hours (and hence UK temperatures) seems quite solid. Am I missing something?
You are correct. Because it is basically surrounded by the Atlantic Ocean, the weather of the UK is very dependent on the weather of the Atlantic, including the effects of the AMO.
w.
Willis thank you for your reply. I’m sorry I didn’t make myself clear.
My question is really a) does the increased sunshine cause the AMO (plausible), or b) does the warm phase of the AMO cause more sunshine (how?), or c) are they both manifestations of a third variable (cosmic rays, higher UV solar output etc)?
Andrew
Andrew, as the current discussion shows, the origins of the AMO are far from clear. However, when the temperature of the surrounding ocean changes, the weather of an island changes. I would hazard a guess that the causation runs AMO —> ocean cooling —> increased UK cloudiness, but causation is an elusive animal in the world of climate.
w.
Thank you; much appreciated!
Andrew