Is climate linear or non-linear? As she says, this is the heart of the scientific debate. But even if it can be modeled as linear, we still don’t understand enough about the interactions to model it with confidence.
22 thoughts on “Simplicity On Complexity”
Comments are closed.
You’d think the last 15 or so years would give the linear model side great pause. I suppose it would nice to have nature simplify your task – the real world tends to not do that for you, however. Non-linear dynamics is just plain hard.
I suppose I’m just as guilty of being simplistic, but I really don’t take the AGW too seriously. If CO2 was really that much of a driving force (with positive feedback mechanisms), we’d have long since gone the way of Venus.
That it is nonlinear is not in question – T^4, after all, ain’t linear. The question is more linearizability.
And, I don’t that is so much the question as having the right linearized model.
I was giving the sensitivity question some thought this weekend, and came up with what I think are some interesting conclusions. I posted these thoughts here for anyone who is interested. I think the comparison to an automobile cooling system is fairly accessible and straightforward.
That was a very interesting comment. 🙂
Yeah, a rifle firing a bullet can be thought of as going straight for short distances with an inverse squared force, but T^4 when people are talking about raising T 1% or more and the amount of forcing we’re talking about is to drop the emissivity by 4% when all of human heat production is down around 0.01% of solar heating of the Earth requires extraordinary proof.
“we still don’t understand enough about the interactions to model it with confidence.”
Do we know enough to call people who study climate “academic frauds”?
No. You moron.
Doesn’t seem to stop you from referring to Mann as an Academic Fraud.
I have never referred to Mann as an “Academic Fraud,” you moron.
I once was stopped from referring to Mann as an academic fraud when I got distracted by an attack by a raccoon that carried rabies, but I beat the animal over the head with a hoe as it swam madly around in my pond (sadly breaking the hoe with the violence of my counter-attack). That done, I went back to referring to Mann as an academic fraud, because some truths are just undeniable no matter how many rabid animals attack.
We do know that the models are not backed up by real world observations. They have no predictive value.
Apparently, there still exists some people who believe that if it comes from a computer, it must be right. Poor deluded fools.
It’s worse than that. Today models can really only be invalidated. That’s because the data is so bad and so short there’s just not enough of it to justifiably validate any model. Proxies are crap. Surface temp data is crap. Sea-level data is crap. And so on. We have a narrow window of data we can actually trust, and it has huge gaps geographically and for certain parameters.
Anyone who thinks we could possibly model a system as complex as the Earth’s climaate using only temperature data is delusional. And even temperature data is crap. The degree to which climate scientist put forward this perception of certitude is appaling. What’s worse is that the crap models don’t even produce predictions within the error bars of the crap data and even then the climatologist hesitate at admitting their models aren’t accurate.
The fact that the models include cycles (such as the earth’s distance from the sun following a yearly cycle or the 11 year sunspot cycle) is enough to indicate nonlinearity. Obviously.
Weather and climate manifest themselves to us as the motion and temperature (and phase, in the case of water) of the atmosphere and the water on the planet. These two substances constitute the main energy and biomass transport mechanisms on the planet (radiation being the main mechanism for energy in and out). The best tools we have to predict the motion of fluids are the Navier Stokes equations, a set of coupled, highly non-linear partial differential equations. Linearized solutions may be found for simple boundary conditions, ideal fluids, and steady, laminar flow. None of those simplifications are available on earth, even as a crude approximation. So there isn’t any question that the problem is non-linear. A discussion like this occurring at this stage of climate “science” indicates to me that the climate model problem is so far over these people’s heads that they’ll never find the bottom of it.
yet the predictions made 20-40 years ago seem to be coming to pass.
Uh, no, sorry DN guy, but none of the predictions from 20 years ago have come to pass, unless you’re talking about the earlier predictions of a coming ice age due to mankind’s production of sulfates which will block out the sun.
The Navier Stokes equations are invalid if evaporation or condensation is occurring. I’ve seen some very impressive mathematical techniques to handle condensation in the final turbine stages of a steam ship, but so far nothing to indicate that climate modelers have a clue that to handle condensation roughly enough for a tramp steamer requires a ten or hundredfold step-up of their game. Weather is driven by the stuff in the roundoff error of normal aerodynamic equations, in a chaotic feedback cycle, and I don’t think many climate scientists aside from Judith Curry (who is a dreamboat) have a clue as to how badly the simplified flow models that are pretty valid for a few milliseconds of violent flow over an airfoil section will do when extend to whole seconds, much less minutes, hours, weeks, months, years, decades, and eons. All the action of importance is in the roundoff errors and invalid states of the usual equations, yet the people sucking down the funding are pointing at things airliners as proof that their science works, while putting total faith in the validity of a number decades out that any competent engineer would think was invalid by the time the airstream from the wing passes by the tail section – due to turbulence, non-laminar flow, condensation, eddy formation, and other factors.
I’ve seen lots of bad N-S runs, and yep, they sure can produce some crappy results.
however, even in the 1920’s , NACA was gathering simple empirical data by
modeling wing sections in small tunnels enough to handle basic aircraft design.
You want to avoid pushing Mach Number too hard or Reynolds number too hard,
but, with the data from that era we designed all the aircraft of WW2.
N-S lets you do elegant design, but, if you stick to simpler design approaches it will still fly.
My basic problem remains: the data comes in too slowly. Regional climate variables, no one claims that they can predict them – just too chaotic. Global climate variables, such as surface temperature, you get on the order of one data point per month. You can adjust all your models now to reflect the last fifteen years, and you should, but it’s going to take decades to validate them.
Note that all the suggestions like heat going into the deep oceans are not explanations why the models aren’t really failing. They are new models, and as such need to be validated. You can’t even try to back-validate them, as we don’t have any data for the last century on the temperature in the deep oceans.
It seems almost hopeless. I don’t blame the modelers, but I don’t see a good way forward. I know that some of them are trying to work on modeling regional values. There at least you get your new data at a reasonable rate. The system is known to be chaotic, you can’t predict weather very far forward at all, but maybe they can identify some regional variables that are actually predictable accurately, and see how far in the future they can push them.
I guess I would add that the fact that the current models do a pretty fair job of predicting the last century’s surface temperatures is actually a bad sign. When you see a model that predicts in-sample data well, and out-of-sample data very badly, you know that the model is over-fitted. It has been tuned to the in-sample data. I know that climate modelers like to say that the models are just physics, are not tuned, but that is shown not to be correct by the results: they began to diverge from real-world data just as soon as the models were fixed, and have continued to do so. Some modelers acknowledge this:
http://rankexploits.com/musings/2012/tuning-climate-models-a-group-discusses-how-its-done/
“Climate models ability to simulate the 20th century
temperature increase with fidelity has become
something of a show-stopper as a model unable to
reproduce the 20th century would probably not see
publication, and as such it has effectively lost its purpose
as a model quality measure. Most other observational
datasets sooner or later meet the same destiny, at least
beyond the first time they are applied for model evaluation.
That is not to say that climate models can be
readily adapted to fit any dataset, but once aware of the
data we will compare with model output and invariably
make decisions in the model development on the basis of
the results.”
“Note that all the suggestions like heat going into the deep oceans are not explanations why the models aren’t really failing. They are new models…”
Such an excellent observation. And just one other example of the sad state that climate science is in today. It’s OK to not understand the climate, it’s OK to still be learning. It’s not OK to pretend that the data is better than it is or that the models are good when they are just utter, utter crap.
At the end of the day the “models” end up being multi-variate regressions fitted to a subset of temperature data from the last half of the 20th century or so. And then cherry picked so that no “wrong” models which fail to predict increasing temperatures are left. It’s all just crap. Wall to wall crap. The amount of science in there is almost non existent. The amount of good data in there is equally hard to find. It’s no wonder the models don’t have predictive value. They don’t represent any significant degree of actual understanding of the workings of climate.
I’m aware of a couple very strongly non-linear climate effects:
– Ice formation and melting takes place at zero C, mostly in polar regions. As the poles cool you get more ice, which radiates heat away more slowly.
– In the last few months on Watts a detail of cloud formation over tropical oceans has been described where clouds form very quickly once the air reaches ~30 C, reflecting away further solar heating. Last I heard the major climate models can’t handle this kind of detail.
A couple strongly non-linear effects which work against large temperature deviations.