Often-Cited Climate Skeptic Refuses To Provide Computer Code He Used To "Prove" Sun Behind Warming

[hatrack]

A physicist whose work is often highlighted by climate-change sceptics is refusing to provide the software he used to other climate researchers attempting to replicate his results. Nicola Scafetta, a physicist at Duke University in Durham, North Carolina, has published a series of papers over the past few years that suggest the sun played a much bigger role in warming over the 20th century than is generally accepted. In particular, one 2006 paper he co-authored concluded that: "The sun might have contributed approximately 50 per cent of the observed global warming since 1900" (Geophysical Research Letters, DOI: 10.1029/2006GL027142).

This paper has been widely cited by those seeking to cast doubt on the scientific consensus on the cause of climate change, including US senator James Inhofe. Scafetta has also contributed to a book that claimed that "carbon dioxide probably is not the driving factor behind climate change".

Many researchers in the field (PDF), however, regard Scafetta's scientific papers on the sun's role in global warming as incorrect, despite their publication in peer-reviewed journals.
Playing down the sun

Earlier this year, for instance, Rasmus Benestad of the Norwegian Meteorological Institute in Oslo and Gavin Schmidt of NASAs Goddard Institute for Space Studies in New York published a paper claiming that Scafetta's methodology produces larger errors than he acknowledges. Benestad and Schmidt concluded that the sun is responsible for only about 7 per cent of the warming in the 20th century and none after 1980. Scafetta claims Benestad and Schmidt have made many errors of their own. However, he is now refusing to give them the program code he used to allow them to try to replicate his results.

EDIT

http://www.newscientist.com/article/dn18307-sceptical-climate-researcher-wont-divulge-key-program.html

[FBaggins]

1. Hard to call something "peer reviewed"...

... if you won't give peers the ability to actually review your work.

And that goes for both sides of the debate.

[joshcryer]

5. The key here is that it seems the paper doesn't support the results.

If a paper supports graphs then it becomes clear that the code behind the paper is most likely sound (though requesting it from one researcher to another is still often done).

The code would obviously show that it doesn't support the evidence in the paper, which is why it is not being published.

[FBaggins]

6. Obviously that's the matter under debate

The problem is that the same thing can be said for both "sides" here.

A paper can't fairly be "peer reviewed" if your peers aren't given enough data/process to reproduce your results. Your statement that it "obviously wouldn't support the conclusion" is exactly the suspicion that many will have in these cases. I don't know that it's "obvious" though, because both sides do it and one of they can't both be wrong.

[joshcryer]

7. Ahh, but, see, Gavin's code is available.

In fact, all American, government funded climate scientists have to make their code and data available.

[FBaggins]

8. Not necessarily the case.

Sometimes they accept data as factual (and include it in their own analysis) that is produced by other groups that don't release their method and/or the raw data. Whether, for instance, the CRU "scandal" is much ado about not very much... they certainly provided one large piece of the puzzle globally and we now know that none of their raw data is available.

It may be that Gavin's code is available... and/or that "government funded" scientists must make it available... but that doesn't mean that both "sides" of the debate rest part of their argument on data and/or analysis that is not properly peer reviewed.

I'm also not sure that even the "government funded" claim is correct. Mann on recently agreed to release much of his work and he was certainly "government funded" by any reasonable definition.

[joshcryer]

9. Certainly a good chunk of the data is locked up, but that is really the doing of our EU counterparts

Here in the US the data is not locked up. So, let's say for the sake of argument, that we only wanted to use data that was not locked up, we could do that without problems at all.

The key is that NCDC has all of the station data (including histories) with which GISS raw and adjusted is based. The methods are all there, and available to scrutiny, by the public, and by other scientists. I have looked at some of the controversial data sets (such as Darwin Zero), and have concluded that the objections by pundits and bloggers are essentially denialism, and not based on anything scientific.

But yeah, I completely agree that open code and transparency is necessary. In the US, though, we really do have it to an alarming extent.

[rfranklin]

2. where are those who were so vocal about the other group hiding data?

Cue the crickets!

[TxRider]

3. Right here

He should release the code and data as used.

[joshcryer]

4. Researchers are happy to release code to one another, that was never a problem.

The issue was that people were complaining that UK specialists weren't / aren't releasing their code.

All US code is released, in the public domain, anyone saying otherwise is a liar.

Code sharing is common practice between researchers.

[TxRider]

10. Umm not quite

It was not only UK "specialists". Unless places like Penn state and NASA are in the U.K.

Can you point me to the US based GHCN source code used for the commonly used GHCN adjusted temperature data set?

IMO if you publish in a peer review journal, data and code should be released as a set as used as part of the publishing process.

If the world is being told to spend tens of trillions of dollars over decades, and change the way we live in almost every aspect of our society, the science should be absolutely transparent.

[joshcryer]

12. GHCN temperature data comes from many commercial sources.

You *can* access most of it from the NCDC for free, though. So while it would be good to have their data analysis software (arguably many lines of source code and many different applications), it probably wouldn't be very useful. The data is in flux, GHCN v2 had very different data gathering and analysis processes than GHCN v1, and v3 will likely have the same.

However, since the vast majority of denialist problems stem from a few cherry picked stations, it is trivial to go to the NCDC for station histories, go to the GHCN paper, and reproduce the station temperature directly. So while I agree we would benefit from having their source code (which may or may not be outdated), I believe the real benefits are having the papers on which their source code is be based.

Note that you couldn't do the homogenization process without having the full dataset, so you'd just take GHCN raw for that part of the step (and assume that their data gathering process with the NCDC is legit; though you could spend months going over the data, if you were so inclined, or write up your own value comparison software). Also of note is that the fact that TOD adjustments account for the vast majority of all data set adjustments, so you can do trivial adjustments quite easily.

Here's the controversial "Darwin Zero" that people were complaining about:

The station move from a hot roof to a cool grassy area, so they had to adjust it to maintain trend integrity.

[TxRider]

17. Beside the point

Deflecting the issue with an argument that since most deniers cherry pick or that data is in flux is irrelevant really.

As is Darwin.

The point is that this science that is being used to transform society at enormous cost to a level unseen since the industrial revolution should be totally transparent.

There is no excuse for non transparency whatsoever considering the gravity of what this science is being used to justify on a global policy level. Same applies to anyone on any side of this issue.

[joshcryer]

21. The primary thing keeping transparency out is profit.

So while we can complain about the current system, we can still recognize that a good part of the evidence is open and transparent, and that the whole of climate science, if wrong, could be shown by looking at this open data.

I posted an image of a guy doing just that.

His conclusions were obvious.

[Nederland]

11. I'm not so sure

I've read so many contradictory things on this subject it's hard to say. However, Wikipedia has a fairly strong record of cleaning out denialist bullshit, and their article on the Hockey Stick Controversy reports that Mann initially refused to share his code.

http://en.wikipedia.org/wiki/Hockey_stick_controversy

Links in this wiki article indicate that according to the January 2005 issue of Energy & Environment, a peer reviewed scientific journal, some of information necessary to replicate Mann's work was not been made fully available to researchers upon request.

[joshcryer]

13. McIntyre has few fans with climate scientists. Mann didn't want to give the data to him...

...because it was clear McIntyre was using the data to attempt to discredit other scientists (it's one thing to say "you're wrong, here's why" and another thing entirely to imply "you're lying" etc).

Note that McIntyre himself doesn't actually make these claims, he simply allows commentators on his blog to make them and run with it. So guys like Mann get a whole shitton of pressure and attacks simply for doing their job.

[Nederland]

14. I would agree

However, you can't have a system where data and code is shared only with people that the researcher thinks are "worthy". You need to allow everyone access to everything and trust the system of peer review to sort out the legitimate objections from the ridiculous ones. It's a bit like free speech--you just have to accept all the bullshit that goes with it because the alternative is worse.

[joshcryer]

19. Look at open source. Everyone has access.

But the truly breathtaking projects have dozens of people on it. When you look at, say, GISS, you know it's just one or two guys at most. Hundreds of thousands of lines of code, by one or two, maybe, just maybe three guys. Now they actually got asinine comments in the past to have all of the code made user friendly.

So do you spend your time doing science or do you spend your time making code user friendly? Code that most likely won't even be used by anyone who claims to be a skeptic.

I suspect the system would sort it all out in the end, the peer review has done great so far, but when you give "skeptics" fuel to their fire, they will run with it. And as climate gate has shown, these guys are terrified that untruths are going to win the day. They want the public to be educated about the truth of global warming, but they are constantly hammered on a regular basis with innuendo and outright lies.

*Using the very data that they provided freely.*

[Nederland]

23. They are inseperably connected

So do you spend your time doing science or do you spend your time making code user friendly?

You cannot have one without the other. If your code is written in such a way that other people cannot understand it you are not doing good science. Why? Because good science is science that can be replicated and verified.

[TxRider]

25. Just like NASA GISTemp

Grudgingly released code, worked on by apparently one guy.

Who now that it has been released gets free help and the code has improved.

If there aren't proper resources on coding the software that the world is basing such far reaching changes to society on then shouldn't that be clearly pointed out, screamed about, so more resources are put to the task?

Obfuscating and hiding the fact that the resources are not adequate by hiding the code is not the proper solution.

The system working it out in the end is all well and good for many areas of science, but governments are proposing the biggest tax increases in history around the planet, and trillions are being taxed and spent -right now- on the basis of this science. We simply do not have the luxury of waiting to see if the system sorts it out eventually or not.

To defend not having total transparency, given the gravity of the situation, is ridiculous.

[TxRider]

16. So let me get this straight

Withholding data and methods is justified because someone wants to check it for correctness? And if found to be in error the paper would be discredited? I thought that was the point of peer review? What the peanut gallery on a blog makes of it is irrelevant in the long run.

Seems Mann thought that using data/methods to attempt to discredit others was perfectly justified, as long as it was MM or someone else he disagreed with being discredited and not him or his associates.

McIntyre seems to allow commentators to run in both directions on his blog.

Total transparency removes all this nonsense. Withholding only makes it seem there is something to hide.

[joshcryer]

18. No, no.

You look at the tree ring divergence. It's a perfectly acceptable scientific practice. But what happens? It is used to "discredit" the science by bloggers and pundits. This is distinct from actually writing a peer reviewed paper about why they should not be included. I'm talking to a guy on another forum who thinks tree ring proxies (which have to be derived from the temperature record), are more accurate than the temperature record. The only reason he would think that nonsense is if he intuited it from reading silly innuendo on other websites.

Note that RC allows commentators to "run both directions" on their blog, people saying otherwise are liars.

Yes, I agree that science should be as transparent as possible. But as least with regards to American publicly funded NASA scientists it is. And that's where a whole huge chunk of the evidence comes from.

[TxRider]

22. Still missing the point

Those people will discredit if the data and methods are released, they will also discredit if the data and methods are not released. Either way bloggers and pundits will do what they do. There is only one difference really.

If everything is transparent they have an argument that can be scientifically shown to be false, if data and methods are held secret all they have to do is point a finger and say scientists are hiding the truth, and nobody can disprove them.

Even NASA GISS did not release code until it was shown to be in error badly enough that a person pointed out the errors without even having the code, and put enough egg on their face they released the code.

There are three big homogenized and adjusted temp sets, GISS, CRU and GHCN adusted.

Of the three only GISS code is available to my knowledge, and only after it was grudgingly released after being shown to have bugs.

Yes most of the raw data is available, but it shouldn't require back tracing adjusted sets station by station to work out each individual adjustment to try to check the temp sets.

Then there are peer reviewed papers that are used to justify the massive global changes being demanded, without transparency.

It is inexcusable under the circumstances.


If this was science on the latest silicon wafer technology it would be one thing, I could see your point. But when science steps up and says the world must change from it very core outward to every fringe, science should expect to make a totally transparent case.

[wtmusic]

24. It's always counterproductive to withhold data or methodology

and bloggers/pundits have a field day when it is. Let them "discredit" the science; they in turn will be "discredited".

You overestimate the significance of bloggers with regards to public policy.

[struggle4progress]

15. If he doesn't want to share his code, he should at least make available a complete

description of the model he used, together with a full discussion of whatever numerical methods he used: for example, he needs to tell us what errors tolerances he used, what (if any) regularization techniques he applied, and so on -- so that someone else can verify his results

[joshcryer]

20. That was RCs objection, they needed more information, or code, whichever was suitable.

He refused on both accounts, because his paper doesn't provide enough math for his model, and therefore is unsupportable.