Guest essay by Charles G. Battig, M.D.
Some say that “God” might reside in a computer…the Deus ex Machina, literally means “god from the machine”. Amongst those individuals are those divining climate with climate computers in which are embedded general circulation models. This has generated a belief system…belief that all variables which drive global climate at all time scales have been identified, quantified as to individual contribution and interactions, and that chaotic variability is foreseeable. At the current state of scientific knowledge, such belief is intellectual hubris masquerading as achieved scientific endeavor.
Massachusetts Institute of Technology pioneering meteorologist and mathematician Professor Edward Lorenz doubted this ability in the 1960’s. The serendipitous discoverer of chaos theory postulated “is there such a thing as a climate?” Is there a definable “normal global climate” from which deviations might be termed abnormal? His 1965 paper includes “…if in addition the present state or the present and past states are not known with complete accuracy, any forecasting procedure will lead to poorer and poorer forecasts as the range of prediction increases, until ultimately only the periodic component can be predicted in the far distant future.” His statement is a description of what has become known as chaotic behavior. Such systems are characterized by the fact that tiny changes in initial conditions may result in wildly different final outcomes over a period of time. Climate behavior aptly fits the definition. Short term changes are known as weather, and weather prediction accuracy has improved out to a week or so over the decades.
Now Judith Curry has had the courage to note the absence of clothes on the climate-computer emperor. Professor Curry, the author of over 180 scientific papers on weather and climate, recently retired from the Georgia Institute of Technology where she held the position of Professor and Chair of the School of Earth and Atmospheric Sciences. She has authored (via the GWPF who reformatted it) “Climate Models for the Layman” in which the fundamental problems inherent in computer modeling are laid bare. These problems are serious enough to cast doubt on the ability to construct such a climate forecasting system. Current climate model predictions diverge from historic reality when viewed over decadal time-scales. Yet these fallible predictions, otherwise known as scenarios, are used by politicians, environmental advocacy groups, and energy firms to set public policy and future energy plans.
Professor Curry:
“It’s not just the fact that climate simulations are tuned that is problematic. It may well be that it is impossible to make long-term predictions about the climate – it’s a chaotic system after all. If that’s the case, then we are probably trying to redesign the global economy for nothing.”
However, there have been those amongst the “we” who have learned to profit mightily from trying to “redesign the global economy.” Billions of dollars have been spent by governments to control energy production and use in their attempt to control the climate. Billions of taxpayer money have been directed to the promoters of numerous such schemes. Controlling energy means controlling all aspects of modern life, including personal freedom. Man-made dangerous climate change is a prime example of “false bad news”, a term coined by economist Julian Simon to define the use of false news scare tactics in the media in his treatise “Hoodwinking the Nation.”
Kudos to Professor Curry for exposing this false god.
Charles G. Battig, M.S., M.D., Heartland Institute policy expert on environment; VA-Scientists and Engineers for Energy and Environment (VA-SEEE). His website is Climate Reality
For a non linear, chaotic fractal iterative system as is climate, if the computer model produces the same answer twice, it’s wrong.
My understanding is that the computer models do not use explicitly chaotic models, though.
Which is why the computer would be wrong, if it produced the same answer twice.
TH, actually they have to. The Navier Stokes equations for convective flow are inherently chaotic. In fact it was Lorenz simulation of a convection cell (Tstorm is a good example) that first discovered mathematical chaos including sensitive dependence on initial conditions and strange attractors.
Any nonlinear dynamic system is chaotic. Nonlinear means feedbacks. Dynamic means the feedbacks are not instantaneous. Climate obviously has feedbacks (clouds, water vapor) and they do not work instantaneously. So any mathematical representation of them will be technically chaotic.
ristvan sure they have to, but they dont really. huge swathes are linearised;we dont have the ability to actually solve chaotic models with any degree of accuracy.
The computer models used by AGW people are trash. They don’t have the law of thermodynamics for solving temperature of compressed fluids in them, they have the mathematics for solids and liquids: Stefan-Boltzmann,
but not the step for ascertaining density of the compressible atmosphere. This makes them give off a temperature that is 33 degrees below what the actual atmospheric global temperature is.
If we cannot forecast weather, how can we possibly forecast “climate?” This is particularly problematic since there is no “climate” data whatsoever: daily temperatures – weather, daily rainfall – weather, annual varves – smoothed weather effects, tree rings – weather (rainfall and soil water), sea temperatures – well mixed, smoothed weather, ice core data – the same. The data used to discuss and model “climate” is weather data. We create “climate” out of weather observations. So, unless we can forecast weather, and Lorenz concluded that was failure at ranges of about two weeks IIRC, then how could we meaningfully discuss “future climate?”
If the program using the same inputs and sequence of numbers as a source of “randomness” doesn’t produce the same results, the program is broken. Reproducibility is key not only in science but in computer science.
My only point was that a valid chaotic systems has an underling deterministic (say programmable) mechanism but the output is unpredictable.
TOM not quite. Multiple runs of the same program with the same inputs will diverge unless the inputs are infinitely precise. But they will all be within the strange attractor envelope. For example, the Mandelbrot set is such a strange attractor envelope.
Jax. True, but that is not lorenz discovery. Vary the starting inputs only slightly and the result varies greatly even though the program is the same.
jaxad, as ristvan says below computers, much less measurements, cannot be infinitely precise. And it’s not just numbers. Some fractions, which work perfectly in arithmentic(1/2 x 1/3= 1/6) only come out correct on computers because they are “fixed”. (1/3=.333333….) Others never end pi x r^2 gives the area of a circle exactly, but 3.14159…..X 2 can never be an exact answer.
Look up the video “How to Believe Six Impossible Things Before Breakfast” by Christopher Essex. He shows a marvelous graphic of what zero looks like to a computer!
ristvan, philohippous, how is that relevant to reproducibility?
I misread TOM as saying a program that simulates the atmosphere must never do the same thing twice, which is false (“if the computer model produces the same answer twice”). You should always get the same answer with the same exact inputs or the program is broken. Varying inputs or source of randomness (an input, really), could give different results, yes. Precision doesn’t matter.
Some of the climate models used can never produce the same output twice because of programming errors. Others output the same thing with wildly variant inputs (like the model that turned Mars into Venus). The former are broken programs; the latter are incorrect programs doing the wrong thing, yes. Neither should be used.
@philohippus
It’s actually worse than we thought. We know that the decimal representation of 1/3 is 0.3333…(neverending), but perfectly well-behaved decimals become infinitely repetitive when translated into binary. This leads to rounding errors. Even double precision only preserves 15-17 base 10 digits. I don’t know if anyone has ever done a sequence of runs to explore what happens when you drop a couple or four digits.
The problem is the inherent lack of infinite precision, and the translation of numerical data between bases. Add to that the finite precision of actual observation data. Lorenz discovered that in the 1960s. Instead of a dense weave of mathematical work you are really dealing with a cobweb. It is not the program but the actual physical limitations of the data and storage systems and the logical limits of the mathematics. If you doubt mathematics has real and important limitations, read up on the “incompleteness theorems.”
Meh. Accuracy, shmaccuracy.
““The weaker the data available upon which to base one’s conclusions, the greater the precision which should be quoted in order to give the data authenticity.”“
To close out my blabbing here, my run in with the froth line many years ago, and my affair with Mandelbrot’s DNA as @Rud Istvan reminded me above..
https://notonmywatch.com/?p=679
All climate models lose or gain energy with each iteration due simply to round-off error, even when perfectly programmed. This happens even when there is zero change in the forcing. This does not happen on the real planet earth because energy cannot be created or destroyed. This artificiality created and destroyed energy is not allowed for by any laws known to science. So when someone suggests climate models describe the real world, they do not. At a very fundamental level. They violate known physical laws.
Isn’t all this sophistry – given that the physical chemistry of carbon is well understood, icebergs are breaking off glaciers at both ends of Earth, every year sets a new record high for global surface temperature, and sea level rise is accelerating at a rate we haven’t seen in thousands of years?
The folks forced out of drowned cities around the globe are not going to be arguing about mathematics, they’re just going to say it was bleeding obvious!
So sad to to see such inane comments still being made, Jack. You are so factually incorrect, so far from reality, that it is not worth anybody’s while pointing out your errors in detail. You might be better off in one of those Alarmist echo chambers.
Deus ex Machina comes from the Greek theatre (yeah I know it’s Latin) where, at the climax of the play a “god” would be lowered on stage to make all things right. Nothing could be more appropriate than a term from theatre for the theory that we’ll doom the planet by feeding the plants. After all the primary component of theatre is the suspension of disbelief, the willingness of the audience to suspend their reality and accept the ‘reality’ presented on stage by actors.
Could ‘Trump’ = Deus in the theatrical production ‘Global warming’?
Regarding “climate models”, would
“Feces ex Machina” be more appropriate?
The machine was, of course, the stage machinery. The term was one of contempt for playwrights who could not resolve the tangles of their plots by any other means. “It was all a dream” is just about as lame.
But no God seems to have emerged to resolve the tangle of Global Warming Theory, and the world has yet to wake from the horrid dreams of the Alarmists.
Charles, I beg to disagree with your statement that climate scientists that “all variables…have been identified”. IMO The IPCC dictum that human activities are the primary cause of “climate change” acts to prevent many researchers from considering a range of non-human variables, such as natural cycles related to our solar system, changes in sunspot activity, etc. for example the Australian bureau Of meteorology still says that it doesn’t study sunspots (try using the term ‘sunspots’ in their search bar!). Maybe some skeptical scientists do, but they are probably a small minority.
He was referring to the ones who designed the computer models. ” Amongst those individuals are those divining climate with climate computers in which are embedded general circulation models. This has generated a belief system…belief that all variables which drive global climate at all time scales have been identified, quantified as to individual contribution and interactions, and that chaotic variability is foreseeable. “
Quantified?
They are not even sure if the water vapor term should have a plus sign or a minus sign.
You should add the implied /sarc to several of the above comments.
Boy, do the climate models still start with a disk in formulating the equations rather than a sphere?
Do they still specifically exclude clouds from their model’s because they’re too hard?
They have too, until the get the size of the grid boxes down by several orders of magnitude.
Clouds are parameterized rather than explicity modeled from first principles. They are not expressly excluded.
Yes, as ristvan says, the clouds are parameterized.
In other words, estimated. Much like you might use 3 for the value of PI when you only need a quick “ball park” figure. Parameterization, much like estimation, is often called “making an educated guess”.
The grid boxes would have to be on the order of 1mm for the oceans- that is ~ where vortexes disappear into the Brownian motion of water. Vortexes in the air may be a bit bigger. That’s about 8 or 9 orders of magnitude smaller than a 200km grid box.
Climate models handle non-linear functions somewhat like what they taught in introductory calculus- By splitting up the area beneath a curve it’s possible to estimate the area under a curve quite accurately. By solving the integral equation it’s possible to get an exact solution.
“Current climate model predictions diverge from historic reality when viewed over decadal time-scales.”
________________
The IPCC AR5 (2013) assessed CMIP5 climate model performance as annual anomalies against a 1986-2005 base period, ending in 2012. They set this out clearly in their fig. 11.25: http://www.climatechange2013.org/images/figures/WGI_AR5_Fig11-25.jpg
The middle chart in that group has been faithfully updated each year by climate scientist (and all round nice guy besides) Ed Hawkins from the University of Reading, UK:
looking at Ed’s chart I don’t really see how the claim that “Current climate model predictions diverge from historic reality when viewed over decadal time-scales” can be justified.
Is there something I’m missing that other folks here can see?
DWR, if you look at the 1998 El Nino peak, it was at the top of the confidence range. The recent peak hit the average of the predictions and is heading down. It is likely that observations will be near or below prediction bottom levels soon.
The models are running hot.
It seems as though you missed the link to the statement
Yes. See John Christy’s Congressional testimony chart from 2 feb 2016 at docs.house.gov. Google Christy congressional testimony 2016. First hit. Ed Hoskins presents the best models case possible, but it contains logical fallacies exposed by RGBatDuke. Christy presents a simpler, clearer picture.
I haven’t seem any comments from RGB lately.
JK, one possible reason is that he is right and there is nothing left to sau.
What I see from your figures is that from 1985 – 1998 (est), the actual temps rode the top half of CMIP5 estimates (light gray). From 1999 on, they have ridden the bottom half, indeed going as far as the “min low estimate” of all models.
Even with the El Nino of 2016, temps have not broken into the top half of the light gray estimates.
This suggests that the models are trending hotter faster than the actual temps.
The newer estimates (dark gray) do better, but they are a clear concession to the fact that the models needed retuning.
Further, the model performance from 1999 – 2014 is much worse than 2015 and 2016; whether this is the result of El Nino OR a manifestation of Skep Sci’s “step function” model remains to be seen.
Yes, you are presenting CMIP5 data that predicts the past and not the future. And shows no historical predictive value whatsoever, especially taking into account the recent El Nino event.
That’s what your are missing.
No Climate model has ever predicted anything accurately. Ever.
My dyslexia turns CMIP into CHIMP.
I think Judith is referring to the whole of the 20th Century. Perhaps even the 19th Century as well. The models are tuned to increasing CO2, so they show no temperature chance in those earlier decades, when the temperatures were indeed changing.
They great myth of climate modelling is the belief that the computers enlighten us about the world, giving us information that we did not know. But the results of the models are predetermined in the programming and have no connection to reality.
The last Julian Simon book is a treasure of enlightened thought.
The ‘machine’ referred to in ‘Deus ex Machina’ is stage machinery, basically a winch used to lower someone into the middle of the ‘action’.
King Charles 2nd employed such methods in his Court Masques…descending from high into the turmoil of life to restore heavenly peace and tranquility to his kingdom.
And we all know how that turned out!
Pretty well for Charles II and Nell Gwynn. Followed soon after by the Glorious Revolution and the Bill of Rights so not all bad!
It took a bloody war.
And that’s an adjective not an expletive.
I think you might be confusing Charles I and Charles II. Charles I was deposed and executed, not his son.
Did I say Charles 2nd? Tsk. Well at least you know what I’m talking about!
Back-tuning models to agree with reconstructions that diverge from instrument records is one of many 800 lbs gorillas in the room.
The CMIP5 models forecast from 2006 onwards I think. The cut-off between hind-cast and forecast is shown on this chart as the vertical dashed line:
On an annual basis at least the forecast projections have been reasonably good.
DWR54
“On an annual basis at least the forecast projections have been reasonably good”.
I have a hard time with that statement – which of the lines on the graph is the unadjusted (or least adjusted) actual data? Exactly what are the models being compared to?
Good to know. Of course, this means the reconstructions are useless,
and or, the models cannot be tuned to mimic past, current and future conditions. The latter of the two seems most probable.
On an annual basis at least the forecast projections have been reasonably good.
So summer is warmer than winter?
Temperatures have been going up 0.7c per century for the past 150 years. Long before CO2 could have been the cause. The models say this rate will accelerate due to CO2. History says this rate will continue. Mathematics tells us that the model results cannot be trusted. Observation is telling us that warming is continuing at a rate of 0.7C, with no acceleration in warming.
It is the lack of any observed acceleration in warming that is telling us the models are wrong. Their predictions have failed. Both for warming and the hot spot. In all other sciences a single failed prediction means your theory is wrong.
Perhaps a dumb question – do they back-tune to the raw data from the past or the adjusted data?
It would be somewhat ironic if the models were tuned to the adjusted data and that’s why they are so poor at forecasting!
Which adjusted data? Because GISS and others continue to adjust the past, this in itself invalidates all previous training and thus invalidates all previous predictions based on this training.
Thanks Dr. Battig, for An erudite look at reality!
Deus ex Sugarcubes…
charles nelson February 22, 2017 at 12:55 pm
King Charles 2nd employed such methods in his Court Masques…descending from high into the turmoil of life to restore heavenly peace and tranquility to his kingdom.
And we all know how that turned out!
Turned out rather well I think. Not so good for his brother who succeeded him.
Sorry, didn’t spot this before posting. James II captured in Faversham while trying to flee to France.
http://www.faversham.org/history/wall_plaques_-_people/james_ii_(2).aspx
I did mean Charles 1st…I’d just gotten up!
easy now, chuck.. you’ll strain something. I too knew what you meant. I’m dyslexic at the proper noun and modifier substitution level
The main quote from Dr. Curry’s paper is logically flawed. Just because we might not be able to
predict the future climate does not mean that we are not changing it for the worst and that we should not be taking steps to prevent it. And in the article Dr. Curry accepts that increasing CO2 will increase the temperature and states that the issue is the uncertainty in the equilibrium climate sensitivity – values over 2 are dangerous values under 1 are harmless. Since Dr. Curry states that we cannot predict the value with certainty then the actual value could be much higher (some estimates based on past climates suggest values of 10 for example) and so there might be an urgent need to act.
Furthermore Dr. Curry’s paper seems to do nothing more than state the obvious – all models have associated errors. But nowhere does she try to quantify the error in climate models. Based on a
comparison with the past the error would be less than 0.3% (0.5 degrees out of about 300 K). So does this mean that we can predict the future temperature with the same level of accuracy?
“Dr. Curry states that we cannot predict the value with certainty then the actual value could be much higher (some estimates based on past climates suggest values of 10 for example) and so there might be an urgent need to act.”
But how would we know? That is the nub. How much sweat and treasure do we put into attempting to alter an unknowable future?
+10!
Not just “how do we know”. What direction is also relevant.
It is acknowledged that an ice age in not unlikely What are we doing about it?
Do we even have plans?
The fundamental problem is that the future is not predictable. Adding CO2 likely increases the odds of warming, the way that lowering interest rates increases the odds of increasing economic growth. Like rolling the dice, just because the odds favor rolling 7, doesn’t mean you will roll 7, because there is still a greater chance of rolling something other than 7.
This is not allowed for when a computer says warming will be 2c. What the computer should be saying is 2C +/- 4C.
There is no evidence that we are changing the climate, much less changing it for the worse.
Any change by man is so far below natural variability that it can’t be discerned yet, if ever.
The claim that an increase of 2C means harm has no basis in reality. Even the guy who first put it forward admits that he made it up on the spot.
The world has been as much as 2 to 3C warmer than it is today in the last 5000 years, with no harm to anyone.
The world has been as much as 5 to 7C warmer within the last 20000 years, with no harm to anyone. In fact that period is referred to as the Holocene Optimum precisely because conditions were so good for life.
There’s not a single person or group of people in existence who know how to reduce the temperature of this planet.Or who even knows what effect reducing C02 would have. No one.
I’ll be damned if I want my government spending trillions of dollars chasing this chimera.
Just because we might not be able to predict the future gravitational constant does not mean that we are not changing it for the worst and that we should not be taking steps to prevent it.
After all, it might be changing. There is uncertainty as to the measurement. And the Universe has more energy than we can identify without a variable gravitational constant.
Now you might think that is pure speculation. You would be right. So I won’t ask for the world to change the whole economic system to prevent it.
Curry has not made a logical flaw.
Unless you will also accept my wild fantasies as world-changing too. And also any other madman who has an hypothesis too.
You are restating the foolish precautionary principle without evidence. Observational TCR is ~1.3, ECS ~1.5-1.65, half of models than run too hot. Data table in Curry’s paper. Read it. See also comment just posted below. In thirty years of CAGW alarm, there is zero evidence for actual alarm. Except for the now rapidly cooling El Nino blip,of 2015-16, there has been no warming this century except by Karlization– a period during which about 1/3 of the entire increase in CO2 concentration has occured. SLR has not accelerated. Polar bears are thriving. Planet is greening. English children have plenty of snow. Normal California drought now emphatically broken. Meanwhile South Australia rushed to high renewable penetration with insufficient grid inertia and has suffered one major and 5 rolling blackouts since October. Today SA lost another 250 jobs as Coca-Cola is shutting its bottling plant there; high cost unreliable electricity a factor in deciding against modernization investment.
As for model error, a number from CMIP5 in a referenced chart in essay Models all the way Down. Anomalies hide the fact that the models disagree about actual GAST by +/- 3C. So they cannot even agree about water phase state changes. Another example. The CMIP5 ensemble mean is running 3.5 times hotter than balloon and satellite observation of the tropical troposphere. 350%, not 0.3%.
SLR?
Single Lens Reflex?
Standard Liars Revival?
MS, you are either a newby or a troll. SLR = sea level rise, as everybody in these environs already knows.
Per CAGW, was supposed to increase. Hasn’t. See my books for specific and quite amusing details. Recommend especially essay PseudoPrecision in ebook Blowing Smoke.
Touche’! Anyone who looks at the absolute temperature outputs will find that the computer models put out nonsense that is completely hidden by converting to “anomalies”.
Ristvan:
Mostly you are a font of knowledge :).
Although some reports suggested Coke is closing in part due to power the spokesperson made it clear that it was not due to power issues. Bottling operations often shutdown on a day basis and power is not a high proportion of costs.
The site has no room for expansion. Its location is high value real estate so the economics favour bringing in product from other plants.
Think you meant 250% higher than the ensemble mean by comparison.
M Simon – Sea Level Rise
So the world might be warming. Or it might not be. What is the basis for taking any action? And even if the world is warming, why would you want to stop it (even assuming you could)? One thing we do know for sure is that we are in an interglacial period right now. The ice will return, we just don’t know when. This is not honestly disputed. So any respite we get now is a good thing. Maybe it will give us time to prepare so we can survive the return of the ice. Unless we de-industrialize, then we are doomed.
the logical flaw is in your reasoning that in the absence of any knowledge whatsoever we should assume that we are doing harm and doing nothing would be infinitely better.
This is utter gobshite of course.In the absence of any knowledge doing and not doing have equal probabilities of being the correct course
The essence of the skeptics positions is simple:since we have absolutely no idea whether or not CO2 makes any difference let alone whether its a good or a bad difference, logic dictates we shouldn’t spend money trying to make a difference that we neither know we can achieve, nor yet whether its a good or a bad difference.
spending money inefficiently (green schemes) does harm already … to do harm in an attempt to prevent a theoretical possible harm (a warmer earth has not been shown to be harmful) is simply ignorance … thus far the only harm to mankind related to CO2 is the money wasted trying to decrease our use of CO2 …
” (some estimates based on past climates suggest values of 10 for example)”
To the contrary. There is NO evidence, from any past climate, that CO2 has EVER paced warming.
The ice age evidence shows temperatures falling when CO2 is high, which means CO2 is causing cooling.
Or something else is causing cooling and CO2 is just along for the ride.
Weaknesses in climate models are a major Achilles heel for CAGW. These weaknesses need to be turned into simple talking points and sound bites whose repetition slowly drives them into public consciousness:
Computational intractability forces large grids.
Large grids mean key processes like clouds and thunderstorms cannot be simulated; they have to be parameterized.
Such Parameters have to be tuned to best hindcast (for CMIP5, from YE2005 back to 1975.
Tuning drags in the attribution problem.
The attribution problem is that the warming from ~1920-1945 is essentially indistinguishable from the warming ~1975-2000 over the tuning period. Yet IPCC AR4 WG1 SPM figure 8.2 says the earlier period is mostly natural; simply not enough change in CO2. So how much of the later period is also natural? The models assume its all CO2; that is why they are running hot compared to observation.
also add:
– parameter tuning allows any value, or range of values, for climate CO2 sensitivity. Confirmation bias ensures sensivity meets expectation.
rist- actchurly, the models assume most of the warming comes from water vapor- caused by CO2 warming the surface, increasing evaporation and humidity- an effect that has not been documented.
How about “Models reflect the opinions of the modelers, not reality.”
Simple enough. If clarification is requested, clarification can be provided.
It should also be pointed out that the fact that climate is chaotic in a mathematical sense allows it to be predicted. In the Lorentz system there is a strange attractor which means that all trajectories end up in the same region of phase space irrespective of their starting points (within a limited region of course). Hence you can make very precise accurate predictions about the model if you ask the right questions. The climate is the same — it is always going to be colder in winter than in summer and
climate models get that right. Similarly there is a huge difference between predicting the average rainfall over a year and predicting on which days it will rain. The first can be done the second is
essentially impossible.
…are they still predicting California’s permanent drought
they’ve moved beyond traditional notions of drought defined by historical precipitation, snow pack, or soil moisture. California drought now means a permanent deficit of water supply vs human demands.
So if everyone leaves California there can never be a drought.
You are correct. Except some climate models dont do seasons at all because not ‘relevant’. Acting on CAGW means knowing how much different in winter, how much different in summer, and whether that matters. Seasonality is not trivial; in temperate zones like Chicago (I have a house in Chicago) the average in July is 23C, and January is minus 6C– an average seasonal swing of 29C. Yet we are told by warmunists that an annual change of 2C (ok, chicago july 25C and january minus 4C) will be catastrophic. Nonsense. My permanent residence is now Fort Lauderdale, Florida where the average July AND August temp is 32C. The average in January is 24C. And now you know why I relocated to Florida. Needed climate change.
Except the whole problem is ill posed. “Climate” is not well defined, and global average temperature is not physically real, it is completely synthetic. Because we don’t have a good definition of past or current “climate”, there is no valid way to compute the boundaries of the system. Not to mention that the parameterizations are not valid chaotic processes. And I haven’t even gotten to all the software issues. Do you have any idea what the complexity scores are for these programs? From one analysis I saw, they were off the charts, which means the are nearly certain to have bugs. How much do they affect the output? Who knows, but I wouldn’t trust them any more than an uncalibrated thermometer (or any lab instrument for that matter).
And the coding was in large part done by academic, not by professional coders. In effect those building the program were working on a task outside their level of skill. Like having the neighbors teenage boy come over to tune-up your new car.
Quite right!
A sensible look at the past would indicate that the climatic attractors would glacials and inter-glacials!
I guess that in addition to a major meteor hit and a major volcanic eruption the only natural disaster that could result in mass extinction would be a new ice age, and based on the earths history, my guess would be that it is extremely likely it will happen again.
All this confusion about software and the infinite correctness of the most holy “computer”.
Reminds me of the first time I tried to run payroll.
Now there is a good definition of confusion.
Lorenz was one of the fathers of climate modelling. He is arguably the father of chaos theory. It is gobsmacking that climate modellers have chosen to ignore his discoveries.
Lorenz was quite critical of climate models as they are currently done.
In other words, if we just use the physics and a set of original conditions then we might have a valid model. Any tuning or curve fitting renders the model invalid. I have never seen any climate modeller explain why Lorenz was wrong. They, and those who rely on their models, just ignore him.
It is enlightening to see how Lorenz learned what he did.
Many years ago, I heard a radio interview with Lorenz. He was running a model and he needed the results for an upcoming conference. The computer crashed and he had incomplete results. He didn’t have time to rerun the model. To speed things up, he ran the model with half as many significant digits. That would enable it to complete in time to be ready for the conference. His expectation was that he would get less accurate, but still useable, results. Imagine his surprise when the results were completely different from those he already had for the previous run. Here’s another version of the same story.
On investigation, Lorenz discovered that a tiny change in initial conditions would completely change the outputs of a system. The term Butterfly Effect was coined to describe this. Here’s a quote by Lorenz:
when all you have is a hammer, everything looks like a nail.
Chaotic systems are in essence harder than random systems. Random systems can be analysed statistically. Concepts like mean and median and standrad deviation have meaning.
Not in chaotic systems
Todays outlier is tomorrows average.
To accept that climate is chaotic is to accept the impossibility of prediction.
I did the exact same thing running a RANS back in university. As an under-graduate student I was only allowed one minute of CPU per run on the Univac, so I made the program write to memory the intermediate results before the time was up and used the dump as starting conditions for a new run initiated as the last action of the current run, only to find out that the path of the results differed widely as I started with a low precision data dump and worked my way up to double precision.
Even a long time ago, it was too late to discover chaos-theory, and I didn’t see the significance anyway.
Mr. Watts, I may have a great project for your Blog. The IPCC Models provide the evidence to shoot down the AGW Theory. The IPCC Models most likely have a very very very low R-Squared, that is why they never publish the R-Squared of the models. You can host a project to beat the IPCC Climate Model R-Squareds. All you would need to do is create a repository for valid climate data. Dr. Spencer and Christy could provide the Satellite Data, Dr. Willie Soon could provide the solar data, someone else could provide the data for El Ninos and Ninas, others could provide data for clouds, albido, etc etc. The CO2 data is readily available. Once all the data sets are collected, multivariable regressions could be run on the data to identify the most significant variables, as well as establishing the R-Squared for the Temp=f(CO2) model. Once that data is collected, and the models run, it would provide great evidence for a court case. The Climate Alarmists would have to defend why a bunch of bloggers were able to create a climate model with a much higher R-Squared than the IPCC was able to do after spending billion of dollars. The models are the key to debunking this nonsense, and your website as the ability to reach the people that are needed to pull this off.
Here is a more detailed explanation of the project.
Climate “Science” on Trial; The Criminal Case Against the Alarmists
https://co2islife.wordpress.com/2017/02/21/climate-science-on-trial-the-criminal-case-against-the-alarmists/
The notion of r^2 fails for climate models. Isn’t applicable.
Please explain.
statistics cannot analyse chaos adequately
Others have wasted huge amounts of time and money building bigger super computers to predict the weather. The results have always been disappointing.
The climate is orders of magnitude more complex than local weather. We shouldn’t fall into that particular rabbit hole.
My point isn’t to build a great climate model, it is to simply show that the existing models are pure garbage, and there are far better explanations than CO2. Once again, R-Squared will do the talking.
1 – If you wish to criticize the R-Squared values of the alarmist climate models, you will have to produce better R-Squared values for your own alternative. The minute you write even the simplest formula, you have created a model. link
2 – If you wish to be on the same page as everyone else, you might use Mean Squared Error (MSE) rather than R-Squared.
I suggest (again) that the term “chaos” be ditched, since the word has a very long history, and what is being discussed in “chaos theory” does not conform to that meaning (disorder). The Earth’s climate systems are not chaotic, but (I feel obviously) hyper-orderly. There is unpredictability involved, but so to in our bodies, and all manner of other hyper-orderly systems.
We can name this quality anything we want to (speaking universally), and the term that’s been chosen is confusing and indeed often misleading to “civilians”. It allows for scary implications that can and have, I am quite sure, been exploited by the CAGW clan, unnecessarily.
(I suggest ‘chaosh’, since it retains enough “flavor” to be easily apprehended/recognized by those familiar with the technical aspects of the theory, making a gradual transition no big whoop, and is available, according to my quick check via search engine).
Too late. For better or worse, it is now Chaos Theory. Just like Dark Matter. If it is eventually discovered to actually exist, no matter what it is composed of, it will always be known as Dark Matter. Even it it’s not truly dark. Either get over your dislike of the term, or learn to be annoyed every time it pops up. I don’t see any other recourse.
When’s the last time you heard ‘Hertzian waves’ discussed, Paul? . . The last time you heard electromagnetic radiation discussed ; )
I’m fond of Dark Energy.
Dark Energy = the amount of energy solar cells collect at night.
Chaos does not supersede the laws of nature. It just means that we puny humans can’t see the relationships. Things are unpredictable by us.
If we don’t perceive chaos, it means we are under the illusion that we understand what is going on and can predict how things will turn out. Chaos is a fine word and, in my humble opinion, it is underused.
cha·os
[ˈkāˌäs]
NOUN
complete disorder and confusion:
synonyms: disorder · disarray · disorganization · confusion · mayhem · bedlam
Sound familiar? That’s what most people think you’re talking about if you speak of the climate being chaotic . .
“If we don’t perceive chaos, it means we are under the illusion that we understand what is going on and can predict how things will turn out.”
If I flip a coin, I “don’t know what will happen” . . but I “don’t perceive chaos”. I get what the newer meaning is referring to, but see no good reason not to employ a new term for it. If you’ve got one, by all means make your case . . but; *That’s what I was taught in university* is not going to do the trick. . these days ; )
PS~
“Chaos does not supersede the laws of nature. It just means that we puny humans can’t see the relationships. Things are unpredictable by us.”
I don’t think that’s what “chaos” theory is espousing/dealing with. It’s about a “world” that is not like infinitesimal ricocheting pool balls all obeying strict/precise rules of motion, such as might be predictable if only we had more refined observational capabilities. It’s proposing that no matter how refined our observations might be, we couldn’t predict various things.
That’s why he picked the term ‘chaos’ it seems to me. So you would not think he meant that sort of ultimately predictable “mechanical” sort of precision, merely “out of our reach” in a technological sense. Rather, something more akin to the case of quantum physics “laws”, which are not describing a “world” that works like ours, merely really really small. The laws at that level, while extremely consistent (hence inviable law-like) do not resemble the “laws of nature” that we can see in operation in our day to day existence. (Or any other sort of existence we can imagine/comprehend, after more than century of many very smart folks trying to).
I find it counter productive to try to use weather models, with ever closer resolution to overcome the problem of chaos. Climate is anyway the average and variation of a lot of variables over time and space, so it might be better to just keep at the energy budget, and let the weather be what it is.
From electronic simulations i have learnt that you have to keep the circuit so simple that you know whats in it and what is most important. Else you would not know why it behaves as it does, and you would not know how to change it in a preferred direction (if possible).
That higher average temperature should give more extremes is not what i see. The summer time in Denmark and northern Europe has less extremes than the winter times. And the difference in temperature is 15K at least. Just look out your window or even worse go outside. How should a minor 2K temperature change make any difference, when we in a week kan have a change from -10C to +5C. Such a change of 15K is never seen in the summer.
You are then told that higher temperature means more potential energy in the air. Could that be the reason for the terrible weather in Antactica?
There are several types of models that are in use in climate studies. As part of my Ph.D., I tried review them. They were published [1] Climatic Classification: The semi-arid Tropics and its environment – A review”, pesq. agropec. bras., Brasilia, 18(8):823-847, ago, 1983; and [2] “Agroclimatic Classification: Numerical Taxonomic procedures – A Review”, pesq. agropec. bras., Brasilia, 18(5):435-457, maio, 1983. Also, models applied under practical application such as crop-weather models and crop-soil-weather models — published in Agricultural Meteorology and Agricultural Forest Meteorology [revised named] in 1983-85. Also, models relating solar radiation, evaporation, etc. Here, the models were tested with ground realities and tried to focus on better results. But in the global warming models, there is no such thing happened.
Dr. S. Jeevananda Reddy
The Butterfly Effect
If a butterfly flapping its wings can affect weather thousands of miles away,
(see chaos theory) how much are these giant wind turbine farms affecting the weather. Good, bad or ugly? See Law of
Unintended Consequences.
….. the major Lorenz point went unnoticed by the bloggers:
Lorenz: …..any forecasting will lead to poorer and poorer
results…[…]…ONLY THE PERIODIC COMPONENT can be
predicted in the far distant future” ….
For this reason, we only employ periodic components in
the climate patter recognition grid in
http://www.knowledgeminer.eu/climate-papers.html
The latest Holocene paper, part 6, deals with the time span 1AD to
1150 AD, the part 7 and 8 will go to the future End of the Holocene,
ONLY PERIODIC COMPONENTS are used.
True, but that still does not help you in making any useful predictions since you still can’t predict when a cycle will start or stop, let alone what the min or max will be for any particular property. We already know that the climate cycles in a quasi-periodic way, from ice age to ice age.
If it has rained on average 200 days a year in Vancouver for the past 100 years, a useful prediction is that tomorrow you should wear a waterproof jacket.
This sort of prediction is more useful than the massive weather computer telling you there is a 30% chance of measurable precipitation.
Periodic is predictable
=========
Exactly. Ocean tides are chaotic, yet we predict them with great accuracy. Not from first principles as is done in climate models. Rather by observation. When past conditions repeat, so do the tides. The same holds for climate. It is raining in California not because of CO2, but rather because it has rained there before. If it had never rained in California before, it would be foolish to predict it would rain anytime in the future. Similarly if it had rained in the past it would be bad science to predict it would not rain in the future.
Digital Moloch
“The opposite of the butterfly fallacy is closer to the truth. The limit to predictability of a nearly deterministic system is highly sensitive to the background noise. Isolation and extremely cleanliness are the lifeblood of modern science and technology. Chaotic systems, on the other hand, are already rather unpredictable. Usually the best forecast for the future state is very statistical. “Hot in the summer and cold in the winter” is the most reliable long-range weather forecast. No conceivable number of butterflies is going to change that, and the total absence of butterflies will not improve that forecast a wit.”
The Art of Modeling Dynamic Systems
Foster Morrison
p. 271
Given the 800,000 years we have approximate temp proxies for, anything fine scaled, if we could do such a zoom in, to any one spot on that data series would look scary. So once again I say we are supposed to be warm, and may get a bit warmer before we fall off the cliff. Again.
I am grateful the Judith Curry is speaking the truth about Climate Science and that she apparently has some gravitas, but she is only reiterating what has been said by countless others for the past 25-30 years.
It is said that the lotto is a tax on those who do not understand math. Similarly, climate models are only convincing for those who also do not understand math. Unfortunately, that is a large percentage of the population.
Even if we had a great understanding of how the climate worked, the models would still suck at predicting the distant future of climate, because of the inherent weakness of the mathematics. Throw in the fact that the modelers are intentionally ignoring almost all the observed natural climate variability of the past, and you get a completely useless product.
This was pointed out by some before the air conditioning was turned back on in that congressional hearing room where James Hansen trotted out his doom-and-gloom dog and pony show in June of 1988. This obvious and inconvenient truth has been summarily and aggressively ignored for nearly 30 decades.
What a load of disinformation. Exxon earns over $475 million dollars per day AND they earn hefty government subsidies. Don’t talk to the public about where tax dollars are going when those dollars go straight into the pockets of the wealthy elite.
America is the richest country in the world yet has the worst health care system, most expensive drug prices and the biggest military budget ever and you want to disregard climate change based on some belief that it is a tax scam?
Shame on you and your toxic spam.
[you should come here and try it, rather than writing slams from the comfort and safety of France -mod]
Would the US be better off with 0 Exxon’s or 1000 Exxon’s? How about if the world economy had 0 Exxon’s or 1 million Exxon’s?
For my part I believe most of us would be much better of if the world economy had 1 million companies the size of Exxon. The wealth created is what is required if we for example want to colonize our solar system. Right now no one has enough money to make it happen in a big way.
Fusion Rockets would solve the money problem.
Moronic post.
Just wait till Marine Le Pen wins the Presidential election and the EUSSR collapses.
What do you think will happen then?
Mark. While some cities (Grenoble, Paris, Lyon, Strasbourg etc) cool air about 0.01 °C by whipping the taxpaying carbon based lifeforms with anything from tax increases to commuting obstructions during Holland’s term: how has French healthcare and military evolved? It’s not like the former would be dispatching preventive cancer treatment patients directly to the cemetery? Or the latter mobilised on the streets heavily armed. Right?
mark, so you think depreciation and expense are tax subsidies and skip over the taxes Exxon pays. Exxon earns a profit annualized that is about ten times less than your claim, by the way. That is called “deception” or “lying”. Unless you are so ignorant as to think gross revenue equals earnings? And America’s health system is the worst you claim. Yes Obamacare has made things worse, but you have no evidence our health system is the worst. Because it isn’t. So you lied twice. Climate hype depends on liars like you to imbed untruth and ignorance in the public square. You are the toxic spewing twit.
How can you argue otherwise – (super) computers are the new God(s)
They are untouchable, they are always correct, they are tended by a select few in what you can only call Temples.
They dispense rules that us mortals must live and also guilt, in equal measures.
We already have (anywhere with a monarchy) representatives of God down here on Earth – it says so on every UK coin you look at. They all have the letters FD – Defender of the Faith.
King Knut worked that out, set yourself as Christian king (God’s earthly representative) so that if you upset the king, you upset God and hence, you pay the price. (No Heaven for you bonny lad)
So, what have we now, Prince Chuckles here in the UK has swallowed the climate guff, hook, line and sinker.
What other evidence if not the CO2 data from Hawaii, immutable words of great precision that we must all live by, coming down off a mountain – a mountain of fire and brimstone no less!
There was a ‘great’ civilisation somewhere very dry, a desert basically. somewhere in the Middle East I think – possibly The Himyar.
Their bureaucracy built, in their capital city, an epic water feature. This was to show how rich and powerful they were and hence could effectively squander a rare and valuable resource (water) – just because they could.
The Himyar civilisation did not last very long after its construction (I wonder why – Brexit anyone?)
Are these new supercomputers our equivalent to the Himyar, in a desert, building water fountains, pools and constant splashing water?
I watch a video lecture by Pat Frank and a related paper where he applied to climate model projections the point that Edward Lorenz made about the propagation of errors. He argues that errors are not only propagated but amplified.
Worth looking for again.
Error can be considered noise of the form 1/f^n, where n=0,1,2…
n=0 is random noise, the coin toss. It converges to give us the law of large numbers. It is the 2 body problem, a planet orbiting a single attractor, with a mean and deviation. We have mathematical solutions to these problems.
N=1,2…n gives us pink and brown noise. The noise does not converge. It is the 3,4..n body problem. There is no meaningful average, only a local average, depending on which bodies you are currently orbiting. These types of systems are beyond our capability to solve mathematically from first principles.
Judging by where the demands of climate scientist and activists alike seem to converge, the “god” at work here is the spirit of hive-mind socialism. Every solution that we must urgently impose on society involves higher taxes, bigger government, less freedom, less prosperity, less free speech and fewer human comforts. But to what end?
This is not a mystery. Consider the following quote:
“The more we come to know about the gnosis of antiquity, the more it becomes certain that modern movements of thought, such as progressivism, positivism, Hegelianism, and Marxism, are variants of Gnosticism.”
— Eric Voegelin, Science Politics and Gnosticism, Two Essays, 1968.
“This has generated a belief system…belief that all variables which drive global climate at all time scales have been identified, quantified as to individual contribution and interactions, and that chaotic variability is foreseeable. At the current state of scientific knowledge, such belief is intellectual hubris masquerading as achieved scientific endeavor.”
Nobody believes this. In fact we know the opposite to be true.
Strawman is not a good way to start an argument.
Just saying
Starting in 1896 scientists have developed models, mathematical abstractions, in order to understand and predict the earths climate.
Without exception those models have improved over time and will continue to improve. They will never be perfect; they will always be wrong. wrong in small areas, wrong over small periods of time, but increasingly less wrong.
They will, without exception, be scientifically superior to A) claims we cant know. B) claims it will never be known. C) claims that fail to represent the Green house effect.
The question is this: CAN models, however flawed, INFORM policy. You have, for example, historical data about sea level rise.. To this you can apply a statistical MODEL. The statistical model says that over the next hundred years you can expect sea level to rise 1 foot. The statistical model knows only about prior data.
It assumes the future will be like the past.. the rate of rise will continue. You also have a Physics model. Its incomplete. And it tends to be biased high . It predicts a 3 foot rise.
You are building a road by the ocean.
Do you.
A) Plan for the road to last?
B) Build it such that it is only 6 inches above current sea level and ignore the past?
C) Accept the statistical model as certain and build it at 1 foot above sea level?
D) Add an arbitrary safety factor to the statistical model?
E) Trust the Physics model and be on the safe side since it is biased high?
F) Listen to what random bloggers say?
G) Bias adjust the physics model
You cant avoid decision unless you decide that you are not building roads to last.
Who decides what information, what models, will inform the decision?
Who decides what weight to give each piece of information.?
So where do you build the road? at 6 inches above Sea level? 1 foot? 2 feet? 3 feet?
6 feet?
Suppose you are building a sea wall in Japan? how high?
Suppose you are building a dam and have to decide how to build an emergency spillway?
Steven
I realise that the ‘road’ is an analogy to demonstrate your point, but….
The rise in SLR of either 1 or 3 feet is projected/predicted over the next 100 years. There are two standards of road building – standard A has a design life of 30 years, Standard B for 100 years but costs three times as much.
So to build the road one foot above current to A will cost 1 unit of resource. To build it to B to a height of 3 feet will cost 9 units.
We only have i unit of funding available. More will require increasing taxation, either directly or to fund borrowing.
Since both projections/predictions could be found 30 years from now to be inaccurate, why would you commit the 9 units now?
Mosh @ 11:20 – “The question is this: CAN models, however flawed, INFORM policy.”
The answer is simple. Yes, they CAN inform policy and all too often are used for that very purpose. But, as comedian Chris Rock once said in one of his HBO specials, “You CAN drive a car with your feet if you want to. That don’t make it a good f#*%ing idea!” Using flawed models to try to inform policy is like basing a marriage on a lie; Like deciding whether or not you need to wear a winter coat based on whether it is sunny or cloudy and disregarding the thermometer; Like building a house on a flood plain because it hasn’t flooded for a few years so you expect it will never flood again. Or like building a road “to last” when you know the sea will soon permanently flood it no matter whether you go with historical trend or unproven theoretical model that is flawed. It’s a bad idea no matter how you slice it. Your analogy is not even apples to oranges. It’s more like apples to pixie dust.
What decision would you make in this scenario: You go to the doctor for a routine check up and he says he’s concerned about your blood pressure, although it has never been high and is not high now. But, he says that things might be different soon because you are getting older and you might decide to start smoking or change your diet and exercise routine in the near future. He wants to prescribe a drug that is extremely expensive and known to have severe side effects, but will keep your blood pressure from getting higher than it is. His recommendation is based on his model (that tends to run high) of how your blood pressure MIGHT be, however flawed that model is, rather than your “statistical model” based on your health history. Additionally, you know your doctor will receive vacations and other various gifts/incentives if he convinces you that you need this medicine to prevent the rise in blood pressure. He assures you this increase will happen sometime in the future and he’s the expert, so you need to trust him. Would you take the medication without question and be happy that you averted disaster while you suffer from the side effects that keep you constantly miserable and broke? Do you seek a second opinion somewhere else, or just dismiss him as a quack and find a new doctor who?
Note: I understand it might not be a perfect analogy, but I believe it is a lot closer than yours.
Steve, the answer of course is to let the present road fail by neglect, spend the money on something foolish like a train no one will ride or even better on windmills, and when the road fails blame it on climate change.
Steven Mosher, your assertion that climate models are improving does not seem to be supported by evidence. In fact it seems Dr. Curry and many other qualified observers are right on target and that it is fair to point out that despite many billions of dollars and many man years of work and huge computer time the models are not significantly better at all.
*slow claps* Excellent post! Always thought of it like this, and glad to see lots of my thoughts on it (and more) are written down and expanded upon. People think we can do anything, when they don’t realize the limitations of computers, at least currently. Who knows if we could do better with Quantum Computers, though I’d think it wouldn’t make much difference because of our ability to measure things to so many decimals for initial conditions.
Dr Battig,
Deum de dolo.
To remain consistent with the latin popular phase “Deus ex machina”
Deum de dolo means God from a deception.
or
Deum ex machina mendacii. God from a deceptive machine.