Data Comparisons

January: Temperature Leaps to +0.51C!

97 Comments

Roy Posted the January UAH temperature anomaly. Wow! That’s some high jumping! Up to +0.506C from +0.206C! With that kind of start out of the gates, maybe temperature will break a record this year.

No one (sort of) guessed the temperature would rise so dramatically, so the three top winners are EdS who bet temperatures would rise to 0.45C, JohnNorris who bet 0.42C and MikeP who bet 0.41C. (I say sort of because there were two bets of 2.0 and 2.8C, I assumed those were data entry errors and corrected. Note: Because I forgot to change the “date” field, many bets did not display previously.)

Everyone’s bets are shown below.

Winnings in Quatloos for UAH TTL January, 2013 Predictions.
Rank Name Prediction (C) Bet Won
Gross Net
— Observed +0.506 (C)
1 EdS 0.45 5 84.751 79.751
2 JohnNorris 0.42 5 67.801 62.801
3 MikeP 0.41 5 37.949 32.949
4 GeorgeTobin 0.313 3 0 -3
5 denny 0.312 3 0 -3
6 DaveE 0.28 5 0 -5
7 pdm 0.277 5 0 -5
8 ArfurBryant 0.272 5 0 -5
9 TimTheToolMan 0.27 5 0 -5
10 RobertLeyland 0.262 4 0 -4
11 Lance 0.253 5 0 -5
12 Freezedried 0.24 4 0 -4
13 AMac 0.226 3 0 -3
14 ivp0 0.222 5 0 -5
15 BobW 0.216 3 0 -3
16 JohnFPittman 0.211 5 0 -5
17 TimW. 0.2 5 0 -5
18 EdForbes 0.2 5 0 -5
19 EdForbes 0.2 5 0 -5
20 RIHo08 0.199 3 0 -3
21 Ben 0.187 5 0 -5
22 Ray 0.185 5 0 -5
23 Cassanders 0.183 5 0 -5
24 ScottBasinger 0.18 5 0 -5
25 DeNihilist 0.18 5 0 -5
26 Anteros 0.177 5 0 -5
27 SteveT 0.17 3.5 0 -3.5
28 mwgrant 0.166 3 0 -3
29 Skeptikal 0.159 4 0 -4
30 dunna 0.157 3 0 -3
31 Pieter 0.157 5 0 -5
32 SteveF 0.143 4 0 -4
33 Tamara 0.142 5 0 -5
34 PavelPanenka 0.14 3 0 -3
35 March 0.123 5 0 -5
36 Perfekt 0.119 4 0 -4
37 CoRev 0.11 5 0 -5
38 DocMartyn 0.1 5 0 -5
39 KreKristiansen 0.07 4 0 -4
40 BobKoss 0.055 5 0 -5
41 Hal 0.05 5 0 -5
42 Pieter 0 1 0 -1

The net winnings for each member of the ensemble will be added to their accounts.

97 thoughts on “January: Temperature Leaps to +0.51C!”

  1. One can wager that if the pause ends, skeptics will shift arguments.
    With solar max approaching the next 5 years will prove interesting.

  4. steven mosher–
    Sure. I’m going to be watching to see if the annual average manages to bust through the multi-model mean.

  5. Looks like UAH dataset became the worst dataset. It’s very likely that none of other datasets will show similar jump. It’s also very likely that these values will significantly change in future with no clue, how and why these changes were made.

  6. Spencer has noted that other NOAA satellites were in line with Aqua, so I would expect RSS to be high as well. Also, Spencer posits a lower SST due to heat discharge to the atmosphere (like in an El Ni~o). Per Tisdale, this did happen.

    The thlot pickens.

    Lucia, all you have to do is plot your annual graph NOW using calendar year means instead of running means. Simple.

  8. The really interesting thing would be a combination of a solar max and a strong El Nino. That would help putting bounds on how much these two factors really affect year-to-year temperatures (and thus, how much of the current slowdown they explain).

    Unfortunately forecasters seem to predict a neutral ENSO for early 2013.

  9. Like last month, this post prompted me to do an early run of TempLS. In Dec there was a big dive, though moderated by later results. It went from 0.51 to 0.25. This month, based on very early data, it is back up to 0.4°C.

    In Dec the freeze was in Russia and NW N America. This month there was a reversal in NW America and warm in Central Asia and Australia.

  10. Weather ain’t climate, as we all know, but January had a massive Sudden Stratospheric Warming (SSW) event. Wonder whether that could have any effect, given what the satellite is measuring?

  11. cui bono–
    Maybe. But if almost 1/2 was some ‘reversion to (Month-2)’ as Nic suggests, it’s just the sort of thing you expect with ‘noise’. Since it’s January, we know this one contributes to the 2013 mean while December didn’t. That’s why I comment on the record. 🙂

  12. According to my look at Environment Canada data there was a big jump from Dec to Jan in Canada, but Jan 2013 is colder than Jan 2012.

    2012 Jan
    http://sunshinehours.files.wordpress.com/2013/02/ec-normals-0x0-2012-01.png

    2013 Jan
    http://sunshinehours.files.wordpress.com/2013/02/ec-normals-0x0-2013-01.png

    Dec 2012

    http://sunshinehours.files.wordpress.com/2013/02/ec-normals-0x0-2012-12.png

    What few stations there are that EC calculates anomalies for, the arctic was slightly colder than normal in Jan 2013.

  13. Is it possible that in December they underestimated polar warming, and in January they have underestimated polar cooling?

  15. Nick:

    Certainly possible, but there are big swings on well-measured land.

    It’s also important to note these are anomalized temperatures. If there’s a shift in seasonal patterns (e.g., due to the late refreezing of the Arctic), then that will show up as these sorts of apparent warming & cooling (but have no net effect on the global trend over > annual periods).

    But it’s an artifact in that case of not knowing the “new norm.” (And until the great Arctic Melt-Off [tm]” is finished, there may not even be a well-defined norm.)

  17. Sudden Stratospheric Warming event. Pretty large event but most influential for the high northern latitudes and not at all for the other latitudes.

    http://www.cpc.ncep.noaa.gov/products/stratosphere/strat-trop/gif_files/time_pres_TEMP_ANOM_JFM_NH_2013.gif

    These events build down to the lower atmosphere over the next one to two months. Some other location in the high latitudes gets the extreme cold instead. Looks like northern Asia got it this time (so far).

  19. Two first places and a second in 13 tries over 18 months. Significant? Six top tens over same period.

  20. Pay no attention to my performance for Feb 2012 and Aug 2012. Nothing to see there. Move along.

  21. “One can wager that if the pause ends, skeptics will shift arguments.”
    Yes, they could borrow an argument from the alarmists. They can call it a pause, from the pause.

    I was going to guess +.50, or was it minus. Anyway, I don’t have any quatloos.

  22. “One can wager that if the pause ends, skeptics will shift arguments.”

    Yes, they could borrow an argument from the alarmists. They can call it a pause, from the pause.
    @@@@@@@@

    no, they wont make that shift. Currently they are keeping track of the number of years in the pause for every time series. As it warms ( or rather when it warms) they will drop this line of argument and they will shift back to “who is cooking the books”
    Spenser makes it hard for them to do this across the board, cause they will never throw him under the bus.

    That part of the reason why I think the pause argument is dumb strategically.(trying to help you guys out ) If you base your doubt on it and it turns around..and you forget that you made that argument well thats UCT un reflexive counter factual thinking.( haha see lweandowsky’s latest ). Put another way, there are two ways to think about the pause or position the pause..

    A) it proves AGW false!
    B) hmm, maybe sensitivity is lower than thought.. the data makes me less certain. hmm make natural varition is higher than I thought.. hmm maybe there are other forcings… hmm

    If you think about the pause as in A, then you face potential UCT if the record turns around. If you look at the pause as B, just a another piece of evidence that shifts your certainty around, then you avoid taking a position you have to walk back from.

    Lets put it another way. Take all your “nail in the coffin arguments” and put them away. There are no nails.

  23. But note that Spencer says:

    “The most common cause of such warm spikes (when there is no El Nino to blame) is a temporary increase in convective heat transfer from the ocean to the atmosphere. This would suggest that the global average sea surface temperature anomaly might have actually cooled in January, but I have not checked to see if that is the case.”

    So, this could result in a net cooling, longer term.

    http://www.drroyspencer.com/2013/02/uah-global-temperature-update-for-january-2012-0-51-deg-c/

  24. Steven Mosher (Comment #109698)
    February 5th, 2013 at 8:50 pm
    “One can wager that if the pause ends, skeptics will shift arguments.”
    Yes, they could borrow an argument from the alarmists. They can call it a pause, from the pause.
    @@@@@@@@

    Actually, I was going to borrow from the alarmists’ arctic ice playbook and point to the 1998 + 0.66, 2010 + 0.58 and 2013 + 0.51 “death-spiral” of temperatures 😛

  26. p (Comment #109671),

    Looks like UAH dataset became the worst dataset. It’s very likely that none of other datasets will show similar jump.

    RSS has published their anomaly, and they DO show a similar jump.

    UAH lost a bit of credibility when the magnitude of the errors in v5.4 were revealed, but v5.5 appears to be a much better dataset. While the jump in the anomaly was surprising and certainly raised a few eyebrows, I think you were too quick to pass judgement on the UAH dataset… which now makes you look a bit foolish.

  28. pdjakow:

    In fact, the difference between these datasets are increasing, especially after 2005.

    Any difference is certainly not statistically significant. My judgement—

    you’ve lost it

    (or maybe you never had it.)

  29. Using the numbers from woodfortrees, here’s what I find for UAH-RSS.

    Isn’t the average difference between them getting smaller over time (closer to a zero anomaly), not larger?

  30. @Carrick
    I don’t think so. Average difference in last 36 months is largest on record. Last three months have difference ~ 0.2 deg C, largest on record.

    First 2-3 years in dataset also looks weird.

  31. What I don’t understand is when you look at UAH ch5 plots for January 2010 and January 2013 you can see that 2013 was warmer than 2010 every single day of the month, and by quite a margin.
    But then January 2013 gets reported as being less warm than January 2010.
    ?
    Can someone explain how that works?
    Thanks

  34. As it would appear that nobody guessed (sorry, I mean estimated) that the temp. would be as high as, or higher than it really was, I guess this proves that the so-called “luke-warmers” that frequent this site are really a bunch of den**rs!

    I can’t see an entry from Lucia – were you off the bottom?

  35. pdjakow, thanks… I admit I did it in a hurry. But it doesn’t look like shifting the baselines changes the interpretation very much (other than as you say the difference is has been growing since 2005).

    I think the case can be made that there is a long-term drift between the two series, but that’s almost certainly due to the difference in the way the two series “hand off” between satellites.
    Since you have two thermometers, so to speak, why do you pick one as right and the other as wrong?

    How would you go about testing whether this difference that bothers you is statistically significant? That’s the real test of course. Differences in central values are meaningless without an uncertainty estimate.

  36. RE: stevetta-uk (Comment #109715)

    Well, since global SSTs fell during Jan while UAH spiked it looks a lot more like short-term weather than climate. Don’t get it twisted.

  37. Mosh, how unlikely of you to get it so wrong (and to pass up an opportunity for the perfect bon mot).

    Don’t say there are no nails.

    Say there is no coffin.

  39. Steven, I didn’t say anything about nails and coffins. I was merely alluding to the fact that skeptics don’t have a corner on the use of dumb arguments. I am sure you can think of a few dumb alarmist arguments. Maybe there is another book in it for you.

    The “who is cooking the books” meme is actually an effective weapon for the skeptics to employ in their propaganda war against the often bogus propaganda of the alarmists. We already learned of some shady activities going on in the climate science, when we got that peak behind the curtain with the C’gate emails. Remember that, Steven? And the whitewash investigations?

    Oh, but we got honest people in our government and that FOIA thing. Some of the top officials who are protecting us from environmental doom are using alias email accounts to conduct our business, but they are not trying to hide anything. It would take a determined conspiracy involving a lot of people to cover up malfeasance with the temperature adjustments. Something like this:

    http://www.chicagotribune.com/services/newspaper/eedition/chi-liberty_tuesoct02,0,43090.story

    Most of our civilian citizens are clueless about this. Not that they would much care about the abandonment and dishonoring of a few dozen expendables, a long time ago. Ask them what they think about Benghazi, and they say “Ben who?” Someday they will look around for somebody to send in harm’s way, and nobody will step forward.

    Anyway, my point is that basically good people, who believe that they face an existential threat, will go so far as to commit mass murder against their nominal friends and steadfastly lie about it. And the leaders of the offended nation will bend over and cover it up, for political expediency.

    No, I am not accusing NCDC of mass murder. I am saying that working the distrust of government angle is effective for the climate skeptics. It’s a serviceable counterpoint to the alarmists’ Big Lie that Big Oil sends checks out to all the skeptics. I am getting more militant, since I looked at my recent PG&E bill to discover that under the CA left loonie progressive carbon tax scheme, we are paying 30 cents a kilowatt hour for a lot of our juice. They are trying to drive us “rich people” out of this state.

  40. With that kind of start out of the gates, maybe temperature will break a record this year.

    Or maybe it will break a record for cold, or for precip (or lack of), or cloudiness (here in Denver it has been unusually un-sunny since last Sept – gray/overcast most of the time. It makes me S.A.D.), or maybe just totally average.

    Weather is like that.

  41. Don.

    “The “who is cooking the books” meme is actually an effective weapon for the skeptics to employ in their propaganda war against the often bogus propaganda of the alarmists. We already learned of some shady activities going on in the climate science, when we got that peak behind the curtain with the C’gate emails. Remember that, Steven? And the whitewash investigations?”

    It’s not effective when you have looked at the books yourself.
    It’s as effective as calling mcIntyre an Oil shill. It’s not effective when you accuse the innocent.
    Climategate was about 3 guys.
    There was an opportunity to grab the high ground after climategate. squandered.

  42. Don Montfort: comne on now, ‘fess up. You’re really a mole working for Stephan Lewandowsky, aren’t you? 😉

  43. Steven, Steven

    Sometimes I have to wonder how you were able to contribute significantly to my conversion from uninformed skeptic to barely literate semi-lukewarmer. Or should I say quasi-lukewarmer?

    OK, you have looked at the books. You, Zeke and a handful of others, with way too much time on their hands. The masses are not inclined to spend their FREE time that way.

    I don’t recall that McIntyre sued anyone for calling him an oil shill. I am pretty sure that nobody lost his tenure or got stripped of his alleged Nobel Prize, for defaming the Mc’auditor. Was Anthony’s call to jail all NCDC scientists retaliation for all that oil shill BS? Does anybody really care?

    I don’t see how Climategate was a squandered opportunity to grab the high ground, if it was only about 3 guys, who were exonerated in whitewashes perpetrated by the powers that be.

    Skeptics and consensus alarmists are fighting for the hearts and minds of people who really don’t want to be bothered. The public mostly ignores this stuff, like they do just about everything that doesn’t involve some prospect for immediate gratification. Those who do intermittently pay some attention just seem to blow with the wind, run hot then cold, depending on the weather.

    http://www.usnews.com/news/articles/2013/02/06/study-opinions-on-climate-change-rise-and-fall-with-the-temperatures

    That is why the alarmists have seized on the extreme weather events scare tactic. I think it works for them, around the margins. So to be fair, I think it is fine for the skeptics to keep using the “who’s cooking the books” gambit.

  45. Don.

    Alarmists have seized on extreme weather for a different reason.
    Right before the increase in extreme weather stories, the team was out getting advice on communication. They were told that you cannot sell future fear. People tire of it. You can sell immediate fear. What they dont see is that the extreme weather meme actually plays into an argument for near term adaptation. Smart skeptics would seize on this.

    with regards to the rest of your comments, sorry you feel that way.

  46. @pdjakow
    “Daily ch05 values”

    Thanks for the reply but that still doesn’t answer the question.
    The graph you linked to clearly shows that jan 2013 was warmer than jan 2010.
    So does the Spencer’s AMSU site.
    I still don’t get it. How can the sattelite data show 2013 as warmer but Spencer reports it as cooler?
    Are the data adjusted for something? What am I missing here?

  47. Leslie Graham (Comment #109744),

    I still don’t get it. How can the sattelite data show 2013 as warmer but Spencer reports it as cooler?

    AQUA became unreliable. You can read about it on Spencers blogpost here.

    I haven’t read anything about the problems with AQUA being fixed yet, and UAH removed the AQUA instrument readings when they released their version5.5 dataset so you can no longer compare AQUA to the published anomaly.

  49. Differences between UAH and RSS (december 2012, january datafile for UAH not abailable yet):

    meteomodel.pl/klimat/rss/uah201212.glb.png
    meteomodel.pl/klimat/rss/rss201212.glb.png

  50. Leslie Graham (Comment #109744),

    While the recently developed AQUA technical problems may be part of the issue, even perfectly functioning satellites which all give identical readings wouldn’t mean a clear relationship between the daily figures on that site and final monthly TLT.

    The first point is that Channel 5 relates directly to TMT, not TLT. TLT is the result of extra processing which involves weighting satellite retrievals differently depending on the angle of their view. This is done to reduce Stratospheric influence on the raw channel 5 data.

    The TMT monthly datasets should therefore provide a closer match with the daily channel 5 data. However, TMT datasets are attempting to track the change since 1979, which means they need to merge the older MSU channel 2 observations with newer AMSU channel 5. These are very similar in their measurement scope but not identical so there is some adjustment to account for the difference and ensure current measurements are relatable to older ones.

  51. Mosher and Monfort have been discussing the tactics of warmists and skeptics. Mosher for the most part defends the work of institutionally supported researchers. (many of whom are warmists.) I believe that any scientific community that would tolerate the likes of James Hansen (who has called critics “court jesters” and who has advocated the violation of international and and constitutional law by seeking the ex post facto prosecution of energy executives) is inherently untrustworthy. If the temperature records are indeed accurate, it is more by accident than design. If the institutionally supported researchers were interested in their credibility, they would disown people like Hansen and Mann and separate themselves from political advocacy. They don’t. So, they have the duty of constantly defending the accuracy and integrity of their work.

    What is particularly offensive to me are the “lost” records that Warwick Hughes requested from UEA. Why aren’t institutional researchers demanding a thorough explanation of this fiasco? Why is Warwick Hughes put through the ringer? The answers don’t speak highly of Hansenite science. That being said, there is significant evidence that the weather records are not as bad as one would expect when people such as Mann and Hansen are prominent contributors to the field.

    JD

  52. Don Monfort (Comment #109731)

    I think we see discussions of AGW and the potential effects of it at several levels. The one that most interested me is at the science level and my interest there is purely and unadulteratedly selfish. I want to learn and I find the science (and lack thereof) interesting.

    At another level is that of the scientist/advocate and determining for my own self satisfaction how much climate science is affected by advocacy.

    At a third level is the methods whereby governments and the advocates for big government go about convincing the masses of its need and how that can be applied to AGW and government mitigation. There I see the standard methods will probably be applied whereby a crisis and emergency will be used to at least start the ball rolling towards attempts at government mitigation. Extreme weather seems to fill that bill nicely. Most if not all large governments projects including wars are initiated through crisis whether real or manufactured and I do not see AGW being any different.

    Government action based on potential detrimental (or beneficial) effects that are predicted for the not so near future creates a bit of a contradiction for our current crop of politicians and intellectuals in that we have a couple of failing major government programs in Social Security and Medicare that these people are not willing to face up to. Government indebtedness and unfunded liabilities are also problems that these people lay off to future generations. The argument that we need to do something about potential and very much less well understood effects of AGW for future generations comes across as out of step with the current Keynesian mantra that in the end we are all dead. The lazy intellectual and politician will surely chose the path of an immediate crisis to justify actions and without a whole lot of forethought going into the process.

  53. “Mosher for the most part defends the work of institutionally supported researchers.”

    I work on a case by case basis. Knee jerk attacks and knee jerk defenses of either institutions or institutional outsiders is lazy.

    Let’s take temperature series. I have found Menne’s work to be excellent. Jones? not as good. i won’t make any generalizations about institutions. cause I know they would be wrong.

  54. JD,

    Mosher is correct that Climategate was a squandered opportunity to grab the high ground. But the squandering was done by the climate science establishment and their allies, not the skeptics. Instead of doing some constructive introspection and cleaning house, they got out the whitewash and tried to cover up the dirt. I don’t think it worked for them.

  55. Kenneth,

    Yes AGW, whether real or imagined, has been added to the list of impending crises that our leaders don’t have the guts to deal with.

  56. Steven,

    If you read your posts recently, I think you will notice that you generalize/stereotype/pigeonhole, when you are talking about skeptics. As if they are a monolithic herd of morons. Try to be a little more patient with them.

  58. Mosher: “I work on a case by case basis. Knee jerk attacks and knee jerk defenses of either institutions or institutional outsiders is lazy.

    Let’s take temperature series. I have found Menne’s work to be excellent. Jones? not as good. i won’t make any generalizations about institutions. cause I know they would be wrong.”

    I agree that individual work should be evaluated independently and even biased people can do good work. On the other hand, good scientists, when they make mistakes, welcome information concerning the mistakes, so they can be corrected. Traditional warmists typically state that even though we made a mistake, it doesn’t matter — instead of trying to get to the root of the mistake. (See Mann & Tiljander) That is one of my fundamental problem with Hansenite science.

    JD

  59. “Traditional warmists typically state that even though we made a mistake, it doesn’t matter — instead of trying to get to the root of the mistake. (See Mann & Tiljander) That is one of my fundamental problem with Hansenite science.

    JD
    #######
    Traditional warmists?

    Let’s take Hansen.
    Peter O neill has forwarded several corrections to hansen WRT the temperature record ( pete does great work nobody reads his blog however> he has shared some code with me. Anyway, hansen made a number of corrections and credited peter.
    Zeke and Sebastian wrote a couple of memos somewhat critical of hansens work. I wish I could publish the nice mail he sent to zeke and sebastian and other on the team.

  60. Heraclitus said you can never put your foot into the same river twice. There is so much noise in this system that prediction is impossible. Bohr also comes to mind when he said predictions are difficult, especially about the future.

  61. MrE: So if I put an ice cube in your coffee, you would philosophically abstain from predicting whether it’s going to be colder or hotter?

    Slightly more seriously: if I actually put an ice cube in your coffee and stir it, I will not be able to predict the exact transient dynamics of temperature changes across the volume of coffee. But if you give me the initial numbers (masses, temperature, etc) I will certainly be able to make decent predictions about the overall final temperature.

  63. Wrong, Mike. If they believe or suspect that someone is cooking the books, they are not lying. They are expressing an opinion. First amendment kind of stuff, Mike. And is it not possible that someone is cooking the books?

  65. “Zeke and Sebastian wrote a couple of memos somewhat critical of hansens work. I wish I could publish the nice mail he sent to zeke and sebastian and other on the team.”

    And I have heard that Hansen is kind to puppies and helps old ladies cross the street.

  66. Using a Feb-Jan year, UAH global temps now show a statistically significant positive regression slope (at 95% confidence level) for the last 13 years, last 14 years, last 16 years, and all greater timespans.

    So I guess, as always, global warming stopped in _____ (insert moving goalpost here.)

  67. KAP: So you picked the rather strange Feb-Jan year so that you could include the very large uptick in Jan 2013, thereby claiming a statistically significant slope for more years than you could the month before. In addition to your selection issue, you might also want to look at the difference between “significant” and “statistically significant”. And you might want to look at issues of linear regression for time series.

    By the way, you filled in your own blank with “2001”.

  68. @ KAP

    Using a February to January “year”??? Even governments don’t do that.

    The year is either January 1 – December 31, or maybe July 1 – June 30 or maybe even October 1 – September 30 (those last 2 examples are governmental fiscal years for various governments.

    However, when you decided to use a February 1 – January 31 “year” I think it was YOU who “moved the goalposts”.

  69. It is very interesting to compare the surface temperature data for January 2013 with the satellite data for January 2013. Kinda makes me think that the theory of the oceans losing heat to the atmosphere might have some merit. If so, will a significant portion of that heat escape through TOA, or will it be retained in the atmosphere?

    My personal feeling is that UAH/RSS temperatures may well be positive anomalies in the same range (+0.4 to +0.6) for the next 3 months or so and then the anomalies will be dropping and may even go negative by the end of 2013, but that is just based on my own personal theory. I am not a “climate scientist” nor do I play one on TV, but I am an environmental chemist and have some idea of how complex systems equilibrate, so at least I can follow along this year and see if my own prediction pans out.

    So personally, I don’t think that 2013 will be a “record year”, but I could certainly be wrong!

  70. @Wayne2,

    My biggest issue with linear regression for a time series is that no one – to my own personal satisfaction anyway – has been able to justify attempting to fit straight lines to data from a non-linear coupled chaotic system.

    Sure, you can do it, but by definition, the system is NON-LINEAR, so fitting straight lines to it is silly, IMHO.

    I have brought this up before and been swatted down with comments such as “fitting straight lines to a portion of data from such a system is perfectly valid, and often quite useful!”

    In my opinion, it certainly can be “useful”, but I still have doubts about the “perfectly valid” part….

  71. Toto,

    Unfortunately, the same principles of the square-cube law makes your cup of coffee experiment irrelevant.

  72. Don, reading #109739, you say Mosher, Zeke and others have looked at the books, but others won’t, so it’s OK to keep accusing of cooking the books. Lying because you won’t get caught. I’d rather have the other side lie, and let the lies be pointed out to win an argument. Upside-down Tiljander basically wins every argument.

  73. Sorry I don’t have time to teach you how to read, Mike. Your interpretation of what I have actually said is really foolish.

  74. MikeN – how do you read the Tiljander measures? It is clear for most people, including the person who created them. that Mann read them incorrectly.

  75. @PeterB: The problem with linear fits to time series include:

    1. Time series have autocorrelations, etc, which OLS has issues with.

    2. Where you begin and end your time series can make a huge difference.

    3. Time series tend to have cycles (seasonality) and quasi-cycles, so any linear fit will probably not extrapolate well beyond the data at hand. And the temptation to extrapolate is nearly irresistible.

    4. In climate research, there are several drivers (clouds, etc) that aren’t well-understood, but the usual assumption is that once a few token (quasi-) cyclical factors are accounted for, everything else is “the global warming signal”.

    F&R is the poster child for items 2, 3 and 4. KAP is the poster child for 1 and 2. It’s understandable, since you can grab any statistics package and feed a time series in and get a fitted linear model out with enough asterisks to make your mother proud. Publish — featuring your p-values prominently — and you’re golden.

  76. PeterB:

    My biggest issue with linear regression for a time series is that no one – to my own personal satisfaction anyway – has been able to justify attempting to fit straight lines to data from a non-linear coupled chaotic system.

    The justification has nothing to do with the data you’re applying it to. It’s just a method for estimating the rate of change of a quantity (“the trend”) that is relatively robust.

    This is a useful thing to estimate if you can compute uncertainty bounds for the associated with the internal noise in the system. For example, Lucia estimates the internal noise using various autoregressive models, I use a spectral Monte Carlo method.

    When using a best-fit linear trend, you have to specify what optimization function (typically unweighted least square) and what the interval that you’re fitting to, but it is a meaningful thing to do as long as the internal noise doesn’t dominate your estimate over the period of interest.

    Longer periods give shorter errors.. This is from my Monte Carlo estimate.

    So how large the trend in the signal is or is predicted to be matters in telling you how long you have to integrate for (e.g., duration of interval used) in order to test a particular hypothesis.

    For example, to test the null hypothesis that your observed temperature series is not distinguishable from from a zero trend, when you were expecting 2°C/century, you’d need at least 10-years using my estimate (slightly longer for Lucia’s method).

  77. Wayne2, see my comments to PeterB. The properties you’ve described are prevalent in almost all experimental data. Yet we progress.

    As I suggested to PeterB, the conditions for whether a statistical method is meaningful is whether it is telling you something useful (rate of change of a quantity is) and whether you can establish meaningful uncertainty bounds on it. Beyond that what the data is in particular, is a bit of a red herring, other than in how the internal noise of that system, measurement error, etc, affect the central value and uncertainty range. That’s a generic property of any data though.

  78. @Carrick: I didn’t say that issues with time series make it impossible to do statistics with them, but rather that the issues need to be addressed. KAS, Foster and Rahmstorff, etc, don’t address the issues and hence get erroneous answers.

    I’d also guess that a lot of experimental data is not significantly time series data, and in that case linear regression is easier to use and less likely to go horribly wrong.

    In either case, I think you misunderstand my point. I’m not waving my hands, saying “time series are too hard!” and saying we don’t analyze them. Just that KAS, F&R, etc, treat time series as non-time-series and get the answers they want.

  79. @Carrick: P.S. Your response to PeterB mainly addresses my point #1. F&R still fit a 30-year time series, then in the final sentence of their paper extrapolate their results 30 years into the future. And they ignore several quasi-cycles with periods longer than 30 years, crediting it all to a straight-line “global warming signal”.

    We may progress, but their paper is not contributing to it. Nor is KAS’s linear regression.

  80. Wayne2, I agree that F&R contribute little to the discussion. I suppose you meant “not significantly correlated” ?

    Outdoor (boundary layer) turbulence is such a problem, and suffers from many of the same warts as met data. (However, you have control over the instrumentation in with boundary layer measurements, and it much easier to model accurately, so it’s a much simple problem.)

  81. Lucia… according to the antipodean NBR:

    http://www.nbr.co.nz/article/too-much-hot-air-about-global-warming-says-researcher-rv-1

    I find this quote interesting:

    “Vilified by global warming zealots, Canadian Steve McIntyre, who was passing through Auckland this week, told NBR ONLINE the impact of global warming is likely to be “about half” of what current scientific models are showing.”

    In relation to the surface temperature indices and your analysis of the multi-model mean, I figure a factor of maybe 0.85. If Steve prefers RSS or UAH, multiply by another factor of ~0.75 to get a total factor of ~0.64.

    Would you concur with the “about half” statement? Personally I would go with “about two thirds” or maybe “about four fifths”.

    Sincerely… AJ

  82. There is no way that CO2 could have caused the recent dramatic warming because CO2 concentrations change so slowly. What we are looking at is “Noise”.

    By my calculations, if CO2 affects global temperature at all, the maximum possible warming delivered is 3.1 K and that would take one hundred times the present CO2 concentration. Universities around the world tell us that Earth’s average temperature (sans atmosphere) would be 255 K. They also tell us that the observed global average temperature is 288 K. So the GHE = 33 K.

    The Moon’s average temperature is 154.3 K so why would an airless Earth have a mean temperature of 255 K as all those Climate 101 courses claim?
    http://www.diviner.ucla.edu/science.shtml

    A GHE based on CO2 is absurd. CO2 can only explain a 3 K GHE whereas James Hansen says the GHE is 33 K. Nikolov and Zeller say the GHE is 134 K. My money is on N&K.

  83. PeterB in Indinapolis (Comment #109775),

    It is very interesting to compare the surface temperature data for January 2013 with the satellite data for January 2013. Kinda makes me think that the theory of the oceans losing heat to the atmosphere might have some merit.

    I’ve been looking at the oceans for a while now. What I can’t understand is why the atmosphere would suddenly scavenge so much heat from the oceans, which appear to be cooling.

    My personal feeling is that UAH/RSS temperatures may well be positive anomalies in the same range (+0.4 to +0.6) for the next 3 months or so and then the anomalies will be dropping and may even go negative by the end of 2013

    I’m thinking along the same lines. I can’t see anything which would cause this high anomaly to be sustained for any great length of time.

  85. Skeptikal:

    I’m thinking along the same lines. I can’t see anything which would cause this high anomaly to be sustained for any great length of time.

    I think it’s important to remember that they are supposed to represent anomalies, which basically mean they’ve been “deseasonalized”.

    However, shifts in climatic patterns affects seasonal patterns too, so it’s likely that part of what you see with a few months that are very warm followed by a few that are very cold (or vice versus) is not a real local change in temperature, but a shift in seasonal patterns.

    When you look at subannual variations, there is inevitably a conflation of effects due to long-period trends in warming/cooling and changes in sub-annual variability.

  86. Skeptikal,

    Like you I don’t see one phenomenon causing this anomaly. You have to assume factors that cause large, sudden and prolonged jumps in temperature should be known and would probably have a name we’re all familiar with, like ENSO. In fact ENSO is the only one I can think of (maybe volcanos aswell). This isn’t an El Nino event beginning. You really have to assume that it’s going to be short lived. Maybe Spencer’s idea that the oceans are giving up heat to the air goes some way to explaining this but the fact the heat is spread over much of the globe, high/mid latitudes of both the NH and SH and maybe most strangely the tropics suggests to me multiple processes in action at one time, all on the hot side.

    WRT Lucia’s comment “With that kind of start …..”, then sure if the annual value is an average of 12 data points and the first is on the warm side then it makes a ‘cool’ year less likely but I don’t (yet) see any reason that Jan’s value is a prelude for what’s to come.

    (BTW Mosher’s first comment is bizarre. If the debate has been reduced to imagining what strange things people might say in the future, and all based on one data point, then we are in for an even more depressing time. Mosher maybe stick with the rubbish people are actually spouting not what your own bias leads you to believe they will be saying in the future.)

  87. @Carrick: By “I’d also guess that a lot of experimental data is not significantly time series data”, I was trying to say that almost all experimental data is in some sense time series data, since it’s measured over time, but in a practical (“significant”) sense most of it can be analyzed with no statistical concern for time.

    Time tends to be cyclical in ways that other variables are not, and it’s way too easy to put a straight line through Apple stock prices and say that I’ll be a billionaire (or flat broke) by the end of the year. For very short time periods and specific purposes, linear regression can be useful, but it’s not meaningful the way F&R or KAS applied it.

  88. Wayne: Since when does using all the data become a “selection issue”? Doesn’t the “selection issue” belong to those who don’t use all the data?

    Peter: Earth’s orbit is a closed curve and nearly circular. The beginning and end points of any such curve are entirely arbitrary. February 1 is just as valid a start date for a year as January 1, or March 25 (as was the case in England before 1751), or March 14 (the Sikh new year), or April 1 (Assyrian), or September 11 (Coptic Orthodox new year). As long as each year analyzed has 12 months worth of data (and they do), you have nothing to complain about.

  89. AJ (#109787) –
    Regarding the “about half” of models comment…I suspect you’re considering the OLS trend over the last 30 years as the “true” warming rate. AR4 multi-model mean has the warming rate a tad over 0.2 K/decade. Recent GISS 30-year trend is around 0.17 K/decade; HadCrut4 & NCDC a little less at 0.16 or so.
    .
    But is all of that due to greenhouse gases (and other anthropogenic effects)? Look beyond the last 30 years. Look at the running 30-year trends, shown here. There’s a strong suggestion of a ~60-year natural oscillation which enhanced warming in the 80s and 90s. If so, the current 30-year trend is not entirely due to anthropogenic effects. The amplitude of the presumably natural variation is such that it’s not unreasonable to suspect that the anthropogenic warming rate is about 0.1 K/decade. That is, half of the multi-model mean.
    .
    Now I don’t speak for Steve McIntyre, and he may have entirely different — likely better — reasons for believing that the models overstate ghg warming by about a factor of two. But the above is my back-of-the-envelope reason for believing it.

  90. Re: gallopingcamel (Feb 8 23:51),

    The Moon’s average temperature is 154.3 K so why would an airless Earth have a mean temperature of 255 K as all those Climate 101 courses claim?

    The 255K number refers to an isothermal effectively superconducting or uniformly illuminated surface with an albedo of 0.3. As such, it’s an upper limit. The lower limit is a surface with zero heat capacity and zero thermal conductivity illuminated with a radiation field like that from the sun to the Earth such that you get latitudinal and meridional temperature variation because the angle between the incident radiation and the local surface varies with latitude and longitude and the local surface temperature drops instantaneously to 0K (or 2.725K if you assume the cosmic microwave background exists) as soon as the sun goes below the horizon. In this case, the sphere can be phase locked to the sun and it won’t matter.

    Needless to say the lower limit is quite a bit lower than 255K even if you assume an albedo of zero. The governing mathematical principle is Hölder’s Inequality (see for example Gerlich and Tscheuchner page 64. This part they get right. or Science of Doom, Lunar Madness…). The calculated average temperature for such a sphere with an albedo of zero is 158K. The lunar surface has an albedo of ~0.1 so it would have a somewhat lower temperature, but it also has a finite surface heat capacity and the temperature of the side facing the Earth is also affected by LW radiation emitted by the Earth when the sun is below the horizon.

  91. KAP: OLS regression is sensitive to endpoints and to outliers. You jumped onto the data in a month that saw an unusually large uptick.

    That’s not exactly cherry-picking — a phrase I avoided using — but it’s related.

    You use a linear regression because you believe it shows a long-term trend, but take advantage of a short-term jump to prove a point. Not to mention the other issues with naive regression on a time series full of cycles and quasi-cycles.

  92. Carrick (Comment #109791),

    I think it’s important to remember that they are supposed to represent anomalies, which basically mean they’ve been “deseasonalized”.

    Yes, I understand that they are anomalies and not absolute temperatures, but it’s still a fairly spectacular jump and as HR has pointed out…

    but the fact the heat is spread over much of the globe, high/mid latitudes of both the NH and SH and maybe most strangely the tropics suggests to me multiple processes in action at one time, all on the hot side.

    I can’t see a shift in seasonal patterns causing this, particularly in the tropics where they only have wet and dry seasons.

  93. Wayne2:

    KAP: OLS regression is sensitive to endpoints and to outliers. You jumped onto the data in a month that saw an unusually large uptick.

    That’s why I often use minimum absolute deviation as an alternative, more robust measure of central tendency. In any case, unless you have a true outlier (like e.g. a volcano that pushes the system well away from its normal behavior), then the Monte Carlo method already handles this issue (your uncertainty gets inflated slightly by using an OLS instead of a more robust measure).

  94. @KAP: I decided to do a little analysis and get specific instead of addressing just the philosophical problems with your analysis. (Which are significant problems, none-the-less.)

    The first issue is your eagerness to use January 2013, which turns out to be tied for the fourth-largest jump since UAH started (410 months of data). Not cherry-picking exactly, but not really concerned with trend, as compared to noise.

    The second issue is how you state your finding about “the last 13 years, last 14 years, last 16 years, and all greater timespans.” I could turn things around and say that in order to find a Feb-Jan year in which the trend was statistically significant and positive starting at every month in that year, you have to go back 16 years or more.

    (You’re not the only one who can stir the data until the tea leaves read well.)

  97. HaroldW (Comment #109801)

    Thanks for the reply. I don’t disagree (too much). Considering I got my fractions upside down, I can only conclude that you knew what I meant.

    Thanks again,

    AJ

Comments are closed.