Is It Time To Stop The Insanity Of Wasting Time and Money On More Climate Models?

Guest Opinion: Dr. Tim Ball

Nearly every single climate model prediction, projection or whatever else they want to call them has been wrong. Weather forecasts beyond 72 hours typically deteriorate into their error bands. The UK Met Office summer forecast was wrong again. I have lost track of the number of times they were wrong. Apparently, the British Broadcasting Corporation had enough as they stopped using their services. They are not just marginally wrong. Invariably, the weather is the inverse of their forecast.

Short, medium, and long-term climate forecasts are wrong more than 50 percent of the time so that a correct one is a no better than a random event. Global and or regional forecasts are often equally incorrect. If there were a climate model that made even 60 percent accurate forecasts, everybody would use it. Since there is no single accurate climate model forecast, the IPCC resorts to averaging out their model forecasts as if, somehow, the errors would cancel each other out and the average of forecasts would be representative. Climate models and their forecasts have been unmitigated failures that would cause an automatic cessation in any other enterprise. Unless, of course, it was another government funded, fiasco. Daily weather forecasts are improved from when modern forecasting began in World War I. However, even short term climate forecasts appear no better than the Old Farmers Almanac, which appeared in 1792, using moon, sun, and other astronomical and terrestrial indicators.

I have written and often spoken about the key role of the models in creating and perpetuating the catastrophic AGW mythology. People were shocked by the leaked emails from the Climatic Research Unit (CRU), but most don’t know that the actual instructions to “hide the decline” in the tree ring portion of the hockey stick graph were in the computer code. It is one reason that people translate the Garbage In, Garbage Out (GIGO) acronym as Gospel in, Gospel Out when speaking of climate models.

I am tired of the continued pretense that climate models can produce accurate forecasts in a chaotic system. Sadly, the pretense occurs on both sides of the scientific debate. The reality is the models don’t work and can’t work for many reasons, including the most fundamental; lack of data, lack of knowledge of major mechanisms, lack of knowledge of basic physical processes, lack of ability to represent physical mechanisms like turbulence in mathematical form, and lack of computer capacity. Bob Tisdale summarized the problems in his 2013 book Climate Models Fail. It is time to stop wasting time and money and put people and computers to more important uses.

The only thing that keeps people working on the models is government funding, either at weather offices or in academia. Without this funding computer modelers would not dominate the study of climate. Without the funding, the Intergovernmental Panel on Climate Change could not exist. Many of the people involved in climate modeling were not familiar with or had no training in climatology or climate science. They were graduates of computer modeling programs looking for a challenging opportunity with large amounts of funding available and access to large computers. The atmosphere and later the oceans fit the bill. Now they put the two together to continue the fiasco. Unfortunately, it is all at massive expense to society. Those expenses include the computers and the modeling time but worse the cost of applying the failed results to global energy and environmental issues.

Let’s stop pretending and wasting money and time. Remove that funding and nobody would spend private money to work on climate forecast models.

I used to argue that there was some small value in playing with climate models in a laboratory, with only a scientific responsibility for the accuracy, feasibility, and applicability. It is clear they do not fulfill those responsibilities. Now I realize that position was wrong. When model results are used as the sole basis for government policy, there is no value. It is a massive cost and detriment to society, which is what the Intergovernmental Panel on Climate Change (IPCC) was specifically designed to do.

The IPCC has one small value. It illustrates all the problems identified in the previous comments. Laboratory-generated climate models are manipulated outside of even basic scientific rigor in government weather offices or academia, and then become the basis of public policy through the Summary for Policymakers (SPM).

Another value of the IPCC Physical Science Basis Reports is they provide a detailed listing of why models can’t and don’t work. Too bad few read or understand them. If they did, they would realize the limitations are such that they preclude any chance of success. Just a partial examination illustrates the point.


The IPCC people knew of the data limitations from the start, but it didn’t stop them building models.

In 1993, Stephen Schneider, a primary player in the anthropogenic global warming hypothesis and the use of models went beyond doubt to certainty when he said,

“Uncertainty about important feedback mechanisms is one reason why the ultimate goal of climate modeling – forecasting reliably the future of key variables such as temperature and rainfall patterns – is not realizable.”

A February 3, 1999, US National Research Council Report said,

Deficiencies in the accuracy, quality and continuity of the records place serious limitations on the confidence that can be placed in the research results.

To which Kevin Trenberth responded,

It’s very clear we do not have a climate observing system….This may come as a shock to many people who assume that we do know adequately what’s going on with the climate, but we don’t.

Two Directors of the CRU, Tom Wigley, and Phil Jones said,

Many of the uncertainties surrounding the causes of climate change will never be resolved because the necessary data are lacking.

70% of the world is oceans and there are virtually no stations. The Poles are critical in the dynamics of driving the atmosphere and creating climate yet there are virtually no stations in 15 million km2 of the Arctic Ocean or for the 14 million km2 of Antarctica. Approximately 85% of the surface has no weather data. The IPCC acknowledge the limitations by claiming a single station data are representative of conditions within a 1200km radius. Is that a valid assumption? I don’t think it is.

But it isn’t just lack of data at the surface. Actually, it is not data for the surface, but for a range of altitudes above the surface between 1.25 to 2 m and as researchers from Geiger (Climate Near the Ground) on show this is markedly different from actual surface temperatures as measured at the few microclimate stations that exist.  Arguably US surface stations are best, but Anthony Watts diligent study shows that only 7.9 percent of them accurate to less than 1°C. (Figure 1) To put that in perspective, in the 2001 IPCC Report Jones claimed a 0.6°C increase over 120 years was beyond a natural increase. That also underscores the fact that most of the instrumental record temperatures were measured to 0.5°C.


Figure 1

Other basic data, including precipitation, barometric pressure, wind speed, and direction are worse than the temperature data. For example, in Africa there are only 1152 weather watch stations, which are one-eighth the World Meteorological Organization (WMO) recommended minimum density. As I noted in an earlier paper, lack of data for all phases of water alone guarantees the failure of IPCC projections.

The models attempt to simulate a three-dimensional atmosphere, but there is virtually no data above the surface. The modelers think we are foolish enough to believe the argument that more layers in the model will solve the problem, but it doesn’t matter if you have no data.

Major Mechanisms

During my career as a climatologist, several mechanism of weather and climate were either discovered or measured, supposedly with sufficient accuracy for application in a model. These include, El Nino/La Nina (ENSO), the Pacific Decadal Oscillation (PDO), the Atlantic Multidecadal Oscillation (AMO), the Antarctic Oscillation (AAO), the North Atlantic Oscillation (NAO), Dansgaard-Oeschger Oscillation (D-O), Madden-Julian Oscillation (MJO), Indian Ocean Dipole (IOD), among others.

Despite this, we are still unclear about the mechanisms associated with the Hadley Cell and the Inter-tropical Convergence Zone (ITCZ), which are essentially the entire tropical climate mechanisms. The Milankovitch Effect remains controversial and is not included in IPCC models. The Cosmic Theory appears to provide an answer to the relationship between sunspots, global temperature, and precipitation but is similarly ignored by the IPCC. They do not deal with the Monsoon mechanism well as they note,

In short, most AOGCMs do not simulate the spatial or intra-seasonal variation of monsoon precipitation accurately.

There is very limited knowledge of the major oceanic circulations at the surface and in the depths. There are virtually no measures of the volumes of heat transferred or how they change over time, including measures of geothermal heat.

Physical Mechanisms.

The IPCC acknowledge that,

“In climate research and modeling, we should recognize that we are dealing with a coupled non-linear chaotic system, and therefore that the long-term prediction of future climate states is not possible.”

That comment is sufficient to argue for cessation of the waste of time and money. Add the second and related problem identified by Essex and McKitrick in Taken By Storm and it is confirmed.

Climate research is anything but a routine application of classical theories like fluid mechanics, even though some may be tempted to think it is. It has to be regarded in the “exotic’ category of scientific problems in part because we are trying to look for scientifically meaningful structure that no one can see or has ever seen, and may not even exist.

In this regard it is crucial to bear in mind that there is no experimental set up for global climate, so all we really have are those first principles. You can take all the measurements you want today, fill terabytes of disk space if you want, but that does not serve as an experimental apparatus. Engineering apparatus can be controlled, and those running them can make measurements of known variables over a range of controlled physically relevant conditions. In contrast, we have only today’s climate to sample directly, provided we are clever enough to even know how to average middle realm data in a physically meaningful way to represent climate. In short, global climate is not treatable by any conventional means.

Computer capacity

Modelers claim computers are getting better, and all they need are bigger, faster computers. It can’t make any difference, but they continue to waste money. In 2012, Cray introduced the promotionally named Gaea supercomputer (Figure 2). It has a 1.1 petaflops capacity. FLOPS means Floating-Point Operations per Second, and peta is 1016 (or a thousand) million floating-point operations per second. Jagadish Shukla says the challenge is

We must be able to run climate models at the same resolution as weather prediction models, which may have horizontal resolutions of 3-5 km within the next 5 years. This will require computers with peak capability of about 100 petaflops

Regardless of the computer capacity it is meaningless without data for the model.


Figure 2: Cray’s Gaea Computer with the environmental image.

Failed Forecasts, (Predictions, Projections)

Figure 3 shows the IPCC failed forecast. They call them projections, but the public believes they are forecasts. Either way, they are consistently wrong. Notice the labels added to Hayden’s graph taken from the Summary for Policymakers. As the error range increase in the actual data the Summary claims it is improving. One of the computer models used for the IPCC forecast belongs to Environment Canada. Their forecasts are the worst of all of those averaged results used by the IPCC (Figure 4).


Figure 3


Figure 4 Source; Ken Gregory

The Canadian disaster is not surprising as their one-year forecast assessment indicates. They make a one –year forecast and provide a map indicating the percentage of accuracy against the average for the period 1981-2010 (Figure 5).


Figure 5

The Canadian average accuracy percentage is shown in the bottom left as 41.5 percent. That is the best they can achieve after some thirty years of developing the models. Other countries results are no better.

In a New Scientist report Tim Palmer, a leading climate modeller at the European Centre for Medium-Range Weather Forecasts in Reading England said:

I don’t want to undermine the IPCC, but the forecasts, especially for regional climate change, are immensely uncertain.

The Cost

Joanne Nova has done most research on the cost of climate research to the US government.

In total, over the last 20 years, by the end of fiscal year 2009, the US government will have poured in $32 billion for climate research—and another $36 billion for development of climate-related technologies. These are actual dollars, obtained from government reports, and not adjusted for inflation. It does not include funding from other governments. The real total can only grow.

There is no doubt that number grew, and the world total is likely double the US amount as this commentator claims.

However, at least I can add a reliable half-billion pounds to Joanne Nova’s $79 billion – plus we know already that the EU Framework 7 programme includes €1.9 billion on direct climate change research. Framework 6 runs to €769 million. If we take all the Annex 1 countries, the sum expended must be well over $100 billion.

These are just the computer modeling costs. The economic and social costs are much higher and virtually impossible to calculate. As Paul Driessen explains

As with its polar counterparts, 90% of the titanic climate funding iceberg is invisible to most citizens, businessmen and politicians.

It’s no wonder Larry Bell can say,

The U.S. Government Accounting Office (GAO) can’t figure out what benefits taxpayers are getting from the many billions of dollars spent each year on policies that are purportedly aimed at addressing climate change.

If it is impossible for a supposedly sophisticated agency like US GAO to determine the costs, then there is no hope for a global assessment. There is little doubt the direct cost is measured in trillions of dollars. That does not include the lost opportunities for development and lives continuing in poverty. All this because of the falsified results from completely failed computer model prediction, projections or whatever they want to call them.

It is time to stop the insanity, which in climate science is the repetition of creating computer models that don’t and can’t work? I think so.

“Those who have knowledge don’t predict. Those who do predict don’t have knowledge.” Tzu, Lao (6th Century BC)

Note: this article was updated shortly after publication to fix a text formatting error.


newest oldest most voted
Notify of

They can spend as much money as they want on climate models but the bottom line is they will never work because they do not put in correct, and complete data.
Let them keep wasting and knocking themselves out so when the day of reckoning comes they fall that much harder. That day of reckoning probably before this decade ends.

Mickey Reno

I’d stop them from wasting, immediately, if I had my druthers. And why do you suppose that if you allow them to continue to waste you’re money, that things will collapse? And that collapse is a manageable event, that the collapse would lead to a new direction you’d favor? What if they collapse in a direction that the Cloward’s and Piven’s of this world envision?
Let’s send Kevin and Gavin to the unemployment line, now. It galls me to think of the pensions they’re going to draw for a lifetime of “service” that helped no one.

Mickey Reno

your money – not you’re money. [blush]

Are you advocating that a person in a position of authority inform the climate modeler’s:




Stormy… I see what you did there. ^_^


Unfortunately, El Nino will most likely drive up temps in the short term and lessen skeptics ability to influence anything. My guess is more money will be put into climate models, not less.


They will never work because of the Lyapunov exponents. End of story.


Exactly right. Let them spend and model to their heart’s content. We will have the real world to compare and contrast with their imaginary pink unicorns.


I predict ( with a 99.9% accuracy ) that the Glo.Bull Warming hoax will come to a crashing halt in 2017 !!!!


As I understand it from Dr. Christopher Essex – University of Western Ontario – climate models can never be correct because of: 1) calculation residual, 2) machine epsilon, and 3) model parameterization.


If you have an hour this is a must!! Clear and concise in what is a very complex discussion!


4) incorrect assumptions (assuming the Earth is like an onion – ie “radiative forcing”)
5) incomplete physics (they can’t even predict the Sahara Desert, they need to program it in instead)
6) a need to force them to fit existing (to 1978) data, instead of making genuine predictions

Steve Case

The day of reckoning by 2020?
I wouldn’t bet on that one. They are getting away with it and they know it. The current game plan seems to be that every weather calamity is blamed on Climate Change. The longer they keep that up, the more believable it becomes. As personal employment becomes more and more dependent on combating Climate Change, the more intrenched it will become. Any day of reckoning will only occur when the economy will no longer support the scheme.

Steve Case

I forgot to point out that a week or so ago there was a post about the Climate Change industry being a $1.5 Trillion business world-wide.
With that kind of money driving it, I wouldn’t bet on a quick and decisive demise any time soon.

george e. smith

Right on Dr. Tim !!
And add this to your list of ” well we just took a wild a*** guess because we can’t possibly get enough meaningful samples (from the ground), so we just made it up and some of it is something like something that happens sometimes. ”
…. … for the full version of historian emeritus Spencer Weart’s concise history of climate science. He wrote a book called “The discovery of global warming” or something close to that” so he pushes the gospel to sell his book.
From what you are saying Tim; this isn’t even a good approximation for throwing darts at a wall to see if anything sticks.
PS Weart does have an impressive resume in Physics. I’m not in any way suggesting his credentials are unsound; quite the contrary.
But you will find the word ” consensus ” in his essay, more times than I care to count. And mostly it is about the hidden inner machinations of committees, rather than scientific evidence supported by peer reviewed experimental data.
It certainly adds to the pile of shameful substitute for scientific rigor.

george e. smith

I noticed that fig. 3 is a Dr Roy Spencergraph, with subtitles.
I did like the fact that the further they go back in the past on their wayback machine, they get less and less confident, that they even know what happened back then.
Well that’s in line with Spencer Weart’s assertion that they basically took a guess.
I guess this whole line of research is fully in keeping with Lord Rutherford’s admonition:
” If you have to use statistics, you should have done a better experiment. ”
Or in this case, made better wild arse guesses.

george e. smith

I always thought that statistics could only tell you what fraction of some large number of experiments would meet some criterion.
And most of the math is only valid if the variables show a normal (izzat Gaussian ??) distribution.
As far as I know any single experiment could come up with ANY physically possible result; which cannot be predicted.
Hey news flash, earth to scientists : This is not a dress rehearsal; we are only going to get ONE SHOT at the future; there will not be any instant replay , and no refunds will be issued if you don’t like the outcome.
So stop wasting our time with statistics. Observe and report what happens; and don’t do any 13 month running averages, or nine point filters, or any other information destroying filtering.
Read the meter, and write down the reading !!


Re: “they do not put in correct, and complete data”.
But, but, but…those are the official global temperatures as measured by real professional temperature measuring sailors in the open ocean, using real buckets on ropes. As then assessed, interpreted and suitably adjusted by the very trustworthy Phil Jones et al in the bucket modelling dept at East Angular.
Then readjusted later for various spurious reasons.
It is fortunate for the individual sailor who performed this bizarre duty, the he was never told that the output of his diligently (or not so diligently) performed duty would one day form the basis for an attempt to exactly assess the average temperature of the globe during the first half of the 20th century to within ONE TENTH OF A DEGREE.
Not that such an onerous responsibility would necessarily have bothered the sailor in question.
But, simply because he would have become, in that moment, aware that his own descendants were destined to be a bunch of complete idiots.
I maintain some temperature recording sheets as a part of my job. Since we do not regard the temperature recording to be of any great significance, the numbers recorded are mostly fictional. It is easier to concoct a realistic looking number and write that down, than to actually go to the trouble to reading the usually faulty thermometer.
Occasionally, for my own amusement I have attempted to check the accuracy of the thermometers by laying two of them side by side. When I have done this I have usually discovered a very significant discrepancy.
Sometimes I attempt to point this out. But nobody is at all interested.
I expect that a sailor in the Southern Ocean was beset by a similar dilemma. Whether to diligently record the real ocean temperature or whether to just guess something and write it in the log – thereby saving himself the bother of slinging a bucket overboard and hauling it back onto the deck etc.
The poor man never imagined that the future of the human race lay in his hands.

Brett Keane

Well put! Having done that job, I doubt the samples would be too far wrong, on average. Water doesn’t change T quickly, and the drying outside of bucket will supply slight evaporative cooling, But, ahem, much depends on Officer and Petty Officer vigilance. Not much reason for crewmen to care much, otherwise. Yes, real malevolent stupidity is left for the present day climate clowns like Karl and Coy.


Salvatore, they will never work because they keep adjusting the past data…
BTW…modeling climate should be magnitudes simpler than modeling weather

It is not just the wrong data, but the wrong process coded into the programming as well. Crap assumptions in crap programs will just give more crap faster on bigger computers.


If it were play money, I would agree.
But that kind of cash could do real good. Total funding for novel fusion reactors in the US this year is less than $100M. For a technology that could bring virtually free energy to everyone and make space exploration practical, this is a travesty.

James Francisco

Maybe they can use the climate modeling computers to figure out the total amount of money that was wasted on this fiasco.

george e. smith

Tim poses a lot of arguments as to the sparcity of the ” surface / near surface / wherever data.”
No stations for 85% of the earth surface; few stations in the Arctic.
I seem to recall from some years back when I started following the climate horror story at Tech Central Station, reading an account, that said that around the turn of the century; which at that time meant 19 going on 30, there was something like 76 “weather stations ” in the Arctic, AKA > +60 deg. lat , but that number had grown to a bit less than 100, and stations had moved; but with the collapse and implosion of the Soviet Union, the number of Arctic weather stations had dropped to about 12 today.
Those numbers may not be the right ones, but it was that order of a progression that I recall.
But it is actually much worse than Dr. Ball suggests; because most of that ” data ” is not data at all, but simply noise.
The assumption of sampled data theory, is that a multivariable continuous but band limited (in all variables) function, can be represented by a set of ” snapshots ” of the state of the function; the contents of that snapshot being the value of each variable at the point of recording the snapshot.
Now time is not necessarily a variable in all sampled data systems; but it certainly is one of the most common variables.
In the case of global weather which integrates over time to climate, the independent variables are essentially time and position.
So a ” datum ” of this two independent variable continuous function consists of the value of the dependent variable ( maybe Temperature ) at a specific instant of time for ALL sample values of the spatial variable (measuring locations), and also for each measuring station, is needed a reading for each value of the time variable.
That means you must get simultaneous readings for all sampled locations all at the same time, and for all stations their readings all need to be made at the same set of time intervals.
If you think about the ” weather ” in your back yard, and the weather over at your uncle’s house at the same instant, it is differences in the weather variables such as Temperature, that will determine from the laws of physics, just what energy exchanges are going on between those locations. Is the wind blowing from your house over to his, or verse vicea.
If you don’t both look simultaneously, it is meaningless. So your uncle got rained on last week , but today you got some sprinkles. Can’t tell a damn thing about what is happening from that. Your house may have burned down since last week, so rain today isn’t going to help you.
A photograph is a map that shows you everything that is in a certain physical space all at the same moment. A video will show you a sequence of photographs each showing everything that is wherever, all at the same instant when the frame was recorded.
So this hodge podge of each station recording a Temperature whenever they feel like it, so long as the get two numbers some time each day, is simply garbage it isn’t a video of anything. It’s more like the neighborhood garbage dump with all kinds of bric a brac at a land fill, with nobody knowing where all the junk came from or when it was dumped at the land fill.
Prof John Christy reported in Jan 2001 the results of some simultaneous ocean water and ocean near surface air temperatures ( from about -1 m and + 3 m respectively ) over about 20 years from some ocean buoys, and found that water and air Temperatures aren’t the same, and they aren’t correlated so you can’t regenerate one from the other.
So prior to about 1980, none of the ocean water based temperature numbers are of any use for plotting on a global lower troposphere map; it’s just garbage at a land fill.
So it is time that all these otherwise unemployable statisticians, started boning up on the general theory of sampled data systems, and then try to comply with those rules.

Alan the Brit

The problem is this, because they are in the public employ, paid by taxpayers generous funding, they are answerable to no one, therefore are responisble to no one, therefore they carry on as they do. Until the money is withdrawn, they continue. The system is geared up so that they will be allowed to take early retirement, on great pensions, or allowed to move sideways into less prominent positions, cozy, cushy, stress free, with the only repost available to them, “We did what we thought was right, based on the best available science!”. The really big lie!

It doesn’t matter that the data is sparse and inaccurate, because even if the data was perfect, the computer would convert the numbers into an approximation called a floating point arithmetic and once the numerical representation is corrupted in the most minute way, you fly headlong into chaos.


It doesn’t matter what data they put in, the models will always fail.
Anyone who claims that an effectively infinitely large open-ended non-linear feedback-driven (where we don’t know all the feedbacks, and even the ones we do know, we are unsure of the signs of some critical ones) chaotic system – hence subject to inter alia extreme sensitivity to initial conditions – is capable of making meaningful predictions over any significant time period is either a charlatan or a computer salesman.
Ironically, the first person to point this out was Edward Lorenz – a climate scientist.
You can add as much computing power as you like, the result is purely to produce the wrong answer faster.

Reblogged this on WeatherAction News and commented:
When model results are used as the sole basis for government policy, there is no value. It is a massive cost and detriment to society


The Great Wall of China was a similarly scaled failure of trust and public policy.

Bill 2

Dr. Ball, why is your twitter account a spambot?

My wife set it up and operates it. Thanks for advising of the problem, we will get it corrected. Thanks

We use Apple and Spambot is not supposed to be a problem, although some are now reporting the problem. Will continue to resolve the problem.


“Remove that funding and nobody would spend private money to work on climate forecast models.”
We should follow the Human Embryonic Stem Cell model for funding this. When the federal government reinforced its moratorium, California stepped forward with $30 billion to fund the research, saying it would make California the world leader for research and innovation in the area.
California is already vowing to spend billions and inflict billions more dollars of damage to their own economy to support and promote the global warming ideology. Let them pick up the banner and devote a few billion to modeling.
As a note about the whole HESC issue, that work is now all scientifically obsolete due to the development of IPS cells that have no ethical concerns and much more clinical promise.


There was never a federal moratorium on Human Embryonic Stem Cell research.
There was a ban on federal funding for research using any embryonic stem cell lines created after a certain date. Funding on stem cell lines created prior to that date were always allowed.

The IPCC Figure 4 (the first graph in this post) has a problem in that new measurements are being obscured by the (Observations) dialogue box … this setup will remain in effect until ~2035.

Crispin in Waterloo

The Canadian model living virtually on Vancouver Island is an embarrassment to science and industry. Its predictions are laughable, sophomoric. It is ‘kept’ to keep up the IPCC’s ‘average’ predictions. We are paying for this junk science. Eliminating (defunding) the worst climate model each year would bring a sense of rigour to the forecasting industry and a sensitivity number below 1 degree per doubling of CO2.
If any P.Eng ran their operations like that they would be defrocked. Instead, the fabricators of junk climate science are canonized.


I think the issue with climate models is that they are tuned to hindcast the past (from 2005 IIRC for the CMIP5 models) as best as they can, with the assumption that none of the rapid warming from the early 1970s to 2005 or so is from any natural cycles other than changes in volcanic emissions. If a contribution to that warming period by multidecadal cycles is determined and accounted for, and the models retuned accordingly, then I think they will become accurate.

Regarding: “It has a 1.1 petaflops capacity. FLOPS means Floating-Point Operations per Second, and peta is 1016 (or a thousand) million floating-point operations per second.” A petaflop is 10^15 floating point operations per second.


I might be wrong, but that is not capacity. It is speed or potency.

george e. smith

Peta is a thousand trillion; that is with a T not a thousand million with an m or even with a B.
Why did I always think that the climate models started with the known laws of physics; including the supposition (which becomes obvious every morning, when the sun rises in the east) that the earth rotates on its axis every 24 hours or so ??
just asking !


I’m pretty sure it’s 1024 million flops, not 1016. 2 to the 10th power is 1024.

it’s 1*10^15 flops, your pc is probably hitting 1024 million flops, 10^9 flops, 4 cores doing 4 flops/ cycle at 2.4Ghz.. I believe that it’s fairly straight forward to download a GCM, compile it and run one on your PC, it would just run really slow.

Sam Prather

This is an excellent article!! One minor correction: a petaflop is a quadrillion (thousand trillion) floating point operations per second (FLOPS) which is a thousand teraflops, or 10 to the 15th power FLOPS.


When the 8080 first came out, it’s clock speed was around 100KHz, at it took several seconds to do a single FLOP.


(When the 8080 first came out, it’s clock speed was around 100KHz, at it took several seconds to do a single FLOP.)
Mark, the 8080 was the successor to the 8008. it ran at 2 Mhz clock speed and performed close to 300,000 flops per second.Microprocessor architecture and bus width were different in those early chips and it is difficult to compare their operation with today’s designs.

Leo Smith

Neutronman. An 8080 had no native floating point abilities whatsoever. You had to code them yourself, and a floating point divide was a brute of a thing.


Not sure why I can’t reply to neutronman2014, but you are both about 3 orders of magnitude off, but in opposite directions. The 8080 did not have any floating point instructions, so required use of a software library. A single-precision add or subtract took about 0.7 milliseconds, multiply was about 1.5 msec, and divide took about 3.6msec. (that last one would be nearly 278 FLOPS!).
Times are taken from the Intel 8080/8085 Floating Point Arithmetic Library User’s Manual, Appendix C.


Nuetronman2014, the 8008 was the successor to the 4004, which was built to run a 4 function calculator.


Leo, the 8080 had the ability to do add and subtract directly. There was no native multiply or divide op-code, you had to build up those functions from the add and subtract functions.

george e. smith

Where do people come up with these numbers. I seem to recall that the original IBM PC had a microprocessor with a 4.7 MHz clock frequency.
Before there was an 8080 or an 8008 or a 4004, there were already calculators that could do floating point math; even the cordic algorithm of the HP 35 hand held calculator.
I used to use a Wang calculator, that did multiplication using logs (1965). So it had all the math functions that were on the HP-35 in a weird desktop package. It had a weird magnetic core ROM made by stringing a bunch of wires through (or not through) a stack of ferrite cores. The sequence of cores that a wire went through or bypassed determined the 1-0 pattern of the word stored on that wire.
I never was able to determine whether Wang had figured out the cordic algorithm, or not because I never could find any literature that described that process, until it appeared in the HP-35 (and in the HP-Journal.


The first IBM PC used the 8086, two generations past the 8080.
The 4004 was the first chip designed specifically for a calculator, and it was a simple 4 function calculator with no memory.
The HP-35 was introduced in 1980, which is about 5 years AFTER the 8080 came out.


Altho Dr. Ball was too polite to say it succinctly, more peta flops means “garbage in, garbage out faster”.

F. Ross

Excellent article Dr. Ball.


The $ value of being able to predict the growing season weather a few months in advance is an astronomical number.


No comment.

The Key Role of Heavy Precipitation Events in Climate Model Disagreements of Future Annual Precipitation Changes in California
Climate model simulations disagree on whether future precipitation will increase or decrease over California, which has impeded efforts to anticipate and adapt to human-induced climate change……..Between these conflicting tendencies, 12 projections show drier annual conditions by the 2060s and 13 show wetter. These results are obtained from 16 global general circulation models downscaled with different combinations of dynamical methods…

Fred Singer: “Successive IPCC summaries have claimed increasing certainty [from 50% in 1996, rising to >95% in 2013] about a human cause of global warming — even as the disparity between observations and IPCC models continues to grow year by year –now for more than 18 years. This is becoming somewhat ridiculous…


Well obviously that is why the climate models are 100% correct. Since they forecast all possibilities they can never be wrong.
In other words I bet any climate modeler $100 that a independently supervised coin flip will be either heads or tails.


Some people may not take what you said seriously, but it is a truism. Everytime we get a result that contradics the narrative they say ‘but the models predicted it.’ Well of course they did!

Well, it’s one thing to argue that studying the climate through models should be discontinued because it hasn’t produced verifiable predictions yet (or that the predictions the models have produced have failed verification). Problem is, as soon as you start torturing your reasoning by conflating climate modelling and weather forecasts, you’ve lost all credibility for the rest of your argument.

No you haven’t


bregmata, you beg the question.
Climate can only be more predictable than the weather it is composed of if, and only if, an understandably small number of factors dominate the weather over the long-term.
That’s never happened in the past. So why believe it will happen now? We’ve had forest fires and volcanos and CO2 hasn’t dominated.
You need to justify why you believe climate will be simpler than the weather that happens.

Clime forecasting vs. weather forecasting: Weather forecasting is like giving a microsecond-by-microsecond forecast of the states of the switching transistors in a class D amplifier. Climate forecasting is like predicting the duty cycle of the switching transistors as a function of input voltage and component changes. This means that a 50-year climate forecast should be much easier than a one-month day-by-day weather forecast. The biggest problem I see now with climate forecasting (such as with a composite of many models) is tuning the models to hindcast the past, with an incorrect understanding of the amounts of past temperature change caused by each of several factors, such as changes of manmade aerosols, volcanic emissions, greenhouse gases, any cloud effects from changes of solar activity such as indirectly through change of cosmic rays, other effects from changes of solar activity, and multidecadal oscillations. Using an incorrect understanding of how much past warming was due to increase of greenhouse gases leads to incorrect determinations of at least some of the feedbacks. Another issue is incorrect consideration (or none at all) of the feedback magnitudes changing with temperature and/or greenhouse gas presence.

Peter Sable

The biggest problem I see now with climate forecasting (such as with a composite of many models) is tuning the models to hindcast the past, with an incorrect understanding of the amounts of past temperature change caused by each of several factors, such as changes of manmade aerosols, volcanic emissions, greenhouse gases, any cloud effects from changes of solar activity such as indirectly through change of cosmic rays, other effects from changes of solar activity, and multidecadal oscillations.

With four parameters I can fit an elephant, and with five I can make him wiggle his trunk. – John Von Neumann.
You have substantially more than 5 variables on your list. None of which we can measure very accurately.
From an information theory viewpoint the entire idea of climate modeling is doomed from the start…

I can not predict where an individual bird will fly. I can not even predict the exact location where a skein of geese will fly on a given day for each of the next 5 days. I can, however, predict the pattern that the geese will fly south each fall and return each sprint.
I’m not saying computerized climate models are right (or wrong). I’m saying that to argue they’re wrong because some other unrelated computerized models do not give completely accurate predictions is just a straw man, and a very poor one that in this case only taps into people’s deep-seated ignorance of the fundamental concepts involved (ie. confounding climate and weather).
One certainly does not do rational argument any favours by demonstrating one’s preference for avoiding it.


I’m not saying computerized climate models are right (or wrong). I’m saying that to argue they’re wrong because some other unrelated computerized models do not give completely accurate predictions is just a straw man, and a very poor one that in this case only taps into people’s deep-seated ignorance of the fundamental concepts involved (ie. confounding climate and weather).

But, consider the following. EVERY YEAR since 1988, 23 of 23 “geese flying models” have predicted the geese will fly southeast towards the Azores Islands and Bermuda Islands, winter there over the cold months, then fly back to the north. And, every fall, the geese actually fly south-southwest towards the TX and Mississippi and Louisiana gulf coast swamps.
Now, should we discount those “geese flying models” models just because they are “consistently” and “almost right” and spend 92 billion dollars to bring our guns and cameras to the Azores to look for geese? the geese did, after all, on average “fly south.”


Weather models can ignore most of the problems that break the climate models.


Not that tired old line again? Sheesh, can’t you guys come up with new material?


Global climate models disagree among themselves by 600% and have not improved in 27 years of gargantuan not to mention incredibly expensive efforts. Is this not prima facie evidence that further progress might not be expected.
Dr Ball details the reasons why.


Yes we know the story, weather is not climate and climate is not weather, except when an individual catastrophic weather event like Sandy is used as proof of climate change.
Can play the same game with forecasts by calling it trending, projections, forecasts, future cast anomaly patterns or whatever. Doesn’t matter what you call it or the context, a prediction is a prediction that can be proven accurate or not and in the case of climate models they are consistently wrong.
How they are consistently wrong is itself an issue. That climate models have wildly varying results but yet deliver a consistent warming result only proves built-in biases. If there were no biases the trends would be wrong both above and below reality.
Regardless, even if modelers got lucky and their predictions or projections actually occurred they are still highly suspect due to lack of understanding of what they are modeling and the uncertainty and scarcity of global data.

Lauren R.

[q]peta is 1016 (or a thousand) million floating-point operations per second[/q}
Actually a petaflop is a million billion floating-point operations per second, not a “thousand million” (which is a billion). The Gaea supercomputer that NOAA uses can do 1.1 petaflops, which is 1.1 million times faster than a “thousand million” (or billion).
1,000,000,000 is a billion (“giga”)
1,000,000,000,000 is a trillion (“tera”) or 1,000 billion
1,000,000,000,000,000 is a quadrillion (“peta”) or 1,000,000 (a million) billion
1,000,000,000,000,000,000 is a quintillion (“exa”) or 1,000,000,000 (a billion) billion


1024, not 1016


[Fake email address. ~mod.]


What about this article?
Can you find a fault with this?
Or can we assume that he is now so expert you can’t find any fault with his current work?
Friendly advice. The science isn’t Dr Ball’s weakness. Attack his political understanding if you want to have a go.


[Fake email address. ~mod.]

Stephen Richards

S bloody what. What is your point? How does it refer to this post?


Magma, forget Tim Ball and his expertise. How accurate have the IPCC surface temperature projections been? FAIL. That is the only qualification one needs, the ability to see projection V observations.
Here is someone with extensive model training and teaching. He is even more stinging than Ball.
Here is a friend.

Abstract – 1994
Naomi Oreskes et al
Verification, validation, and confirmation of numerical models in the earth sciences
Verification and validation of numerical models of natural systems is impossible. This is because natural systems are never closed and because model results are always non-unique. Models can be confirmed by the demonstration of agreement between observation and prediction, but confirmation is inherently partial. Complete confirmation is logically precluded by the fallacy of affirming the consequent and by incomplete access to natural phenomena. Models can only be evaluated in relative terms, and their predictive value is always open to question. The primary value of models is heuristic…….
In some cases, the predictions generated by these models are considered as a basis for public policy decisions: Global circulation models are being used to predict the behavior of the Earth’s climate in response to increased CO2 concentrations;…….
Finally, we must admit that
a model may confirm our biases and support incorrect intuitions. Therefore, models are most useful when they are used to challenge existing formulations, rather than to validate or verify them. Any scientist who is asked to use a model to verify or validate a predetermined result should be suspicious.

4 eyes

Attack and dispute the article, not the author. I used to think AGW was something to worry about but about 15 years ago it became apparent the science was not rigorous. And things haven’t improved despite all the money spent. If it was your money wouldn’t you occasionally audit the progress being made. And if no progress had been made I am dead sure you would consider withdrawing support.

Svend Ferdinandsen

The problem with lack of data is made worse by the fact, that the data available is constant adjusted to show warming. If you base a climate theory on fabricated warming, you get a useless theory.
Most papers try to corrolate the warming with observations in nature, but when there are hardly any warming the results point in all directions. Try to corrolate anything with a flat line that just wiggles a bit.


This aspect has been bothering me for years. The fact that the input data are constantly being revised means previous work is in need of review. I don’t see how you can take climate science seriously if the core data are in question. Another thing I find hilarious, is the desire to model ever smaller cells in the models, despite the temperature data being tortured to represent areas orders of magnitude larger than those modelled. I mean if you want to make pretty pictures, just run a simple Mandelbrot program. Essentially that is all these multi-billion dollar projects are… an expensive way of making abstract graphics that have no practical use whatsoever, unless you count their use in pushing policy agendas.

Reblogged this on The Arts Mechanical and commented:
Using computers to model anything chaotic is a crap shot at best. A computer model is at best a guess on boundary conditions and chaotic turbulent systems don’t behave in a linear boundary contained fashion.


The increasing confidence noted in Figure 3 indicates that by 2020 they should be ‘Absolutely Sure’. And once AS is achieved much bigger budgets will be needed as it becomes harder to become even more sure.
Being able to accurately forecast the growing season would be of great $ value if it can be done. If not, not.


“Short, medium, and long-term climate forecasts are wrong more than 50 percent of the time so that a correct one is a no better than a random event.”
With all due respect, Dr. Ball, a short-term climate forecast seems an oxymoron.

Berényi Péter

Computational general circulation climate models are clearly nothing but a waste of time and money, but they are actually worse than that. They are diverting attention and resources from fundamental scientific questions and attract the wrong kind of minds to the field, who are unable to do experimental work to solve basic riddles; whose the lack of understanding prohibits any further progress.
We have a fairly comprehensive understanding of reproducible non equilibrium thermodynamic systems. Unfortunately the terrestrial climate system does not belong to this class. It is not reproducible in the thermodynamic sense, i.e. microstates belonging to the same macrostate can evolve to different macrostates in a short time. Of course, it is nothing, but the other side of the coin called “chaos”, a.k.a. “butterfly effect”.
Such systems are not understood at all, their Jaynes entropy can’t even be defined.
If the climate system were reproducible, it would lend itself to the Maximum Entropy Production Principle. However, it clearly does not.
As energy exchange between the climate system and its cosmic environment goes exclusively by electromagnetic radiation, one hardly has to do more than to count incoming vs. outgoing photons, with a small correction for the different angular distribution of their momenta and deviations of their spectra from that of pure thermal radiation, to calculate net entropy production of the system. Turns out it would be very easy to increase entropy production by making Earth darker, that is, by decreasing its albedo.
However, Earth’s albedo is what it is, the planet is very far from being pitch black. In spite of this, albedo is strictly regulated, only its sweet spot is different. Due to a peculiar property of Keplerian orbits, annual average insolation of the two hemispheres is the same. It is a curious fact, that annual average reflected shortwave radiation is also the same though, which can’t be explained by simple astronomy. It is an emergent property of the climate system, for clear sky albedo of the Southern Hemisphere is much lower due to prevalence of oceans there, still, its all sky albedo is the same. The difference is, of course, due to clouds.
This simple symmetry property is neither understood nor replicated by computational general circulation climate models.
Under these circumstances any rational scientist would leave the climate system alone for a while, go back to the lab and study the behavior of non reproducible (chaotic) non equilibrium thermodynamic systems in general, until a breakthrough is achieved.
Plenty of such systems would fit into a lab happily, the terrestrial climate system being an exception in this respect.

BP, Feynman did that for a significant portion of his career (middle third). Lectures on Physics, V2, Chapter 41, delightfully titled the Flow of Wet Water. (Insider joke, chapter 40 is the Flow of Dry Water, only omitting a simple small ‘detail’–viscosity (see paragraph 1 of V2:41 for confirmation). The last paragraph of this very famous chapter is also known as his sermon on the Mysteries of Mathematics. As famous in some circles as the Sermon on the Mount. We await the next Feynman.
And, both V2 chapters should be required reading for every climate modeler. To their shame.
Just a little historical context. None of this is new. Lectures is Caltech Physics 101, 1961-1962. (and of course, much more…Feynman’s then statement about everything he understood about everything then existing in physics). Pity the poor CalTech freshman and sophomores then. It is said that by the second year, only genius undergrads, grad students, and post docs were following along. The other CalTech physics professors were too terrified to show up. Lectures is also the origins of his aphorism, ‘if you cannot explain it simply, then you do not understand it yourself’.


It doesn’t matter how much money they sink into their high tech crystal ball, it still just shows what the developers want to see.


The IPCC AR5 report itself has graphics showing the divergence of climate models from measurements.
From the AR5 Technical Summary, pg. 64:
And detail of figure 1-4:
And here:comment image
When a climate alarmists tells you the climate models are accurate, point them to these 3 graphics from the IPCC’s own report. A picture is worth a thousand words.


What do you mean, Dr. Ball?
Adjusted hind casting and raw-data manipulation show the models are working great!
Everyone knows empirical evidence must be adjusted to match CAGW hypothetical projections, because the empirical evidence is obviously off, because 97% of scientists agree the models are correct.
How can you fault that logic?
Always remember that, War is peace, freedom is slavery and ignorance is strength, because the consensus says so.
Again, how can you fault that impeccable logic?

David in Cal

“The Poles are critical in the dynamics of driving the atmosphere and creating climate”
Doubly true, as Poland often describes EU climate policy as “costly, overambitious, unrealistic in terms of targets, and disproportionally burdensome for the Polish economy.”


And the Pole guy living at the south is increasing its ice extension.


I prefer the guy living at the north pole.

Gary Pearse

Actually I think the models could do better than 40% if they just accepted that their CO2 theory of warming is just incorrect. They know it and they won’t change it. They must see that their ceteris parabus CO2 theory of warming is countered by some negative feedback phenomena, even if they don’t know what it is. A reading of the virtually universal Le Chatelier Principle (initially identified for chemical reactions and later found to be much broader, is for all intents and purposes a law). If an agent perturbs a system, the system changes such a way as to resist the change. It doesn’t succeed in stopping the change, but less change happens.
This works in the test tube but also anticipates such broader phenomena as Newton’s laws of motion, back-emf in a motor, supply and demand price behavior (push up the price and demand falls, or push up the price and supply increases while demand is falling, causing the price to fall again if supply increase goes too far or is badly timed), friction opposing motion, heating ice water – the ice has to disappear before the temperature begins to go up, buffering ocean ‘acidification’ – adding CO2 re adjusts, the equilibrium of carbonate reaction to resist the reduction in pH – we are having the same kind of fun with this looming ‘disaster’.
I think it suitable to add at least an unknown negative feedback to depress the ceteris paribus CO2 doubling effect to what it appears to be in fact about ~1 C per doubling. It would certainly bring the models into better congruency with the observational records. What is wrong with this idea? Now it still wouldn’t work for ever in a chaotic system but it might work for a couple of decades with some skill. The idiocy is doing it over and over and expecting something different. I think Einstein weighed in on this practice in an unkindly way.

Stuart Jones

if they were to adjust the CO2 factor from 3.something to 1 then the models would be closer to reality. but that would mean that CO2 has no effect on the climate. has anyone done this? taken just the estimated CO2 feedback out of the model results and see where the “predictions” end up after that, simplistic I know but just a back of an envelope estimation shows that the output of the models will come down in temperature, closer to the actual temp and therefore will be more useful, they may even be right, after all those billions of dollars must have produced some sort of theoretical model that may have some usefulness (minus the cO2 fudge factor)

I did not hear of LeChatelier’s name in my econ course or my physics courses, but I did in my chemistry course. Meanwhile, Earth’s feedbacks to a climate forcing can be positive and were for surges and ebbings of ice age glaciations. However, once the Earth is so ice-covered or so free of snow and ice that a temperature change does not cause significant albedo change, then the total feedback is less positive (or more negative). Also, the lapse rate feedback (a negative one) gets more negative as the Earth gets warmer or greenhouse gases increase.


What ‘unknown’ negative feedback. Any forcing of rise in surface temperature will increase water vaporization. Nature in her infinite wisdom utilizes water vapor to lift heat energy upward to altitudes where water vapor can radiate energy to space. In spite of the random chaotic processes which have entertained physicists for decades, nature has used water vapor to provide a NET cooling. We know this is true since, aside from the direct IR window radiation to space (also a negative feedback) there is no other physical phenomenon available for natural processes to use to rid the planet of the balance of the solar energy being absorbed. Thus evaporated water vapor provides a NET cooling. This is by definition a negative feedback to any rise in surface temperature. To assert that some increment to extra water vapor would result in positive feedback is ludicrous. Nature has already figured this out since it is the only mechanism for ridding the planet from any excess heat from whatever forcing source.


In any case, trying to forecast the behavior of 3 (atmosphere, ocean, sun) chaotic systems is doomed to failure.


I like to call them the 5 spheres.
Unless your model can successfully juggle all 5 spheres and the constantly changing interactions between them. Then it is useless.


What about that sphere 93 million miles away?


Heliosphere (#6)


magnetosphere very possible also, MarkW.
(hmm… the ‘magic number’ seven?)


#6 -The big sphere that contains all the other spheres.

No. Just no. This site needs to stop giving untrue information to its readers. It says:

People were shocked by the leaked emails from the Climatic Research Unit (CRU), but most don’t know that the actual instructions to “hide the decline” in the tree ring portion of the hockey stick graph were in the computer code.

Most people don’t know that because it is basically untrue. The HARRY_READ_ME file had nothing to do with the hockey stick graph. It was all about the CRU TS2.1/3.0 data set. Despite that, something like half the text in the link offered is about that file.
Other text is like:

FOIA\documents\osborn-tree6\mann\mxdgrid2ascii.proprintf,1,’Osborn et al. (2004) gridded reconstruction of warm-season’
printf,1,’(April-September) temperature anomalies (from the 1961-1990 mean).’
printf,1,’Reconstruction is based on tree-ring density records.’
printf,1,’NOTE: recent decline in tree-ring density has been ARTIFICIALLY’
printf,1,’REMOVED to facilitate calibration. THEREFORE, post-1960 values’
printf,1,’will be much closer to observed temperatures then they should be,’
printf,1,’which will incorrectly imply the reconstruction is more skilful’
printf,1,’than it actually is. See Osborn et al. (2004).’

Which is to print warning statements about what was done to the data, hardly a damning thing. Especially since some of the code being referenced is for papers which were never published, papers which went to great length to discuss what they did. Here is an excerpt from a draft of one:

Warm-season temperature reconstructions with extended spatial coverage have also been developed, making use of the spatial correlation evident in temperature variability to predict pasttemperatures even in grid boxes without any tree-ring density data. The calibration was undertaken on a box-by-box basis, and each grid-box temperature series was predicted using multiple linear regression against the leading principal components (PCs) of the calibrated, gridded reconstructions described in section 4.4. The PCs were computed from the correlation matrix of the reconstructions, so the calibration was in effect removed and similar results would have been obtained if the PCs of the raw, gridded density data had been used instead. The only difference is that the calibrated data with the artificial removal of the recent decline were used for the PCA. Using the adjusted data avoids the problems otherwise introduced by the existence of the decline (see section 4), though all reconstructions after 1930 will be artificially closer to the real temperatures because of the adjustment(the adjustment is quite small until about 1960 – Figure 5c). Tests with the unadjusted data show that none of the spatial patterns associated with the leading PCs are affected by the adjustment, and theonly PC time series that is affected is the leading PC and then only during the post-1930 period. Inother words, the adjustment pattern is very similar to the leading EOF pattern, and orthogonal to theothers, and thus only influences the first PC time series.

The draft discussed exactly what changes were made to the data and why, then showed what effect the changes had. There’s nothing dishonest or wrong about. I personally don’t think the changes were justified, but I could never claim someone is hiding things by telling me what they’re doing and showing me what effect it has.
If you’re going to say “the actual instructions to “hide the decline” in the tree ring portion of the hockey stick graph were in the computer code,” you need to do things like make sure the code you’re talking about is actually for graphs that were published. And were for hockey stick graphs. And was for hiding a decline. Because most of what readers are linked to wasn’t.
In fact, I’m not sure any of it was. I can’t rule it out though. There was a bit of code there I’m not familiar with. So hey, maybe 5% of it does something to support what this post says?


Tim seems to be engaged lately in a “crusade” to validate the climate models…..
The main logical flaw with this is that it supposes to be the other way around……..that the climate models actually are and exist to validate Tim’s and every one else knowledge of climate and climate system, including the mainstream orthodox climatology and climate knowledge.
Probably somewhere on the line the models do hurt Tim’s feelings in the issue of climate change……
And I think that is the case with many other so called sceptiks.
Still Tim, as any body else, has the right to take any position with this but never the less that does not mean that he will be correct or right about his take in this one.
I do not know and can not even begin to contemplate the possibility of a climate science and progress in climatology without climate models, but some like Tim seem to not have a problem at all with such a non realistic, regressive and backward position…..and the only thing I can say is: “good luck to all of you with your non realistic and “blind” running towards what you may call “knowledge”..”
After all chaos exist in the absence of the knowledge.

Dr. Ball’s position is not regressive or backward. It is simply realistic. Read my previous illustrated guest post here on this here a few months ago. You evidence severe lack of knowledge about climate models.
1.1 petaflops is 7 orders of magnitude, not 2, below the minimum resolution to approximate convective processes using computational fluid dynamics as is done in weather forecasting. Therefore essential processes like tropical thunderstorms(and therefore water vapor feedback, Lindzens adaptive iris and Eschenbacks albedo governor) cannot be simulated, so must be parameterized. Until the attribution problem can be resolved (anthropogenic forcing v. natural variation), no aproximately correct model parameterization is possible. GCM attribution has been essentially all anthropogenic except for part of the erroneously tuned high aerosols used to cool models to get reasonable hindcasts. Which is why all the models run hot now, and likely will for something like another 15 years if the Curry/Wyatt stadium wave, the Akasofu Arctic ice cycle, the PDO, and the AMO are any indication. It will take at least another half full cycle, about another 30 years of ARGO and UAH, to begin to untangle attribution. And until then, all funding should be stopped as a provable waste of time and money.
Those resources would be far better spent researching basic empirical climate science and weather forecasting (why the absence of Atlantic hurricanes, why the Arctic ice cycle), energy storage, and 4 gen nuclear like MSRs (which will need plenty of detailed engineering design simulation).

Robert B

Maybe I misunderstood, Whiten. Did you say that we can’t learn to predict the future if we don’t compare our predictions with chicken entrails?

I think you are failing to distinguish prognostic models from process models. As far as I can tell, Dr. Ball’s complaints are mostly if not entirely about prognostic models, and your assertions are, ISTM, most likely to be about process models. If not, then ISTM that your objection to Dr. Ball’s “crusade” is mere assertion, to which you are of course entitled, but which may also be dismissed by mere assertion.
BTW, I think Dr. Ball has given the prognostic models an undeserved pass on their poor hindcasting, but I do know that one can only write about so much at one time.


In your opinion, it’s better to have models that are always wrong, than to have no models at all?


When the climate STOPS changing , THEN we should start worrying !!!!


Just a minor point:
By law the BBC is required from time to time to renew the contracts that it has for all the services that it uses, and this time another forecasting company undercut the Met Office offer.
I understand, though, that the new forecasting company will still be using the information from the Met Office computer to make their forecasts.

Stephen Richards

That is also my understanding and seems perfectly feasible when you realise that only the UKMO collect data in the UK. HJow much they will have to pay for it and if that cost was factored into their tender only time will tell but these secondary contracts very often fail in the UK

This brings up “Lies, damned lies, Statistics and Climate Forecasting.” A song.
With apologies to Tennessee Ernie Ford:
Some people say people are made outta mud
Global warmists they are, they are chewing their cud,
Chewing their cud and follow Al Gore
A mind that’s a-weak can you ask for much more?
More than one megawatt, and what do you get?
Another prognosis and deeper in debt
Saint Peter don’t you call ’em ’cause you must let ‘em be
They sold their souls to the IPCC.
They came in one mornin’ when the sun didn’t shine
They picked up their papers and continued the grind
They had sixteen conditions, mostly falsified bull
And the straw boss said “Well, a-bless my soul”.
More than one megawatt, and what do you get?
Another prognosis and deeper in debt
Saint Peter don’t you call ’em ’cause you must let ‘em be
They sold their souls to the IPCC.
They came in one mornin’, it was drizzlin’ rain
the prognosis had failed them again and again
The boss harshly told them, You will do many more
Do as I tell you, and agree with Al Gore.
More than one megawatt, and what do you get?
Another prognosis and deeper in debt
Saint Peter don’t you call ’em ’cause you must let ‘em be
They sold their souls to the IPCC.
The cold snap we’re having now, it just cannot last
and hidin’ the warming that occurred in the past
Their ol’ man Mann and his hockey stick.
With conditions like this nothing ever will click.
More than one megawatt, and what do you get?
Another prognosis and deeper in debt
Saint Peter don’t you call ’em ’cause you must let ‘em be
They sold their souls to the IPCC.


*grin* Nice!


As a catastrophic global warming, er, warmist, I want to say a big “No!” to the headline: Is It Time To Stop The Insanity Of Wasting Time and Money On More Climate Models? It is NOT!!!! In fact, we should be even more insane and waste even more money, if that’s possible. Oh sure, we had to drop “hide the decline” and settle for “realign the decline”. And we may even be pushed one day to “redefine the decline”, but, “malign the decline”? — never!!! There is NO place for sanity or belt-tightening when it comes to something as important as the climate.Too many lives (and careers) depend on it. And I am unanimous in this.

Stephen Richards

errr /sarc

Cookie Krumbles

Remind me never to agree with anyone who disagrees with me.


How many supercomputers would it take to provide real time simulation and prediction of the AMO? Start the grant writing. It could be the next virtual high speed rail project.


“Those who have knowledge don’t predict. Those who do predict don’t have knowledge.” Tzu, Lao (6th Century BC)

Dr. Ball, you may have stretched Laozi beyond the breaking point. 😉

So um… yeah. I hadn’t read the entire post before submitting my last comment, since I wasn’t worried about the post as a whole. I’ve always been interested in the hockey stick debate, but global warming as a whole mostly bores me. Still, since I had commented on the post, I thought I should read the post as a whole.
I haven’t managed to do so though. You see, I got kind of stuck when I found out Tim Ball misquoted the IPCC in a rather severe way. He claims:

The IPCC acknowledge that,

“In climate research and modeling, we should recognize that we are dealing with a coupled non-linear chaotic system, and therefore that the long-term prediction of future climate states is not possible.”

That comment is sufficient to argue for cessation of the waste of time and money.

And had to stop. Why in the world would anyone think that a sufficient reason to stop spending money trying to forecast the weather? Here’s a fuller quote:

In climate research and modelling, we should recognise that we are dealing with a coupled non-linear chaotic system, and therefore that the long-term prediction of future climate states is not possible. The most we can expect to achieve is the prediction of the probability distribution of the system’s future possible states by the generation of ensembles of model solutions

Or from another part of the IPCC report being quoted:

The climate system is a coupled non-linear chaotic system, and therefore the long-term prediction of future exact climate states is not possible. Rather the focus must be upon the prediction of the probability distribution of the system’s future possible states by the generation of ensembles of model solutions.

All these quotes are saying is we cannot hope to predict exactly what future weather/climate will be via a model. That’s not surprising or remarkable. Nobody thinks we could. We don’t create weather forecasts by running a single model and taking its results. We run models over and over and see which results are the most likely.
That doesn’t prove weather models are useless, yet that’s exactly what this post claims. It claims because the IPCC acknowledges it’s impossible to predict exact climate states, models should be scrapped. That’s beyond silly. Maybe models aren’t worth their costs, but this argument does nothing to show that. If we accept this argument, every weather and climate model in existence would have to be scrapped, including ones many businesses rely upon.

Like your comment — it is better than mine that follows.
With regard to: “. . . {chaos insures] the long-term prediction of future climate states is not possible.”
Too vague about “long term.” Is it a day, a week, a year, a century? Where is the reference showing this chaotic behavior? Don’t appeal to generalities about chaotic systems because:
1) Some chaotic systems have areas of stable behavior for which predictions may be possible and our climate may be in one. Where are the studies showing ore climate isn’t stable?;
2) The displayed temperature behavior may be in error, but doesn’t appear chaotic, so where is the evidence that these errors aren’t due to parameter errors rather than chaos?

“our climate” not “ore climate”


Except the predictions are uniformly way off base from reality and dead wrong. Any business relying on these bad predictions would go bankrupt.

Robert B

The Australian BOM gave us a prediction that went something like – For the rest of the country, the chance of above average rainfall is 50%.
How much would you pay for a prediction like that?

Stuart Jones

and they were wrong!!!

Leo Smith

I think you misunderstand the mathematical subtleties involved. That IPCC complete statement is skating on very very thin ice.
Let’s take an example: a chaotic system with three attractors. So at any given point it can flip from one attractor, with one sort of average-ish value, to a completely different attractor with another longish term but very different average.
Computing the overall average of all the states is useless, because depending on micro differences, it might stay stuck around an attractor for several centuries, before wandering off to a new one.
I.e it is of no political or economic use to say ‘there are three possible states for the climate in 20 years, it might be 3 degrees warmer, three degrees colder or pretty much the same as it is now, on average with an equal probability of all three’
But that sort of prediction is exactly what prediction of the probability distribution of the system’s future possible states means.
We have established that its bounded – and we have assigned a probability distribution to some possible future states.
The fact that the answer, though accurate, is pragmatically totally useless, is the point being made.

Not “totally useless” — such a prediction might show that warming was limited and not dangerous.


Simply make their funding contingent upon results. For example, if last year’s forecast/projections/predictions are correct, they get money. If not then, bad scientists, no cookie — they get exactly zero.

The Mathematics of Turbulence

Accuracy and Precision

Gary Pearse

Van Gogh had visual problems. I had cataracts removed that had me seeing blurred clusters of stars from one star and several blurred moons distributed about a centre. I kept one cataract for about 10 years because I could read fine print and see my computer clearly without glasses. Now with ocular implants I can read a licence plate at more than 100 ft, but need reading glasses. Several other painters experienced eye disorders, especially with age, that could be diagnosed from their paintings. Van Gogh’s “math” was simply the chaos of pathological light dispersion.


The problem is the people who believe in AGW will not react to facts and reason. As Aristotle said, ‘some people just cannot be taught’. You can show them graphs, p-value’s, error bars, photo’s of thousands of polar bears enjoying the sun on an oil rig. They will not change their minds with Reason. Until something changes their ‘feelings’, they will go on believing.

Extremely likely that it’s all a fraud?


In a related story by the WSJ on zombie data servers, roughly 10 gw of power is being wasted on them being powered on but not connected or doing anything. I wonder how much CO2 footprint that is.

The 1/14 APS workshop minutes revealed serious doubts about IPCC and GCMs, just didn’t have the cajones to outright say so.

David S

The role of confirmation bias I think overwhelms and has controlled the argument s on global warming. As a non scientist I actually don’t believe there is any correlation between CO 2 and temperature / climate. There would be better correlation between temperature and business cycles. When you start with the aim of establishing whether a correlation exists you can’t start with the assumption that it does theoretically or otherwise. Even many sceptics seem to concede that there is a relationship between CO 2 and temperature. They could’ve just as easily established a theory based on oxygen. The amount of energy and money wasted on proving or disproving correlation with CO 2 has actually perpetuate the scam. By arguing about why models don’t work sceptics have fallen for the trap of assuming that scientifically there should be a correlation between temp and CO 2. I think as a layman there is none both theoretically or otherwise.

Gary Pearse

CO2 does what physicists say it does but the interaction with negative feedbacks largely neutralizes the effect. The atmosphere/land/water/ice/biophere butts in to counteract the effect. The recent observation by NASA of greening of the planet is an example. Growth in plants is endothermic (it takes heat out of the system). If the plants are growing on a desert, albedo is reduced which would tend to warm things up, but it also anchors moisture and itself emits water vapor, cooling things down. The net effect is to cool the day down and warm up the otherwise cold desert night. This in turn reacts to create more precipitation and…..
The well established physics of CO2’s absorption of long wave radiation gives so much comfort and encouragement to CAGW proponents that isn’t deserved, because the planet reacts in a negative response to it in multiple ways, not only in a part of the biosphere as I have described above.

as said elsewhere:
In engineering and design models are essential tools. In the climate ‘science’ models are essentially for the WUWT’s punters amusement.

In engineering and design models are essential tools.
Yeah, but an engineer’s model has to actually work.

Ric Haldane

Only because engineers know they have to deal with reality. Climate modelers try to create reality.


Tim Ball says….”Weather forecasts beyond 72 hours typically deteriorate into their error bands.”
So you are saying weather forecasting and predicting climate are the same thing? I’m not so sure they are.

“…weather forecasting and predicting climate are the same thing?”
IPCC AR5 and WMO define climate as weather averaged over thirty years. So, yeah, they are.


Not long term they are not and that is what we are worried about. Very different science.


I agree with you, Simon. He at first seems to conflate weather and climate.


As far as the U.S. goes, I don’t think any congress-critter ever actually funds climate models. Congress funds Departments, Bureaus, and Agencies which fund climate modeling. Only in Washington D.C., when a 12% increase in budget is submitted, is a 7% budget increase called a budget cut (horrors!). From their ever-increasing budgets, the various TLAs allocate funding, some of which is for climate modeling.
Keep in mind that the potential for catastrophe or the need to be doing “sumpthin’ ’bout sumpthin” – no one remembers the original purpose of most of the TLAs anyway – keeps the bureaucracy in place. The way to make the bureaucracy bigger and thus more powerful, is to increase funding on all the various things currently being done and find new things to do, whether or not they are already being done by other agencies.
IMHO, our elected political ‘leaders’ have been overtaken by the entrenched bureaucracies, and climate modeling and climate science are entrenched in the bureaucracies.
Climate modeling will go on being funded and receive increased funding in the U.S. regardless of the wisdom or merit in pursuing a reasonably working climate model and despite the abject failures of the current crop of models. And if by chance a perfect climate was accidentally produced tomorrow, all the agencies would ask for a budget increase to improve it!


“And if by chance a perfect climate model was accidentally produced tomorrow, all the agencies would ask for a budget increase to improve it!”

Isn’t this what they’re trying to accomplish? If they accidentally create a perfect model, then they will know just what to do to create a perfect climate. Then they can ask for a budget increase to improve it! 😉 /sarc


I don’t want to stop funding climate models because basic research is vital. What I do want is merit based allocation of funds. So 99% of the models lose their entire budget and 1% get more money. It is easy to pick them out. The one or two closest to reality win. This would encourage a lot of accuracy and pit one team against others trying to “fix” the numbers.


“I don’t want to stop funding climate models because basic research is vital.”
This is not in evidence. You are begging the question.

Gary Pearse

How do we know the ‘close’ models are related to reality? I’ve been thankful that they didn’t by accident match the actual climate at the height of the hysteria over global warming. Getting killed by falling windmill parts or dead geese would become like traffic accidents by now.


Basic research is vital, models aren’t basic research.
Basic research is going out in the field and getting actual data.


peta is 1016 (or a thousand) million

Not quite. The number should be 1024, but even if we stick to to conventional 1000’s, peta refers to one quadrillion. Terra is one trillion, Giga one billion, and mega one million. We will have to start using SI notation with computer sizes soon!

Peta gram = giga ton metric


…But Petabytes sounds like a treat cookie for abused dogs.

Leo Smith

That is not a bad description of harassed sysadmins.
Of course the BOFH strikes back…


I was interested in the second sentence of the post:
‘Weather forecasts beyond 72 hours typically deteriorate into their error bands.’
A reference would be very useful here. Certainly the scientific establishment claim that their performance in weather forecasting is much better than this.
An example comes from Peter Bauer, Alan Thorpe and Gilbert Brunet writing in Nature last week (The quiet revolution of numerical weather prediction, Nature 525, 47–55, 03 September 2015).
The abstract is freely available here:
You can also see figure 1 for free at that link (click to enlarge).
According to their figure 1, they are claiming that forecasts have significant value out to 7 days (= 168 hours, more than twice what is claimed by Tim Ball here).
Furthermore, they show a significant improvement in forcast performance over time. According to their statistics you would have to go back to around 1980 before useful performance fell to 72 hours. In the article they claim that both improved data collection and bigger computers have contributed to this improvement.
While Tim Ball’s focus here is on climate, he does kick off with an attack on weather forecasting (‘They are not just marginally wrong. Invariably, the weather is the inverse of their forecast’), and does later refer to workers in ‘weather offices’.
It would be useful if he could clarify the extent to which he really means to critique weather forecasting. If he does, then I think his piece as written is very weakly supported. As it stands I just don’t see the numbers or the supporting references that could be used to demolish an article such as Bauer et al.
If Ball wants to go on the offensive with an openning line, such as ‘forecasts beyond 72 hours typically deteriorate into their error bands’ then he should expect that people will ask him to justify that.

Gunga Din

Perhaps he’s referring TWC’s “Weather on the 8’s”. 😉
(I’ve seen them forecast thunderstorms for the day at 5:08 AM and at 6:18 AM forecast clear skies. (Later that day, it rained.))


I have seen that many, many times in Canada !!! They even put out ” Flash Flood ” warnings for my area , then we get a sprinkle of 2 MM !!!


Who needs skepticism when you have anecdotes.

A weather forecaster at WTAE-TV Channel 4 in Pittsburgh has started issuing a “4-degree guarantee”: “Every weekday during the 5pm newscast, Mike Harvey will give viewers his 4-Degree Guarantee. He guarantees the next day’s high temperature will be within 4 degrees of what he predicts.”
As far as I can tell, that’s +/- 4 degrees (F). In other words, this weather forecaster is claiming bragging rights if he gets a day-ahead high-temp forecast within a nine-degree range.

Peter Sable

‘Weather forecasts beyond 72 hours typically deteriorate into their error bands.’

Certainly the scientific establishment claim that their performance in weather forecasting is much better than this.
Deteriorate: “become progressively worse.”
“Deteriorate into their error bands” doesn’t mean fall to zero instantly, it means there’s less accuracy at 4 days, even less at 5, with very minimum utility at 7 days per your referenced paper.
I’m a surfer, I know quite well how long and even under what conditions a forecast is good for. (e.g. if there’s an approaching low, plus/ minus 100km on the center will dramatically alter surfing conditions)