Saturday, January 03, 2009

The baseline game
Ethon came flying in from Boulder for the New Year. He brought news of the second coming of Willis Eschenbach. You might remember Willis, he is the guy who, as Tim Lambert gently put it used some slight of hand on the Hansen 1988 prediction

Well, there are only two things you can do to make Hansen look bad -- you can misrepresent the results of his model, or you can misrepresent the instrumental record.
Willis took door two over at Climate Fraudit when he drew in the HADCRUT3 data.
"The trick Eschenbach used was to use a single year for the baseline (1958-ER) instead of the thirty year average that is normally used (Hansen used 1951-1980-ER). Yes, it's another version of the disingenous baseline game that produced all those bogus "global warming ended in 1998" claims. Given the year to year variability of climate, by choosing the right year to use as a baseline you can manufacture almost any result you want."
If you use the same thirty year average for the baselines, you get the temperature anomaly graph on the right. (Since Hansen et al. 1988 used 1951-1980 and HADCRUT3 used 1961-1990 you do have to be a bit careful). Lambert has a much sexier way of showing this, but Eli is only a humble hare.

There was lots of patting on the back over at CA, Willis was the hero of the day and then he made the mistake of getting into it with the crew at Deltoid and starting to blather about forcings and scenarios. It was pretty amusing when it turned out that he was using greenhouse gas growth scenarios from a 2006 Hansen paper and ascribing them to the 1988 one. Quite cross with Eli he was.

If you thought that was a class act, look at what Roger Pielke Jr. has done to address the question if global warming has stopped


Will someone go over there and mention that his act is wearing very thin.

UPDATE: Roger, points to a letter he got into Nature in April 08 with another version of his graph, one at least that uses the right baseline. . . . . . . except for the 2007 IPCC report which starts in 1990 about 0.05 oC above the others. If you move the green line down, Eli did, it lies just about on the 2001 IPCC projection which is fairly good. Roger, of course, only stated the 2007 projection in 2000. You can blow this figure up to see it better by clicking on it.

Oh yes, you can buy Ethon T Shirts over at the Miserable Shop of Prometheus. Tell em Eli sent you. If you are in Boulder, wear them to class.

Comments?

26 comments:

Anonymous said...

If you take a trend of the GISS data from 1990 - present, it has a slope of about 0.2C/decade. In other words, the same as the trend shown on RP's graph. But of course, the way RP shows it, it looks like the IPCC projection greatly overshoots the GISS data.

The difference is the y intercept, of course. In other words, the GISS trend line does not pass through the temp for 1990.

No surprise there (or at least there should be no surprise there -- at least not to anyone who understands linear regression.

Linear regression involves "splitting the difference", so there is no way that it would pass through that value unless all the temps fell perfectly along a straight line.

Because the temp values include the effect of noise, the "most probable" temp for 1990 if the noise could somehow be removed would be below the actual temp by about 0.15C. The drop in temp at the start of the sequence is due to the pinatubo volcanic erruption (ie, noise on the signal)

I really find it hard to believe that Pielke does not understand this.

Magnus said...

Any one left who can post there?

Anonymous said...

Of course,

Kevin Vranes (but only on odd Tuesdays of the month)

Anonymous said...

It's easy to see how dumb/misleading RP's zeroing scheme is by using his method but instead of starting in 1990, starting in 1989.

In 1989, the GISS temp anomaly value was 0.2 deg C below what it was in 1990 (0.2 vs 0.38)

Adding 1 year to the GISS data for the period iun question 1990-present has very little effect on the slope of the trend 9it's still essentially 0.2C/decade, same as IPCC 2001 projection for the period 1990 - present)

But if RP started his graph in 1989 and had used the same zeroing convention -- ie, made the starting point for the projected IPCC trend for 2001 (1990 - present) coincident with the GISS temp value for 1989, the effect would have been completely different. In that case, most of the GISS values beyond 1990 would fall above the IPCC projection rather than below it as in RP's graph.

What RP is doing is simply stupid -- AND misleading.

Anonymous said...

I think RPJr as well as at least Lucia Liljegren, Anthony Watts and Michael Asher don't realize just how stupid they are. Ie don't assign to malice what is just stupidity. They really think they are doing something important and worthwhile and discovering new things.

This after having done some messaging with all of them and not seeing them understanding very basic points about series analysis.

So I think it's uncurable. Either they should stop making claims (the claims wouldn't be of better quality if they were pro-IPCC either, though their effect wouldn't be as harmful as it is now) or people should ignore them. There is no capability for improvement.

Funny how in this new world, with the "marketplace of ideas", you can grab such excrement they produce and present it in politics with a straight face.

Anonymous said...

Oh wait I mixed up Michael Asher from Techdaily and Steven Goddard from The Register.
I haven't messaged with the former.

Anonymous said...

"They really think they are doing something important and worthwhile and discovering new things."

I think the primary problem is the belief that "blog science" is real science. It's not. It does not have any of the built in checks that publication in the peer reviewed journals has.

One can say pretty much anything one wants on a blog and there is ALWAYS someone who will pick up on and repeat the claim as if it were true.

With "blog science", it is also easy to ignore or dismiss criticism that one does not like -- because blogs are very much an ego thing.

Anonymous said...

It took a little staring to understand he'd actually done what he did.

If you're like me before coffee you'll stare for a while before you figure out what's the problem with

"... the GISS trend line does not pass through the temp for 1990. "

So let me try for anyone as challenged as I was, here's the game.

PSr shows straight lines, the longterm trends. But he starts them from a high point in the variation of the actual observations.

PSr shows wiggly lines, the observations.

But if you were to draw the longterm trends _for_the_wiggly_lines_ where would they fall?

Answer: they'd be crossing left to right more or less in the middle of the wiggly lines, rising from left to right -- and they wouldn't reach either end of the wiggly lines like PSr's straight lines do.

Maybe someone with a crayon or lipstick can improve on PSr's chart there.

Anonymous said...

Here's a graph showing the linear trend of UAH monthly anomaly for 1979-2008 and two sub-periods.

The trend is lower (0.13 deg C per decade) than the other satellite and surface data sets. But even in UAH the 1979-2008 trend is greater than the 1979-2000 trend. So the observed trends simply don't support the "global warming has stopped" blather.

http://deepclimate.files.wordpress.com/2009/01/uah-tlt-global-1979-2008.gif

Anonymous said...

The elder Prof P. gets into the act with some chopped beef liver:
Rural Myth Busting: Breaking Wind

skanky said...

"What is it about the climate change debate that causes previously excellent scholars to go absolutely insane and disregard all standards of research integrity?"

http://tinyurl.com/a6u7l6

I'll leave it up to others to discuss RPJr's previous scholarship, but this isn't the first time he's been shown to be clueless on this subject (see JEB passim).

Anonymous said...

This guy has a Bachelors in math?

Good grief.

That graph is just idiotic.

It's should be embarrassing to anyone who calls himself a "professor of science".

Anonymous said...

I posted above that
if RP started his graph in 1989 and had used the same zeroing convention ...most of the GISS values beyond 1990 would fall above the IPCC projection rather than below it as in RP's graph.

But I just realized something.

The IPCC projections started at the beginning of 1990 and the observed temperature data RP shows (for GISS, for example) is for the end of 1990.


In other words, RP made an(other) error.

He should actually should have started his graph using the data values for end of 1989 (not end of 1990).

The data swings upward by almost 0.2 deg C from end 1989 to end 1990. (At end of 1989, for example, the GISS temp anomaly value was 0.2 (vs 0.38 at end of 1990) )

Though the offset is different, you can see the uptick in temperature over that year clearly by looking at Tamino's graph here.


In other words, not only is RP's zeroing convention bunk, but the data he shows does not even match the proper year!

And funniest of all , if his graph were adjusted so that the data did match the proper year, it would imply just the opposite of what RP apparently thinks it does. For example, virtually all GISS values from 1990 on would lie above the IPCC projection (for 2001, started in 1990)

Dano said...

Look at the commenters on that RP Jr thread - Biggs, Clarke, Bahner.

They can't comment elsewhere because they get laughed off the thread. There, they are safe from derive laughter (and Dano's Bahner klaxon).

Comedy gold.

Best,

D

Anonymous said...

A bit off topic, but interesting nonetheless
International Energy Agency 'Blocking Global Switch to Renewables'

International Energy Agency accused of consistently underestimating potential of wind, solar and sea power while promoting oil, coal and nuclear as 'irreplaceable' technologies


by David Adam

Anonymous said...

The Id,

Hey rabbit, I just read your weak assed comment calling me stupid.

No class at all, eh big guy.

EliRabett said...

=:>

Anonymous said...

Isn't it
=:x

You see anatomically more correct:
http://www.thereddragonhood.com/images/rabbit.jpg

Anonymous said...

Where are all of Roger's fanboys (and fangirls) defending him from this scurrilous attack funded by the carrot industry?

Will it be the Gavin did it gambit or the Webster's New World defense? Or perhaps it will be the famed--Hey, look! It's cold outside!

Anonymous said...

Their way of dealing with quite legitimate criticism is to simply ignore it.

On the web, it is actually quite an effective strategy.

In real life, not so much.

What RP does not seem to realize is that there are consequences for posting stuff that is just dumb (even if few read it).

After the nth time, (where n is fairly small) scientists stop taking you seriously.

Perhaps he does not care?

Anonymous said...

I understand that RPjr's graph plots observations relative a different baseline (1990) than the projections (1980-1999), and that the appropriate mid-point should have been the end of 1989.

But why is the IPCC 2007 projection above that of IPCC 2001? This seems to be based on the idea that the 2007 projection starts from 2000, and implies that the preceding ten-year period had a significantly higher trend than 0.2 deg. per decade. It looks to be set from an intercept 0.04 deg. higher than the IPCC 2001 trend, implying 0.24 deg. rise in the previous decade. I don't that this can be jusitifed - or am I missing something?

By the way, in a parting shot over at CEJournal, RPjr said:
"Oh yeah, I forgot to mention that you can see the same figure through 2007 with the observations plotted against a 1980-1999 baseline here:"

http://sciencepolicy.colorado.edu/admin/publication_files/resource-2592-2008.07.pdf

Of course, that "same figure", unlike the one above, shows several observations above the trend lines, and would presumably show even more if the other corrections suggested in this thread were applied.

EliRabett said...

Thanks for the heads up.

However, as we say in the rabett hutch, that raises the point about why did the dear boy not use that figure from Nature in his blog post.

Eli agrees, the 2007 line looks a bit higher than it should be too, as it does not extrapolate back to a zero at 1990. What it looks like is that it is parallel to the 2001 line pretty much and was moved up so that you can see it. which would answer your first question.

Auditing is such fun.

EliRabett said...

Honest broker, or three card baseline player. Can’t you just see Roger on the street in downtown Boulder with a cardboard box, a set of crayons and some graph paper looking for some action

Anonymous said...

Second try ...

Eli, you said:

"However, as we say in the rabett hutch, that raises the point about why did the dear boy not use that figure from Nature in his blog post."

There's a couple of different possible answers to that question. But notice that RP claimed that the Nature graph "had the observations plotted against a 1980-1999 baseline", but that the "trends are with respect to a 1990 baseline." In other words, he implies that the newer graph is comparing "apples" with "apples" (i.e. 1990 baseline all round) - totally ludicrous.

As you've already noticed, the IPCC 2007 trend is even more displaced in the Nature graph, so the offset was and is surely deliberate.

Finally I note that RP claimed in his year end post that "it is obvious that global temperatures have not increased over the past 8, 11 or 18 years ...." But this claim was missing from the Nature commentary which only went to 2007. Can there be any clearer indication that the contention is based on the addition of data for one single year, a year that happens to have featured a strong La Nina?

Anonymous said...

All this offset stuff is a sideshow.

The real problem with what RP is doing is that he is comparing observed temps to projected temperatures , rather than the trend in one case to the trend in the other. Sure, he is using an IPCC projection, but what is actually comparing on his graph is the observed temp and projected temp in each individual year.

That's nonsense.

The trend is the only thing that is meaningful in this case.

Anonymous said...

Why resort just to baselines

http://www.ncbi.nlm.nih.gov/pubmed/18070041

ok, that's only local.