![no-button1[1]](http://wattsupwiththat.files.wordpress.com/2012/06/no-button11.jpg?w=300&resize=300%2C225)
Over at Climate Audit there is renewed interest in data availability with McIntyre asking whether journals that don’t guarantee data archiving (The Holocene in this case) should be cited in IPCC reports.
It happened that yesterday Phil Jones of CRU gave a talk at KNMI in De Bilt, The Netherlands, where he also talked about availability of data, in this case the data behind the recently published Crutem4 and Hadcrut4 graphs. The talk itself was pretty neutral, just explaining what had been done to produce these two datasets. However at the end Jones made a statement that is relevant to the long lasting discussions about data availability:
For raw temperature data you have to contact the NMS’s.
NMS’s stands for National Meteorological Services, like the Met Office in the UK or KNMI in The Netherlands. The good news is, as we can also read in his latest Crutem4 paper, that CRU will make all data available. However the bad news is that these data have already been homogenized by the NMS’s and the original data are not available at the Crutem4 webpage.
Jones explained that he and Moberg concluded already in 2003 that the homogenisation of temperature data is best done at the NMS’s. So for Crutem4 whenever possible they used these homogenized data of the NMS’s directly, as can be seen at their webpage.
Now in itself this is a fair approach. If the NMS’s cannot figure out what happened with their stations in the past and how best to control for station moves and instrument changes who else can?
…
We know that in some countries adjustments to raw data determine a large part of the trend. In New Zealand sceptics fight (see also here) with NIWA (the NMS of New Zealand) over the adjustments made to the raw data. The temperature trend in the raw data is only 0,3 degrees per century while the adjusted data show a trend of 1 degree per century. Jones uses the adjusted NIWA data in Crutem4. Later this year the High Court in New Zealand will consider this case.
Jones seemed satisfied with the new situation. Anyone asking him for the raw data in the future will be referred to the NMS’s.
================================================================
Anthony: This is particularly worrisome, because as we’ve seen, metadata for GHCN global stations is very poor to virtually non-existent, and from what we know, GHCN and CRU takes very little metadata into account. While the NMS’s may have a better handle on metadata, given the disparity of quality of met services globally, this pretty much ensures that no individual researcher is going to get their hands on a complete set of all data. Phil Jones is essentially blowing off the issue saying “let them figure it out, not our responsibility”. Whatta guy!
================================================================
Crok continues:
In his future answers to sceptics asking for data he can almost copy this paragraph of Joelle Gergis blowing off McIntyre when he requested some tree ring data from her:
This list allows any researcher who wants to access non publically available records to follow the appropriate protocol of contacting the original authors to obtain the necessary permission to use the record, take the time needed to process the data into a format suitable for data analysis etc, just as we have done. This is commonly referred to as ‘research’.
In the case of Crutem4 the raw data in many cases is also not publicly available and anyone interested has to contact each of the NMS’s trying to get these data. There is no guarantee at all that they will release the data.
Now although Jones was very obstructive to data requests from sceptics in the past, I don’t say that Jones is to blame for the current situation. At least he tried to get permission from the NMS’s to release the data as he promised in this Nature article in 2009, which was also covered in several Climate Audit posts (see here for example). In the Nature article Jones said:
“We’re trying to make them all available,” says Jones. “We’re consulting with all the meteorological services — about 150 members [of the World Meteorological Organization] — and will ask them if they are happy to release the data.” A spokesperson for the Met Office confirmed this, saying “we are happy for CRU to take the lead on this, as they are their data”.
But getting the all-clear from other nations won’t be without its challenges, says Jones, who estimates that it could take several months. In addition, some nations may object if they make money by selling their wind, sunshine and precipitation data.
In his new paper on Crutem4 he reports back on this attempt:
In November 2009, the UK Met Office wrote on our behalf to all NMSs to determine if we could release the versions of their monthly temperature series that we held. Of the about 180 letters, we received 62 positive replies, 5 negative replies, and the remainder did not reply.
These results are worrisome in itself. Almost two-thirds of the NMS’s didn’t even bother to answer to a request concerning one of the most important climate graphs in the world. For these countries Crutem4 uses the GHCN data.
===============================================================
The need for a journal that demands all data, (used and excluded) up front, along with methodology, code, and supplementary material to ensure the work is reproducible, before even considering a paper for review is becoming clearly obvious. – Anthony
h/t to Dr. Roger Pielke Sr.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Stephen Brown says:
June 1, 2012 at 1:24 pm
Without the raw data being available to all, no reliance can be placed on any conclusion drawn from any study using the unavailable data,
End of story.
++++++++++++++++++++++++++++++++++++++++++++++
Now they can do what they want, say what they want, and NO ONE CAN PROVE THEM WRONG…
Just like Obama’s EO on scientific integrity, no scientist (or fraud) who is employed by the government need fear an FOIA request… Agenda trumps Truth….
WE ARE THE GOVERNMENT AND YOUR THINKING HAS BEEN DONE….
now do as we say slaves….
Who says the UN Agenda 21 can not be forced up everyone…
scary indeed.!
IT just dawned on me… The Government can adjust the data at will because it is no longer in the public eye as an unchanged file…
this is much worse than hide the decline… IT’s HIDE THE DATA
I thought the berkeley study did include all data. The criticisms that some data was cherry picked was something that this study was supposed to prove. The author of the study was reputed to be a skeptic of the pre-existing studies for having cherry picked data.
In any case, 0.6 or 1.0 or 0.3 whatever the rise it isn’t enough to be worried about because the trend has been essentially zero for 20 years it may be another degree in the 21st century but it is hard to get anybody convinced the temperature will rise 3 degrees since that would require that temperatures rise every decade for the rest of the 21st century the amount of the entire 20th century’s rise. It’s hard to imagine that happening with no evidence temperatures have ever gone up that much and not during this time when every molecule of co2 will have much more effect than a molecule of co2 100 years from now will have.
If everyone in the whole world swears it is true, if it is based on a thousand papers, if it is known to be true for a thousand years but is a lie, it will be found out. A fool whispering the truth can bring down the greatest theories of the most renown thinkers and show them to be the real fools. The truth will out! It is just that it takes so darn long.
“Because it isn’t simply a question of the data, homogenised or otherwise. It’s a question of the methodology used to arrive at the end result. If an individual cannot replicate the results, how can they be confirmed or improved upon? Not refuted, please understand, but replicated. Where, exactly, is the problem with that?
#############
many people here fail to understand or remember the issues involved in the request for CRU data.
The issue was twofold
1. CRU claimed to make “value added adjustments” so we asked for the raw data, and all the data that went into the final numbers.
2. we asked for their code.
CRU has responded with CRU4. in CRU4 they use the data AS PROVIDED after homogenization.
This effectively removes the question about which adjustments CRU makes. They dont anymore.
If you compare the CRU versions of data with those provided by NWS they match.
The question about code is also answered. Actually that was answered a long time ago when I replicated their method and match their answers exactly.
There are NO issues of interest surrounding various methodologies. Changing methodologies has little impact on numbers. There is no issue surrounding CRU adjustments. They dont. They use homogenized data as provided.
Now, the issue is the homogenization process. Folks here will not like what has been found over and over again.
1. Homogenization is required
2. Different methods produce the same results
3. the job of checking got much harder
The raw data is out there. its a royal pain to work with because it is full of gross errors.
Steven,
Anthony already countered this in his original thoughts on this.
Anthony: This is particularly worrisome, because as we’ve seen, metadata for GHCN global stations is very poor to virtually non-existent, and from what we know, GHCN and CRU takes very little metadata into account. While the NMS’s may have a better handle on metadata, given the disparity of quality of met services globally, this pretty much ensures that no individual researcher is going to get their hands on a complete set of all data. Phil Jones is essentially blowing off the issue saying “let them figure it out, not our responsibility”. Whatta guy!
Not sure what you are arguing against, as the original post by elfstone is still correct despite these wild thoughts and rather off-topic strawmen you threw up.
And as a side note, I would love to see how you obtained the data from over 100 Government agencies (MET’s) and were able to reconcile this using what is provided and otherwise create the same output as Dr. Jones using his methods as provided.
Let me give you a hint on your problem here, you are claiming here: Actually that was answered a long time ago when I replicated their method and match their answers exactly.
that you basically replicated their methods. So how did you get them all to respond to you since you “replicated their method and matched their answers exactly.”
Please, feel free to prove me wrong and provide all the data and methodology used and how you were able to communicate with the MET’s so I could duplicate the experiment. I would love to know how to obtain raw “homogenized” data to work with. So feel free to share with us all.
Steven Mosher says:
June 2, 2012 at 5:11 pm
many people here fail to understand or remember the issues involved in the request for CRU data.
The issue was twofold
1. CRU claimed to make “value added adjustments” so we asked for the raw data, and all the data that went into the final numbers.
2. we asked for their code.
CRU has responded with CRU4. in CRU4 they use the data AS PROVIDED after homogenization.
This effectively removes the question about which adjustments CRU makes. They dont anymore.
If you compare the CRU versions of data with those provided by NWS they match.
The question about code is also answered. Actually that was answered a long time ago when I replicated their method and match their answers exactly.
There are NO issues of interest surrounding various methodologies. Changing methodologies has little impact on numbers. There is no issue surrounding CRU adjustments. They dont. They use homogenized data as provided.
Now, the issue is the homogenization process. Folks here will not like what has been found over and over again.
1. Homogenization is required
2. Different methods produce the same results
3. the job of checking got much harder
The raw data is out there. its a royal pain to work with because it is full of gross errors.
Let’s take this a step at a time.
Did they respond with the code they had used, together with links to the actual data they had used?
Did they explain why they had used someone else’s homogenised data?
Did they examine the methodologies used in homogenising said data?
Why, exactly, should any researcher have to reproduce, from scratch, their methodology?
Yes, homogenisation is required. Do all methods produce the same results? (I know, a broad question, but I believe it to be valid.) Checking what, precisely?
If the raw data is full of gross errors, why is one method of determining what is correct and what isn’t better than another? Who determines what errors are gross or otherwise?
Matching answers is a good thing, as long as it’s reproducible again and again. Methodology is important, however, and without the means to reproduce the “experiment”, it seems to be reduced to the level of wiggle-matching.
In the end, it comes down to whether or not one is able to reproduce the results, using the same methods originally employed. If that is not possible, then one has to assume that there’s something missing… simple logic dictates that. Openness is the only way, surely?
Homogenization of data should produce a neutral result on trends. This is not the case (0.5 ° C additional warming for the twentieth century on the data of NMS). This bias must absolutely be explained so that the methodology has any validity. Steven Mosher, instead of always saying the same thing, explain us this bias.
Steven Mosher could be inspired by Hansen et al 2001 Chapter 4 and Figure 1:
http://pubs.giss.nasa.gov/docs/2001/2001_Hansen_etal.pdf
Incomplete but a good start to explain.
Crispin in Waterloo says:
June 2, 2012 at 9:18 am
“Anthony, I have a suggestion.
We need two on-line publications: the Book of Reproducible Results and the Book of Unreproducible Results.
It is hard to keep track which papers are published with no supporting data or methods, or with partial data and incomplete methods (whether by design or incompetence). The point is that there are too many important-to-the-field papers being published reporting results are not reproducible by other researchers.
When one gets into a discussion there are frequent references to papers that are definitely in the Journals, but whose result are not reproducible. This is an affront to science but the authors do no seem to be bothered by this, for whatever reasons they have.
On your links to the right of these blogs there is a list of sites binned according to temperament and content. Would it be possible to create there a link that goes to a list, at least two lists actually, of important-to-the-discussion papers divided into two major groups: those which contain archived data and methods and which can be used to reproduce results, and those which do not. A third list might be papers for which the data is promised or is available on request.
Before allowing debaters to get away with citing-in-argument the result of works that are not reproducible, we can have a quick look-up of the paper by simple CTRL+F for part of the name.”
Crispin, that is a very good idea. Sooner or later research should move in this direction.
Such list would greatly facilitate real science advance and better understanding but would require real work to be maintained.
Nevertheless even if in the beginning it would contain 1 or 2 articles it would still be a strong reference for those articles.
Crispin in Waterloo says: @ur momisugly June 2, 2012 at 9:18 am
Anthony, I have a suggestion.
We need two on-line publications: the Book of Reproducible Results and the Book of Unreproducible Results.
It is hard to keep track which papers are published with no supporting data or methods, or with partial data and incomplete methods (whether by design or incompetence). The point is that there are too many important-to-the-field papers being published reporting results are not reproducible by other researchers…..
_____________________________________
THAT is what the Union of Concerned Scientists SHOULD be doing instead of using the name to promote “Post Normal Psycience”
It is also what scientific societies such as ACS (AMERICAN CHEMICAL SOCIETY) should be doing for chemical papers. ACS should not be a CAGW advocacy group. It is NOT their field and they should just keep their mouth SHUT.
So if you get adjusted data from 150+ NMS and each NMS uses its own method of adjustment, how can you combine that data into a global temperature mean with any confidence?
I would think you would need the adjusted and unadjusted data from each NMS so you could make sure that the adjustment methods used were at least reasonable. I mean if Outer Frajakastan, where astrology is the national religion, adjusts their data based on the position of Venus would you reasonably use that adjusted data in your research?
ddpalmer: Only if it agreed with my preconceived grant application guaranteed result /sarc (for those who need it)
“Get the data directly from every NMS yourself.”
Translation:
“Go pound sand.”