By Steven Goddard
I recently had the opportunity to attend a meeting of some top weather modelers. Weather models differ from climate models in that they have to work and are verified every hour of every day around the planet. If a weather model is broken, it becomes obvious immediately. By contrast, climate modelers have the advantage that they will be long since retired when their predictions don’t come to pass.
Weather and climate models are at the core very similar, but climate models also consider additional parameters that vary over time, like atmospheric composition. Climate models iterate over very long time periods, and thus compound error. Weather modelers understand that 72 hours is about the limit which they can claim accuracy. Climate modelers on the other hand are happy to run simulations for decades (because they know that they will be retired and no one will remember what they said) and because it provides an excuse to sink money into really cool HPC (High Performance Computing) clusters.
But enough gossip. I learned a few very interesting things at this meeting.
1. Weather modelers consider the realm of climate calculation to be “months to seasons.” Not the 30 year minimum we hear quoted all the time by AGW groupies. That is why NOAA’s “Climate Prediction Center” generates their seasonal forecasts, rather than the National Weather Service.
2. The two most important boundary conditions (inputs) to seasonal forecasts are sea surface temperatures and soil moisture. No one has shown any skill at modeling either of those, so no surprise that The Met Office Seasonal forecasts were consistently wrong.
For example, just a few months ago the odds of La Niña were considered very low. Compare the December forecast with the May version. How quickly things change!
SST modeling capabilities are very limited, and as a result seasonal weather forecasts (climate) are little more than academic exercises.
Oh and by the way, Colorado will be exactly 8.72 degrees warmer in 100 years. But they can’t tell you what the temperature will be next week.
“If I don’t understand it, it must be simple.”
– Dilbert Principle
In the top picture, which boxer is weather and which one is climate? What do readers think?
Climate models are deterministic, not random. They use random numbers for appropriate Monte Carlo issues, like locations of clouds within a grid cell. But that is a far cry from being based on randomness.
carrot eater says:
July 1, 2010 at 9:52 am
If you accept that we can predict that a sun with much lower radiation output would lead to a colder climate, and that a sun with a much higher radiation output would lead to a warmer climate, then you are accepting that it is possible to use physics to make some estimates of how climate may change over the long term in response to changes in the energy flows into and out of the system.
Not really, no. Climate is far more complex than that. At present, we can only study what has happened in the past wrt the sun, and try to make projections based on the history. We know that solar activity is related to, and somehow tied with sunspots, as shown, for example, by the Maunder Minimum. We know too that it has something to do with clouds, but so far, they can’t be modeled (and may never be).
The “energy budget” concept of Warmists is a nice idea, but it conveniently diminishes some influences (like the sun) and grossly exaggerates the role of C02.
There is a basic flaw in the climate forecast models and the defense of them here.
We have the coin toss comparison:
““Weather is chaotic and unpredictable. Climate is weather averaged out over time. Simple analogy: you can’t predict whether a coin will land heads or tails, but you can predict statistically how thousands or millions of coin tosses will go.”
and the similar but more complex argument:
“Climate is maybe more like the ring than the fighters who stumble about, inside the boundaries… Actually, seriously, “climate” is more like predicting the probability distribution of the outcomes of 1 million fights, whereas “weather” is predicting the outcome of the next round of a single fight.
Both of these make the same mistake – climate is not a set of “Markov” daily weather models – climate has no Markov properties a cold day affects the following day, heavy rain one day leads to higher soil moisture the next etc. As forecasters know there is a stochastic property in weather that can at times appear to be a Levy Flight although the likely bounds of the dispersion about the expectation are known (the boxing ring) where the boxer will next step may be chaotic and the step after that depends on the previous step. The ring will move to be the likely Levy Flight bounds from each boxer and could be said to represent the climate. After a round lasting a century it is difficult to forecast the position of the ring – the climate – with any accuracy.
Added to the problem of the non-Markov Levy Flight is the lack of full understanding of the inputs that affect the weather or even knowledge that they exist. It is difficult enough attempting to forecast the behavior of a chaotic system of chaotic systems but when not all the variables are known and the effects of the known variables are uncertain it becomes gifted guesswork at best.
Venter (July 1, 2010 at 7:57 am), thanks for the Pielke link. Though brief, the discussion of “initial value problem” vs “boundary value problem” really puts the issue in perspective.
Climate is not exactly an average over weather, is it a wise average over weather.
In the case of the flip of a coin, it is easy to do a wise average because the probability is 50/50 for each flip. In the case of climate, the probability to have a temperature under 40F in New-York is not the same all the time. Meteorologists are aware of several cycles wich modulate the temperature: day-night, summer-winter, el nino-la nina. Now how many other cycles are present in the system? If you calibrate a model during a positive phase of one of these cycles, you will automatically predict a positive trend that does not exist.
I find it very difficult to believe a wise average of climate can be done when the existence or nonexistence of the medieval warm period is not supposed to have any effect on the predictions of the model.
carrot eater: July 1, 2010 at 11:21 am
Actually, I know perfectly well what I am describing, and it substantially changes what you said. You are making the standard mistake that has led the the entire ‘climate industry’ astray, namely the categorizing of physical processes as being either ‘forcings’ or ‘feedbacks’. The equations don’t neatly divide things up that way, and don’t ‘know’ which is supposed to be which. This is true even in very simple systems that are easy to solve exactly. See here, for example. If you add to that model the possibility of the ‘output hole’ altering itself in opposition to a change in the input rate (which is easy to do – you just have to add a genuine ‘feedback’ term ☺☺), you obtain a steady-state system that can be almost impervious to change. But feel free to rave on.
/dr.bill
Global warming/climate change has splitted the personalities of many former scientists, who in order to survive in such a competitive environment have surrendered their science replacing it for a creed. This is why some show two or more personalities at the same time, playing the eternal drama of good vs. evil, appearing like Dr.Jekyll and Mr.Hyde.
carrot eater says:
July 1, 2010 at 11:21 am
“dr. bill:
None of that changes what I said. If the solar output increased by 20%, and all other forcings remained the same, then the climate on earth would on average be warmer.
What you are describing are various feedbacks. The feedbacks determine how sensitive the climate would be to the imposed change in solar – precisely how much warmer it would be, as the system responds and reacts. Some feedbacks could strengthen the initial warming, some would counteract it. But overall, it’d still be warmer, than had the sun not done that.”
That assumes that the feedbacks are an invariant set. If solar output remains pretty much constant, and yet we have changes of 4, 5, 6C, then something is changing amongst the set of feedback mechanisms.
“I think some pretty impressive things happen within your limits. The difference between ice age and interglacial is ~ 6 C. I’d say an ice age and an interglacial are fairly different circumstances for the earth to find itself in.”
So… small differences in CO2 are enough to completely throw the various feedbacks way out of kilter? What rearranges the feedbacks to swing global temperatures back the other way?
And back to the topic, has anyone developed a model that holds the earth’s temperature within the 12-22C range and reasonably closely models the temperature history over the past 600-800 million years? (Rhetorical of course; not that I’m aware of.)
The point you were making in response to dr. bill is pretty clear. I’m just trying to point out the the set of the state feedback mechanisms is not constant over time nor are all the feedback mechanisms well understood or even if all feedback mechanisms completely known.
R.Gates
Both have chaotic dynamic elements
Chaos’ explanation, or rather, justification, appears where knowledge is lacking.
Knowledge is not so hard to attain than hard to accept.
Pitagoras with his simple monochord attained it, how about you?
“If I don’t understand it, it must be a hoax.“
– Goddard Principle
😛
Re: the boxers, the guy on the left is me, the guy on the right is the IPCC. 🙂
“Climate models do have their limitations and modelers are constantly improving their models with newer data as the understanding of climate processes improves with research. According to the IPCC (ibid): “Nevertheless, models still show significant errors. Although these are generally greater at smaller scales, important large-scale problems also remain.”
I agree with this and haven’t asserted anything to the contrary. I also agree with the sentence preceding the “nevertheless,” which was not quoted: “Model global temperature projections made over the last two decades have also been in overall agreement with subsequent observations over that period.”
“Climate models iterate over very long time periods, and thus compound error. ”
Non sequitur.
“Weather modelers consider the realm of climate calculation to be “months to seasons.””
Whichever arbitrary and unspecified subset of weather modellers you spoke to were gravely mistaken. You can no more measure climate over months or seasons than you can measure continental drift in a minute.
Günther Kirschbaum
Don’t pollute this thread with garbage. I have never said anything about a hoax and I have an excellent understanding of climate models.
Take it back to Romm or Tamino’s site where BS is the norm.
“That assumes that the feedbacks are an invariant set. ”
At this point, time scales matter. There are some feedbacks that operate on the shorter term. There are some feedbacks that operate over the longer term.
For example, for a sloooooow feedback, there’s the volcano/rock weathering one. Or, similarly, all the carbon dioxide being introduced into the climate system now will eventually end up in rocks somewhere. But just because all the carbon dioxide will eventually go away if you just wait many thousands of years does not mean you can’t predict it will warm in the meantime.
Just as there might be some feedback, perhaps now unimagined, that would over a long time counteract a huge change in the sun, it doesn’t mean you can’t predict that a huge change in the sun would have a climate impact in the meanwhile.
To make an extreme example to prove the point: if the sun magically ceased to exist tomorrow, I really don’t care that we can’t predict weather more than a week ahead of time. I know that I’m going to be cold and dead. Actually, if we wait long enough, we’re going the other way with the sun, but you get the point.
“you obtain a steady-state system that can be almost impervious to change. ”
You can write whatever you want into your equation, but the actual Earth clearly isn’t impervious to change. There are ice age cycles, after all. There may have been a snowball earth or two. There was the PETM. And so on.
Steven:
“People run weather models out to two weeks, but they are pretty useless after 72 hours. I check the accuweather two week forecast almost every day – it changes radically from day to day”
That is correct, but that’s the point: run the model out for a year, and you will see the seasonal cycle. Any given day’s forecast at a particular location (i.e. the weather forecast) will probably be wrong, but I assure you the model will predict that it will be warmer during the month of July at, say, Boston, than in the month of January (i.e. a climate forecast). One quantity may be predictable only out to a few days, but other quantities may be predictable on much longer time scales.
Do you agree with the above statement? If not, please explain. If you do, then you understand why the logic of this article is flawed.
Thanks,
Dan
dr.bill says: July 1, 2010 at 12:28 pm
“…..But feel free to rave on.”
If you are real Dr. (someone ?), then for the above choice of words, the real name rather than the ellipsis would be more appropriate, so readers can judge the real command of authority you articulate.
Paul Daniel Ash mentions that some models got something right. Paul Daniel Ash: When your model does not get the cloud distribution right but in the end the temperature fits, does this not give you the feeling that it might be right for the wrong reasons? And has there ever been a climate model that gets cloud distribution over latitudes halfway right? I think it has yet to be invented.
“I have an excellent understanding of climate models. ”
Not once have you shown that you understand that weather prediction is an initial value problem, and climate is a boundary value problem.
I agree with carrot eater.
Translate, please?
/dr.bill
To Anthony Watts:
Who is this “Steve Goddard” character, who now pretty much dominates your website? Do you support his posting style, which seems very far from genuine scientific scepticism? His biases are manifestly clear in his posts (he can’t seem to ever resist slipping in a snide comment at the end, even if the rest looks pseudo-scientific). Who is he? We deserve to know.
When your model does not get the cloud distribution right but in the end the temperature fits, does this not give you the feeling that it might be right for the wrong reasons? And has there ever been a climate model that gets cloud distribution over latitudes halfway right? I think it has yet to be invented.
First off, they’re not my models, and secondly yes, as I’ve been saying throughout the thread, there’s lots to criticize in how models are run and in the certainty ranges. And absolutely: parameterization of clouds and aerosols go near the top of that list. “Failing to accurately predict anything worthwhile” doesn’t even make the list.
Past a certain point, as the goalposts get moved from “wrong” to “right but for the wrong reasons,” one has to wonder if we’re talking about science or belief. And i don’t get into religious arguments, on principle.
Dan
If all inputs were static, then every year would indeed average out to be the same.
But the whole point of the climate models is to evaluate the feedback from changes to the environment. Feedback is calculated through iteration, and if the feedback assumptions are incorrect, the model output will become progressively divergent from reality. There is no magic averaging function which will come to the models’ rescue.
Paul Daniel Ash
I quoted verbatim from Scotts site and did not leave anything out preceding
‘ nevertheless’ it must have been Scott himself that truncated that quote-not me.
Tonyb