Where most of us live (with apologies to southern-hemisphere readers)

Almost all of us live on land, not the ocean. And, most of us live in the northern hemisphere, not the southern. For the benefit of most of us, let’s take a closer look at how temperature has changed, in the northern hemisphere, on land.

There are several data sets to choose from. I’ll focus on two of them, from NASA and NOAA. It’s bound to arouse the ire of the usual deniers; whenever data show that global warming might be a serious problem, their main tactic is to slander both the data and the scientists who release it.

Some of you might even remember some years back when the Berkeley Earth Surface Temperature project was organized by Berkeley physicist Richard Muller because he suspected that something might be fishy with global temperature data sets; the climate scientists might be just plain mistaken through lack of competence, or might even be “fudging” the data to make global warming look serious. He was determined to create a new, comprehensive but uncompromised, global temperature estimate. The denier community was thrilled; they eagerly awaited confirmation from a skeptic that all those other guys tracking global temperature had got it wrong. Anthony Watts, one of the most adamant data assassins, said that he would accept the results no matter what they were.

Unfortunately for them, not only did the Berkeley Earth data show that the other guys got it right, Richard Muller turned out to be a genuine skeptic rather than denier. When he saw the results with his own eyes, he believed what he saw. All the deniers who had so adored and embraced Muller, and swore to believe what was discovered, turned on him like a pack of wolves.

But the Berkeley Earth temperature data isn’t kept constantly up-to-date like that from NASA and NOAA. It will, however, be useful to display what I was able to get of it for northern hemisphere land in order to show how my two choices, NASA and NOAA, compare to the Berkeley data — the one the deniers said they’d love until they found out what it shows.

First let’s compare Berkeley data to NASA in those months for which both report results, with Berkeley data in blue and NASA data in red:


The match is excellent, the main difference being that the Berkeley data, the one the deniers said would settle the issue once and for all, actually show a wee bit more warming than the NASA data. Here’s a comparison of Berkeley and NOAA data, with Berkeley again in blue and NOAA in red:


The match is, quite simply put, remarkable enough to remark on. The main difference, if you study the differences closely, is that again the Berkeley data show a wee bit more warming that NOAA data.

We want to know what has happened up to the present, which for monthly averages means through June of this year. To make things easier to see I’ll compute yearly averages, and even though 2015 is only half complete I’ll include this year’s year-so-far average so we can see how 2015 is shaping up by comparison. Let’s start with NASA:


Several things are worth noticing. One is that since about 1970 temperature has risen steadily and consistently. It certainly shows a lot of fluctuations, jiggling and wiggling up and down in addition to its steady long-term rise, but the trend is consistent. It has consistently gotten warmer for over 40 years now, it hasn’t stopped, it hasn’t even slowed down.

The same is true of global (not just northern hemisphere) temperature combining land and ocean (not just land), but that hasn’t stopped deniers from clinging to the idea that earth’s temperature has stopped rising (they call it a “pause,” I call them “pausemaniacs”) or at least slowed down (they call it a “hiatus”). Their chosen “pause time” (which I call “denial time”) varies, but it usually starts in 1998 and ends with 2013 or 2014:


To get an idea of the trend, apart from those up-and-down fluctuations, we can apply a bit of mathematics. One good way to identify real trend changes, not just wiggling around that might look like a trend change, is to apply the technique called change-point analysis. It identifies several times in the past when we can have some confidence that the actual trend, the rate of global warming, changed. There isn’t any change point in at least 40 years, temperature really has kept going up and not slowed down. The estimated trend it gives is the type of mathematical function which is referred to as piece-wise linear.

Another useful analysis is to smooth the data, i.e. find a smooth curve that approximates the trend not the fluctuations. For these data the two methods give very similar results (piecewise linear in blue, smooth in red):


Another noteworthy sign is that this year, so far, is on track to beat the pants off any previous year for hottest on record.

That’s what the NASA data say, what about NOAA? It shows much the same story, but estimates that this year isn’t just on track to beat the pants off the hottest on record. It is, so far, (to use the vernacular) “wicked hot”:


We can apply the same mathematical techniques to NOAA data, which again finds no change point (no trend change) for over 40 years, and again the two trend estimates are quite similar:


Kinda puts “denial time” in the trash bin.

Finally, let’s zoom in on more recent times, since 1970. Here’s the NASA data:


The red line is a fit to the data by least-squares regression. This makes abundantly clear that in addition to its rising trend, temperature shows constant fluctuations up and down. It also shows how silly it is to claim that temperature has “paused” — only those who aren’t willing or able to face the truth believe that those fluctuations somehow make a genuine change in trend.

NOAA data show the same thing, as well as just how hot this year-so-far is:


Analysis of the NASA data suggest that the land area of the northern hemisphere is warming at a whopping 2.87 deg.C/century. On a scale more familiar to most Americans, that’s 5.2 degrees Fahrenheit per century. Imagine, if you will, the entire northern hemisphere land area getting over 5 degrees Fahrenheit hotter. That’s what NASA data put the present trend at. And that’s just the present trend; we expect within a decade or two that warming will get even faster.

NOAA data, however, put the trend at 3.12 deg.C/century, which for most Americans translates to 5.6 degrees Fahrenheit per century. Now, imagine the whole northern hemisphere land area getting even hotter than the NASA trend suggests.

Then imagine the warming rate getting even faster. Suddenly the phrase “wicked hot” seems like a frighteningly apt description.


39 responses to “Where most of us live (with apologies to southern-hemisphere readers)

  1. Dan Andrews

    I think I’ve been a bit complacent. Those numbers (i.e. wicked hot) are a kick in the pants.

  2. Philippe Chantreau

    I can imagine the deniers catchphrase for next year: “There has been only one year warmer than 1998 by a statistically significant margin…”

  3. This post will spread like a Canadian wild fire

  4. Denial time.
    It will become common usage by Friday.

  5. David B. Benson

    Wow. Thanks.

  6. What is the baseline period used for the anomaly? I think normally NASA and NOAA use different baselines but I presume you applied the same one to all data sets.

    [Response: NASA uses 1951-1980, NOAA uses the 20th century. Ordinarily I’d have aligned them, but didn’t bother because those two baselines are already extremely close.]

  7. What smoothing function did you use? Low pass filter or something?

    [Response: Slightly modified lowess smooth.]

    • “Slightly modified” :-o …Omigod. You mean you are hoaxing us by fiddling with the “raw numbers”!!! For shame, for shame!

      I’ve never understood the mind that could pick a local max of 2 to 3 sds (depending on the source distribution) and then “infer” a nonsignificant era for later points before the actual trend in that distribution has a chance to catch up. It simply makes no sense to me. There is obviously something wrong in my education.

  8. Southerners are probably a bit sensitive at the present, (Didn’t I hear somewhere that they have had their flag banned.) so to cheer them up a bit – Southerners may be few in number (10% of the population – see graph of population by longitude here.) but that means they have about ten-times more globe than us Northerners do. That’s over four-times more land as well as a whopping 50-times more ocean.

  9. HTML mistake in my previous comment feel free to delete it.


    The excess of land mass in the Northern Hemisphere (NH) in comparison with the Southern Hemisphere (SH), coupled with the greater uptake of heat by the Southern Ocean in comparison with northern ocean basins means that the NH generally warms more than the SH. Arctic warming is much greater than in the Antarctic, due to the presence of the Antarctic ice sheet and differences in local responses in snow and ice. Mechanisms behind these features of warming are discussed in Section 12.4.3.

    Land warms faster than the ocean because more of the additional heat goes to warming the air and less to evaporation. This especially when the soil is dry and because the NH has so much land, it is also on average far away from the ocean and dry. This effect will always be there.

    A second main reason why the oceans warm slower is their large heat capacity, their large heat uptake. This effect is in principle transient, but on very long time-scales.

  10. Reblogged this on Hypergeometric and commented:
    Love the comparison with Berkeley Earth Surface Temperatures.

  11. Harold Brooks

    Nit-Muller is a Cal-Berkeley physicist. I’m not sure whether Cal or Stanford would be more upset with him being labelled Stanford.

  12. I predict with high confidence that this year marks the beginning of a new pause…


    • Horatio Algeranon

      “The Trend of an Error”

      The birth of a pause
      And trend of an error
      With physical laws
      Suspended forever

    • michael sweet

      Global temperature rise from an El Nino often lags the El Nino by a few months link . Since this El Nino is forecast to last until next spring, you may have to wait until 2016 to mark the start of a new pause. We will see how the El Nino holds out.

    • Maybe. By analogy with ’97-’98, though, it could turn out that ’16 is the ‘new pause year’–if any such eventuates.

  13. Horatio Algeranon

    I was out on a boat yesterday in Long Island sound.

    Though it was very hot on the shore, it was very nice on the water.

    I think I will move to the water
    Cuz that is where it is cool
    To live and swim like an otter
    Avoid the heat as a rule

  14. The map produced by GISS should NH Land trend predominately (ie modal rather than mean) of 1 to 2C per century somewhat lower than your figures. I suspect you land trend is biased upwards by arctic trends where most of us don’t live.
    The GISS maps indicate a better case for most of us although this will be no comfort for those knee deep in melting permafrost! Not that this should make anyone complacent without mitigation of our emissions we should expect an increasing rate in line with CO2 forcing.

  15. Global Feb 2014, the NH land was the 53rd warmest in the NOAA record. The pause held so much promise.

  16. Am I correct that we can expect satellite data to show greater response to the current el Nino and that the 2015 satellite data may well show higher temperatures than GISS, NOAA, Berkeley, HadCrut or other surface thermometer based datasets?

    • It hasn’t been so far; it’s been high but not comparable to ’97-’98. We’ll see what happens as the Nino progresses. But you’re right in general; the satellite datasets are much more responsive to El Ninos than the surface record, as can be seen in this graph:


      (Since WFT doesn’t afford an ENSO index among it’s data choices, you’ll have to just look at the big Nino peaks in the UAH data in ’86-88, ’91, ’95, ’97-’98, and 2010.)

      If I were betting, I’d bet on your expectation to turn out to be correct.

    • Regarding the lags… I found Tamino’s earlier post calculating the temperature index (surface indexes and lower troposophere) lags for El Nino, volcanoes and solar intensity…

      Please see:

      Best-fit lags are different for different data sets, but the lag for el Nino is 3 or 4 months surface, 5 months lower-troposphere. For solar, 2 months except GISS 3 months. For volcanic, 8 or 9 months surface, 3 or 5 months lower-troposphere. Hence the lower troposphere temperature responds sooner to volcanic eruptions, later to el Nino, and at about the same time to solar.

      How Fast is Earth Warming?
      Tamino, January 20, 2011

      The article includes this chart:

      Surface temperatures will peak one to two months before lower troposphere.

      The same sort of calculations make it possible to subtract these influences from the trendlines and here are his results:

      The Real Global Warming Signal
      Tamino, December 6, 2011

      • PS

        Sorry, the chart you will want to look at (for the El Nino surface and lower troposphere temperature index lags) is in the first article I link to, and it is the first chart in the “Update” section…

  17. Timothy (likes zebras)

    “It also shows how silly it is to claim that temperature has “paused” — only those who aren’t willing or able to face the truth believe that those fluctuations somehow make a genuine change in trend.”

    This also includes prominent mainstream climate scientists such as Richard Betts of the Met Office Hadley Centre. All humans can get mislead by seeing patterns in what turns out to be short-term noise, but this issue of the “pause”, and how it managed to get its way into the IPCC reports as a “hiatus”, is a worrying sign of the success of the fake sceptics in framing the narrative.

    • Partly. But if you look at the hiatus box in Chapter 9 of AR5, WG1 you will notice that hiatus is placed in scare quotes. Of course deniers either leave them out of any quote, generally.

  18. The Antarctic Sea-Ice supposedly acts as a barrier to stop heat escaping from the ocean. Resource bottlenecks are the main problem predicted when we all decide at once that it’s on like Donkey Kong!!

    If I may, the Southern Hemisphere(Australia) is just coming to grips with a parliament that doesn’t work: we know there are long term rainfall changes and the fact we are doing nothing is simply political shennanigans. How is the drought in California going btw?

  19. We are bad at eyeballing trends in noisy data. It is hard or even impossible to visually spot where the measure of central tendency is. This is why you need to do regressions in such data.
    What you tend to do when eyeballing noisy data is to look at the envelope, the pattern that you can see in the points towards the edge of the data cloud. Unfortunately being extreme values they are highly variable and you can easily see a pattern generated by chance. In the global data set your eye is drawn to the apparent small tend of the maximum temperatures after 1998. But this is an illusion in the GJSS data set largely created by one point, the high temperature in 2005.
    Tamino tested broken stick and polynomial models and found their departures from linearity not to be significant. His tests were conservative. Since autocorrelation was not accounted for and tests were made for multiple models the chance of finding a significant departure from linearity even if there actually was not one was higher. Anything significant on his simplified analysis would be significant on a more complex analysis and what he did is easier to explain than a more complex analysis.
    What I though was missing in his analysis was an examination of whether there were any departures from normality that the parametric models that he used would nor efficiently capture. To do this I fitted non parametric smoothers to the data. I used penalized regression splines. I fitted them using the gam function in the mgcv package in R. The smoothing parameter was chosen by generalized cross-validation.
    This approach is also conservative. It does not account for autocorrelation and a moving average process might lead to a spurious curvature.
    When I used the annual global data from 1970 to 2013 and from 1970 to 2014 the fitted curve was a straight line. There is no evidence of curvature over these time periods. If you choose shorter time periods you can find curvature but for the reasons that I gave above I would need to undertake a more detailed analysis. I found curvature in the 1980 to 2013 data set but not in the 1980 to 2014 set.

  20. Ooops! I meant departures from linearity, not normality.