how is it possible for climate models to be as reliable as the models used in other branches of science?
In other branches of science, through observation and trial and error one can refine and reduce the uncertainties. In climate science it is acknowledged that the observations of our climate system are inadequate. Trenberth called it a "travesty".
He thinks it is a travesty that we lack proper observations, not that he thinks that the models are inadequate.
We used models to gauge the structural integrity of houses at my last job. They knew every detail and it could be very accurately estimated when a particular structure would fail. When they do that for more complicated systems like like airplanes, they often fail in the real world and it requires teams of engineers to correct whatever flaw existed. They have gotten pretty good at over the years but testing the structure of a wing is very simple when you consider the number of unknowns is very low. That isn’t at all the case with climate. It is pretty silly to think you can model the climate when you can’t accurately model the cloud formation, ocean currents, sun spots, volcanoes, CO2 output 10 years down the line, water feedback mechanism and other significant contributions. Them pretending that the climate is purely governed by CO2 content is obviously obfuscation of that undeniable fact. Then they used the climate models as gospel. They deserve the derision that they get.
There not because they have built in to them the fact that it is going to warm and no other outcomes.
most models make assumptions. for example, one poster said,
<<We used models to gauge the structural integrity of houses at my last job. They knew every detail and it could be very accurately estimated when a particular structure would fail.>>
consider, when modeling a house for example, do you model the nails that didn’t go in straight and where hammered over? Do you model the stud that’s partially cracked? Do you model the joist that wasn’t cut right but was used anyway?
Obviously not. But the "average" is still good enough to be used to predict how the house will perform. Over time, the house models get better and better. They’re not using the same one today that they used 10 years ago. Today’s is better.
Climate is the same, although several orders of magnitude more complicated. However, the computers that are available for testing climate models are some orders of magnitude more powerful than those that test construction models.
Quite clearly there are those who will point to just about anything and say, "Well that’s not good enough." The tactic is to delay any action, when clearly action is justified at this point. There’s something like 97% agreement among climate scientists on what’s happening with the environment. For such a complicated system, with so much money being spent to halt any action, that’s an extremely high threshold.
If this were a situation where one were deciding whether to go purchase automobile insurance, no-one would think twice. You’d be looking for the best insurance you could find, as soon as you could find it.
Clearly it would be nice if there were more data, and more computer power, and more reporting stations, and many centuries of observations. However, to point to those and say that we need to wait until all that becomes available is nonsense. The problem is clear, and the fact that it’s getting exponentially worse is also clear. Any delay now is just stealing from our children and grandchildren. It is morally indefensible.
Models are only as good as the information feed into them ! If a bit of info is added that should not be or a bit of info is not added that should be added then the entire model is worthless and will only show that which was provided.
I once took the time to read thru many pages of the results of a climate model, I will never waste the time again as I realized that they never added in earths changes of tilt, changes to the universe (sun getting hotter and larger or the moons position changes, etc).
Models only show what they are fed, without all info being used then a model is junk science at best !
The climate is an extremely complex system – remember the Chaos Theory example "something as small as the flutter of a butterfly’s wing can ultimately cause a typhoon halfway around the world". The climate model is improving, but it still ignores many factors and contains large numbers of assumptions.
An important factor in improving a model is feedback. Predictions are tested against results and then the model can be adjusted. We have not had enough time to do this.
If we understood the climate, we would have reliable weather forecasts for days, weeks or even months ahead.
We will not have a reliable model for
>"In other branches of science, through observation and trial and error one can refine and reduce the uncertainties"
And that is exactly how climate science has proceeded.
>"In climate science it is acknowledged that the observations of our climate system are inadequate."
No, it is acknowledged that we do not control 100% of the variance. Scientific theories of the physical universe and mathematical theories of probability and statistical inference all support the "adequacy" of our observations and the reliability of our models.
Here’s what Feynman said about climate models:
"If you’ve ever programmed computers you can imagine the development of those models. You write a program to simulate the climate. You run it the first time and it shows that the Earth is a frozen ball of ice right now. "Oops. That’s not right." Then you fiddle with it so it’s warmer and run it again. This time it predicts the planet will get so hot the oceans boil away next week. "Oops. Now it’s too hot." And so you tweak it until it gives you the answer you expect, the answer that conforms to your current set of prejudices and the current publishing environment.
That can be published in scientific Journals, but that’s not science."
All Black,
That is not a quote from Feynman, it is a quote from Les Jones, a blogger. Jones is trying to interpret something that Feynman has said, but it is not a Feynman quote.
Jeez.
Edit: Meadow, if you can find a quote where Feynman talks about computer models, I would love to see it. The quote in question is a blogger trying to make an association about a problem with the Mulliken experiment to climate models. Let’s not attribute something to Feynman that he did not say, just because some blogger wants to associate climate models with some unrelated Feynman quote.
@ Beren:
Maybe, but the point is still valid. Feynman was right to point out that models – unsurprisingly – model. They don’t have, in and of themselves, predictive capabilities. They can only take inputs and generate an output according to the rules that constitute them.
Feynman’s point was that this is not science – which is isn’t.
I don’t think anything is impossible. In this case it just hasn’t been achieved yet. The problem is that some people pretend that it has – either because we want that to be the case or because we want to pretend that the case for man-made global warming is as certain as the case for man-made acid rain.
I don’t think you really mean "reliable," I think you mean "accurate." A model could easily be reliable and completely inaccurate, and an accurate model could be unreliable.
To answer what you should have asked, how is it possible for climate models to be as accurate as the models used in other branches of science, the answer is that the question is ill-conceived: climate models should be compared against other climate models, not against models from other branches of science. If I have a model of a transistor and I can estimate the power consumption to 10%, and have a climate model that can estimate sea level pressure to 5%, does that make the climate model more accurate? I would say absolutely not, estimating sea level pressure to 5% is downright lousy, but knowing power consumption to 10% is not really all that bad. So when you talk about the accuracy of models I think it is foolish to try and compare between different fields, since it is impossible (or at least extremely difficult) to find objective criteria that are useful.
However, there are ways to test any sort of model, and it is important for modelers to test the models that they create. More observations are always better, although I should point out that when deniers speak of all the money being spent on "global warming" they are mostly talking about money being spent on improving the observation platforms, so if you want better data you should not advocate cutting funding for atmospheric research.
If you want to talk about real travesties, I think it is a travesty that deniers invoke the name of Feynman, one of my personal heroes, and then attribute blogger quotes to him–and then even support the misattribution when it is pointed out!