NCEP

Anthony Watts has published one of his most foolish posts ever, courtesy of Erl Happ. If this is the level of “science” from fake skeptics, then God help us all. But rather than rip the arguments to shreds (taking candy from a baby), I’d like to examine some of the data used, which mostly come from the NCAR/NCEP reanalysis data set.


The NCEP reanalysis data set is the output of computer models which take as input a vast array of meteorological data, including temperature, pressure, wind speed and direction, and many others, and simulate the weather in order to estimate a wide variety of data over the entire globe. As such it provides world-wide coverage of a great many interesting quantities. All quantities are provided at monthly time resolution, and at a variety of levels of the atmosphere where appropriate.

These are computer model output rather than directly observed data, but they are based on (and therefore constrained by) a great quantity of direct observations. The NCEP data are also internally consistent, based on the same model for the entire period from January 1948 to the present, which makes them suitable for examining climate change. Let’s take a close-up view of just one of the available variables, the estimated surface air temperature.

One of the striking facts is that the size of the seasonal temperature cycle is dramatically different bewteen the two hemispheres:

Note that the northern hemisphere has a much larger seasonal cycle than the southern. This is almost surely due to the greater thermal inertia of the southern hemisphere, which in turn is due to the greater proportion of ocean to land in that hemisphere. We can also see this in a “boxplot” of monthly temperatures by month of the year. For the northern hemisphere we see a range from winter to summer of over 12 deg.C:

But for the southern hemisphere, the annual range is only a bit over 6 deg.C, about half that of the north:

We can of course transform temperature to temperature anomaly (the difference between temperature at a given time, and its average for the same time of year) to remove the annual cycle. This allows an interesting comparison between the overall temperature changes in the two hemispheres (smoothed temperature anomaly):

Over the period from 1948 to the present, the two hemispheres have warmed by about the same amount. However, during the 1948-1975 period the southern hemisphere warmed while the northen cooled, making global average temperature roughly stable. This is consistent with the idea that northern cooling was due to sulfate aerosol emissions; those emissions were predominantly a northern-hemisphere phenomenon, so the north cooled while the south continued to warm due to increasing greenhouse-gas concentrations. Since 1975 both hemispheres have warmed, but the northern has warmed considerably faster than the southern.

Another interesting point is changes in the size of the annual cycle. We can estimate this with a wavelet analysis, which yields the time-varying amplitude (actually semi-amplitude, which is simply half the amplitude) of the annual cycle (I’ve plotted semi-amplitude anomaly so they’ll be roughly on the same scale):

There’s only slight indication of a decrease in the amplitude in the north, but great reduction of the amplitude in the south. This tells us that in the southern hemisphere, winter has warmed much faster than summer, but in the northern hemisphere the two seasons have warmed at nearly the same rate. The annual cycle has actually increased globally, because the offsetting influence of the southern hemisphere has been much reduced while the northern influence has remained strong.

I also extracted data for each 10-degree latitude band, 18 in all from the south pole to the north. Here are smoothed temperature anomalies for each, with each latitude band offset by half a degree so that the lines go from the southernmost latitude band (90S to 80S) at the bottom to the northernmost (80N to 90N) at the top:

Both poles have shown markedly faster warming than the global average since about 1990. A close-up on the latitude bands of the polar regions (both north and south) shows this is detail, and reveals that the north polar regions have warmed faster than the southern, although both poles appear to have warmed rapidly over the last two decades:

If we look at the equatorial region, we see that not only was the 1948-1975 cooling absent from the southern hemisphere, it was also absent from the equatorial region at least as far north as latitude 20N:

The mid-century cooling is confined to regions north of latitude 20N, but is clearly present in the northern hemisphere extratropics, which again is consistent with mid-century cooling being due to industrial emissions of sulfate aerosols:

There are more data fields in the NCAR/NCEP reanalysis data set, including temperature at multiple atmospheric levels, humidity (both relative and absolute), and many others. In future posts we’ll take a look at some of the interesting changes that have occurred in some of those other quantities.

22 responses to “NCEP

  1. “There are more data fields in the NCAR/NCEP reanalysis data set, including temperature at multiple atmospheric levels, humidity (both relative and absolute), and many others. In future posts we’ll take a look at some of the interesting changes that have occurred in some of those other quantities.”

    Watch out for the upper tropospheric humidity in that particular reanalysis.
    http://www.agu.org/pubs/crossref/2010/2010JD014192.shtml

  2. Rob Honeycutt

    Please take the candy away from the baby. The resulting temper tantrum is always a sight to behold.

    • I just finished reading the comments at Watt’s and Mr Happ got spanked pretty bad.
      Surprised am I.
      And about that candy…..
      It seems the baby was into the wine.

  3. David B. Benson

    Thanks, Tamino. That was of interest!

  4. Woah. The third comment is Watts himself disagreeing with a key premise of the post.

    The question I would like answered is… why would Watts post this if he himself knows there are errors in it? We know he doesn’t uphold high scientific standards, and you can excuse at least some of that because he’s not a scientist, but why post something by someone else that he knows is wrong?

    I’ve said it before and I’ll say it again: best science blog my arse!

  5. Turns out that Erl was on the sauce when he wrote the post. Tony has expelled him for writing while loaded.

    • JDR,

      Under no circumstances would I minimize the contribution that agriculturists make to climate science. In this case I have to say though that you may be right.

      Yeah the clouds come and they go and temps go up then they go down……

      and then from his bio:

      “I think there is room for a subject called historical climatology. We can learn a lot just by looking at what has happened over long periods of time. Good data sets are readily available on the net. All one needs is a little curiosity, a facility with spreadsheets and a lot of determination.”

      When I first looked at the post I thought it was Watts who knocked down a couple of cases of the guys wine and in his stupor let the post go through. Then when he came to He realized what he did and had to attack the post to try and salvage his “Worlds Greatest Science Site” reputation.

      Man this is bad. I need to take Ray Ladbury’s advice and quit clicking on those links to Watts’s site. Thing is you’re always tempted to see “how bad did it get this time”, then you go over there you get a feeling like you’re some kind of voyeur peeking behind the curtain at some perverse sideshow performance.

      Gotta show more discipline.

  6. I always enjoy your analyses however I should warn you that there are problems with taking reanalyses as observational datasets for climate change purposes which I’m sure you are aware of but which need to be stated either way. It is true that they represent our best synthesis of as much observational data as possible (depending on the analysis system) but you have to be careful when you look at trends because the observational database changes in time and the best reanalysis system in the world can’t avoid that. In particular, changes in the early 70’s are suspect because of the advent of satellite data, which is now the main source of data for weather analyses and hence the later stages of the reanalyses. There are almost certainly other observational system changes in the mix as well. Having said that it is true that there are uncertainties in all observational data that need to be kept in mind and reanalyses are no different. Also there are a number of global reanalyses out there and the quality varies considerably. It is probably fair to say that currently the best is the ERA-INTERIM reanalysis from the ECMWF.

  7. I found this comment intriguing, re changes in the amplitude of annual temperature swing over the long term.

    There’s only slight indication of a decrease in the amplitude in the north, but great reduction of the amplitude in the south. This tells us that in the southern hemisphere, winter has warmed much faster than summer, but in the northern hemisphere the two seasons have warmed at nearly the same rate. The annual cycle has actually increased globally, because the offsetting influence of the southern hemisphere has been much reduced while the northern influence has remained strong.

    This gives the impression that globally, summers have warmed faster than winters, counter to expectations. I’m struggling to see how a global increase arises from decrease in both hemispheres. It’s on the tip of my brain (Summer in the North while Winter in the South), but could you clarify a little to plant it more firmly for me?

    One (naive) interpretation here is that, globally, winters have not warmed faster than summers, and therefore there is something wrong with the theory of GHG warming, and I’ve read informal comments here and there to that effect. Seeing the amplitude change for each hemisphere provides some very useful context.

    When Fall et al came out and Anthony Watts and colleagues emphasised a diurnal trend for the US that ran counter to expectations, I went a-hunting for papers on the global diurnal trend. As far as I read, the global diurnal trend is negative, as expected, as long as the analysis commences from mid-20th century or earlier, but flattish if the analysis proceeds from ~1970 or later. IIRC this was the case for the Northern Hemisphere, too. I’m wondering if a similar kind of hemispheric offsetting (daily, rather than semi-annual) may impact on diurnal range trends.

    This is a not-so roundabout way of requesting a post (2nd time lucky) on diurnal range trends – though I see that NCEP is not fit for that task.

    [Response: The seasonal cycles are of course reversed in the two hemispheres. But the northern-hemisphere cycle is stronger than the southern, so the global cycle — which is pretty much the average of the two — is hottest when the northern hemisphere is hottest (northern summer, southern winter) and coldest during norther winter (southern summer). If the two hemispheres had equal-size cycles, the global cycle would be nil. In fact the size of the global cycle is proportional to the “size of northern cycle” minus “size of southern cycle.”

    Now reduce the size of the southern cycle significantly, and the northern cycle is even more dominant. The global cycle is still proportional to “size of northern cycle” (almost the same as before) minus “size of southern cycle” (greatly reduced). Since you’re subtracting less, the global-cycle size is greater.]

    • Aha – I almost queried in the former comment whether you had averaged or computed the difference between N and S hemisphere. It seems, unless I’m still mistaken, that they amount to the same thing for these two oppositely phased cycles.

    • Why inverting the SH plot isn’t an option? Anyway, the way you’ve presented the semi-amplitude anomaly plot is clear, though it requires a bit more thought. Thank for this, shared.

    • Maybe the clearest way to say it is this: Let x = N.Hem temp, y = S.Hem temp. They are “out of phase” so when one is low the other high, and vice versa. In fact suppose
      x = -A \cos(2\pi t)
      and
      y = +B \cos(2\pi t)
      with amplitudes A and B both positive, but A greater than B since the N.Hem cycle is larger than the S.Hem cycle. The global temperature is the average of the two hemispheres, or
      g = (x+y)/2 = (-A \cos(2\pi t) + B \cos(2\pi t))/2
      = \frac{1}{2} (B-A) \cos(2\pi t),
      so by averaging the hemispheres we difference the cycle amplitudes.

  8. Oh dear.

    Earl Happ’s notions of ‘science’ in his comments following his original post are even more outlandish. The fellow might actually be an unwitting comic genius in scientific parody.

    It is fortunate indeed that I was not imbibing a glass “of very nice Cabernet Franc” whilst reading that thread, else I would had to have bought a new laptop.

  9. God, even the creationist and far right ex-marine Dave Scott Springer dumps on Happ.

    That’s something …

  10. Thanks Tamino, very understandable.

  11. Many NCEP/NCAR reanalyses variables contains false trends due to discontinuity at the introduction of a new observation system, not just tropospheric humidity.
    For instance i see a graph with 200hpa(20-30S) temperature that “shift around 1976-1980” according to WUWT post, actually that increase is just a spurious step change that occurs at the introduction of satellite data in jan 1979 as demonstrated here(appendix B):
    http://pubs.giss.nasa.gov/docs/2008/2008_Chen_etal_1.pdf

    see also the ccsp report for a discussion of reanalyses false trends:
    http://downloads.climatescience.gov/sap/sap1-3/sap1-3-final-ch2.pdf

  12. Ugh, the stupid it hurts.

    Jeff D says:
    October 6, 2011 at 12:06 pm

    This is why I love this site. There will be more peer review here about a hour then has been done on the team in years.

    Really? Really!? Some people’s bar for what constitutes peer review is set incredibly, pathetically low.

  13. Lars Karlsson

    The changes in albedo due to the winter snow cover should also have an impact on the seasonal cycle of the northern hemisphere.
    Here is a nice animation from NASA showing that:
    http://earthobservatory.nasa.gov/GlobalMaps/view.php?d1=MOD10C1_M_SNOW&d2=CERES_NETFLUX_M