From the “our expert is smarter than your expert” department and Tallbloke’s Talkshop:
How NIWA added lots of warming in New Zealand – and got away with it – so far
Something is rotten in the state of Denmark New Zealand. A judge has disregarded as inadmissible expert evidence from a statistician who showed that the adjustment method NIWA claimed they used only give a 0.3C/Century rise in temperature and ruled that NIWA (now a limited company) can adjust the temperature record as they see fit without having to demonstrate the use of a method based on any accepted science. here’s the graph:
NIWA are now claiming $118,000 from two named individuals in costs. This looks like a vindictive move.
NZCSET, the organisation which brought the review case to court are examining their options for appeal.
A comment on Jo Nova’s site summarizes the judgement.
Jaymez
September 7, 2012 at 10:17 pm · Reply
It is a very disappointing judgement. I have read the entire judgement and at first pass it seems to me that Justice Venning found himself an escape hatch. Judges do not want their verdicts to be controversial or overruled, and you can bet he knew if he had found in favour of the ‘Coalition’ his ruling would have been attacked.
His escape hatch was two pronged:
1. Venning found that two of the key ‘expert witnesses’ for the Coalition, did not satisfy his definition of ‘expert witness’. That allowed him to simply disregard their arguments and evidence rather than look at them with as much weight as he applied to the testimony of NIWA’s experts which he appeared to accept without question.
[51] Section 23 of the Evidence Act 2006 provides that a statement of opinion is not admissible except as provided by ss 24 or 25. Opinion is defined in the Evidence Act at s 4 as: “A statement of opinion that tends to prove or disprove a fact.” I accept Mr Smith’s submission that there are substantial portions of Mr Dunleavy’s original and reply affidavits where he proffers opinions on matters in issue in the proceedings, particularly on scientific practices and the validity of the scientific practices of NIWA. Such evidence could only be admissible under s 24 or s 25. Section 24 is not applicable in the circumstances. Section 25 could only apply if Mr Dunleavy was an expert in the particular area of the science of meteorology and/or climate. He is not. He has no applicable qualifications. His interest in the area does not sufficiently qualify him as an expert. I also accept Mr Smith’s further point that Mr Dunleavy’s views are not capable of offering substantial help to this Court on the issue that it has to determine. To that extent I agree that substantial passages of Mr Dunleavy’s evidence are inadmissible.
[53] Similar issues (as to the limited nature of his expertise), apply to the evidence of Mr Dedekind. Although in his affidavit in reply Mr Dedekind purported to comply with r 9.43, Mr Dedekind’s expertise is in relation to computer modelling and statistical analysis.
In fact Vennings makes the following outrageous statement which reminds me of the same criticisms which were made of Ross McKitrick and Stephen McIntyre who were ‘unqualified’ to critique Michael Mann’s infamous Hockey Stick graph, with errors which had escaped the notice of thousands of supposedly qualified climate experts!
More here:
Thanks for highlighting this one Anthony. To try to help our friends in NZ, I’ve been emailing John Christy today, and hopefully he’s now onboard to assist with his expertise and status as a bona-fide card-carrying climatologist the appeal judge will have to listen to.
The NZ statistician who has done the bulk of the analysis work has called in at the talkshop to discuss the issues, so come on over, everyone welcome.
Amazing expert statistician can’t provide evidenced on the use of statistic in climate science but bad statics from climate scientists themselves must be taken on face value.
Th fact that NIWA still will not claim that they produce a ‘official’ warming value tells you how much they really trust their own work to stand up to a real review from those that on their payroll .
So according to the judge climate statistics and statistics are two separate and unrelated fields.
Its all starting to make sense now.
“… Mr Dedekind’s expertise is in relation to computer modelling and statistical analysis.”
Yet these are exactly the same areas within which his testimony was to be applied.
Modern science has a problem in that it is now trivially easy to apply various statistical tools to data using readily available computer ‘statistical tool kits’, without necessarily understanding the techniques involved, and without needing to verify whether the techniques be appropriate to the data or to the conclusions being drawn from the processing.
When I went to University something like a linear regression could take days, weeks, or even months of laborious calculations, depending on the size of the dataset, using a slide rule or log tables. So before embarking on such an arduous task, one made bloody well sure that the method was going to be appropriate and applicable.
A researcher nowadays can toss the data into 50 or 100 different computerized ‘black box’ algorithms until one set of such algorithms appears to excrete the results that are expected or desired.
And it may require an expert in statistical analysis to be able to confirm whether the method is valid for the conclusions drawn.
Should’ve asked the BOM to give their expert opinion on NIWA methods and data use..now that would’ve been fun to see and hear.How can a judge dtermine who’s more expert on somthing he has no expertise himself is mystifying.appeal appeal…and retire this numpty judge
At the end of the day the judiciary will rule that the expert giving evidence that conflicts with NIWA is not an expert climatologist or statistician. Put up a climatologist and be told that he is not a statistician. Put up a statistician and be told that he is not a climatologist. In NZ new evidence is not normally accepted in appeals which are supposed to consider only matters of law. The success rate on appeal is negligeable here. NZ judges are most loth to over-rule their mates. In a society of four million these buggers all know each other! The only way to get a crazy verdict overturned is to get a judge in from Australia to review the whole case. This was done in the Arthur Allan Thomas case by the Prime Minister of the time, Muldoon. That being about the only decent decision the bugger made in his whole political life.
The judge’s ruling is a gift to sceptics. He is basically saying that the climate record in NZ need have no basis in fact or science. So he is basically saying that the record has no meaning except the opinion of the keepers of the record (pure opinion, not peer-reviewed scientifically-justified information). That is huge. That should be shouted from the rooftops by every sceptic in the world.
Legal process has become one of the casualties of main stream climate science.
Here is a completely independent analysis of the New Zealand temperature record: http://berkeleyearth.lbl.gov/regions/new-zealand
Seems to correspond pretty well to that of NIWA.
hey Zeke did you see my email to you last week? Never got a reply.
Zeke Hausfather says:
October 25, 2012 at 3:49 pm
Here is a completely independent analysis of the New Zealand temperature record: http://berkeleyearth.lbl.gov/regions/new-zealand
With 1.5 trillion dollars of tax “revenues” at stake if the government is wrong, are you going to trust a .gov site to give you accurate results when checking government-modified figures to audit another part of the government-needed “research” on the climate??
Give them a cheque for $1.53 and tell them it can be “adjusted” in 50 years time to whatever amount they like.
Zeke:
“Seems to correspond pretty well to that of NIWA.”
Not really. The BEST trend from 1910 is 0.68°C/century, while NIWA’s from 1909 is 0.91°C/century.
When we applied the RS93 method rigorously on the same stations, we came out with 0.34°C/century.
Bob D,
You are correct that Berkeley has a lower trend (0.68 vs 0.91 for 1910-2010). I was simply eyeballing the graphs, as the NIWA site doesn’t make it particularly easy to download the anomalies. Mea Culpa on my part for a quick post before digging into the data (in my defense, its been a long day at work :-p )
That said, the Berkeley results are still more in line with NIWA than the pre-adjustment record.
RACookPE1978,
I’m rather partial to that particular .gov site, given that I’m a member of the team who is responsible for it.
This is rather an old, though shocking, story – here is the Jo Nova thread :
http://joannenova.com.au/2012/09/nz-justice-shows-courts-are-useless-in-a-science-debate/
There is a certain wry amusement in that he considered a statistician to be unqualified to comment on the statistical methods used by climate scientists as they are unique to climate ‘science’ – they certainly are!
Zeke,
Great stuff….the earlier data is based on assumptions based on Hobart in australia less the number we first thought of …maybe a bit more or less…… .check…http://www.climateconversation.wordshine.co.nz/2012/10/niwa-says-it-wasnt-about-climate-change/comment-page-1/#comment-129735
Zeke Hausfather says:
That said, the Berkeley results are still more in line with NIWA than the pre-adjustment record.
————————————-
You realize that the BEST data is adjusted (homogenized, normalized, maladjusted and then molested) as well?
You’re looking at two counterfeit notes (bills) and saying, “Look, they match. They must be real.”
BTW It’s been over a year, has the BEST study passed peer review yet? How long does is take to rubber stamp these things?
Reg Nelson,
Yes, BEST fiddles with the data like Michael Mann taught. Compare the before and after:
http://i.dailymail.co.uk/i/pix/2011/10/30/article-2055191-0E974B4300000578-6_634x639.jpg
Another incredible hokey stick chart.
I am not a qualified chef, but I can smell a rotten egg. Although two Universities have suffered me, I am not an “expert” in climatology or statistics, but I can recognise a massaged result when I see one. Possibly my main qualification is that I do not rely on the system for my pay packet, so I can beg to differ from the official line.
Zeke:
October 25, 2012 at 5:23 pm
“…in my defense, its been a long day at work :-p”
No worries.
“That said, the Berkeley results are still more in line with NIWA than the pre-adjustment record.”
If you mean that BEST is also high, then yes. I don’t have access to the individual station records used in BEST, nor do I know how each site adjustment was made, so I can’t comment on their results. If they use an automatic homogenising algorithm, then they will certainly introduce errors in the NZ record, simply because most NZ sites are urban (UHI) and/or sheltering affected (see Hansen, 2001). A good example is the transition from Albert Park to Auckland Aero. See here (pg 57):
http://www.climateconversation.wordshine.co.nz/docs/Statistical%20Audit%20of%20the%20NIWA%207-Station%20Review%20Aug%202011%20SI.pdf
For a summary, see here:
http://www.climateconversation.wordshine.co.nz/docs/Statistical%20Audit%20of%20the%20NIWA%207-Station%20Review%20Aug%202011.pdf
On the other hand, I am fairly well acquainted with each site adjustment made on the NZ 7SS, and am confident that the BEST team is unlikely to have gone into the detail I have in checking the NZ adjustments specifically. That’s not to say the “true” NZ trend (1909-2009) is exactly 0.34°C/century, but I have more confidence in that number than, say, BEST’s 0.68°C/century.
Regarding the adjusted vs unadjusted trends, if we are prepared to accept that NZ covers a reasonable spatial scale, we would expect (Peterson 1995, 1998) that homogeneity adjustments should largely balance out, implying that the further you are away from the unadjusted trend, the more skeptical you should be about the result, initially at least.
When we excluded the two sites most affected by UHI and sheltering (Auckland and Wellington – see Hessel, 1980 & Fouhy, 1992) the trend became 0.26±0.30°C/century, and the homogenised sites all conformed pretty closely to that number (0.21 to 0.27°C/century, with one outlier, Masterton at 0.36 – also subject to UHI) Unadjusted trend, for reference, is 0.23°C/century. Contrast that with NIWA’s results, where the trends vary greatly over the 7SS (0.61 to 1.53°C/century) and the final result of 0.91°C/century is very far from the unadjusted number, and also significantly above the global average for the same period.
Bob D;
Is it just a coincidence that your 0.34 is exactly ½ of their 0.68?
>:)
“Albert Park to Auckland Aero”
Sorry, that should read Albert Park to Mangere (right next door to Auckland Aero).
Brian H says:
October 25, 2012 at 7:28 pm
“Is it just a coincidence that your 0.34 is exactly ½ of their 0.68? >:)”
Sadly, yes. 🙂
I find it odd we get so wrapped up in 100 to 200 year trends, when we have much longer trends that are able to show a bigger picture that does not necessarily depict anything other than a small general interglacial period of temperature variability that is insignificant. Yet there is such a cascade of effort to discuss basic nothingness. $,$$$,$$$,$$$
peterg says:
October 25, 2012 at 2:36 pm
So according to the judge climate statistics and statistics are two separate and unrelated fields.
Its all starting to make sense now.
Of course, “statistics” relate to the accuracy of a trend, “climate statistics” relate to the absence of money in a treasury.
Surely, in a civil case, there is no presumption of innocence or guilt and no presumption of truth or falsity?
In presuming that NIWA told the truth, simply because they have status, the judge is wrong in law?
Roger Dewhurst
All so familiar…from the Wegman Report:
“…Generally speaking, the paleoclimatology community has not recognized the validity of the [McIntyre and McKitrick] papers and has tended dismiss their results as being developed by biased amateurs. The paleoclimatology community seems to be tightly coupled as indicated by our social network analysis, has rallied around the [Mann] position, and has issued an extensive series of alternative assessments most of which appear to support the conclusions of MBH98/99… Our findings from this analysis suggest that authors in the area of paleoclimate studies are closely connected and thus ‘independent studies’ may not be as independent as they might appear on the surface.
It is important to note the isolation of the paleoclimate community; even though they rely heavily on statistical methods they do not seem to be interacting with the statistical community. Additionally, we judge that the sharing of research materials, data and results was haphazardly and grudgingly done. In this case we judge that there was too much reliance on peer review, which was not necessarily independent.
Based on the literature we have reviewed, there is no overarching consensus on [Mann’s work]. As analyzed in our social network, there is a tightly knit group of individuals who passionately believe in their thesis. However, our perception is that this group has a self-reinforcing feedback mechanism and, moreover, the work has been sufficiently politicized that they can hardly reassess their public positions without losing credibility…”
http://climateaudit.org/2007/11/06/the-wegman-and-north-reports-for-newbies/
Very few temperature histories have been tampered as much as NZ. To make it simple, NIWA altered the temperature readings of locations, some over a hundred years old. Approximately a dozen benchmarks. It did so after it concealed all the real readings, meticulously kept in logs. These logs were then made unavailable to all, including staff and university researchers. However NIWA was unaware that a copy, and perhaps more, were at the national library. When compared with the reported temperatures, it was found there was a profound difference. The past was much colder in the ‘new’ past than the actual readings. NIWA then went into a bizarre explanation about how a location had been changed for a couple thermometers, a time of day reading was inconsistent and had to be discarded, ad infinitum. It was clearly contrived. Fraudulent.
Roger D.;
You defeat yourself! “No presumption” means they do not get the benefit of the doubt.
Manfred said on October 25, 2012 at 10:27 pm “All so familiar…from the Wegman Report:”
Nice… I hadn’t seen that before. It should be required reading for all politicians, along with a compendium of all the dubious, questionable and downright fraudulent data manipulation that continues to this day.
It is a shame that we don’t have the same graph above, but with addition of the difference between the two curves. (If curves is the right word). Usually a wonderful way to clearly demonstrate that there is no possible justification for the data tampering that has been done.
When judges run away from their responsibilities to deal with contentious issues as in this case. This can only weaken the independence of the judiciary bringing with it potentially negative consequences for society at large.
Bob D,
“If you mean that BEST is also high, then yes. I don’t have access to the individual station records used in BEST, nor do I know how each site adjustment was made, so I can’t comment on their results. If they use an automatic homogenising algorithm, then they will certainly introduce errors in the NZ record, simply because most NZ sites are urban (UHI) and/or sheltering affected (see Hansen, 2001).”
BEST doesn’t make homogenization per se but the scalpel method comes back substantially to automatically homogenize at the aggregation stage. The critique of adjustments contained in Hansen et al. 2001 therefore also applies to BEST.
The only way to obtain regional and global anomalies curves beyond this critique is to use only raw long series. One could also theoretically proceed through short series using raw temperatures and not anomalies but This raises issues particularly in relation with altitudes.
You dont need an “expert” to see whats wrong!
A highschooler is enough!
phi,
I think its fair to call the scalpel a form of homogenization. The Berkeley method also uses correlation-based weighting in the Kreiging process, so records that are significantly at odds with surrounding records (after being cut at detectable breakpoints) will be somewhat down-weighted in the resulting regional averages.
I’m going to be presenting a poster at this year’s AGU comparing how well the Berkeley method and Menne’s PHA do at dealing with synthetic temperature data with known inhomogenities added to it (though we are currently blind to what those are until we complete the analysis). The idea is to compare how well the different methods do at correcting records contaminated with large step changes, small step changes, gradual positive trend biases, gradual negative trend biases, etc. knowing what the “real” temperature record was before these inhomogenities were added. Its very similar to the approach taken with the PHA in this paper: ftp://ftp.ncdc.noaa.gov/pub/data/ushcn/papers/williams-etal2012.pdf
Reg Nelson,
Oddly enough, peer review can take quite some time. 1-2 years is not particularly uncommon for science papers.
Bob D,
I can’t claim to have looked at individual NZ stations in much detail; I’ll spend a bit of time on it over the weekend/next week and see what things look like. It looks like Berkeley is using 48 New Zealand stations (32 current, 16 no longer providing data), plus a few others within 500 km for the early part of the record when little NZ coverage is available (though distant stations are considerably down-weighted relative to nearby stations). You can download the unhomogenized data (though I’d suggest using the quality controlled version) here: http://berkeleyearth.org/data/
Zeke,
I’m pretty skeptical about the ability to simulate and adequately address the major perturbations affecting temperatures. In particular, UHI which not only has a continuing character but can still affect more or less parallely the majority of stations. In all cases it is clear that practically continuous effects are not corrected (see http://wattsupwiththat.com/2012/10/15/circularity-of-homogenization-methods/#comment-1113793).
What is particularly comical is that stations moves represent a partial correction and that the main effect of homogenization is to remove these corrections.
Gee they aren’t indictable because they are a private company and not obliged to provide their data and charts to officialdom. Aren’t they suable if they are providing products for sale that are defective or for which they are making false claims? Judge, these guys aren’t in it for fun without profit.
phi,
I agree that the ability of homogenization to correct UHI is a rather interesting and relevant question. Matt Menne and I have a paper under review at the moment that examines this very issue in some detail, but I don’t want to dive too deeply into the details until its published.
Zeke,
It will be a pleasure to read you in time.
Zeke – New Zealand may have an interesting UHI “control experiment” in the form of the city centre of Christchurch (wholly evacuated on 22FEB11 following a major earthquake, still largely unoccupied).
All the best.
Zeke,
You should be highly skeptical in relation to the Berkley graphs that you link to. They are only a sample of the data available within NZ. I have conducted a detailed examination of the NZ national climate database nosted by NIWA (cliflo database). As an example for the year 1940 I located and downloaded data from 68 climate stations reporting mean, max, and min temp. These stations reported daily, monthly and annual data. In your graphic there are only 9 stations within NZ in 1940. Similarly for 1983 I located 201 climate stations in the database. This number excludes dozens more stations where there were from one to three months of “missing” data.
In addition I note for everyones benefit that the NZ database is rather incomplete. From the 1870’s through to the 1930’s the database does not include the daily wind, barometric and temperature data collected at numerous shipping ports and lighthouses. That data, collected and used through that entire era for daily weather forecasting (by Captain Robert Edwin and Mr DC Bates of the NZ Met Service) has been lost by NIWA. One of the “data streams”, the 9 am temperature, was reported daily by several NZ Newspapers (delivered by telegram from the Met Service). This data covers the entire country and could be assembled from 1887 through to 1940 with reference to as few as two newspapers. Unfortunately neither was published on Sunday, but that probably would not influence the temperature trend. Please note that there were no changes to the time of observation from the 1860’s through to 1927.
There are also some metadata issues that have not yet been recognised by NIWA. One of these has a material impact on any temperature trend calculated using the early data from Nelson. If the Nelson data is incorporated in the Berkley subsample then that assemblage is contaminated as well.
phi:
“What is particularly comical is that stations moves represent a partial correction and that the main effect of homogenization is to remove these corrections.”
Exactly correct. I would love to see how BEST handles the NZ station moves I’m familiar with. I doubt very much it’s done properly, especially dealing with gradual effects like UHI and shelter growth.
Shelter cut-backs will provide the algorithm with a clear discontinuity, but it will probably miss the growth of the shelter in the first place, and introduce a completely spurious warming adjustment. Same with UHI, where one site is closed (eg: Albert Park) specifically because of UHI and shelter, and another new site (reading cooler) is opened. A discontinuity is detected, and wham! we have false warming added (Hansen, 2001)
Notice that in both these cases the homogenization has introduced false trends, and both in the same direction (false warming introduced).
There are very few mechanisms for the balancing case: where cooling has occurred gradually over time followed by a sudden step-like warming.
Therefore the automatically homogenizing algorithms are very likely to be inherently useless.
Of course, it may be that they take all this into account in BEST, but I’ll believe it when I see it.
Zeke,
I don’t have time at the moment to go through all the BEST data to extract the NZ stations, but if you have a quick method to get the data I’ll be very happy to work with you to look specifically at the NZ region.
We have the advantage here in dealing with a fairly well-known and well-analyzed set of sites, with reasonable metadata. Since NIWA and the NZCSC have published their reviews of the 7SS, we have both an “official” and a “skeptical” set of adjustments using different methods.
I think it would be valuable to see how well BEST performs on homogenizing this small data set.
Bob D,
“Of course, it may be that they take all this into account in BEST, but I’ll believe it when I see it.”
There is no claim from BEST to adequately address gradual biases.
You know if they had done it in this order it may have been a little more convincing:
1. Hey, those older temperatures might be wrong, let’s correct and adjust them.
2. Hey, will ya take a look at that! … Its gettin’ warmer!
3. Must be that CO2! Let’s do something! Anything!
But it was really done like this:
1*. Hey, all this CO2 must be causing trouble, it could be making things warmer!
2. Um…it’s not really, you know, but wadyareckon about the older temperature records? They look wrong, so let’s adjust ’em!
3. NOW it’s warming! What did we tell ya?! Ha! Do Something! Anything! But do it now!!
(1* … perhaps with a little addendum: Political animal in background mutters to self, “Hey, carbon dioxide, a trace gas, easily measurable, output by most factories and vehicles, farms and… just about anything…now THERE is something we can tax and trade…that’s even better than our attempts to commoditize water!”)
If someone described NIWA as the UEA of NZ, would both organisations sue for libel?
>If someone described NIWA as the UEA of NZ, would both organisations sue for libel?
🙂 Actually NIWA couldn’t, apparently. I’ve heard that under NZ law you can only libel a person, not an organisation.
Of course, the links between UEA and NIWA are many. The Climategate emails show the level of collusion between the various members, particularly in the sordid affair surrounding the ganging up on Chris de Freitas for daring to publish a paper that went against the group-think of these people.
J.W.D. Hessel of the New Zealand Meteorological Service, Wellington NZ. published a paper in the N Z. Journal of Science 1980, Vol 23,1-9 entitled “Apparent trends of mean temperature in New Zealand since 1930”. He says that a systematic analysis of all New Zealand climatological stations with sufficient length of record reveals that no important change in annual mean temperature since 1930 has been found at stations not affected by shelter, screen-age, and/or urbanisation. It might seem that J. Salinger using homogenisation techniques was responsible for any other impression..
Pat H. Pat Heenan says:
October 27, 2012 at 8:33 pm
“J.W.D. Hessel…”
Hessell’s paper is quite central to our argument against NIWA. He classes both Albert Park (Auckland) and Kelburn (Wellington) sites as “stations that have been assessed to have increased sheltering from trees during the second period and/or significant urbanisation and/or screen changes”. These sites make up the bulk of the Auckland and Wellington station histories.
On top of that, he specifically mentioned Hokitika and Dunedin as “unworthy of investigation due to periods of unsatisfactory records between 1930 and 1970”. These two stations were also included in NIWA’s 7SS.