2012-02-22 07:37:20Scafetta's Widget Problems
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Post on Scafetta's widget (comparing his prediction to IPCC's).  Should probably get this up before it's too old news.

Scafetta's Widget Problems

2012-02-23 04:57:31
Dikran Marsupial
Gavin Cawley
gcc@cmp.uea.ac...
139.222.14.107

Dana, looks good, just a few additional issues:

I'd replace

"If Scafetta wants to use the 1900-2000 baseline, an argument can certainly be made to do so, but then he should also show the data and models starting in 1900 in his widget, not 2000."

with

"If Scafetta wants to use the 1900-2000 baseline, an argument can certainly be made to do so, but he should also update the uncertainty of the model projections to account for the change in baseline (which was not done and is likely to further biases the test against the IPCC models)"

Essentially the real problem with changing the baseline is that it would increase the spread of the model runds and hence the uncertainty of the IPCC projections, which he didn't do.  A baseline period of 1980-1999 basically brings togethether the model runs as tightly as possible at the start of the projection, giving the narrowest range opf uncertainty (the skeptics wouldn't want to mention that point as it makes the model projection more easily falsified).  If a longer baseline is used, this will necessarily increase the variance in the 1980-99 period and so the 2 sigma error bars will be wider in 2000 than depicted using the original baseline.

"We have made these adjustments to Scafetta's widget in Figure 6 - showing both 1-sigma and 2-sigma IPCC uncertainties, using annual NCDC temperature data, and using a 1980-1999 baseline." (add "on which the IPCC projections are based", and change the "a" to a "the")

"And his model actually fits the post-2000 data more accurately without that arbitrary adjustment."

We really can't say this without the caveat that the difference is not statistically significant.  In fact it would be better if we didn't say it at all as I seem to be forever telling skeptics that decadal trends are unreliable, and this comparison is unreliable for exactly the same reason, so it would lay us (me) laible to accusation of double-standards.

2012-02-23 05:01:48
Dikran Marsupial
Gavin Cawley
gcc@cmp.uea.ac...
139.222.14.107

BTW still haven't heard any news from the editors.  If I have the time I'll start a short one page comment just on that diagram.

2012-03-13 03:50:38
KR

k-ryan@comcast...
216.185.0.2

Scafetta has updated his widget with the latest data, and attempted to respond to Dana's post. I can't say he succeeds, though...

http://wattsupwiththat.com/2012/03/11/scafetta-prediction-widget-update/

He flatly disagrees with the baseline alignment, the display of 2-sigma ranges, significance of HadCRUT4, and calls demands for back-projecting "evident that it is stupid, before than being dishonest, to extrapolate it for 2000 years and claim that our result is contradicted by the data" - as they were according to him not looking at millenial/longer term cycles.

Ptolemaic epicycles, anyone?

2012-03-13 04:21:15
Dikran Marsupial
Gavin Cawley
gcc@cmp.uea.ac...
139.222.14.107

I just posted this over a WUWT, inviting Scafetta to discuss the criticisms in depth, one by one, here at SKS.  Probability he will accept?  Small I would have thought.

Dear Prof. Scafetta. rather than try and estimate the error bars on the IPCC projections from a magnified diagram from the IPCC report which doesn’t have the resolution to give a reasonable estimate, why not do what I did and go and get the A1B model runs from the publically available archives and plot them, along with the temperature data? If you do, you will get an image like this one
http://www.skepticalscience.com/pics/sresA1B.png
which shows the IPCC model runs project that temperatures both warmer and colder than observed during the past decade. I note also that the error bars you have estimated from the IPCC diagram are for annual data, which has a substantially lower variance than the monthly data that you plot. I would be happy to discuss your criticisms in depth, one by one, over at Skeptical Science.

http://www.skepticalscience.com/scafetta-widget-problems.html

best regards

Dikran Marsupial

2012-03-13 04:32:11
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Basically he just repeats the same errors, doesn't really address any of our criticisms (though he pretends he does), tosses out a bunch of ad hominems, and makes some statistically idiotic claims.

Epic fail.

2012-03-13 04:46:18
Dikran Marsupial
Gavin Cawley
gcc@cmp.uea.ac...
139.222.14.107

I have sent a reminder to the editor and he has basically invited me (us Dana?) to submit a comment.  I am trying to argue that a very brief comment needs to appear very queickly addressing just that one diagram, leaving the other problems to another paper (the editor hinted that other groups have contacted him wanting to send in comments).

2012-03-13 04:48:54
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Thanks Dikran.  If you do get a response, I'd be happy to help with the comment in question.