Climate Model Uncertainty

Whatever I read I stick to my basic premise - we (all the billions of us) are not doing nice things to our one & only world. That we have to make less of a mess is only common sense.

 

I also happen to believe we are a major cause of climate change. 

 

I don't know enough about it to argue about stats but I learned very well over my working days that you can prove or disprove most anything if you twist and turn enough. Some people will always believe you. Or to paraphrase PT Barnum, "there's one born every minute".

The article reads a little Spencery, but I made it through the whole thing. The main point, in case someone doesn't feel like reading the entirety of it, is that "uncertainties in the energetic responses of Earth climate systems are more than 10 times larger than the entire energetic effect of increased CO2"

 

Here's how he got that: 

 

 

This 10.1% cloud uncertainty number is based on a study that examined the ability of climate models to match a 7 year (1983-1990) average in actual global cloud cover percentage, which is illustrated graphically here.

 

He then says this, which doesn't really make sense to me: 

 

That being said, he makes his own graph of climate model uncertainty, with the ridiculous temperature range of -120 C to 120 C by 2100.

 

Which he explains: 

 

Anyway, it seems obvious that he is making some horrible assumption somewhere but I can't quite figure out where. The fact that he continuously adds the errors so that they become so wildly wide seems to be a problem.

 

I think this may have something to do with it: 

 

Apparently he's forgetting that these "quantillion-watt feedbacks" have more or less been cancelling each other out for thousands of years.

Edited by dawei - 6/8/2009 at 10:39 pm GMT

Well I didn't read the thing very carefully - it takes the guy a long time to get to the point.  But it does seem to hinge on that first claim

 

 

I don't understand what point he's trying to make here.  The energy impact from clouds may be -27 W/m2, but that's not the uncertainty in the radiative forcing.  The cloud radiative forcing is about -0.7 W/m2.  You can't just take the uncertainty in the energy impact of clouds and compare it to the total radiative forcing.  I mean, total solar irradiance is 1366 W/m2, but that doesn't mean if there was a 10% uncertainty we wouldn't know the radiative forcing to better than +/-137 W/m2.

 

Either I'm misunderstanding his argument or it doesn't make a lick of sense.

 

There was a long exchange between Gavin Schmidt and Frank at RealClimate on this. Gavin handed Frank his head.  Frank is, as Schmidt puts it, naive beyond belief.  Anyway, here's the link to the article, the exchange between Frank and Schmidt is in the discussion.  Enjoy

 

http://www.realclimate.org/index.php/archives/2008/05/what-the-ipcc-models-really-say

The iron ore reduction process (Midrex Process) I worked with for many years serves as an example for climate understanding I believe.

 

The process was developed by lab engineering with lab scale models and then semi-commercial pilot plants and later commercial plants. The process was only a commercial success after maybe 10 years in operation at several locations. 

 

What works on paper may or may not work in the lab - same going from the lab to pilot plants - same from pilot plants to commercial operations.

 

Calculations only get you so far and with process engineering (or even more so for climate science) there are too many variables where there is only a vague understanding of what is happening - many reactions happen without anyone really being aware of what is going on. In the mainly H2+CO (with a little CH4, N2, CO2) atmosphere in the Midrex furnace there are different conditions in different zones of the furnace and many different conditions within each of the zones. An iron ore pellet is maybe 15 mm in diameter - the conditions on the surface and the in the center of the pellets is different. The furnace may hold 1000 mt of the pellets.

 

In the iron ore reduction plants you have a given set of parameters which are considered safe and productive. The plant engineering staffs try to tweak those parameters to make small (and occasionally large) increases in productivity and cost effectiveness.

 

After 40 years of operations no one except a pure fool would claim to know 100% of how the catalytic reactions take place in the reformer or reducing reactions in the furnace work. There are a few fools that do make the claim though.

 

The Midrex process is far more simple than the climate - to me there is no chance it will be fully understood. We have to find the parameters which we can have some affect on (CO2, CFC, CH4, carbon particle, SO2, NOX for example) and work at controlling/reducing those. We need to implement beneficial programs aimed in the 'right' direction.

 

It makes little or no difference how the thing really works - we need to drive it in the right direction. 

I really should have clarified one point - I mean this may apply for myself and those like me.

 

For the ones who are really interested in and enjoy the calculations like Dana, Dawei & others - have fun but firm, undisputable answers are few and far between.

 

A very high percentage of the population believes that the underground temperature at 30 meters is uniform around the world which is quite incorrect and that one is easy to come up with. 

Well I remember this paper for instance that showed that cooling trends for as long as 20 years can be observed under single runs of the models. Almost any model will give these long cooling trends, but the reason that deniers squawk that "no models predict cooling, ever" is because these cooling trends happen at different times with different models, and so they average out to give a steady warming every year. (That's why the blogs always show IPCC averages of several models, never a single model, when comparing prediction to observation.)

Of course, the actual climate does not repeat itself hundreds of times and take an average, it only goes through one single 'run', and so if the models are right then minor cooling trends should be expected during this real world 'run' just as they are in most simulated runs.

Yeah the "models didn't predict the 'global cooling'" refrain is just dumb.  Climate models don't make short-term predictions, they predict long-term trends.  In the short-term, ENSO plays a big role but is highly unpredictable over the long-term.  ENSO modelers can't even very accurately predict what ENSO will do a few months ahead of time, let alone years.

The Hadley Centre recently combined climatology with meteorology to predict global temperatures over the next few years, and said we'd see accelerated warming after about 2010.  Another more recent study came to very similar conclusions - that in the next few years, global temperatures will rise rapidly.

But these sorts of short-term global temperature predictions weren't made 10 years ago.