This month’s open thread. In honour of today’s New York Marathon, we are expecting the fastest of you to read and digest the final IPCC Synthesis report in sub-3 hours. For those who didn’t keep up with the IPCC training regime, the Summary for Policy Makers provides a more accessible target.
Also in the news, follow #ArcticCircle2014 for some great info on the Arctic Circle meeting in Iceland.
#390 and #393.
Thanks so much Kevin, for your very thorough, and respectful, responses. That’s much appreciated. I do realize that linear regression is more sophisticated than simply identifying the “anchors” or the end points, but, as you clearly recognize, the choice of these points does make a difference.
“But the important thing that you *don’t* do is to pick your analysis period based on the result that you ‘want’ to see. That is the very essence of the ‘cherry-pick.’”
I know I’ve used that term myself, and I apologize. I prefer “confirmation bias” to “cherry picking” because in most cases I don’t see any deliberate attempt to deceive. As I’ve said before, it’s just very easy to fall into confirmation bias, even with the most honorable intentions. It’s not so much what you WANT to see, as HOW you see the data in the first place, what you get into the habit of looking for, that creates the problem.
“That’s also the tip-off that the ‘stopped warming in ’98’ meme is not motivated by a search for the truth, but by the desire to ‘debate.'”
Well, I’m a debater for sure, ever since high school, when our team won several championships. I originally thought I wanted to be a lawyer, but thank God that never happened. I understand why that ’98 data point has become controversial and am willing to accept that it could be due to an extreme El Nino event, which certainly could serve to distort the overall picture. For me, however, it’s not so much the claim that warming stopped that impresses me, as the breakdown, beginning roughly in that same year, of what had previously been a very strong correlation between CO2 emissions and temperatures.
Some graphs show continued warming, some show no trend (even after ’98) and some even suggest a cooling trend. What all have in common, however, is the breakdown in that all important correlation. With CO2 emissions continuing to soar one would expect temperatures to soar along with them — and for me that’s the crux of the problem.
Thanks for your explanation of the curving lines, which did puzzle me to some extent. That helps.
I guess my biggest problem may not be with the science itself so much, as with the way I see it presented in so many media reports. What laymen like myself see in those graphs is basically the trend lines, the straight ones, and it’s all too easy to jump to conclusions about what they mean. Then the deniers come along with their own graphs and their own trend lines, which are of course completely different. While you guys are disturbed by what you see as distortions produced by the “deniers,” what bothers me most is the attempt to snow everyone with “scientific evidence” that often looks questionable or even dubious to me. And the latter strikes me as more dangerous because it’s supported by very powerful government officials and very influential individuals. I’m not one of those people paranoid about “big government,” not at all. But I am concerned that in this particular case all the weight of big government has been placed on only one side of the debate, with the other side too often being dismissed. I have no sympathy for the typical “denier,” who accuses climate scientists of being motivated by grant money, etc., which is absurd. But I do think there is room for legitimate debate on this issue.
I’ll add that in my own research, oriented toward the social sciences and semiotics, I’ve time and again had to deal with claims, often from the realm of cognitive science, which I’ve found to be extremely questionable if not obviously wrong (think of the “Mozart effect”), yet in many cases backed up by the most incredibly convoluted math I’ve ever seen, outside of the physics literature. It’s experiences of that sort which have made me especially skeptical regarding the use, and misuse, of statistics. On the other hand, I’ve worked with statisticians and certainly recognize its value.
Enough for now.
I wonder if we can make any projections about this new discovery “Star Trek-like invisible shield found thousands of miles above Earth” http://www.astronomy.com/news/2014/12/star-trek-like-invisible-shield-found-thousands-of-miles-above-earth in light of climate change/ changing atmospheric compositions.
Victor @394: what “logical” analysis did you apply? I am seriously interested to know what analysis you applied. I know the analysis Rayne applied: the Mark IV eyeball. As everyone knows, that is not a “logical” analysis, but a subjective analysis.
Our “failure to accept Rayne’s very straightforward analysis” shows that we are skeptical – in the scientific sense of the word. We would like to know what criteria were used, and how those warrant the conclusion there are (at least) two different trends. Note in this respect that you would also need to show how different those trends really are, i.e., include confidence intervals.
You claimed you were not a troll. Well, here’s your chance to show it. Unfamiliarity with statistical analysis is not a good excuse anymore, since you have then forfeited your claim to have done a “logical analysis”.
Rafael Molina Navas, Madrid @389.
Of course, to fix Paul Berberich’s model with respect to the modelling of historical temperatures is quite a trivial task. You would simply choose a more appropriate mathematical form for his underlying trend. Thus you would use some function similar to Zhou & Tung (graphed here). But from there, this approach becomes impossibly difficult.
One part of the analysis missed by KK Tung is analysis of his ‘residuals’, that is the noise and rubbish left over after the mathmatical fit. KK Tung’s ‘residuals’ are packed full of wobbles that remain as unexplained as the ‘system’ KKTung models.
[The red trace on this graph plots KKTung’s input data into his wobble analysis and the blue trace is the ‘residuals’. The blue remains half the ampitude of the red which brings the whole analysis into serious doubt without serious consideration. (As an analogy, consider how easy anagrams become when you are allowed leftover letters.) Yet KK Tung dismisses the blue trace as “white noise”.]
And it is also wrong to ignore the known forcings (including natural ones) acting on the climate. CO2 is the daddy but he is patriarch to a large extended family. And further, it is wrong to ignore the composition of the wobble in global average temperature. When you consider these two factors, if you account in some manner for the forcings and the regional inputs into the wobbles, the remaining wobbles are regionally and temporally unique. Can there be any thought of a recurring cycle in such a situation?
Thus given there is pretty-much zero basis for the actual existence of a 60-odd-year wobble, it would be very ill-advised to begin serious speculation about the effects of cycles with that sort of period out there in the solar system.
But non-seriously, would the gravitational effects be strong enough to power a 60-year wobble? And if so, how would it evidence itself other than this speculative impact on global average temperature? I think here lies a sort-of locked room murder mystery, something a little too complicated and intricate in form to be given much credibility.
> Victor …
> impresses me …
> the breakdown, beginning roughly in that same year, of what
> had previously been a very strong correlation between CO2
> emissions and temperatures.
Oh, hell, that wasn’t “strong” nor year by year, and didn’t “break”
— emission isn’t atmospheric concentration;
— “temperatures” are many different measures;
— a stead increase in input (CO2) into a complicated multifactor system (climate, ocean, atmosphere, biosphere) changes everything, not a simple highly correlated annual average change in one measure (temperature).
Read the first link under “Science Links” in the right sidebar:
Spencer Weart’s ‘Discovery of Global Warming’
Do you get impatient when people come to you, in whatever field you’re an expert, with certainties they got off of daytime TV or nighttime AM talk radio? What do you tell people who fall for this kind of stuff?
Victor, you believed an old familiar strawman claim.
That’s often rebunked by people who write blog-fake-science.
You know how to look this stuff up.
You never bothered.
You believed the strawman claim.
Then you believed her claim that she knocked it down
You tell us how that impressed you.
https://www.google.com/search?q=correlation+between+CO2+emissions+and+temperatures
Okay, you’ve established you’ve been fooled by commonly rebunked stuff.
You’ve proclaimed your happy ignorance of statistics.
You haven’t sorted out who you can rely on, online.
You say you’re an academic.
You have access to a science department library? Or are you retired?
Have you any way to find a science librarian to ask for help with this?
Now what?
#396–The comment “There are now 14 papers in the peer reviewed journals recalculating Climate Sensitivity Down . Some as low as .30C by 2100, indistinguishable from Natural Variability…”
…pretty much shows the guy is an idiot parroting stuff he doesn’t understand, since comparing climate sensitivity to natural variability of temperature makes no sense. (One is temperature response to a given forcing, one is variation of temperature. And he doesn’t define either term… You might try asking him to define ‘natural variability’–while sweetly pointing out to him that the latter could include both hothouse Earths which extinguished most life, and glaciations which would be a severe challenge to any sort of civilization.
And no, there’s no-one calculating sensitivity of .3 C. Even Craig Loehle, who’s been Quixotically tilting at ‘warmist’ windmills for decades, finds:
As to the MWP stuff, ask him for evidence. It’s certainly not the story found in this 2011 study:
That’s here: http://eprints.uni-kiel.de/11820/
I found that through a simple Google search for “ocean temperature medieval climate optimum.” When the search came up, I clicked on the ‘scholarly articles’ link.
His information on the ‘hottest month’ meme probably comes from here:
http://www.climatechangedispatch.com/noaa-s-hottest-month-claims-are-unscientific.html
Note that nowhere does it say that NOAA admitted that anything was wrong; he apparently made that up out of whole cloth. Even the article doesn’t claim that it was actually wrong; it merely points out that there is a considerable margin of error. And the use it makes of that–to create a specious “8-way tie” is wrong, since the other 7 months have the same margin of error.
As to ‘the satellite record’, there’s not much distinction between 3rd-warmest and warmest; both UAH and NOAA agree that May was a very hot month. Moreover, UAH for October *was* in fact the warmest October in that record, as can be confirmed at the UAH web site:
http://www.nsstc.uah.edu/data/msu/t2lt/uahncdc_lt_5.6.txt
Well, OK, it really *was* in a tie this time–with 2012. But I doubt your opponent will gain much comfort from that.
Oh, this WUWT piece could be the source on low climate sensitivity. If so, he bungled it, since the ‘new blockbuster’ finds .43 C as climate sensitivity. And WUWT seems to bungle, too, calling a study using an ‘advanced two-layer climate model’ “empirically-based.” Probably another fringe effort, but I don’t have time to look at it right now.
Shorter:
> Victor …
> the breakdown in that all important correlation. With CO2 emissions
> continuing to soar one would expect temperatures to soar along with them
– See more at: https://www.realclimate.org/index.php/archives/2014/11/unforced-variations-nov-2014/comment-page-9/#comment-619554
Strawman. Neither emission, nor amount in the air, not _closely_ nor _annually_
http://www.skepticalscience.com/soares-correlation-co2-temperature.htm
> any projections about this new discovery “Star Trek-like invisible shield –
I project you’ll see mention at Climate Etc. as a newly discovered unknown unknown, e.g. see pp. 159-160 of the APS transcript
#403 Marco “I am seriously interested to know what analysis you applied. . . Unfamiliarity with statistical analysis is not a good excuse anymore, since you have then forfeited your claim to have done a “logical analysis”.
Well, first of all, despite what I might have said earlier in the interest of brevity, I’m not really “unfamiliar” with statistical analysis, since I’ve worked with statisticians and evaluated statistical results for many years during the course of my research. It would be more accurate to say that I’m unfamiliar with certain technical details of statistical analysis and can apply only the simplest methods myself. In short, “I have always relied on the kindness of strangers.”
Now to the question of logic “vs.” statistics (and I realize they are not mutually exclusive). The best example I can come up with relates to a theory promoted by a colleague of mine. Based on a statistical analysis of two databases, he discovered a correlation between the treatment of women and certain types of behavior. And since his sampling was worldwide, he concluded that this correlation represented a sociocultural universal: societies where women are clearly relegated to a subservient role tend to exhibit certain types of behavior. The implication being that the treatment of women produces the type of society where these behavioral traits are likely to develop.
His first error, of course, was the assumption that correlation implies causation, but what’s relevant here is his second error. In order to meaningfully argue that one has discovered a universal, one needs to determine not only a clear correlation based on a worldwide sample, but to also demonstrate that the same correlation can be found in each and every region represented in that sample.
Based on my knowledge of the ethnography, I could tell right away that something was wrong, because there was in fact a very large world region where the correlation simply didn’t hold up. I didn’t need statistics to tell me that. A general knowledge of the database (which I had helped to produce) combined with an extensive knowledge of the ethnographic literature was all that was necessary. Despite the sophistication of the statistical methodology he used (aided by a professional, as he himself was no more of a statistician than I), it wasn’t difficult to spot the logical flaw in his methodology.
As I knew very well, the sample from the region that didn’t fit was considerably smaller than the other samples and was thus overwhelmed by the other samples when the correlation was produced. His correlation was indeed significant, in both the statistical and cultural sense, but applicable only in certain societies which in fact were, for the most part, historically related. When assessing the cause of the behaviors in question, it would seem most likely that both the treatment of women AND the associated behaviors were caused by a third factor, based on traditions deeply embedded in the history of all those groups where the correlation was found to be strong.
How does this story apply to the question at hand? Well, first of all it’s an example of how a logical evaluation can cut more deeply than a merely statistical analysis. Which is not to say that statistics aren’t important — they certainly are. But NOT as a substitute for critical analytic thinking. It’s also an example of how focusing on the “big picture,” and dismissing all the smaller elements that make it up as mere “noise,” can produce a misleading result. To return to the lake effect graph in question, the overall trend displayed in the original paper was based on an assumption very much like that of my friend, i.e., the assumption that what works for the whole also works for the parts. Not always.
In my next post I’ll deal with that graph in more detail, to more fully explain why I see it as misleading.
MARogers #382
I have tried several fitting functions (For details download KlimaGlobal.zip from my Website). The parabola+sine worked best. Going 60 years back into the past, the agreement between data and forecast was still within of +- 0.1 °C (See http://www.gigapico.de/Fig1.html ). I don’t want to discuss the results, because you can find many discussions of this in the literature. My aim was to get better estimates of future global temperatures – for my grand children.