Friday, June 21, 2013

Confirmation bias and anomalous anomalies at WUWT



Anthony Watts is wondering if the NOAA latest monthly report is in error in regard to May being the third hottest month.  GISTemp is currently showing May as the 10th hottest May on record, although it's May data is asterisked, meaning provisional I assume.  Not that Anthony went that far in his "investigation".

What Anthony does is compare anomalies.  What he fails to do is adjust those anomalies to reflect the different baselines.  For example:

  • NOAA - baseline is the average of the entire twentieth Century - at least as far as it's monthly reports go.
  • GISTemp - baseline is the 1951 to 1980 mean
  • UAH - baseline is 1981 - 2010
  • Weatherbell / NCEP  - baseline is 1981-2010
  • HadCRUT4 - baseline is 1961-1990

Anthony's confirmation bias


Anthony doesn't tell his readers that UAH and RSS don't correlate well on a monthly basis and he gets very cross with Zeke Hausfather for pointing out this fact.  (I wasn't aware either so it's good to know.) It does weaken his story a bit I suppose.

Zeke Hausfather says:
June 20, 2013 at 12:47 pm  UAH and RSS are not measuring land temperatures, and generally do not correlate that well with land temperatures on a monthly basis (though they correlate pretty well annually). The discrepancy with GISS is a bit more interesting, though there are some methodological differences that can lead to different values (e.g. NCDC doesn’t interpolate nearly as much as GISS); I’ll download the latest GHCN data from the NCDC web site and see how many stations have reported so far.
REPLY: Zeke no need to lecture me on what I already know (and routinely publish about) about UAH/RSS and the lower troposphere. I’m simply pointing out large discrepancies, usually not that large. BTW the 2meter reanalysis temp from WeatherBell has been right on in many occasions, so I tend to trust it as a parallel metric to NCDC. It shows near zero, like UAH/RSS. – Anthony

The thing is that Anthony takes Zeke's comment as a personal affront.  Zeke makes no accusation.  Does not use any coloured language.  All he is doing is making some straight up observations.  He even comments that it is "interesting", which Anthony could have taken as a compliment.

But no.  Anthony accuses Zeke of "lecturing". Anthony could have said nothing, or he could have replied: "Thank you, Zeke.  That extra information is useful for my readers" and maybe he could have added "who are generally extremely ignorant of all things climate."  

Instead he exhibits this.



Temperature series bias


Another cute thing is where he writes in his reply to Zeke above: "It shows near zero, like UAH/RSS", referring to the Weatherbell chart.

It's not the only spot - in the main article Anthony writes: "The RSS temperature anomaly dataset is also much lower than NCDC is reporting". Not only does he ignore the fact the baselines are different, he then shows the RSS data points for December 2012 through to May 2013, as if they have any bearing on May vs May records going back however many years.

There he's talking about anomalies from different baselines.  The "near zero" refers to how much UAH and Weatherbell (NCEP) are from their baselines of 1981 to 2010.  Whereas the NOAA is from the 20th century average.  So of course if NCEP shows "near zero" you'd expect UAH, which has the same baseline to also show "near zero".  But you wouldn't expect "near zero" if the baseline for the anomaly was different.  And the baseline for NOAA is different.

What Anthony should have done was compare ranking - ie 3rd hottest vs 10th hottest May or whatever, not comparing differences from different baselines.

But then Anthony Watts has always had trouble working with temperature anomalies.

Here is a comparison of UAH and GISTemp for the month of May from 1979 to May 2013.  I've set the x and y axis to the same scale.  I've then roughly aligned the two charts.  That's purely for illustrative purposes and is not something viewers should try at home!  Anyway, I think you should be able to see what a difference the different baselines make. (Click to enlarge.)

Correction: Oops! I inadvertently showed GISTemp for March instead of May.  I've replaced it with the corrected eyeballing :D




BBD posted a link to the charts with a properly aligned base year, which prompted me out of my laziness and so here it is.  My eyeballing wasn't too far off the mark, but it pays to do it properly.
 


Tamino does it better.  Although I think he got the Weatherbell observation wrong.  Anthony's chart does look like it's for the whole month of May, and is labelled as 1 May ---> 31 May.


PS I don't know why all the fuss and aggro over a monthly global weather report.  A month tells you nothing about climate.  A year doesn't tell you anything about climate either.  That is, unless the weather is pushing extremes and continues to do so.  It's the trend that counts, not May, June and July.

7 comments:

  1. It's been more than five years since Anthony published one of his masterpieces, "A look at temperature anomalies for all 4 global metrics: Part 1". It seems he still has troubles understanding baselines.

    ReplyDelete
  2. Replies
    1. Thanks, BBD. When I looked I realised I'd used the March data for GISTemp instead of May. Shows the trap that haste can pose, even for a scrupulously meticulous blogger like me :D

      I've corrected the charts above and added one with the propoer alignment.

      Delete
    2. It's something to do with climate blogs. They slowly send you mad. Why, only the other day, I mistook Martin Vermeer for the enemy...

      ;-)

      Delete
  3. Zeke wears a kick me sign. He's nice and knows his stuff and all that, but he wears a kick me sign.

    ReplyDelete
    Replies
    1. You could be right, Martin. But there's nothing in his comment above that reads: "Kick me!" other than the fact the comment was written on WUWT.

      I guess if you write a paper AND an article on realclimate.org, that shows that UHI corrections are adequate and that UHI does not bias the temperature records then you are certain to upset Anthony.

      Delete

Instead of commenting as "Anonymous", please comment using "Name/URL" and your name, initials or pseudonym or whatever. You can leave the "URL" box blank. This isn't mandatory. You can also sign in using your Google ID, Wordpress ID etc as indicated. NOTE: Some Wordpress users are having trouble signing in. If that's you, try signing in using Name/URL. Details here.

Click here to read the HotWhopper comment policy.