Comments on “To Tell The Truth: Will the Real Global Average Temperature Trend Please Rise?

March 18, 2008

I appreciate the detailed write-up that Basil Copelanddid for Anthony Watts’ “Watts Up With That” over the last several days.  However, I think the temperature data is being dangerously over analyzed.  I must agree with Christopher Monckton, as reported here, when considering the recent low tempertures:

“… for goodness sake the one thing we mustn’t do is crow,” he said in an interview. “Yes, the temptation of course is to get excited and say this proves the alarmists wrong. But a single extreme weather event in either direction does not prove anything.” 

We have been rightly critical of the global warming alarmist for years for seizing upon every climate extreme as evidence for climate change.  Let’s not seize on these last few months or low temperatures as the basis for a counter argument.

I have reproduced Basil’s smoothing using the Hodrick-Prescott filter (which I got here) for the UAH temperature data.   The UAH data temperature data runs from 1979 to February of 2008.  But I also repeated the smoothing by ending three months early (Nov. ’07), six months early (Aug. ’07), nine months early (May ’07), and 12 months early (Feb. ’07).  The entire run, starting in 1979 can be seen if figure 1, below.  Figure 2 zooms in on just the last ten years, and shows the temperature trend at the end of each smoothing run.  This simply demonstrates that this smoothing technique, like most others, can be quite unstable near the beginning and end of a time series.


Figure 1.  Reproduction of Basil Copeland’s Hodrick-Prescott smoothed UAH temperature data.


Figure 2.  Smoothed UAH data zoomed in to last 10 years.  Note how the final slope varies as each 3 month period is removed from the data.

If the recent cold temperatures hold, or continue to drop, then we will have plenty of time to point this out as the trend becomes more obvious.  If the temperatures go back up over the next year or so, then the alarmists will gleefully point out how wrong this analysis is, ultimately giving us less credibility. 

 Thanks again to Anthony Watts and Basil Copeland.  Watts’ “Watts Up With That” is on the front-lines in the battle between hysteria and sanity.  I made a donation to his tip jar today, and recommend that readers give what they can.


  1. Tom,

    I’m not fixating on the end point by itself. If people will try to understand better what I’m saying, or doing, then I don’t think there is anything to fear about losing credibility.

    I’ve replicated your results, but then applied them the way I think they should be applied. Taking the extreme case, smoothing only to May ’07, the smoothed series yields a decadal trend of 0.133, versus 0.142 using straight line regression.

    Everybody seems to think my case hinges on the downturn at the end of 2008. It doesn’t. The downturn does make the difference more dramatic, but that’s not the main point.

    I come from a perspective where a climatological normal is based on a 30 year average. We don’t have 30 years of satellite data, but once we do, I’d be quite happy to see a consistent approach applied over a running 30 years, and the trends fall where they may. For now, and for some time to come, any straight line regression through the 1998 El Nino is going to produce a biased estimate of the 30 year normal. Straight line regression does not handle a shock like that well. I’d be happy to see the bias factored out, or moderated, through smoothing, a dummy variable, or whatever, just so the approach is consistent.

    As for what’s going to happen if we were to keep appling the HP smoothing to the data as we move forward from this point, I fully expect the actual anomaly to revert toward the mean, and for this to dampen any downturn in the smoothed series. A little further out, it is anybody’s guess as to what’s going to happen, or what is going on.

    It is bit frustrating that there seems to be so much misunderstanding about what I’m saying, which is: don’t over play the downturn or leveling off in the data since 2001, but do give it proper weight. A straight line trend doesn’t do the latter.


  2. Tom,

    There is a constant linear trend that can be extracted from the HP filtering. If you are interested in how to do it, let me know.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: