2011-07-03 05:55:52Lessons from Past Climate Predictions: Syun-Ichi Akasofu
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.140.0.15

Akasofu is next up in the series.

Lessons from Past Climate Predictions: Syun-Ichi Akasofu

2011-07-03 06:43:43
Alex C

coultera@umich...
67.149.101.148

The download link to the paper takes a tedious amount of time to finish - he did apparently publish a slimmer version in an obscure journal titled "Natural Science" here.  It's only been around, it seems, since 2009.

2011-07-03 06:52:19
Alex C

coultera@umich...
67.149.101.148

>>>It is a logical failure to assume that a warming over nearly two centuries must have the same physical cause throughout the 200 years.

Yeah, but that doesn't drive home the point.  Your ending note shouldn't be on "can't logically conclude that" - illogical predictions or statements can still be true.  Where he really went wrong was that this illogical prediction was not true because the evidence points in the completely opposite direction.

I wonder, since the choice of a trend completely determines the strength of the oscillation that Akasofu points out, as Riccardo also covered in his Astronomical Cycles post, as there is no physical mechanism Akasofu provides to actually identify what the best trend would be, other than "oh, it looks OK," would any further discussion on that be warranted?  Such as, fit a parabolic trend through the data, and a linear trend, calculate the residuals from each, and compare the results?  I would bet the cycle falls apart in magnitude depending on what curve you fit to it, exactly as in Riccardo's post.  If there's no physical basis given for a linear trend, then the trend choice is arbitrary, and if you're doing period analysis that is a completely fatal error.

2011-07-03 07:29:41
Alex C

coultera@umich...
67.149.101.148

If you would like to include any discussion on the trend choice, I just created some graphs in Excel using the global data from HadCRUT3v and could email you either a slimmer Excel file (as I have many more things graphed/charted) or a PDF with the graphs pertinent attached.

Edit: Hm, I don't know how to model equations from a [y = a(x - 1850)n - b] format.  I don't think Excel is capable of performing such a trend analysis...  I have polynomials, that's it.

2011-07-03 08:17:12
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.140.0.15

Ah thanks for the link to the published paper, Alex.  It seems he updated the actual prediction graph to include measured temps starting around 2010, as opposed to 2000, as in the longer, unpublished version.  Good thing we've got the unpublished version on the record :-)

Are you talking about making a graph something like Riccardo's Fig 2?

2011-07-03 08:19:51
Alex C

coultera@umich...
67.149.101.148

More like Figure 3, to compare the residuals.  Figure 2 would be necessary to show as well, I would presume.

Well heck you could just use Riccardo's figures, don't know why that slipped my mind.

2011-07-03 08:53:46
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.140.0.15

Yeah I'll add a discussion of trend selection and make use of Riccardo's figure(s).

2011-07-03 14:23:34
Alex C

coultera@umich...
67.149.101.148

>>>Riccardo found that an exponential trend fits the instrumental temperature data much better than a linear trend (Figure 3).

Uh, I don't think so.  The exponential trends weren't assumed to be better necessarily, just alternatives since the choice of trend was arbitrary as there was no discussion of physics.  From a mathematical perspective, you can choose any power magnitude and find a best fitting curve for that power, but the point of whether one fits better is not as important as the one that the residuals completely change depending on the trend, and thus the cycle(s) appear or disappear.  That's the whole point behind Figure 3 (you posted figure 2, I think): the cycle is only evident in n = 2, not n = 1 or n = 4.

This also just occurred to me, as I'm looking through Akasofu's paper again: he graphs HadCRUT3 data from 1880-on, not the 30 year period before as well.  So, trend comparisons using this data only would be best (as the trends are different including that earlier data).

Edit: *sigh* Riccardo graphed HadCRUT3 data, whereas Akasofu is graphing a smooted version of the data from NCDC here.  That would help to explain the different start times in the data sets; in either case, the same principle of residuals applies.

2011-07-03 14:31:13
Ari Jokimäki

arijmaki@yahoo...
91.154.101.246

It seems that you messed up the title to natural variability section quite badly.

There is possible relevant recent study (Rea et al. 2011) that suggests that Earth's climate doesn't revert back to some mean state.

2011-07-03 16:03:24
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.140.0.15
Alex - yep, I should have read Riccardo's post more carefully. Ari - it's called creative typing, thank you very much. Will have a look at the Rea paper, that sounds relevant.
2011-07-03 18:32:28
Riccardo

riccardoreitano@tiscali...
93.147.82.108

Replotting NCDC anomalies or residuals is a matter of minutes. Smoothing the data would make it more similar to Akasofu's graph.

I find your last sentence rather weak and too CO2-oriented; I'd be more generic. Underline that he assumed an unknown linear trend, an unknown periodic variability and that these two unknown phenomena will continue while disregarding what we know. There's really nothing we can learn from this process.

2011-07-04 00:54:52
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.140.0.15

Agreed Riccardo, good point. I've incorporated the suggested revisions.

2011-07-04 08:21:51
Alex C

coultera@umich...
67.149.101.148

I still think using the NCDC data would be best.  That is the data Akasofu uses, and the trends would be much closer to the approximation that Akasofu came up with (well, the linear one at least).  Also, to reiterate some points:

- it's not important that one trend fits better than another, as - for example - you can find always better fit a higher-order polynomic trend to any data set; without a physical justification for a trend, then the trend power is arbitrary, so what might appear as a cycle in Akasofu's data might disappear with a parabolic trend, or a higher order trend, or become more pronounced in one.  In any case, this demonstrates the necessity of having a physical justification for a trend choice, which was the whole point of Riccardo's post.

- the way you have it phrased now actually makes it seem like while you disagree with Akasofu's choice of a linear trend, you're also leaving it wide open to conclude that the cycle is still there, just with a power 2 trend.  That would be a wrong conclusion, as per my above comments.  In any case, I think this might change if you calculate the residuals for the NCDC data; perhaps linear does show a better cycle with NCDC, I don't know.

-snip-

2011-07-04 08:53:27
Alex C

coultera@umich...
67.149.101.148

Man this paper confuses me.

Are we relying on his "interpretation," this graph, for rebutting against?  That would make sense to me, but he says it's an interpretation of various other figures, such as 1a, 1b, 1d, 1e, and 1f.  He practically uses every data set: GISS, HadCRUT3, NCDC - makes me want to rip my hair out, I'm changing my mind so frequently about what to use....

OK, amended suggestion: anything 1880-on would suffice, as that is the period his "interpretation" extends.

2011-07-04 10:17:38
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.140.0.15

Okay Alex, I'll re-work that section.

The various temp data sets are so similar that I don't think switching from one to the other will make a significant difference.  I think Riccardo's original graph is fine - actually I prefer starting in 1850, because even though Akasofu's starts in 1880, he talks about the trend since 1825.

2011-07-04 12:24:30
Alex C

coultera@umich...
67.149.101.148

OK, that sounds fine.

2011-07-04 20:21:42
MarkR
Mark Richardson
m.t.richardson2@gmail...
192.171.166.144

Climate sensitivity is pretty tough to understand properly. I think it's worth putting it in simpler terms: Akasofu implies that extra heating from CO2 (which we've directly measured with satellites) isn't causing any warming. The heat somehow magically disappears: yet another breed of the 'Harry Potter arguments' so beloved by skeptics.

2011-07-04 20:34:00
nealjking

nealjking@gmail...
84.151.55.66

Seriously, what's wrong with this guy? Did he just go emeritus, or what?

2011-07-05 05:58:41
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.140.0.15

Okay, I made the revisions suggested by Alex and Mark.

neal - no idea, but he is old.  Plus he's a geophysicist in Alaska, so he may be very pro-fossil fuels.

2011-07-05 06:49:09
Alex C

coultera@umich...
67.149.101.148

I like that much better; a grammatical correction here:

"Riccardo fitted various trends to the global temperature data (Figure 3), and found that whether the residuals form a clear cycle depends on the choice of trend"

and here:

"The paper was  trimmed down and published in 2010 by a new journal called Natural Science,"

are my last suggestions, methinks.  Good to go.

Edit: Actually I still think a restating of how trend choice is arbitrary without physical justification one way or another would be warranted, perhaps just amended to this statement to read as:

"Akasofu does not have this justification, and without a physical reason the choice of trend is essentially arbitrary."

2011-07-05 09:04:10
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.140.0.15
Thanks Alex, will make those changes later tonight.
2011-07-05 13:14:47
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.140.0.15

Changes made.

I'm also wondering, is it worth turning these into rebuttals, like "Easterbrook predicted global cooling" and "Akasofu predicted linear warming"?  It might be beneficial if somebody is searching for a rebuttal to their papers, but then again, if a person were to type in "Akasofu" to the search bar, this post would pop up.

2011-07-05 14:21:08
Alex C

coultera@umich...
67.149.101.148

Perhaps.  I know from experience that at least one other blog is making this claim, SBVOR.  He's all over YouTube spouting this nonsense, as well as misrepresenting others' work, you know.  The whole schebang.

Careful though, this site is complete crap.  Pay attention to the article title too, and what ocean cycle Akasofu actually DOES talk about.  I'm not entirely sure if we can rebut a claim that doesn't even properly use its *bad* sources... is that possible?

Edit: and as I type right now, I think it's still yesterday in California, so happy Fourth.

Edit Edit: WUWT also threw out what appears to be Akasofu's op-ed here.  This appears to be a copy republished there, but I'm not sure where the original is.  Perhaps it's right under my nose, linked in the article...