From: "Tim Osborn" To: k.briffa@uea.ac.uk Subject: [Fwd: Re: data request to SCIENCE for 1120514] Date: Tue, 7 Mar 2006 08:22:22 -0000 (GMT) Reply-to: t.osborn@uea.ac.uk Cc: t.osborn@uea.ac.uk Keith - see below. I bet it won't be the end of the episode! - Tim ---------------------------- Original Message ---------------------------- Subject: Re: data request to SCIENCE for 1120514 From: "Jesse Smith" Date: Mon, March 6, 2006 8:03 pm To: t.osborn@uea.ac.uk -------------------------------------------------------------------------- Dear Dr. Osborn, Thank you for your clear and careful response to the requests made by Dr. McIntyre, which we forwarded to you: it was quite satisfactory, we believe, and will greatly help Brooks (Hanson) in crafting his reply to Dr. McIntyre. I hope that this will be the end of this episode, but if it is not, we will be in touch again. Best regards, Jesse Smith ======================= Dr. Jesse Smith Senior Editor ---------------------------------------------- Science 1200 New York Avenue, NW Washington, DC 20005 USA ---------------------------------------------- (202) 326-6556 (202) 408-1256 (FAX) hjsmith@aaas.org ======================= >>> Tim Osborn 3/3/2006 11:22:17 AM >>> Dear Jesse Smith and Brooks Hanson, thank you for your patience while waiting for our reply. Before responding to the specific data requests, we would like to say that it is our view that we should provide sufficient data to enable all the main elements of our analysis to be checked, but that we are not obliged to provide the data that would enable the research reported in other papers to be checked, even if we cite those other papers or use results reported in those other papers. You will see how this view has determined our response to some of the requests. Now to the requests themselves, numbered according to the numbering system of Steve McIntyre's email. (1) As you know, we provided (in advance of publication) the 14 smoothed and normalised proxy records to WDC-Paleo that enable the main parts of our analysis to be replicated. The only part of our analysis for which the unsmoothed data are required is to calculate the correlations against temperature that we reported for some of the series (not those that had already been reported by Mann and Jones, as indicated in our Table S1). These unsmoothed data for all 14 series are now also archived at WDC-Paleo, which will enable those correlations that we reported in Table S1 to be checked. These unsmoothed data were archived on Thursday 23rd February, in response to a request by a different colleague. This should cover this request in full. (2) Our Table S1 provides the full citation to the source of our data, funnily enough given in the column labelled "Data source". Some of these may or may not have publicly archived their data, but our WDC-Paleo entry now contains the series that we were originally provided with (i.e., the unsmoothed data that we refer to in item (1) above). The "Orig source" column in our table was our effort to ensure that original work on collecting/processing these data is acknowledged, because it is important for us to acknowledge that work even when we obtained the data from a secondary study. We did not intend to imply that the data that we had used would match the data in these original sources, because various different versions might exist (due, e.g., to different methods of processing the data, or due to updated measurements, etc.). That is why we made the source of our data clear. (a-c) We have not yet had time to double check the ITRDB citations that we provided for these three records, but we will do so as soon as we have time. Our data source was in fact Esper et al. (2002) and this is correct, so the concern over the accuracy of these ITRDB citations does not limit the ability for others to check our work. (d-f) The original studies that we cite are definitely correct for these two records. We have provided sufficient data for our analysis of these records to be checked. We have not provided extra data to enable other people's studies to be checked, nor do we feel obliged to do so. (g) These series from Esper et al. (2002) were considered by us and then rejected. As we understand it, Esper et al. have made available their site RCS records and therefore these four records could be obtained from Esper et al. If this is not the case, we could provide these four rejected series. (3) D'Arrigo et al. (2006) do *not* report directly opposite findings in respect to the correlations we obtain for Jaemtland and Boreal/Upperwright. Neither paper reports any correlations involving these series versus temperature. Both papers list more than one reason why series might be rejected. For example, our reasons were "We removed series from (S1) that did not correlate positively with their local annual or summer temperatures (Table S1), or which did not extend into the period with instrumental temperature to allow a correlation to be calculated." The latter is our reason for excluding Jaemtland, not the former: the Jaemtland series that we obtained from Esper et al. (2002) has no data after 1827 and so no correlation was calculated. The Jaemtland series used by D'Arrigo et al. continues through to 1978 due to the inclusion of additional data. Similarly, D'Arrigo et al. list a number of reasons for excluding series, but they do not state which one(s) were used to exclude Boreal and Upperwright, though in fact none disagreed with our criteria anyway! We have not separately stored the temperature time series used to obtain the correlations reported in our Table S1 and to do so requires some changes to our program, which we have not done because there does not appear to be a need to do so (given our explanation above of the situation regarding our paper versus D'Arrigo et al., 2006). Because Steve McIntyre has explicitly stated that he is unable to verify our results for the Boreal/Upperwright case, we have extracted the temperatures we used for that case only and attach them here as a text file. We hope that he can use them to reassure himself about the correlations that we obtained. (4) (a) We explicitly state that we did not use the Esper et al. (2002) Jasper series, so there is no expectation that they should be identical. Esper et al. (2002) have, we believe, made their version available and we have made available the series that we used via WDC-Paleo. (b) Similarly, we explicitly state that we did not used the Esper et al. (2002) Tornetrask series and data are available as for (a). (c) We are not obliged to confirm anything that Esper et al. (2002) did. (5) This request is not relevant to our paper, as discussed at the start of this email. (6) Same as (5). We hope that we have dealt with these requests to a more than satisfactory extent, but please let us know if you feel that we should do more. Best regards Tim Osborn and Keith Briffa At 19:30 23/02/2006, you wrote: >Dear Dr. Osborn, > >We have just received an email from Steve McIntyre (pasted below), >with a long and very specific list of alleged deficiencies in the >availability of data by which to evaluate your recent paper, "The >Spatial Extent of 20th-Century Warmth in the Context of the Past >1200 Years," and others. Wishing to deal with this issue in a >conscientious and reasonable way, we are passing the email along to >you as a request for data, without taking a position on the validity >of any particular point. We would like to have your confidential >response to this request, keeping in mind the stated policy of >SCIENCE that "Any reasonable request for materials, methods, or data >necessary to verify the conclusions of the experiments reported must >be honored." Please return your response by email directly to me, >and CC: Brooks Hanson, our Deputy Editor >(bhanson@aaas.org). We appreciate your >cooperation, as well as the time and effort that a reply may >take. Feel free to contact me if you have any questions about this issue. > > >Sincerely, > >Jesse Smith > >*******START OF EMAIL FROM S. MCINTYRE********* >Dear Dr Hanson, > >Thank you for your prompt response to my letter in respect to Osborn >and Briffa [2006], Esper et al [2002] and Thompson et al [1989; >1997]. I appreciate your efforts in this and realize that you are >frustrated at being criticized. However, if you reflect on the >matter, I'm sure that you will agree that the problem stems entirely >from the original authors failing to comply with Science's data >archiving policy. > >It will come as no surprise to you that I do not believe that the >additional data, useful as it is, comes anywhere near discharging >Science's obligations under its data policies for reasons that I >will set out in detail below. I will discuss the shortfalls in >connection with what I understand to be one of Science's governing >policies >http://www.sciencemag.org/feature/contribinfo/prep/gen_info.dtl#datadep) >: > >Science supports the efforts of databases that aggregate published >data for the use of the scientific community. Therefore, before >publication, large data sets . must be deposited in an approved >database and an accession number provided for inclusion in the published paper. > >Since the issue pertains to how Science discharges its policies, it >is my position that you, rather than the original authors, are the >appropriate arbiter of that. (Additionally, the authors have refused >all requests in the past and I see no reason why their behavior >would now differ.) > >Status of Each Request: > > 1. Digital versions of all 14 series as used in their > final compilations; > >I have inspected the archive at >ftp://ftp.ncdc.noaa.gov/pub/data/paleo/contributions_by_author/osborn2006/osborn2006.txt, >to which you directed me. This consists of smoothed (and re-scaled) >versions of the 14 series and is relevant to the request, but does >not satisfy it. The authors specifically discuss correlations of >these series to temperature, which requires consideration of the >pre-smoothed series. Accordingly, I re-iterate my original request >for digital versions of the 14 series. > >2. For each of the tree ring sites analysed (both the 11 >retained and Esper site not used, including Gotland, Jaemtland, >Mackenzie Mts and Zhaschiviersk), an exact data citation to a public >archive (e.g. WDCP) for the data set used; or, in the alternative, >an archive of the data set at the Science website. In cases, where >the publicly archive dataset for a site is related to but different >from the version used by Osborn and Briffa, please archive the data >set as used. > >I was able to reasonably reconcile the smoothed series to original >sources in public archives and accordingly have no issue with data >provenance for the following Osborn and Briffa series: the Mann PC1 >(#1); #5 Chesapeake; - #6 - Fisher's Greenland O18 stack; #7 - >Netherlands documentary; #14 - Yang's China composite (although >there are problems in the Thompson series used in this composite). >For other users less familiar with nuances of series versions, I >recommend that the SI be modified to provide accurate data citations >for these 5 series. > >The problems mostly pertain to tree ring data, which make up the >other 9 series. In three cases, Osborn and Briffa provided data >citations for sites in public archives (#4 - Quebec- cana169; #8 - >Tirol - germ21; #11 - Mangazeja - russ067, russ068). In each of >these 3 cases, the Esper version reconciles to the Osborn version >(up to re-scaling). However, they do not reconcile to the original data sets. > >a) the dataset germ21, cited by Osborn-Briffa for series #8- >Tirol, has values from 1466 to 1837, while the archived version goes >from 1324 to 1975. Obviously the data set has not been cited >accurately or is incomplete. > >b) the series cana169 goes from 1352 to 1989, while the Osborn >version (#4 - Quebec) goes from 1352 to 1947. Again, it appears >that the data set has not been cited accurately or is incomplete. >Additionally, while I have been able to substantially replicate the >features of other RCS chronologies, my efforts to reproduce the >archived result from cana169 lead to a series with a significantly >different shape. > >c) one of the two cited data sets (russ067) does not contain >measurements at WDCP. However, the versions "mangazla" and >"mangazpc" in the Schweingruber section of WDCP appear to have the >data for russ067 and russ068. However, these data sets only yield >values from 1246 to 1969, while the archived Osborn version (#11 - >Mangazeja) goes from 1246 to 1990. Some additional data must exist >somewhere, but has not been archived at WDCP to date. > >Two sites (#9 - Tornetrask; #13 - Mongolia) have WDCP measurement >archives (swed019; mong003 respectively), but there are >inconsistencies between the data as archived and the length of the >Osborn and Briffa versions. > >d) the WDCP archive for Tornetrask ends in 1990, which is >inconsistent with the Osborn version which ends in 1993. This >indicates that the data sets are not the same. > >e) similarly, the WDCP archive for Sol Dav, Mongolia begins in >900, while the Osborn version begins in 800. > >For the following 5 sites, no archive of the measurements exists at >all - a direct breach of Science's archiving policy: > >f) Jasper/Icefields, Boreal, Upper Wright, Taimyr, Yamal, > >Accordingly, I re-iterate my request that the measurement data >consistent with the archived site chronologies be archived for each >of the above items 2(a)- 2(f), as well as corresponding information >for the following 4 sites considered in Osborn and Briffa: > >g) Gotland, Jaemtland, Mackenzie; Zhaschiviersk > >3. Digital versions of the specific gridcell temperature series >used in each of the reported temperature correlations together with >version date. > >As noted in my previous request, D'Arrigo et al [2006] have reported >directly opposite findings in respect to the correlation between >their RCS chronology and gridcell temperature for: Jaemtland and the >two foxtail series. I have specifically been unable to verify their >claim in respect to bristlecones. Accordingly, I re-iterate the >request for the digital versions of the temperature data used in >these calculations. (In connection with a similar request, Nature >required Mann et al. to archive the exact temperature data used in MBH98.) > >4. Exact data citations to a public archive for all datasets >used, or, if such do not exist, an archive of the data set at the >Science website. > >While most Osborn versions match Esper versions up to re-scaling, >they differ in three cases, and a separate Esper version is required >in two of them: > >a) the Esper version for the Jasper data is different than the >Osborn and Briffa version (as noted in Osborn and Briffa) and both >data sets need to be made available; > >b) similarly, there are differences between the version of the >Tornetrask series archived by Esper and the one archived by Osborn, >again requiring examination of both data sets; > >c) the Polar Urals version of Esper differs from the Yamal >version of Briffa. It is possible that the Esper version used a >combination of data sets russ021 and russ176 (if so, would you >please confirm this.) > >5. A clear and operational definition distinguishing "linear" >and "nonlinear" trees, preferably with source code showing any >differences in methodology. > >While the provision of site chronologies for 13 Esper sites is >appreciated, one site (Mongolia) was unaccountably omitted. The >corresponding information is requested. > >While the provision of the site chronologies was interesting and >appreciated, according to my reading of Esper et al [2002], these >site chronologies were not used in the calculations in the article, >which distinguished between "linear" and "nonlinear". No operational >definition is provided. combined with the unavailability of the bulk >of the data, the calculations of "linear" and "nonlinear" >chronologies cannot be replicated even from the recent information >regarding Esper et al [2002] and this remains unresolved. > >6. Thompson provides a complete archive of both Dunde and >Guliya ice cores, including both isotope and chemical data. > >While I appreciate that Thompson has provided sample information on >(only) 2 Kilimanjaro cores, he did not provide the requested >accompanying chemical information necessary for their >interpretation. The Kilimanjaro data is obviously of little help >with the Dunde and Guliya data. > >The U.S. Global Change Research Program required archiving of data >commencing in 1991 and the World Data Center for Paleoclimatology >has been in existence since then and has been online since 1994. >Accordingly there was an adequate facility for the archiving of the >Guliya core when it was published in 1997. > >I realize that the Dunde core was published in 1989, at a time when >your present archiving policies were not in effect. However, >Thompson has published versions of this series in other journals >which are inconsistent with the version published in Science. I >cannot imagine that you are content with such a situation. Even if >you did not have policies at the time, I am sure that you can give a >very firm request to Thompson and I find it difficult to believe >that Thompson would refuse a direct request from Science to provide >this data. If he has refused a direct request, then that too is >relevant information, upon which I would appreciate confirmation. > >Again, I apologize for putting you in the middle of this and for the >public nature of the exchange. However, some of this has been going >on far too long with minimal results, leaving no alternative. >However, I assure you that I will be equally public in commending >you if and when you resolve matters. In my opinion, you should >simply do the following: > >(1) send a copy of your data archiving policy to each >of the authors: Osborn-Briffa; Esper et al. and Thompson; > >(2) tell Osborn-Briffa and Esper et al. that you >expect them to comply with the policy which was in effect at the >time of publication or else you will retract the article. > >(3) tell Thompson that, if he wants to publish at >Science in the future, he should immediately clean up his archive >for the earlier articles. > >Obviously there has been some inadequate housekeeping in the past. I >can understand this and my concern is not with the past. My concern >is with the present. You have an opportunity to remedy the situation >now and no one will criticize Science for ensuring that paleoclimate >authors meet Science's data archiving policies. On the other hand, >you will be justly criticized both by me and others if you don't do so. > > > >Regards, > >Stephen McIntyre > >********END OF EMAIL FROM S. MCINTYRE********** >******************************************* > >======================= >Dr. Jesse Smith >Senior Editor >---------------------------------------------- >Science >1200 New York Avenue, NW >Washington, DC 20005 >USA >---------------------------------------------- >(202) 326-6556 >(202) 408-1256 (FAX) >hjsmith@aaas.org >======================= Attachment Converted: "c:\eudora\attach\untitled-23.htm"