Comments 1 to 28:

1. actually thoughtful at 06:37 AM on 22 January, 2012
   Looking at the first graphic, the 1910-1940 warming looks similar, in slope, to the 1970-2010 warming. Of course one is 30 years, and one is 40.
   
   But wouldn't we expect an increase in the slope as CO2 is driving temp up even more than it was in the 1940s?
   
   Or am I missing something.
   Response:

   [dana1981] See "It warmed just as fast in 1860-1880 and 1910-1940"

2. Albatross at 07:14 AM on 22 January, 2012
   actually thoughtful @1,
   
   You seem to be missing the point. In the absence of a log term underlying warming trend global SAT anomalies in 2011 should have been between -0.1 and -0.2 C. Instead they were over +0.5 C (w.r.t the 1950-1980 baseline) and according to GISTEMP the 9th warmest year on record. That, despite just emerging from a prolonged solar minimum and increased aerosol loading and about 7 of the 12 months having an ONI of <-0.5 (ONI values of -0.5 or less are typically consistent with La Nina conditions.
3. actually thoughtful at 07:19 AM on 22 January, 2012
   Dana - thanks. Still groggy from my time at WUWT.
   
   The evidence I find most compelling, when comparing 1910-1940 to 1970-2010 is the higher solar activity in the former time period, which is notably, stunningly absent in the the last 40 years (even more so for the last 10 years).
4. actually thoughtful at 08:04 AM on 22 January, 2012
   Ironically two threads I am posting on over at WuWT are invoking the 1910-1940 (nothing to do with me or this post). Having just reviewed the solar contribution to the 1910-1940 I was able to quickly set the record straight. (not holding my breath that anyone posting at that site will accept said reality).
   
   In regards to WUWT -
   Must...stop...posting
5. MA Rodger at 08:07 AM on 22 January, 2012
   Another noteworthy event from the 2011 temperature records. The UAH annual temperature anomaly is higher than RSS for the first time ever.
   The differences between them were getting more divergent up to 2000 since when they have converged and now swapped places with the UAH anomaly diverging at a rate of 0.3 deg C per decade!
6. Pete Dunkelberg at 08:17 AM on 22 January, 2012
   The apparent cool spot in south Florida (fig. 4) is just lake Okeechobee.
7. actually thoughtful at 08:44 AM on 22 January, 2012
   I suspect La Ninas will bring us more coldest/driest and El Ninos will bring us more record heat.
   
   That we are getting record heat in a La Nina should give doubters pause. Very big pause.
8. Daniel Bailey at 08:50 AM on 22 January, 2012
   @ actually thoughtful
   
   You will note from the 3rd figure in the OP that the La Niña's are consistently getting warmer.
9. muoncounter at 09:00 AM on 22 January, 2012
   actually thoughtful#1: "1910-1940 warming looks similar, in slope, to the 1970-2010 warming"
   
   Another big difference is seen here:
   
   
   
   Summer warming, driven by solar output, was faster than winter warming in the early 20th century. Latter 20th century (and ongoing) warming shows winters warm more rapidly; a greenhouse effect.
10. Chris G at 09:02 AM on 22 January, 2012
    Curious, why the 133-month average?
    I most commonly use a 132 because that smooths over any 11-year solar cycle effect, but then...I'm thinking that the solar cycle average may not be, actually, is highly unlikely to be, exactly 11 years, on average. So, maybe your rationale is the same, just using a more refined number for the period.
    
    I'd be surprised if it made any noteworthy difference though.
    
    Ah, good old friend Google. Seems the solar cycle average is more accurately 11.1 years; 133 it is.
11. John Hartz at 09:03 AM on 22 January, 2012
    For a portend of what’s on the horizon, see “NASA Sees Repeating La Niña Hitting its Peak” posted on the NASA/Jet Propulsion website.
12. John Hartz at 09:07 AM on 22 January, 2012
    Dana,
    
    Recommend that you move the text caught up in the right-hand margin of the graphics to underneath the graphics.
13. Alex C at 09:17 AM on 22 January, 2012
    @Chris G: The running average I think is given at the center of the time interval, which only exists with an odd number of months. I don't think it has to do with removing the solar cycle.
14. Tom Curtis at 09:40 AM on 22 January, 2012
    Grant Foster (Tamino) has updated the analysis from Foster and Rahmstorf to include (most of) 2011's data. The "most of" is because HadCRUTv3 has not yet released their data for Dec 2011. He has also switched to using sunspot data rather than TSI to allow for the influence of solar activity. For the computer savy, he has also updated his publicly accessible programs.
    
    His summary of the unadjusted data:
    
    
    
    

    "For GISS data, 2011 was the 9th-hottest year on record, for NCDC it was 11th-hottest, for HadCRU (using only data through November) it was 12th-hottest, for RSS it was 12th-hottest, and for UAH 9th-hottest."

    
    
    And of the data adjusted for exogenous factors:
    
    
    
    

    "In the adjusted data, GISS ranks 2011 as the 2nd-hottest year on record (just behind 2010), NCDC ranks it 5th, HadCRU (using only data through November 2010) ranks it 5th, RSS ranks it 2nd, and UAH 2nd. No, global temps are not in a crash — they still fluctuate (for a lot of reasons, including exogenous factors) but the trend continues. It’s called “global warming.” It’s caused by human activity. It’s dangerous."

15. Tom Curtis at 09:41 AM on 22 January, 2012
    Sorry, I forgot to include the link to Tamino's site for those not familiar with it.
16. Tom Curtis at 09:47 AM on 22 January, 2012
    Alex C @13, while the reason for the odd number of months is as you give it, the reason for choosing an approximately 11 year running average is almost certainly to remove the solar cycle. (I suspect your aware of that, and have merely phrased your response to Chris G @10 poorly.)
17. Tom Curtis at 09:54 AM on 22 January, 2012
    It would be interesting if one of the tech savvy people at this site used Tamino's software to adjust for exogenous factors for the periods 1880-1940, 1910 to 1940, and 1880 to 2011. (I have just made the same suggestion to Tamino, so you may want to hold of to see how he responds.)
18. Alex C at 09:55 AM on 22 January, 2012
    Tom Curtis @16: I have typically seen 11 year running averages, and thought that it was for much the same reason as that a centered mean cannot be given for an even number of years. I am aware of the solar cycle length average, but the thought that it was the primary driver behind a choice of 11 years hadn't occurred to me before (I suppose it should have, hm?). Thanks for the clarification.
    
    I personally found Tamino's switch to sunspot number interesting. I thought that TSI data was updated continuously (daily), why use a proxy instead of the real deal?
19. Tom Curtis at 11:02 AM on 22 January, 2012
    Alex C @18, Tamino writes:
    
    

    "For the updated results, rather than use TSI (total solar irradiance) to represent solar variations I used sunspot counts, because they’re easy to get, kept up-to-date, and already available as monthly averages."

    
    
    The "easy to get at" and "available as monthly averages" may be particularly important factors in that his updated software passages included a program which gathers the data from the internet, and formats it for you so that you can easily keep the analysis up to date. It may be that the programming task was easier in that respect with sunspot numbers than TSI. Of course, I am speculating, and if you think it is important, you should ask him.
    
    What I do know is that in Foster and Rahmstorf (2011), Tamino wrote:
    
    

    "To test whether the results might be sensitive to these choices, we also did experiments characterizing el Nino by the southern oscillation index (SOI) rather than ˜MEI, characterizing volcanic aerosols by the volcanic forcing estimate of Ammann et al (2003) rather than the AOD data from Sato et al, and using monthly sunspot numbers as a proxy for solar activity rather than TSI. None of these substitutions affected the results in a significant way, establishing that this analysis is robust to the choice of data to represent exogenous factors."

    
    
    That being the case, the switch to sunspot numbers should make no significant difference.
20. jyyh at 12:40 PM on 22 January, 2012
    "periods of 1880-1940, 1910 to 1940" I think MEI isn't calculated that far in history, as it takes note of many atmospheric variables not commonly and widely measured before war. But, there's SOI, and as he says it's robust then it seems there should be no reason why this couldn't be done. Aerosol loading may have changed for alterations in ff use, though, i'd imagine
21. Tom Curtis at 13:39 PM on 22 January, 2012
    jyh @20, Foster and Rahmstorf have already used the Southern Oscillation Index instead of the Multi-variate ENSO Index to test the robustness of their analysis, so that aspect should be no problem.
    
    Given the removal of exogenous factors, the result should be not a linear trend over the entire century, but a variable curve approximately matching anthropogenic forcings, ie, the green curve in this figure from Lean and Rind (2008):
    
    
    
    Based on the empirical model produced by Lean and Rind (below), we would expect large excursions from the "anthropogenic" signal in the periods 1910-1915, and 1940-1945, and a smaller excursion in the period 1960-1964. On top of that we would expect small, fluctuations about the anthropogenic signal.
    
    
    
    I put "anthropogenic" in "shudder quotes" above because Lean and Rind do not model all anthropogenic factors. In particular they do not model Black Carbon, which I suspect was a major influence in the high temperatures during WW2. It should be noted that in the period 1880 to 1940, BC and sulfates found in Greenland ice cores rise to a peak around 1910, before gradually tailing of. The reduction after 1910 would be because of the transition from coal to oil as a major fuel. Because the sulfates and BC are significantly correlated, they would have to a significant extent counteracted each others effects. In contrast, BC from WW2 would come from the burning of cities, and would not have significant associated sulfate emissions. Because of the direction of the prevailing winds, cities burning in Europe would also not have left a significant record in Greenland.
22. jyyh at 14:08 PM on 22 January, 2012
    Then there's the question of how much data is available from the Arctic and Antarctic during the war times, if it was relatively cold in these locations, this could bring the CRU observational dataset down. The topmost snow/ice from Greenland and Antarctic icecore locations should provide some evidence.
23. Chris G at 15:10 PM on 22 January, 2012
    Alex,
    Hmm, I'm not aware of any difficulty calculating the mean of an even number of points or any requirement that the endpoints of the running mean has to align perfectly with endpoints of the periods of the base measurement. Time can be considered a continuous variable; the middle of a month exists as surely as the start or end. In minutes,
    
    X Y
    00:01:00 1
    00:02:00 2
    00:03:00 3
    00:04:00 4
    
    If you are calculating the mean of these points, and consider that Y is the mean of the period from the start of the period ending at X, you get means of
    
    00:02:30 2.5
    
    IDK, in my mind, that is just as valid a point, and could be graphed just as easily as
    
    00:03:00 3
    
    would have been if there had been 5 data points (following the same pattern).
    
    Hmm, that is actually not correct. If 00:04:00 is the endpoint of a period of 4 minutes, then the middle of that interval is at 00:02:00. You have to consider time not as points, but as intervals. And, you have to know if the point in the dataset represents the start, end, or middle of that interval.
    
    Cosmetically, it could be a little easier to deal with the periods of the base measurement aligning with the period of the calculated mean, but if you can put 2.3 children per couple on a graph, I'm pretty sure you can define a point in the middle of a month.
    
    For that matter, months are kind of a poor choice for period; there is not the same amount of time in all months. Nevermind leap years, and leap year February.
    
    I'm sure that people that deal with temperature record data have figured all this in, but let's not assume that what's convenient for cosmetics or human thought patterns is a requirement of reality.
    
    On the other hand, I'm open to the idea that I missed something. Or, maybe I'm overstating the obvious. Or, maybe I'm just overstating the obscure.
24. Alex C at 16:16 PM on 22 January, 2012
    Chris G @23: True, if you assume continuity along the horizontal axis (which, regarding time, makes physical sense, but the data is not presented as a function of a continuous variable). You could of course align any mean with any fractional abscissa, but cosmetically (to use your word) it would make more sense to have an X-year centered running mean actually centered on a year, and not a piece of a year (or, month, and not a piece of a month).
25. Glenn Tamblyn at 21:31 PM on 22 January, 2012
    AT #1
    
    Possibly missing several things. Between 1910 and 1955 the worlds temperature record was in flux. Initially stations were being added to the Arctic. Then much latter stations were added to the Antarctic. So the temperature record has some significant possible distortions during that period. Also the SST record was distorted due to a discontinuity caused by a shift in the proportion nof nUS vs UK ships involved in SST measurement during the War Years.
    
    So one cannot assume that the 1910/40 curve is really the result of an early warming pattern, rather than simply a result of othet factors not well analysed. And without that curve, the longer term pattern has a different meaning.
26. Chris G at 05:15 AM on 23 January, 2012
    Re: John (#11)
    If the ocean retains relatively more energy (heat) during a La Nina relative to an El Nino, I suppose this means we can expect more verbiage about "missing heat" in 2012. It'll be an easy cherry pick to show a lack of temperature increase in the atmosphere while ignoring the temperature increase of the oceans, in the totality that we can measure, rather than just whatever layer happens to fit the preconceptions.
    
    Re: Tom (#21)
    Thanks for that reference. I have had heard, in the past, and had trouble accepting, that the human signal (GHG) only started about 1950. I always figured it was there, just hard to detect. And then there is the bit of, feedbacks aside, a constant increase of GHGs would produce a logarithmic (decreasing slope) curve, but an increasing rate of growth of GHGs should produce an increasing slope curve. If you add them together, does the result have a positive or negative second derivative? Anyway, that reference clears a few things for me.
27. JoeTheScientist at 14:13 PM on 31 January, 2012
    Thanks Tom C@15. I was trying to find that, without success.
28. Eyal Morag at 06:26 AM on 2 February, 2012
    Simple "back of the envelope calculation" give 2011 as the ~ 2nd warmest year.
    1 I made correction only for ENSO using Multivariate ENSO Index (MEI)
    2 With Excel I got the best fit linear to 1990-2011 yearly ( GISS )
    3 year tamp' - the correction from the line
    4 Average MEI for 14 months Nov Dec Jan-Dec. Because it take some months to the ENSO to spread its influence.
    5 Plot a graph Average MEI to the result from 3
    6 Take out 4 years 3 after Pinatubo + 1 that looks of line
    7 With Excel I got the best fit [PEARSON=0.88]
    8 Use the slope* Average MEI to correct the themp
    9 plot the result . Red , original blue, top red the same little up for clarity , yellow Average MEI
    All the data + calculations at this spreadsheet
    
    

You need to be logged in to post a comment. Login via the left margin or if you're new, register here.