Arguments
Weather vs Climate
Posted on 26 March 2011 by dansat
The "skeptic" claim "scientists can't even predict the weather right" is based more on an appeal to emotion than fact. The inference is that climate predictions, decades into the future, cannot be possibly right when the weather forecast for the next day has some uncertainty.
In spite of the claim in this myth, short term weather forecasts are highly accurate and have improved dramatically over the last three decades. However, slight errors in initial conditions make a forecast beyond two weeks nearly impossible.
Atmospheric science students are taught "weather is what you get and climate is the weather you expect". This is why this common skeptical argument doesn't hold water. Climate models are not predicting day to day weather systems. Instead, they are predicting climate averages.
A change in temperature of 7 degrees Celsius (°C) from one day to the next is barely worth noting when you are discussing weather. Seven degrees, however, make a dramatic difference when talking about climate. When the Earth's average temperature was 7ºC cooler than the present, ice sheets a mile thick were on top of Manhattan!
A good analogy of the difference between weather and climate is to consider a swimming pool. Imagine that the pool is being slowly filled. If someone dives in there will be waves. The waves are weather, and the average water level is the climate. A diver jumping into the pool the next day will create more waves, but the water level (aka the climate) will be higher as more water flows into the pool.
In the atmosphere the water hose is increasing greenhouse gases. They will cause the climate to warm but we will still have changing weather (waves). Climate scientists use models to forecast the average water level in the pool, not the waves. A good basic explanation of climate models is available in Climate Change–A Multidisciplinary Approach by William Burroughs.
Source: AMS Policy Statement on Weather Analysis and Forecasting. Bull. Amer Met. Soc.,79,2161-2163
*Image source: Meehl, G. A., C. Tebaldi, G. Walton, D. Easterling, and L. McDaniel (2009), Relative increase of record high maximum temperatures compared to record low minimum temperatures in the U.S., Geophys. Res. Lett., 36, L23701, doi:10.1029/2009GL040736.
NOTE: This post is also the Basic rebuttal to "scientists can't even predict the weather right"
Putting that pot of water onto a campfire might leave you concluding that it is never going to boil, as I'm sure many experienced outdoors might testify to.
Sorry, feeling the need to be a bit silly.
Saying you can't predict climate change if you can't exactly predict tomorrow's weather is like saying you can't determine who will likely win when the Miami Heat plays the Minnesota Timberwolves (substitute your favorite sports mismatch)unless you know the exact spot on the court where every player will be when there are exactly 8 minutes left in the 4th quarter.
Someone who says that weather forecasts are inaccurate then can only be remembering back to a much earlier time, a time when isobars were based on a few weather stations and drawn by hand in small generalised maps in newspapers; where the forecast of rain could only be refined by what time it rained in the neighbouring town; where a front might peter out and slip away to the south before anyone could be aware of that fact; when conversely a cyclone might arrive out of nowhere; and where the dynamics of interacting weather systems were barely understood. I remember these kinds of forecasts from the 1950s, and I am sure that the deniers are harking back to that time (or even earlier) as well.
The fact that weather forecasts now are extraordinarily accurate is the result of the huge growth in monitoring stations and weather satellites and computer analysis and well developed theories about rainfall and wind behaviour. All of the same factors that make our understanding of climate change so advanced.
Time the deniers got to know which way the wind blows and why and when. Oh and if the weather forecast says take an umbrella, take an umbrella, a hard rain is going to fall.
The ability to make such forecasts accurately is improving, but in terms of progress the mainstream forecasters are perhaps years behind some of the more progressive ones.
Generally you get what you pay for, and some of the free services are simply too costly to follow.
To go one step further on the absurdity index, small earthquakes are generally harmless; in large quantities (magnitude), not so good.
Surely he means groups like IRI or ECMWF when referring to "progressive ones"? ;)
Very nice post dansat-- succinct and informative. Yet another ridiculous "skeptic" myth busted.
However I'm not sure how you will be able to compare his forecasts to others without subscribing to his service, or if you are even in the region he provides services to.
Are you going to subscribe?
However, good article: succinct, as it should be.
A thought, perhaps you could include a hyperlink to where someone might find some info about Burroughs' book? Maybe an abstract or review, not Amazon but something informative.
Most certainly, his forecasts would not be available on-line, it is a subscriber paid for service, tailored according to individual requirements and distributed directly to each subscriber, updated as necessary.
He claims about 80% success rate, keep in mind his forecasts are very specific, in both outlook periods and specified coverage,not your usual BOM general forecasts of 50% chance of above and 50% chance of below outlooks produced for whole states.
Even BOM and CSIRO claim they are several years away from producing useful and reliable forecasts with the government being asked to put up a large sum for purchasing new "super computers", I think that was the they term used.
If you are not going to subscribe to his service, then you may have to be satisfied with reading testimonials from satisfied subscribers.
He originally worked for BOM, but his thinking on what data was needed to make forecasts more accurate differed to theirs, so he left to start his own commercial forecasting service.
His advantage was that he had found that incorporating IO data increased substantially the accuracy of his forecast models putting him about 10 years ahead of BOM in that regard, something I think we have discussed previously.
He has continually added data from all ocean areas around Australia that wasn't previously being used for modeling.
Nonsense. I'll elaborate later, but if someone wishes to debunk this myth please go ahead.
Climate is like rolling a single die roll 1000 times. It’s quite easy to predict the average outcome using a simple model. The average of all those rolls will be 3.5 +/- 0.17.
Future climate is predicted by averaging many runs of a model to eliminate noise or some inherent bias towards certain initial conditions. And sometimes averages of different models are used to further eliminate bias.
As the time window expands, predictions become more and more fuzzy. When the forecast says that it will be raining six days from now I book my golf foursome with better than 50% confidence that the weather will be fine.
When the BBC predicts that the next winter will be mild I (usually correctly) expect the forecasts to be wrong to an embarrassing degree. Likewise, when various university scholars predict how many hurricanes will hit Florida I realize that their guesses made using super-computers are no better than my guess based on waving a damp finger in the air.
Predicting the temperature 100 years from now is much tougher than forecasting just a few months ahead.
This post does say one thing that I believe is probably correct:
"A change in temperature of 7 degrees Celsius (°C) from one day to the next is barely worth noting when you are discussing weather. Seven degrees, however, make a dramatic difference when talking about climate. When the Earth's average temperature was 7ºC cooler than the present, ice sheets a mile thick were on top of Manhattan!"
Many of you seem to be in favor of reducing the average global temperature. Do you hate New York so much that you want to restore that ice sheet?
In other words, although it is clear that climate prediction is not the same as weather prediction, there is nothing in the way of proof that climate predictions have any accuracy or skill.
Climate models are NOT a bunch of weather predictions glued together. Anybody trying to use this kind of argument is deeply confused on the subject.
Charlie A, Skepticalscience has another post dealing with climate models and their accuracy.
Doesn't look too shabby to me.
Do you mind explaining, in detail, exactly what scientists were predicting 4 years from the time of the prediction including some sort of uncertainty they were allowing themselves?
I predict in 4 years time, barring some unpredictable event, that global temperatures will be within 0.3-0.4oC of where they are now. Does this mean I have some skill in climate science?
Such summary judgments should be avoided when you don't know what you are talking about. This is exactly how many industries use computer models - to see the range of things that might happen if there is a small change in the input.
Some call these Monte Carlo methods. Industries that use these type of models (petroleum exploration, nuclear engineering, operations research, military, solid-state physics, fluid dynamics, particle physics, financial analysis, network design, weather forecasting, etc) should immediately cease work because PT says they are inaccurate, irrelevant, worthless and so on. Or due to the consensus that these models are applicable, perhaps PT's assessment is inaccurate, irrelvant, worthless and so on.
We can decide whether to put civilization back 50 years or ignore PT based on a flip a coin, best 2 out of 3.
And the basis of climate models are empirically tested physical laws and the results are tested by back-casting through known climate conditions.
"It is interesting you mention financial analysis of which those models are just as worthless."
I doubt that a corporation would keep throwing money at financial models if they didn't return something useful.
My conclusion is that climate models can't predict weather. It doesn't mean climate models can't predict climate in some simplified way, but not the critical nuances (i.e. weather) that control sensitivity.
His argument is equally as easy to brush aside.
Not always. Monte Carlo methods were introduced for the Manhattan Project (and named by physicists with a sense of humor). There was no experimental verification until the Trinity test; if memory serves, there was some doubt as to whether or not it would set the atmosphere on fire - and they went ahead with the test.
There are no experimental tests in the oil industry, until you pony up and drill a well. A dry hole is merely one of the expected outcomes of a suite of potentially valid models.
Best advice remains: keep your opinions to yourself unless you actually know what you are talking about.
"any result on a computer system that is not 100% accurate ... not to be trusted"
OK, run a computer simulation that reports the location and velocity of an electron in a semi-conductor to 100% accuracy. If you can't, you must therefore not trust anything on your computer, or for that matter any piece of electronics you use. Better turn the lights off, because computers run those too.
Or you could actually try learning about things before you pontificate.
But this thread is about weather and climate. Further off-topic digressions should be deleted.
Mods: I have moved my post to the model thread so please remove my previous post here.
Gilles and HR, I have questions for you here.
On top of that, it's a bit of a car crash every time PT stops by. And like a car crash it's hard for people NOT to watch. It appeals to our more base level instincts.
I don't think anyone wants to see Skeptical Science become just about this kind of car crash. Many may disagree with this but overall, from time to time, a PT car crash serves it's purposes.
I agree with Rob @60, SkepticalScience is above this kind of nonsense and typically has a very high signal-to-noise ratio, that is why it is pretty much the only place that I choose to post on climate issues.
Read and enjoy/learn/weep as so inclined.
Impact of Global Ocean Surface Warming on Seasonal-to-Interannual
Climate Prediction
As to Poptech and Ken Lambart and a few other posters. I hope smart, knowledgeable people will continue to refute the ideas they present when they are wrong. I personally don't have any problem seeing that models work (Hansen 1988 is still mind boggling to me - 23 years ago he knew what would happen to us right now!).
But the claims that Trenberth's travesty is still with us will give me pause until we can irrefutably (within reason) put it to bed.
Perhaps others are the opposite - they understand the limitation of ARGO, have read the studies regarding deep ocean temperatures (and a few lakes which give us fascinating insights into ocean heat content and heat transfer behavior) and see no problem with the supposedly missing heat.
It is for those readers (presumably many X larger than those of us who post - else this is a lot of work for very few eyeballs...) that honest refutations of misinformation (regardless of the sincerity of the poster - wrong is wrong) are so valuable.
Perhaps the moderators have to be even more hard nosed to move the debate to the right thread? That would ensure that those of us taken in by the false claim would read the background/supporting post and hopefully get a better understanding, so as not to be taken in by the same claim next time.
Firstly you probably accept it because it might only impact on whether your take an umbrella or not and so the free, but entertaining forecast attached to the TV news remains useful in providing a topic of conversation, and an excuse to complain.
But for those whose businesses and enterprises depends on accurate short and long range forecasts, that is not only not good enough, but totally unacceptable.
So what do they do? They seek out services that have proven track records and provide real value for money. Free is hardly ever free.
So one of the points that your comment raises is, whilst, as you noted, there are weather forecasters out there that are virtually useless, but probably still retain a large and faithful following, and whilst there are also probably unknown to yourself, weather modelers, that by virtue of providing reliable and useful projections for more astute investors, and are able to charge high fees for their services, is it possible that the same range of expertise might also be within the ranks of those who model climate?
Just for interest:-
(1)How many people who read this thread are totally dependent on accurate forecasts for planning the next year or two forward?
(2)How many people are totally dependent on accurate forecasts to make decisions about umbrellas?
Gallopingcamel, I cant believe after all this time you can seriously say this. Repeating again, concern over climate science isnt about proposing an optimal global temperature (higher or lower), it about reducing the rate of change in the temperature. From little I know of new york geography, the equilibrium sealevel, last time we had 400ppm would put most of NY underwater. I could ask whether you want that instead?
Tell, if you put a large kettle on to hot flame, could you will all the computer modelling in the world accurate predict the convective flow within that pot? Not likely, though you might predict the pattern. (the weather) Could you predict when the kettle will boil? (climate) yes.
You're not supposed to ask questions like this here. :-)
The classical answer back from the AGW camp to your question will always be "deviations are variations in weather and should not be confused with climate that is long term based", combine this with error range that in principle cover every possible future scenario how can one ever be found to have made an incorrect predictions then? ;-)
I read your comment as "Since what I do is a correct procedures, it follows that what climate scientist do is also a correct procedures."
But surely this is not what you really meant to say or?
Back to an “unlikely” event that has been predicted by climate models, if we look at Hansens model from 1988, I repeat what I’ve said on the models are unreliable thread, the global average surface temperature has risen. There were many who predicted the reverse based on overestimates of solar influence, or claimed that any temperature rise to date was not statistically significant, and that this would remain the case (famously for example based on the MSU satellite evidence). Climate science has advanced at least partly as a result.
It is more telling to look at events not adequately predicted by climate models (and we should remember there are many many types of models and not generalise too much) such as the acceleration in Arctic ice loss and in particular the loss of 2007, which it can be argued is a result of localised combination of “weather” events superimposed on background “climate” warming. There is recent work with higher resolution models which provides insight. There are similar stories to tell about eddy resolving ocean circulation models, sometimes the theory is adequate already, but we need much higher resolution to successfully model or forecast unlikely events (or even realistic variability), such as extreme weather, and then any “knock on” effect that this subsequently may have.