Dr. Roy Spencer evaporates Tamino's critique

Joel (05:11:21) The legal requirements about documentation and retention of intellectual property nearly become irrelevant when we are talking about policy decisions of this magnitude. In this sort of circumstance, the motive for not following the moral requirement of complete and utter disclosure and openness becomes germane. Why does Mann persist in the manner he does? Watch the last tape from Rhode Island. Something is not right.
===========================================

“I also don’t think Spencer’s short response to Tamino’s critique constitutes anything like a complete rebuttal.”—What??? Tamino has shown nothing–nada. His math does not apply–he wasted his time. I am still wating for the analysis showing a time varying forcing function.

Bob B (07:51:59) That’s why I suggested Tamino might be breaking pencils over there. Not in anger but in concentration. Too bad he can’t work with Spencer.
=================================

That the GCM models cannot be correct in “projecting” far into the future is inevitable.
1) The bands around each “projection” are not error bands. They are estimates of bands for what errors should be according to the feelings of the modeler. This is written in black and white in the AR4 report. If true errors are used the “projections” will be all over the temperature space. Have a go at this toy model from Junkscience.com
http://junkscience.com/Greenhouse/Earth_temp.html .
If you change the albedo by 3% the temperature changes by 1 degree. Can anybody insist that the albedo is known to 3% 1 sigma error?
2) And more importantly, what I have said several times here, is that the GCMs are using linear approximations of the putative solutions of the many coupled differential equations that control the thermodynamic engine that is the climate/weather.
Linear approximations work up to a point. The GCM models when used for weather predictions show us up to what point they do work: some days, i.e. an N number of time steps through the code. When these programs are transformed into climate models, i.e. more linear approximations (averages) the effect does not disappear. It changes scale, but still there will be an N1 number of steps after which the non linearity of the true solutions will appear in force.
Even if all the differential equations that apply have been taken into account, (which is not true) still linear approximations will inevitably fail.
—
The simplest point of clarification though comes from the errors. It is inconceivable that the true errors are not plotted for the “projections” so that the hoi polloi can see what nonsense is being pushed down their throats at the expense of trillions and million of lives.
The AR4 waffles on the errors:
Let me repeat the direct quote from the AR literature, from chapter 8 that is supposed to evaluate the models:
” 8.1.2.2 Metrics of Model Reliability from the AR:
“The above studies show promise
that quantitative metrics for the likelihood of model projections
may be developed, but because the development of robust
metrics is still at an early stage, the model evaluations presented
in this chapter are based primarily on experience and physical
reasoning, as has been the norm in the past.””
We should insist on seeing true errors on the “projections”. That is where statisticians with their hands on a model can come in strong.
The toy climate model above is frightening, in that it shows on what flimsy foundations the world is asked to destroy its economies ( though it seems we are on the way of doing that without the help of GCMs).

kim (07:52:49) :
“anna v. (22:09:00) If the globe’s climate regulating mechanism is a gigantic analog computer processing all the inputs, can it be modeled successfully on a smaller scale?”
I do not know. It seems that an analogue computer might do the job. Analogue computers are computers that use electric ( resistances, inductors and capacitors) elements in a net simulating coupled differential equations, and were invented in parallel with digital computers back then. They are orders of magnitude faster than decimal computers in solving differential equations ( look it up). In the beginnings, I met them in the 1960s, they were a strong part of research, but then the all purpose utility of the digital won out.
I do not know whether any research is going on still, what with chaos and complexity problems.