"But every time the data gets revised and homogenized, the trends keep increasing."

That's because it is WORSE THAN WE THOUGHT!! ;>)

This seems to have been conclusively refuted. Giorgio Gilestro says he has done the right calculation. He has looked at all the stations in the GHCN set for which adjustments were performed (not just Darwin). He shows the distribution of the effects of the adjustment on trend. It looks quite symmetric. Adjustments are just as likely to cool as to heat. And his Python code is available.

PS: I consider changing station numbers, dropping and adding stations as adjustments also.

Nick,

seems to have been

says he has done

distribution of the effects of the adjustment on trend

just as likely to cool as to heat

Just to be bitchy, if the adjustments have 0 effect why do them?? You are playing with numbers here. There is NOTHING that shows whether any of these adjustments are useful to actually IMPROVE our understanding of temps.

The real issues are ignored. He does a run on ALL the stations in the GHCN records that pass their scrutiny. He then says the adjustments make no difference. That would be true IF GISS AND HADCRUT USED THE SAME SET OF STATIONS!!!!

GHCN MAY be doing a good, but pointless, job of adjustments. GISS and Hadley aren't.

Quit throwing trash at us Nick!!!

Willis chose the GHCN adjustments, looked at just one (Darwin) and a huge fuss was made, echoed in this thread. His underlined words:

Those, dear friends, are the clumsy fingerprints of someone messing with the data Egyptian style … they are indisputable evidence that the “homogenized” data has been changed to fit someone’s preconceptions about whether the earth is warming.

Now GG has shown that if you look at all the stations in the set, there is no bias - adjustments are about as likely to shift the trend down as up.

And the silly thing is that Hadcrut, which is the set relevant to the IPCC pix that Willis showed, didn't even have a big Darwin adjustment at all. Nor did GISS.

Why do adjustments? Because you have to. I can just imagine the complaints from you guys if no adjustment was made.

Well Nick,

It looks like someone a lot smarter than me found an answer to why they do those adjustments that average to 0!!!

http://statpad.wordpress.com/2009/12/12/ghcn-an...

Sorry I've been a bit unresponsive, but I've been doing my own R programming. I've verified Giorgio's calcs. I've tried to post at his site, and Romanm's, but they haven't shown yet. They are at the Air Vent.

I have been travelling so excuses for being unresponsive too. DISQUS
handles the comments, so they may be held up there. In point form:

1. the adjustments are all ad hoc. Some might affect trend, some not.
The Torok and Nicholls appear not to, but are then adjusted for
homogeneity (again) to make the 'high quality Australian' network,
which does apprear to differ. So why to these later adjustments
produce a trend.

2. while some literature about false negative 'detectability' exists,
little about false positives in adjustments. Its not hard to image
homogeneity adjustments 'in theory' that produce false adjustments of
trend.

The GHCN adjustments are not ad hoc. Read the documentation. They are based on a change detection algorithm of the kind you are fond of. What they apply is virtually a Chow test.

Why are the original data, the magnitude and the reason of the correction not made available in each case?

They are in the computer, right?

I am a mathematical physicist, I read the papers on corrections. They indicate reasons for corrections but there is no standardized algorithm; the magnitude of the correction is left to the adjuster, and based on metadata.

Why is all this carefully hidden by GISS?

Just saying TRUST US, WE KNOW HOW MUCH TO ADJUST THE DATA AND WHY makes it all unverifiable from the outside and certainly NOT SCIENTIFIC DATA (to put it poetically, it gives a mirror into the soul of the adjuster, rather than a prosaic mirror of reality....)

If I had this metadata and correction I could have, e.g. my Australian student who is from Darwin check what was going on with temperature records there, and all would be for the better - that is if GISS doesn't have anything to hide.

Mr. Stokes, do you realize that until you release case by case the magnitude and reason of the adjustment your data IS UNVERIFIABLE thus NOT SCIENCE. People who do not handle their data the way scientists are supposed to do - i.e. make it available and verifiable in full - do not qualify as scientists, no matter what degree they have - I think the term is 'hack'?

Please don't take it personally, but please just post the data, and feel good about it! I am sure that the fellow from CRU who leaked the emails feels much better now...

First, it isn't GISS, it's GHCN. And the pre- and post- adjustment data is posted; that's why this analysis is popular.

GHCN adjustments are not made for specific causes but in response to a change detection algorithm. So there's no reason to record.

If you want to see all this in reverse, I've analysed the case of Coonabarabran on my . Big, jumpy adjustments, but heading down, not up.

My idea of a generic framework would be the kind of thing adoped by
image processing (smoothing).
See http://www.ph.tn.tudelft.nl/Courses/FIP/noframe...
for example of some filters. The algorithms used in homogeneity
adjustments possess elements of a median filter or a kuwahara filter.

I have not seen anything in climatology where the homogeneity
adjustments are analysed from a generic perspective, their pros and
cons relative to statistical abstractions assessed, freedom from bias
proven etc. Because the use some documented method in some part of
their process does not mean they are not ad hoc - I read the
documentation. Look at how many times they apply it, choices of number
of breaks, termination conditions -- all non-standard.

Well Nick, you are disagreeing so I would expect you to show reason to disagree. Please show us where HADCrut and GISS get their data and how they adjust the RAW data to get their final output. If, like GISS, they use adjusted products from an intermediary I would expect you to show these adjustments also.

Thanks in advance for this important information.

""Why do adjustments? Because you have to. I can just imagine the complaints from you guys if no adjustment was made""

Another non-responsive response.

Again, why do adjustments if the net result is 0 as your link points out. If it is NOT net 0, what is it and why?? If it is net 0 on purpose this is ridiculous. If it is net 0 by accident it would seem to be obvious there is a problem with the procedures. We are dealing with a system that if not chaotic, seems to be close, with human systems plagued with problems. It would seem a stretch for all these errors and accidents to net 0.

you also did not respond to the issue that the stations used by GISS is a small subset of the GHCN DB and may be in no way representative of the net 0 of your link.

Thanks for giving me a reason to drop the link to E.M. Smith's work with GISS.

http://chiefio.wordpress.com/

Nick, instead of jousting with a moron like me why not do something useful like auditing E.M. Smith's work??? Are you afraid you might validate the fact that GISS is useless as currently configured??

Vernon

A better phrasing would have been:
"The blatantly bogus GHCN adjustment for this one station NEED NOT mean that the earth is not warming. It also NEED NOT mean that the three records (CRU, GISS, and GHCN) are generally wrong either".
It would be harder for someone to quote that and then wilfuuly misinterpret it, than for the original phrasing.