Home > CRUTEMP, GIStemp, satellites > Did Global Warming Stop After 2000?

Did Global Warming Stop After 2000?

2011 January 1

Second verse, same as the first.
So why bother? Completeness.

I used the data set prepared by O’Day over at Climate Charts & Graphs. Thus, we have charts for GISTEMP, HadCRU, NOAA, UAH, and RSS. The data for Dec 2010 is still missing – but I’m using the ave for Jan-Nov for the 2010 annual data.

The script is here

Trend GISTEMP 2000 30

Trend CRU 2000 30

Trend NOAA 2000 30

Satellite data does not extend far enough back to run 30 year trends. These are 20 year trends.

Trend RSS 2000 20

Trend UAH 2000 20

About these ads
  1. Daniel Bailey
    2011 January 1 at 10:45 pm | #1

    Nice work, Ron.

    Q1. The wider widths for the confidence intervals for UAH and RSS are due to the shorter datasets, correct?

    Q2. Any chance you can make larger versions of the charts?

    These eyes are getting close to 50 years of use & don’t work as well as they used to…

    BTW, lost Open Mind posts through June 2009 now available here.

    The Yooper

  2. 2011 January 2 at 6:28 am | #2

    Hehya, Yooper.

    Since you asked, you get to be the very first alpha-tester. Been working this weekend on porting some stuff to a new webhost. They told me they can’t support R. They didn’t tell me that I couldn’t compile my own static version and upload it. :D

    The naming convention should be self explanatory.
    http://rhinohide.org/gw/trendtester/img/

    If you want any other tweaks, let me know. But, to tell the truth, the script is so dead simple, you should be able to download R and run the script yourself!

  3. 2011 January 2 at 6:44 am | #3

    As to Q1, sample size does play a role, as the inverse of the number of points is a factor in setting the 95% CI. On the other hand, a linear fit may not be the best fit (as in the case where the ‘natural’ or ‘real’ underlying trend is nonlinear), and the CI will increase as more points are added. See the next post where the GISTEMP CI decreases for 10,20, and 30 year intervals, but then starts increasing again at 50 and 60 years.

  4. Girma
    2011 January 14 at 9:35 am | #4

    SIMPLE PREDICTIONS OF GLOBAL MEAN TEMPERATURE

    From the historical global mean temperature data shown below

    http://bit.ly/bUZsBe

    the following patterns can be established:

    a) 30-years of global cooling by 0.2 deg C.
    b) Followed by 30-years of global warming by 0.5 deg C.

    VERIFICATION

    Let us start from the global mean temperature anomaly (GMTA) for the 1880s of -0.3 deg C, which was at the beginning of a cooling phase. As a result, we have:

    1) For 1880s, GMTA = -0.3 deg C
    2) For 1910s, a GMTA of -0.3 – 0.2 = -0.5 deg C
    3) For 1940s, a GMTA of -0.5 + 0.5 = 0 deg C
    4) For 1970s, a GMTA of 0 – 0.2 = -0.2 deg C
    5) For 2000s, a GMTA of -0.2 + 0.5 = + 0.3 deg C

    These results approximately agree with the data given in the link above!

    PREDICTION

    6) For 2030s, an approximate GMTA of 0.3 – 0.2 = + 0.1 deg C

    CONCLUSION

    Global cooling until 2030!

  5. Girma
    2011 January 14 at 9:43 am | #5

    Did Global Warming Stop After 1998?

    Let us look at the data shown below:

    http://bit.ly/dQ8S9i

    This data shows the decadal trend flat at 0.4 deg C.

    CONCLUSION

    Yes, global warming stopped after 1998!

  6. 2011 January 14 at 10:58 am | #6

    Girma, Wood-for-Trees uses an inclusive start and an exclusive end. You are chopping off 2010 instead of including it. Either set the end date to 2011, or just leave off the end date as such:

    1998
    http://www.woodfortrees.org/plot/hadcrut3vgl/from:1998/to:2011/plot/hadcrut3vgl/from:1998/to:2011/trend
    http://www.woodfortrees.org/plot/hadcrut3vgl/from:1998/plot/hadcrut3vgl/from:1998/trend

    2000
    http://www.woodfortrees.org/plot/hadcrut3vgl/from:2000/to:2011/plot/hadcrut3vgl/from:2000/to:2011/trend
    http://www.woodfortrees.org/plot/hadcrut3vgl/from:2000/plot/hadcrut3vgl/from:2000/trend

    Also, you can add the most recent decade to your chunky trends located in the first link so: http://www.woodfortrees.org/plot/hadcrut3vgl/from:1880/to:1910/trend/plot/hadcrut3vgl/from:1910/to:1940/trend/plot/hadcrut3vgl/from:1940/to:1970/trend/plot/hadcrut3vgl/from:1970/to:2011/trend

    You will notice that your chunked up trends resemble the linear or exponential trends plus a sine fitted to the residuals that can be found here:
    http://rhinohide.wordpress.com/2011/01/14/lines-sines-and-curve-fitting-6-backcast-and-forecast/

  7. Girma
    2011 January 14 at 11:27 am | #7

    Ron

    How about this one?

    http://bit.ly/cO94in

  8. 2011 January 14 at 11:39 am | #8

    Interesting, Girma.

    How did the author select the sine parameters (amplitude, period, phase shift)?

  9. Girma
    2011 January 14 at 6:10 pm | #9

    Ron

    Here are the details:

    http://bit.ly/gaA9kS

    Regards

  10. 2011 January 14 at 10:08 pm | #10

    For the oscillating anomaly, the sinusoidal function cosine meets the requirements listed in Table 1. From Figure 2, the amplitude of the oscillating anomaly is given by the vertical distance in deg C from the central linear anomaly line to either the top or bottom parallel lines, and it is about 0.3 deg C. From Figure 2, the oscillating anomaly was at its maximum in the 1880s, 1940s, & 2000s; it was at its minimum in the 1910s and 1970s. The years between successive maxima or minima of the oscillating anomaly is the period of the cosine function, and it is about 1940–1880=1970–1910=60 years. For the cosine function, once its amplitude of 0.3 deg C and its period of 60 years are determined, the mathematical equation for the oscillating anomaly, for the years starting from 1880, can be written as

    Oscillating anomaly in deg C = 0.3*Cos(((Year-1880)/60)*2*3.1416) Equation 2

    It appears that you kind of fit it by hand. Did you ever check to see if 55 years or 65 years made a better fit for the period? If 0.25C or 0.35C for amplitude fit better than 0.3C?

  11. Girma
    2011 January 15 at 4:36 am | #11

    The main issue is the model has a high correlation coefficient of 0.88 as shown below:

    http://bit.ly/f7VYQH

    As a result, the global mean temperature pattern is cyclic with a slight overall warming of 0.6 deg C per century.

    !!!Global Cooling until 2030!!!

  12. 2011 January 15 at 8:48 am | #12

    Dr Orssengo, I asked you a couple of simple questions which you have not answered.

    I’ll ask them again.

    1. Did you fit the curve by hand?
    2. How did you check to see if there were better fits?

    And now a third question, given your comment about ‘the main issue.’

    3. If I can find a simple equation like “line + sine” that provides a better correlation, would that be a better model than your “line + sine” model?

  13. Girma
    2011 January 15 at 9:43 am | #13

    Ron

    From the following global mean temperature data of the CRU

    http://bit.ly/fizsCE,

    I selected the following values in deg C to fit the curve:

    Year => Anomaly => Model
    1880s=> -0.2 => -0.22
    1910s=> -0.6 => -0.64
    1940s=> 0.1 => 0.13
    1970s=> -0.3 => -0.29
    2000s=> 0.5 => 0.48

    Is it not the agreement between data and model very good?

    http://bit.ly/cO94in

  14. 2011 January 15 at 9:49 am | #14

    Yes, Dr Orssengo, the agreement between the data and the model is very good.

    See. I just answered one of your questions. It started with ‘yes.’ It is an easily understood answer. I still don’t know if you selected the values in your equation by hand. I still don’t know if you looked for better fits. And I still don’t know whether you think the model with the best correlation is the best model.

    If I provide a better agreement between data and model with a model just as simple as yours, would my model be better?

  15. Girma
    2011 January 15 at 6:07 pm | #15

    Ron

    “If I provide a better agreement between data and model with a model just as simple as yours, would my model be better?”

    YES

    However, my model has a very high correlation coefficient of 0.88. As a result, any higher correlation is not necessary.

    The main test is, as predicted by the model, will we have global cooling until 2030?

    That is the main test. If we have this cooling then we can say the model is a good model.

    Kind regards

  16. ant1
    2011 January 20 at 3:49 pm | #16

    Are these statistics significant, with respect to IPCC AR4 projections of 0.2C/decade from 2001. I think they use Hadcrut, which is a negative trend from 2001 to Dec 2010.

  17. 2011 January 20 at 5:06 pm | #17

    ant1: are slower than 0.2C more like 0.17C for the last 30 years (taken together)?

    When you ask if there is a significant difference in trends between this and the IPCC projections, I suggest that Lucia’s construction provides a better answer. All the global temp indices easily fall within the 95% confidence interval of the IPCC’s projection, although they are on the low side.

    http://rankexploits.com/musings/2011/hadcrut-december-anomlay-0-251c/

  18. Girma
    2011 January 20 at 7:12 pm | #18

    ant1

    Here is comparison of IPCC projections with data:

    http://bit.ly/cIeBz0

    Note that the new value for 2010 of 0.475 deg C is still BELOW projections had CO2 emission been held constant at the 2000 level.

    As there is no warming of 0.2 deg C per decade, IPCC projections are wrong.

  19. ant1
    2011 January 20 at 8:09 pm | #19

    Thanks Ron and Girma. You have really cleared this up with 2 different graphs and values for A1B scenario. Do they use different baselines? Is such a graph in AR4 as I can’t find it?

    I think Lucia’s other graph is a more pure test of the independant 0.2C/decade projection.
    I was under the impression that the 95% confidence interval would be zero trend. Is that correct?

  20. 2011 January 20 at 9:00 pm | #20

    Ant, the 95%CI is built around a model projection. If the model shows an increase, the CI will increase with it. In a linear model with time invariant variance (heh!), the CI will form a steady +/- envelope around the linear trend. But I don’t think that is the case with an exponential model. I’ll know soon enough. That is a topic that I am working on now.

    To answer one of your questions, Girma, ‘when can we know which model reigns supreme: line+sine or exp+sine?’, we need to be able to place confidence intervals around the two projections. If future temps fall within the CI of one trend but not the other, we can rule out the other. That’s what I am working towards right now.

    And Girma … I wouldn’t presume too much in how RealClimate feels about me or this blog. You might be surprised. :lol:

  21. Girma
    2011 January 20 at 11:27 pm | #21

    Ron

    RealClimate don’t let me post a word!

    They do.

    Try it and you will find out.

  22. 2011 January 21 at 7:47 am | #22

    Girma,

    For what it’s worth: most of your comments look like “drive-bys”. You say what you have to say and you go away, only to return in a later discussion, with more of the same post. You must understand that moderators tend to get a bit ticked off by these kinds of posts, more so when they’re critical without being negotiable.

    You might be serious and well-intentioned, but your posts are not contributing to the discussion, most of the times. I, for one, stop reading drive-bys. I am very skeptical that they work and they invalidate your claim that “yes, but RC moderation.”

    I am saying this because you answered my question in the other thread. That was enough for me to show your willingness to discuss.

Comments are closed.
Follow

Get every new post delivered to your Inbox.

Join 27 other followers