Climate Audit

In today’s post, I’m going to examine Gergis’ dubious screening out of the Law Dome d18O series, a series that has been of long-standing interest at Climate Audit (tag).

Gergis et al 2016 stated that they screened proxies according to significance of the correlation to local gridcell temperature. Law Dome d18O not only had a significant correlation to local temperature, but had a higher correlation to local instrumental temperature than:

• 24 of the 28 proxies retained by Gergis in her screened network;
• a higher t-statistic than 19 of 28 proxies retained by Gergis;
• a higher correlation and t-statistic than either of the other two long proxies (Mt Read, Oroko Swamp tree ring chronologies);

Nonetheless,  the Law Dome d18O series was excluded from the Gergis et al network. Gergis effected her exclusion of Law Dome not because of deficient temperature correlation, but through an additional arbitrary screening criterion, which excluded Law Dome d18O, but no other proxy in the screened network.

This is not the first occasion in which IPCC authors have arbitrarily excluded Law Dome d18O. CA readers may recall Climategate revelations on the contortions of IPCC AR4 Lead Authors to keep Law Dome out of the AR4 diagram illustrating long Southern Hemisphere proxies (see CA post here).

Law Dome d18O is intrinsically an extremely interesting proxy for readers interested in a Southern Hemisphere perspective on the Holocene (balancing the somewhat hackneyed commentary citing the Cuffey-Clow reconstruction based on GISP2 ice core in Greenland). The utility of Law Dome d18O is much reduced by inadequate publishing and archiving by the Australian Antarctic Division,  a criticism that I make somewhat reluctantly since they have been polite in their correspondence with me, but ultimately unresponsive.

Gergis et al 2016 Screening

In my previous post, I criticized the data mining of Gergis et al 2016 for “significant” correlations, a technique that is aptly characterized as data torture (sensu Wagenmakers 2011, 2012), since so many possible combinations of lags and gridcells were examined. Such data mining became necessary for Gergis following the retraction of Gergis et al, 2012, since they could not obtain a network with more than a handful of proxies using the the reported screening criterion of Gergis et al 2012 and the reconstruction using that handful appears to have had a different appearance than the reconstruction reported in Gergis et al 2012.

In their attempt to salvage the earlier article within the constraint of detrended correlation, a constraint which they sought to abandon but which appears to have been rejected by the Journal of Climate, Gergis’ main tactic was a switch to correlations with local gridcells as a target, rather than the Australasian regional composite previously used in Gergis et al 2012.

Gergis et al stated that they then carried out detrended screening over the period 1931-1990, determining the “significance of the correlations” between detrended proxy and detrended gridcell temperature using a t-statistic allowing for autocorrelation. (They carried out such calculations over multiple gridcells and lags, but this data mining is not relevant to today’s issue).

For predictor selection, both proxy climate and instrumental data were linearly detrended over 1931-90. As detailed in appendix A, only records that were significantly (p  <0.05) correlated with temperature variations in at least one grid cell within 500km of the proxy’s location over the 1931-90 period were selected for further analysis. For predictor selection, both proxy climate and instrumental data were linearly detrended over 1931-90…To account for proxies with seasonal definitions other than the target SONDJF season (e.g., calendar year averages), the comparisons were performed using lags of -1, 0, and +1 years for each proxy…

The significance of these correlations was determined using a Student’s t distribution, with the degrees of freedom adjusted for autocorrelation at lag 1 using the method of Bretherton et al. (1999).

For Law Dome d18O over 1931-1990 for the central gridcell at lag zero i.e. without any Gergian data mining or data torture, using the HadCRUT3v version on archive, I obtained a detrended correlation of 0.529, with a t-statistic of 4.71  3.65 (for 37 degrees of freedom after allowing for autocorrelation using the prescribed technique) [updated Sep 10, 2016]. This  was one of the highest t-statistics in the entire network, higher than 24 19 of 28 proxies selected into the screened network and higher than both long proxies included in the network.  It also met any plausible criterion of statistical significance.

So how (and why) did Gergis screen out Law Dome?

Gergis excluded Law Dome through the following, seemingly innocent, additional screening criterion:

This comparison [of detrended proxy to detrended instrumental data] was only performed for cells containing at least 50 years of data between 1921 and 1990.

Because a t-statistic test already allows for the number of observations in determining significance, there isn’t any need for this restriction.   The choice of 50 years excluded Law Dome d18O, but did not impact any other proxy included in the G16 screened network.  Gergis did not provide any justification or explanation for the choice of 50 years (as opposed to 60 years or 35 years), nor am I aware of any principle that would justify this particular choice.

In addition, because the Law Dome d18O series is available up to 2007, there is actually more than 50 years available – just not between 1921 and 1990.  If Gergis really believed that additional measurements were needed to pass the relationship between Law Dome d18O and gridcell temperature, then why didn’t she use data up to 2007? Or even 2001?

Because the Gergis reconstruction has only two proxies with values prior to AD1100, inclusion of the Law Dome series would have a material impact on early values of the reconstruction.  In addition, Gergis also reported results from her “R2” and “R3” reconstructions, consisting of only two and three proxies, results which would be materially impacted by inclusion of Law Dome over their entire record.

The New Law Dome d18O Series

The Law Dome d18O series, which is included in the Gergis et al 2016 prescreening archive,  has interesting points of similarity and difference to the prior version (archived in 2010, after prior use in Mann and Jones 2003).  The two versions are compared in the diagram below: the earlier version in the left panel, the more recent version in the right panel.

Both versions show a noticeable decline over the past 1500 years,  with relatively low values in the late 20th century values and in the early 21st century in the more recent version.  Late first millennium values in the version archived in 2016 are noticeably less than in the version archived in 2010. This is more than a little puzzling since both versions use the same DSS core. It’s hard to understand how the earlier sampling could lead to such different results in this earlier period. Because neither version has been formally published, there is no explanation on the record. Nor  is there any explanation in any of the articles which have used or considered the new version (Neukom et al 2012, PAGES2K, Neukom et al 2014, Gergis et al 2016).

Figure 1. Law Dome d18O versions.  Left – version used in Mann and Jones 2003, archived in 2010 data. Right – version used in PAGES2K. Near identical shorter (1000 on) version in Gergis et al 2016. In the right panel below, annual data is converted to 4-year averages to make comparison more direct.

Re-visiting AR4

Unfortunately, this is not the first occasion where IPCC authors have arbitrarily excluded the Law Dome series. In a previous CA post here, I discussed an earlier dispute, on which further light is shed by the present analysis.

The Second Draft of IPCC AR4 included a figure showing Southern Hemisphere proxies for the past millennium. It showed the Mt Read and Oroko Swamp tree ring series (also used in Gergis et al 2016) even though the Oroko series had been screened out of Mann and Jones 2003, but didn’t show the Law Dome d18O which had been used in Mann and Jones 2003.

In June 2006, as an AR4 reviewer, I asked that they include Law Dome d18O in the  diagram.  In an email to fellow AR4 authors, Tim Osborn of East Anglia,  aware of the high early values of the Law Dome d18O series, reported that “sceptics wanted us to show this [Law Dome] for obvious reasons”.

Conversely, Osborn and fellow IPCC authors clearly didn’t want to show Law Dome in their diagram also, to borrow a phrase, “for obvious reasons”.

While each side may have had “obvious reasons” for wanting to show or hide this series, at the end of the day, the decision ought to be made on scientific grounds. As a supposed rationalization for exclusion of Law Dome, Osborn stated that its interpretation was “ambiguous” on the grounds that he thought, but was “not certain” that “some authors” (here citing Souney et al 2002) had interpreted it as an index of atmospheric circulation changes:

Jones/Mann showed (and Mann/Jones used in their reconstruction) an isotope record from Law Dome that is probably O18 (they say “oxygen isotopes”). This has a “cold” present-day and “warm” MWP (indeed relatively “warm” throughout the 1000-1750 period). The review comments from sceptics wanted us to show this for obvious reasons. But its interpretation is ambiguous and I think (though I’m not certain) that it has been used to indicate atmospheric circulation changes rather than temperature changes by some authors (Souney et al., JGR, 2002).

Osborn proposed that they add a reference to Law Dome in the fine print, but not show the series in the figure itself (thus “hiding the decline” in the Law Dome data.

None of the IPCC authors appears to have bothered checking out Osborn’s speculations about Souney et al 2002. Had they done so, they would have immediately discovered that Souney et al did not consider Law Dome isotope data at all; they considered sea salt (Na+) in the Law Dome core. Nor in a Southern Hemisphere context are “atmospheric circulation changes” independent of temperature changes:  changes in temperature are invariably associated with atmospheric circulation changes, particularly with movements of the southern westerlies towards and away from the equator.

Coordinating Lead Author Overpeck, who was eager to minimize anything to do with the Medieval Warm Period, quickly agreed with Osborn’s recommendation, also sneering at the non-“expert”-ness of the suggestion that the Law Dome series be shown to readers:

your suggestion to leave the figure unchanged makes sense to me. Of course, we need to discuss the Law Dome ambiguity clearly and BRIEFLY in the text, and also in the response to “expert” review comments (sometimes, it is hard to use that term “expert”…). Ricardo, Tim and Keith – can you take care of this please. Nice resolution, thanks.
best, Peck

It is interesting to re-visit this exchange in light of the correlation and t-statistic reported above. The correlation (and t-statistic) between Law Dome d18O and gridcell temperature is exemplary- far better than the two tree ring series accepted by AR4 authors.  There was never any valid reason for IPCC authors to exclude the Law Dome series from their diagram of Southern Hemisphere proxies.

Law Dome in a Holocene Context

Law Dome is one of the very few sites where data is available for both the early 21st century and through the Holocene. Such sites are of unique importance for assessing whether modern temperature increases have escaped the Milankowich boundaries -an issue that I regard as central to  practical interest in proxies. It also provides a welcome Southern Hemisphere perspective to a topic dominated by Northern Hemisphere proxies (and, for “skeptics”, far too much by the Cuffey-Clow temperature reconstruction from GISP2 e.g. recent WUWT discussion here).

Although the deep Law Dome core was drilled between 1987 and 1993, the record is still frustratingly incomplete. The figure below combines available elements (see below figure for explanation):

Figure 1. Law Dome d18O series (Morgan et al 2002; Tas van Ommen, email 2006, archived 2010); insert – dD from Masson et al 2000.

The main panel shows d18O values.  The deglacial portion (left) shows a more or less continuous increase in d18O values from low values around 18000BP to a maximum around 9500BP.  Values over the last two millennia are less than those in the early Holocene.  In this figure, I’ve used the older (2010 archive  version of recent d18O data since the shape seems to me to be more similar to the recent portion in the dD series in the insert figure (from Masson et al 2000). The (unarchived) Masson et al dD series also shows peak Early Holocene warmth with a gradual decline through the Holocene, with low values in the 20th century.   The early Holocene maximum in the Masson et al dD series is somewhat earlier than the early Holocene maximum in the d18O series (Morgan et al 2002), but I presume that this difference in timing arises from differing dating systems for the same core. (In any event, it is not material to this post).

As a caveat on the interpretation of Holocene scale d18O series, Vinther et al 2009 convincingly showed that decreases in ice cap elevation led to long-term relative increases in d18O values (GISP2, GRIP) relative to those observed in locations with more or less constant elevation (Renland, Agassiz).  A Law Dome series adjusted to have constant-elevation would therefore show an even larger decline in d18O values over the Holocene.

Conclusion

The Law Dome d18O series has a stronger correlation to gridcell temperature than 24 of 28 “passing” proxies or either of the long tree ring series used as long proxies in Gergis et al 2016.  It was excluded from the Gergis et al network based on a additional arbitrary screening criterion that excluded Law Dome without impacting any other proxies in the screened network. It is not known whether Gergis et al intentionally added the additional screening criterion in order to exclude Law Dome or whether the criterion had been added without fully understanding the ramifications, with the exclusion of Law Dome being merely a happy coincidence.  In either case, the exclusion is not robust.  And because the Gergis et al 2016 reconstruction (R28) is based on only two proxies in its early portion, neither are its various reconstructions.  The impact will be particularly felt on the R2 and R3 reconstructions, which have only two and three proxies respectively.

Just another day of data torture by the paleoclimate “community”.

The new analysis also sheds new light on AR4’s decision not to show Law Dome d18O to readers. In retrospect, it seems very clear that they didn’t want to show the decline in Law Dome d18O for fear that, to borrow a phrase, they might “dilute the message” or give “fodder to skeptics”.

Law Dome is an important location, since it is one of relatively few proxy datasets in the world which has results both through the late 20th century (actually up to 2007) and through the Holocene.   I am increasingly of the view that there is relatively little purpose in examining proxy data over the last one or even two millennia without placing it in a Holocene perspective.  While there is obviously ongoing “skeptic” interest in a Holocene perspective, such commentary (e.g. recent WUWT posts) is far too often limited to a Northern Hemisphere (Greenland) perspective and, in particular, to the problematic Cuffey-Clow temperature reconstruction from GISP2 (which ends in 1855, though GISP2 isotope data is available to 1987).

Because precessional forcing over the Holocene is opposite for the two hemispheres, both Southern Hemisphere and Northern Hemisphere should be shown.  The Law Dome series is instructive for such comparisons. It yields an entirely different perspective on the relationship between modern and early Holocene values than the bogus Southern extratropic reconstruction of Marcott et al 2013 that has been credulously accepted by many academics.

Unfortunately,  use of Law Dome d18O is compromised by inadequate publication and archiving of Law Dome d18O data by the Australian Antarctic Division.  Tas van Ommen is a polite and cordial correspondent, but that only goes so far when nearly 30 years has passed without a proper publication of Law Dome d18O results.

Source Code

See: http://www.climateaudit.info/scripts/multiproxy/gergis_2016/gergis_cor_final.txt

References:

Masson et al, 2000. Holocene climate variability in Antarctica based on 11 ice-core isotopic records. Quaternary Research.

Morgan, V., M. Delmotte, T. van Ommen, J. Jouzel, J. Chappellaz,
S. Woon, V. Masson-Delmotte, and D. Raynaud. 2002. Relative Timing of Deglacial Climate Events in Antarctica and Greenland.

Neukom and Gergis, 2011. Southern Hemisphere high-resolution palaeoclimate records of the last 2000 years. The Holocene.

Neukom, Gergis et al, 2011. Multiproxy summer and winter surface air temperature field reconstructions for southern South America covering the past centuries. Climate Dynamics.

Postscript

The failure of the Australian Antarctic Division to properly publish this data is really quite puzzling. Ironically, I first sought this data in August 2003, leading to multiple promises to publish the data, none of which were fulfilled.  I haven’t previously documented these efforts and do so below.

In August 2003, shortly after the publication of Mann and Jones 2003 and before anyone had heard of me, I asked Tas van Ommen of the Australian Antarctic Division for the Law Dome d18O data used in Mann and Jones 2003 as follows:

dear Dr. van Ommen, I’m studying the Mann and Jones GRL [2003] article carrying their projections back to 200 AD. Your Law Dome dO18 dataset features prominently in the compilation. I have not located this dataset at NGDC and was wondering if you could direct me to an FTP location or otherwise email me the dataset. Thanks, Steve McIntyre

Van Ommen refused to provide me the data that he had provided to Mann and Jones, politely refusing, but refusing nonetheless on the grounds that they expected to submit an article on the data “in the coming few months”:

As you may have detected in the Mann and Jones article from the way the Law Dome data was cited (acknowledged, to be more correct) Mann and Jones were using unpublished data.  We are nearing completion of a fuller treatment of the Law Dome d18O dataset for publication – it has only recently reached its present state of calibration and dating.  We expect to submit it in the coming few months.  Once it is accepted, we will release the data.  So, I am happy to provide the data but I must postpone delivery until that time.

In February 2004, six months later, I asked for a second time, once again being put off by van Ommen, who said release of the data was controlled by a paper then being finalized:

I [van Ommen] am finalizing a paper that will allow me to release the isotope record more widely. It is this next paper that controls the timeframe for release to you and archives.

The following day, the Climategate dossier shows that Van Ommen notified Phil Jones about my inquiry. (In subsequent FOI requests, various universities have claimed that they regard correspondence as confidential, but no such deference was given to my correspondence which was immediately broadcast among the Team.)

What you will find below is (in reverse chronological order) an email interchange between Steve McIntyre and myself.  He has been asking for LD data for a while (since your GRL paper came out) and to my chagrin, I have put him off once already, for reasons I spell out below.  For your information, I am close to submitting the full LD isotope record, which I hope to present at SCAR Bremen, along with some interesting spectral analyses and comparison to EPICA Dome C.

Jones, who forwarded the correspondence to Mann, indicated his plans not to cooperate with such requests, while Mann vehemently urged that no data be provided to me.

The following year (February 2005), I once again asked van Ommen for the Law Dome data:

Dear Tas, is the dO18 information for Law Dome supplied to Jones and Mann [2004] available yet? Thanks, Steve McIntyre

Once again, van Ommen declined to provide the data:

I will get back to you shortly on this. It has been a busy summer – field trip to Antarctica and then some vacation and I have been out of the office a lot. More soon.

The following year (March 2006), I repeated my annual inquiry for the fourth time:

BTW could I get a digiital version of the Law Dome data about which I inquired a couple of years ago and which was supplied to Jones and Mann.

This time, I was partly successful. Van Ommen finally sent me the requested data, but only on a confidential basis pending publication, which was supposedly “getting near, and along with that, a full open release of the data”:

My apologies for not following that up – things have been rather hectic.

The data set I am sending is, I am displeased to report, still within my internal “mill”. In the interim, I have been involved in a further Antarctic field trip and the publication of the isotope data has been slowed somewhat. Particularly since we have been able to get extra material that will bring our record up to 2004AD (eventually at seasonal/quasi-monthly resolution). You will appreciate more than many that an extra several years greatly improves our statistical power for meteorological calibration (given the shortness of the instrumental record in Antarctica). So, I really wanted to hold off until I had the longer set to work with. This being in place, and analysis nearly complete, publication is getting near, and along with that, a full open release of the data.

Until then, I give you the data that Jones used, for your own purposes. Please do not distribute the data itself.

Van Ommen’s observation that the additional data “greatly improves” statistical power for calibration is an interesting point in connection with Gergis et al 2016, of which van Ommen is listed as a coauthor.

In 2011, Neukom, Gergis and others published a supposed compendium of Southern Hemisphere proxies (Neukom et al, 2011. Clim. Dyn.), which was later relied upon for the Gergis et al 2012 and Gergis et al 2016 networks prior to screening. In January 2011, I asked Neukom to provide the unarchived data, one of which was Law Dome d18O provided to Neukom as a “pers. comm.”(see CA here):

Dear Dr Neukom,
I notice that your recent multiproxy article uses a number of proxies that aren’t publicly archived. Do you plan to provide an archive of the data as used in your study? If not, could you please send me a copy of the data as used. Thanks for your attention.
Regards, Steve McIntyre

Even though there had been a “consensus” post-Climategate that data in multiproxy studies be archived, Neukom refused.

Thanks for your interest in our work. Most of the non-publicly available records were provided to us for use within the PAGES LOTRED-SA initiative only and I am not authorized to further distribute them. You would need to directly contact the authors. I am sorry for that.

This was precisely the sort of daisy chain chase that I had long objected to. My position – both before and after Climategate – was that multiproxy authors should be required to have obtained permission from original authors to publicly disclose data before considering it in a multiproxy study or else not use it.  To its credit, Nature has adopted this policy, but many paleoclimate journals (e.g. Holocene) acquiesce in data obstruction.

Subsequently, I asked Journal of Climate to require Gergis et al (in connection with Gergis et al 2012) to provide the

The compilation of this database represents years of our research effort based on the development of our professional networks. We risk damaging our work relationships by releasing other people’s records against their wishes. Clearly this is something that we are not prepared to do.

We have, however, provided an extensive contact list of all data contributors in the supplementary section of our recent study ‘Southern Hemisphere high-resolution palaeoclimate records of the last 2000 years’ published in The Holocene (Table S3)… 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’.

We will not be entertaining any further correspondence on the matter.

Though I obviously rejected Gergis’ refusal, I followed up with van Ommen about the publication promised many years earlier:

Dear Dr van Ommen,

Some years ago, you wrote me as follows saying that you were anticipating

“The data set I am sending is, I am displeased to report, still within my internal “mill”. In the interim, I have been involved in a further Antarctic field trip and the publication of the isotope data has been slowed somewhat. Particularly since we have been able to get extra material that will bring our record up to 2004AD (eventually at seasonal/quasi-monthly resolution). You will appreciate more than many that an extra several years greatly improves our statistical power for meteorological calibration (given the shortness of the instrumental record in Antarctica). So, I really wanted to hold off until I had the longer set to work with. This being in place, and analysis nearly complete, publication is getting near, and along with that, a fullopen release of the data.”

Has this been published yet?  I notice that the data in Schneider et al 2006 came to 1999, but not to 2004. How far back were you able to extract annual values?

Regards, Steve McIntyre

Van Ommen replied that the papers were “just submitted or about to be in coming week”:

We have not published any further Law Dome d18O results since we last corresponded, with the exception of a study of the last deglacial period (ca. 20ky-10ky BP) for which those data are archived publically.

Our research focus has been on trace chemistry work, snowfall rates and also for me an excursion into ice sheet work.

This will shortly change with papers either just submitted or about to be in coming weeks. These will benefit from improved dating arising from the trace chemistry studies and annual values back to ca. AD170. As soon as any of the publications are accepted we will be archiving the corresponding data set.

I then asked van Ommen for the data that he had provided to Gergis:

Gergis et al cited a newer version of the d18O data, which they considered in their study, which refers to a start date related to the salt studies.  It looks very relevant to their conclusions.  Could I get a copy of the data that they were sent for the purpose of commenting on Gergis et al.  Regards, Steve Mc

Van Ommen had been corresponding with Gergis about my data request.  In order to frustrate my request for the data prior to screening, he had suggested to Gergis that I only be provided with the 1921-1990 portion of the data, only enough to check their screening. He followed this tactic in his reply to me:

As you know, the Gergis study rejected the Law Dome d18O series as a non-predictor. It was not significantly correlated (detrended, as described in the paper) to the instrumental target series over the period 1921-1990. If you wish to check the rejection correlation, then you can do so. The data set through this time period is “as archived” from the 2006 Schneider and Steig paper and publicly available. To save your effort, I reproduce the 1921-90 portion below.

1990 -23.43
1989 -21.91
1988 -21.22
1987 -23.07  ….

I then asked van Ommen about an apparent discepancy between the d18O series of Jones et al 1998 and the more recent versions:

Tas, maybe you can clarify something. In the graph below, I’ve compared the annual data from the Law Dome O18 data used in Jones et al 1998 (red) to the data in the LD2.1kyr file (blue)  that you sent me a number of years ago (which, as you observe, matches the Schneider average in the last 2 centuries.)  The values of the two versions match closely from about 1600-1840, but diverge quite sharply after that.  I haven’t seen any discussion of this and I was wondering what caused the difference in results in the 20^th century? Regards, Steve McIntyre

van Ommen politely answered as follows:

I believe we covered some of this in our exchange back in 2008. I noted that we haven’t published on the data set that Phil used in 1998 and didn’t wish to archive it publicly as it was a very early product that might have been state of the art at that time, but which was superseded by data coming on line in subequent years: noteably the LD2.1kyr data set and the Schneider and Steig data sets (both archived).

Anyway, to the matter of why they are different, I did make some passing comment in 2008 – gaps, newer cores and a refinement of dating and calibration were mentioned. Specifically, the Jones 1998 data uses the upper part of the initial DSS core, thermally drilled in 1987, and analysed in new mass spectrometry facilities. There was a section of core when drills were changed (corresponding to the ~1840-~1880 gap) that was in bad condition and so we didn’t have a series through that portion. These were the early pioneering days, and since then we have improved all aspects of our operation end-to-end, with better technology, calibrations and replicate cores (dry drilled rather than thermally, which helps) over the period in question. That is why the records that we are releasing are different.

I hope that helps  – please don’t hestitate to continue the dialog if you have further questions.

I responded as follows:

Thanks for the clarification.  I have experience in drilling cores (though in the mineral business) and understand that mechanical vagaries occur under much less demanding circumstances than you’re experiencing.

Given both the importance of d18O as a proxy andIt  Law Dome as a high resolution site, I’m surprised that publication of Law Dome O18 results has been so sketchy to date.  BTW contrary to your email, I’m not aware of the LD2.1 dataset being archived anywhere.  It’s not at NOAA nor did it turn up in a AADC search under “law dome”. If I’m mistaken, the location isn’t easy.

Van Ommen responded by pointing to metadata record at AADC which did not actually contain a public archive, but a direction to manually request the data. Van Ommen reflected on this overnight and decided to make the data available online after all:

I just checked this out again and on consideration I will just have the data opened for automated access – should happen within a day.  Although my intent was to be constructive rather than obstructive with access, and I have no evidence that anyone had complaints, I reckon this is a better option.

In this exchange, van Ommen also made the prescient comment that studies using Law Dome as a temperature proxy had assessed correlation to local temperature (which was high), whereas Gergis et al 2012 had assessed correlation to an Australasian regional composite (which was low).  So at least one Gergis et al 2016 coauthor was keenly aware of the strong correlation of Law Dome d18O to local temperature:

One point you may want to consider in your view of various uses of the proxy data – some, like Mann and Jones,  or Schneider et al., have assessed Law Dome for its correlation to _local_  Antarctic temperature. We know for isotopes in polar precipitation that this is to be expected. Gergis et al were assessing the correlation to very distant temperatures in the Australasian sector – their finding: not much. In the absence of a particular teleconnection, this is also expected.

In 2016, the Australian Antarctic Division archived their official Law Dome d18O version over the past two millennia (see here), still without a covering academic publication.

In August 2003, van Ommen had said that they expected to “submit it in the coming few months”. Thirteen years later, the long and multiply promised article remains unpublished.