NIWA’s data confirms: little warming
When it’s calculated correctly
Why did they lie to us?
In December last year, NIWA released their long-awaited review of the NZ temperature record (NZTR). We’ve reviewed that report and found serious errors. NIWA used the wrong method and created strong warming. We used the right method and found mild warming.
There are a few things we need to understand about weather stations. The first is that these stations sit there for a long time. Some of them have been in the same place for 80 years and more. If you sat in one place for that long, you’d see stuff happening around you — same for the weather station.
Trees grow, buildings go up, airport runways get covered in tarseal or concrete, roads appear, and these and other non-climatic influences affect the temperature readings, usually making them warmer, but not always. Sometimes the station gets moved, and it’s always better to keep all that history if you can, so you try to adjust it rather than start again with a new station.
NIWA had to start from scratch
Knowing this, when scientists examine a series of temperature readings they look for what has changed at the different stations. If the changes affected the temperature readings, they adjust the readings.
When we and the Coalition asked NIWA in November 2009 for the adjustments they had made, this was what we were asking for — a list of all the individual changes and the reasons for them. We wanted to independently review them. In the end, they didn’t have that information any more so they had to start from scratch with a brand-new report. That’s what they released just before Christmas and we’ve been reviewing it since then.
Our scientists have now completed their audit and released their findings. They report a number of serious errors in NIWA’s review. They are summarised here, but the details are available in the official Coalition report (see details below). Some of the errors are a bit complex, but the worst of them is very simple: NIWA didn’t do what they claimed.
There is nothing trivial about this failure — it’s not about cosmetics, company image or merely using the wrong typeface — this is a fundamental failure, either of method or communication.
NIWA heavily criticised us for not knowing the methodology used in the original Seven-Station Series (7SS), but now it turns out that they haven’t used it themselves.
There’s plenty wrong with the new NZTR
NIWA stated repeatedly before they undertook the reconstruction that they would use the adjustment methodology specified in Rhoades and Salinger, 1993 (R&S), and Wayne Mapp, their Minister, gave the same assurance in answer to Parliamentary questions. NIWA’s paper itself reinforces the point by making reference to R&S throughout its 169 pages.
But the mathematics is unmistakable: NIWA did not correctly use the R&S method to make the adjustments. Instead, they went way outside the requirements of R&S, changing the method in ways that strongly changed its results. Their changes created a trend that overstates the New Zealand warming over the past hundred years by 168%.
Other mistakes
The R&S method calls for comparisons to be made only between “neighbouring” stations, but in another mistake NIWA uses Dunedin and Dargaville to fill gaps in the data at Albert Park, though they are more than 1100 km apart across several mountain ranges and lie on opposite coasts.
Then, NIWA wrongly uses stations known to be affected by the Urban Heat Island (UHI) effect, which introduces a spurious warming, and makes no adjustments to remove it.
Further errors
- Monthly data – R&S clearly requires the use of monthly data, but NIWA uses annual data.
- Symmetric interval centred on the shift – R&S specifies data from a symmetrical period, but NIWA sometimes uses asymmetrical periods.
- A 1-2 year period before and after the shift – Where R&S requires just one or two years, NIWA uses variable periods of up to 11 years. This can introduce unwanted climatic factors.
- Weighted averages based on correlations with neighbouring stations – NIWA ignores the R&S requirement to weight the averages according to the correlation with neighbouring stations.
- Adjustments only performed if results are significant at the 95% confidence level – NIWA seems to perform all adjustments — there’s no evidence of any significance tests.
Sequence of events
Late in November 2009, the Climate Conversation Group helped to produce, in collaboration with the NZ Climate Science Coalition, Are we feeling warmer yet? a report that some called Niwagate.
The report examined the database of temperature records on NIWA’s web site and asked the reasons for unannounced adjustments we found that NIWA had made to the raw temperature readings from around New Zealand.
It soon became apparent that NIWA were unable to answer our questions; they had lost records, or they never kept records; they delayed, they blustered, they criticised the Coalition, they obfuscated and, most egregiously of all, they pointed us to non-existent references to the scientific methodology which took several of our scientists (volunteers all) many hours to discover were useless.
We just wanted to know why they adjusted the temperature records. The way they squirmed, you would have thought it was a matter of state security. In the end, it transpired they no longer had the information. Why didn’t they just say so?
After continued questions in the Parliament from the ACT party, in February 2010 NIWA provided an “example” set of adjustments to the Hokitika records. A week later they announced they would “reconstruct” the entire New Zealand temperature record. It would take several months and cost about $70,000. This was a victory for the persistence of the Coalition and for all New Zealanders who thought that so simple a thing as the public temperature record should be properly scientific.
A public temperature record beyond dispute?
Do we have a NZTR which is finally beyond dispute? No, we’ve found too much wrong with it. The new disputes are just beginning. This NZTR is excruciatingly awful. If there’s any whisper that this unscientific fiasco has influenced the wider picture in the Pacific or even the globe, we’ll be demanding it be corrected.
There’s a strong possibility that the NZTR has been used across a wide swathe of the Pacific Ocean, falsely inflating temperatures for a large region.
Everything depends now on what NIWA does next. Will they admit to their mistakes? Will they make things right? Will they, like last time, underestimate their opponents and try to fight their way out?
Or will they confess to prefer being corrected than continuing in error? They are an organisation which does a whole lot of good things in New Zealand and it would be a pity to see their reputation soiled by a piece of inept or unethical work.
The Coalition asks for the suspension of the new NZTR until it can be corrected. I would go further and expect whoever made the misguided decisions that resulted in the faulty NZTR to resign from NIWA.
Who lied to NZ public?
Who decided to make substantive changes to the R&S method and lie about it to the New Zealand public? That person should not have a senior role in a reputable organisation.
What now for NIWA’s supporters?
NIWA has enjoyed tremendous support in our dispute with them. Much has been written about it and much obloquy heaped on my and my Coalition colleagues’ heads. This scientific review of NIWA’s report provides justification for our initial scepticism and our perseverance. Seeing what this audit reveals, it’s hard to find reasons to support NIWA.
At first, we hinted strongly at malpractice, to goad the ever-languid NIWA scientists into releasing the data they had withheld for many years. Our goading eventually succeeded in prompting the reconstruction of the NZTR.
Examining the results of that work has uncovered stronger evidence of unscientific and even possibly unethical behaviour than we ever imagined. We certainly never hoped for this. To clarify: this does not please us. We never wanted to find evidence of dishonesty, never imagined that we might and we are disappointed to discover that they failed to do what they said.
Where to get the documents
NIWA released their review document, containing an overview and reports for each individual station in the New Zealand Temperature Record (NZTR), just before Christmas last year. It had been peer-reviewed by the Australian Bureau of Meteorology. You can download those documents as pdf files from the NIWA site. Here’s the overview document.
You can download the official NZCSC audit of NIWA’s review, Statistical Audit of the NIWA 7-Station Review (pdf, 2.7 MB).
There’s a sizeable Supplementary Information document, Statistical Audit of the NIWA 7-Station Review SI (pdf, 6.2 MB), to go with it. It contains the material for the other six stations in the NZTR.
One of the principal messages to be taken from this audit of NIWA’s new NZTR is that New Zealand has seen no temperature rise from global warming so far, and we can expect none in the foreseeable future. A seminal paper from the Coalition’s Chairman, Barry Brill, entitled New Zealand Unaffected by Global Warming, describes the research and its conclusions — handy for students (pdf, 1.3MB).
Views: 85
So, you now acknowledge that adjustments to the raw temperature data, including those associated with elevation, are entirely necessary in order to construct an accurate temperature record. Further, you support the published methodology of Jim Salinger. Excellent progress Richard. How’s the legal claim going?
Mike, I don’t recall Richard or the NZCSC ever saying that corrections weren’t needed, only that they should be stated plainly for the sake of transparency and peer review. I recall that the corrections to the Wellington site were criticized for issues but that was not a criticism of the S&R methodology, rather it was a criticism of the failure to adhere to the methodology.
The same is true here; it’s not the S&R methodology that is being criticized but the failure to apply the correction methodology properly.
Mike, what Huub wrote is spot on.
Nowhere in the NIWA report did they in fact use elevation. That was a red herring completely, and they were forced to admit that early on. I take it you have read their report?
When have we said we did not? The method of Rhoades & Salinger is an internationally-accepted technique (see Aguilar 2003). It also appears to work quite well, when it’s actually used. Why would we not accept it?
Mike P,
Don’t listen to Perrott or Renowden on these matters, because even though I’ve told them numerous times, they insist on not knowing the truth and spreading it around. As others have just told you, I’ve never said adjustments were not necessary — you can read that paper yourself if you haven’t already — it’s linked in the post, too. The altitude thing was strictly a red herring NIWA raised in their first press release, I think, to sound knowledgeable. But Wellington wasn’t altered with any altitude calculations. In the 7SS, no corrections were calculated on the basis of altitude — and in Hokitika, the higher station was actually warmer than the lower! Although altitude changes must have been a factor in some station moves, the corrections were all made by comparing temperatures.
Your comments give off a distinct and unwelcome odour of sarcasm, which I am ignoring, but I wonder if your previous courtesy is forgotten or you dropped it as an unaccustomed strain?
The only Salinger material I’ve criticised is the so-called “methodology” in his thesis. We tried for weeks even to locate a copy, which is hidden away in the library at VUW. Then our scientists tried to understand it. It’s completely unsatisfactory as an adjustment method because nobody can replicate it, which is why it’s never been published, used or acknowledged.
When Jim got together with Rhoades, a statistician, it made all the difference, and they put together a sensible, robust method that anyone can follow. Except NIWA, it seems.
The legal claim? A hearing has been delayed until next year, I don’t know why. NIWA have been cooperative in giving us access to the documentation we asked for, except what’s related to the BoM’s peer review.
Regarding this jibe: “Further, you support the published methodology of Jim Salinger.”
I always did. His only published method is in R&S, which I’ve never disagreed with. I emphasise that the incomprehensible method in his thesis is still unpublished. Did you know that?
Typo, Richard. It was Hokitika where the higher station was warmer.
Ha! Thanks, Bob. Changed.
Richard. OT but I didn’t know where to post this. You should do a post about Prof Murry Salby’s observations and the conclusion that it is temperature change that controls atmospheric CO2. Not the reverse. He is the chair of Climate Science at Macquarie University.
see http://joannenova.com.au/2011/08/blockbuster-planetary-temperature-controls-co2-levels-not-humans/#more-16345
This could be big.
Thanks, John. That looks most interesting. I’ll see what I can do.
Mike Palin completely misses the point and builds a couple of straw men to divert attention. The point, Mike, is that NIWA lied. Period. Suck it up Mike. NIWA lied. They said they would stick to a particular method and they did not.
Perhaps the Minister has some ‘splainin’ to do.
Thanks for this Richard; I’ve linked to Jo Nova’s open thread and Warwick Hughes blog so you might get some Ozzie comments
I fully expected the Auckland, Nelson and Hokitika results from my own clumsy reconstructions but I have to say I’m surprised by the results of the other four. It looks like there’s a whole lotta reading to be done.
I’d walk over broken glass to see a polynomial trend fitted to the R&S composite series too.
Good work guys, does this mean that “unequivocal” warming has been downgraded to “equivocal”?
Here’s a 3rd order poly for you.
Ooooo yeah, nice Bob D.
Doesn’t include the 13.1 C 2010 figure that I suspect would flatten the R&S curve a little but it paints a whole new picture (or should I say cycle).
Two questions Bob. First, what is the uncertainty envelop on your curves. Second, what is the physical basis for fitting a 3rd-order polynomial to the data?
What is the physical basis for fitting a linear trend to the data?
Mike – according to the Coalition paper, the uncertainty envelope for the “corrected” 7SS curve is ±0.29°C/century. This is similar to the error bars shown by NIWA for the NZT7 curve contained in their Review.
A big difference between NIWA and the Coalition, is that the latter tested each adjustment for uncertainty in the manner laid out in Rhoades & Salinger. Where the adjustment failed to attain statistical significance levels (at 95% confidence) it was not adopted. NIWA fully agreed with the principle of testing confidence levels for each adjustment, but said in the text of the Review that this work was not complete. Well, 8 months have since gone by, but nothing further has been published.
None whatsoever. I simply responded to Richard C’s request.
Ok, but he asked for a polynomial of unspecified order. Why did you choose a 3rd-order over, say, a 1st-order?
Because a first-order polynomial is a straight line. A second order is quadratic, which isn’t going to tell anyone very much (not being particularly cyclic), and 3rd order was next.
So the 3rd-order polynomial is a better fit to the data than a straight line. If you didn’t calculate uncertainties, then how did you determine this? I would guess that you assumed a cyclic pattern and calculated the simplest cyclic curve fit. Not very scientific.
Oh for heaven’s sake! Nobody said it was a better fit. I know Richard C has been interested in cyclic effects, so I plotted the lowest order polynomial fit that would show this. Nothing more, nothing less. Get over it – I did it for him, not for you.
Mike P, so you think applying a polynomial is unscientific. You should read Scafetta 2010 and Loehle & Scafetta 2011 in that case. The underlying best fit trend for HadCRUT3 is a quadratic, what then, will replicate the cyclic component better than a higher order polynomial?
But to a degree I agree with you, fitting a linear or polynomial trend is unscientific (they’re external), but it is an indicator and and to not look at other trends (moving average, Fourier etc) is negligent.
I could have waited to get access to the NZCSC data (haven’t got that far yet) and done an analysis myself by looking at a range of trends which would include an EMD trend (internal, inherant) which I’m sure you will agree is perhaps a little more scientific than a linear or polynomial trend. The cycle revealed however, will not be a great deal different than the quick and easy 3rd order polynomial provided.
The linear trend over such a short time scale is bunk though, I’ll agree, isn’t it Mike? Or is it the gold standard in warmer-world.
Any curve fit to data in the absence of a physical basis is nothing more than a mathematical description. It has no predictive capability outside the data range and certainly is not indicative of a process. I think we agree on this.
Who said we were using it in a predictive manner?
Also, please then explain the physical basis for a linear trend. NIWA used it in their original graph, remember. Methinks you protest too much, perhaps to deflect attention away from our findings that NIWA didn’t use R&S correctly.
Here’s “a physical basis” for you Mike:-
An alternative theory of global temperature dynamics
Accumulation Theory of Solar Influence
http://landshape.org/enm/accumulation-theory-of-solar-influence/
Bob – what’s the significance of the fact that both the NIWA and the R7S curve are now heading downwards? Doesn’t NZ temperature know that human-caused CO2 is being added to the atmosphere at record rates?
🙂 No, nobody told it. However, as Richard C pointed out, this graph ends in 2009, and the 2010 mean was on the high side high, so one would expect the end to flatten somewhat.
“Equivocal”? Ha!
Try “deceptive”!!
When the Minister announced that NIWA would receive special funding to undertake its Review of the NZ temperature record, he stated that the methods used in the Review would be the subject of a scientific paper, independently peer-reviewed and published in a learned journal during the 2010-11 fiscal year.
The Review was published in December 2010, but the Minister admitted in March 2011 that the “methods” paper hadn’t been written, and he wasn’t sure when it would be. This week, in answer to a further Parliamentary Question, the Minister said he didn’t expect the paper to be finished until the end of the 2011-12 fiscal year, and it might not be published until well into 2013.
In light of the Coalition’s findings, one has to wonder whether this long-awaited paper has any chance of ever getting past an independent peer-review. Until it does, the authority of Rhoades & Salinger reigns unquestioned – as it has for the past 18 years.
Regarding Salinger:
http://www.times-age.co.nz/local/news/masterton-hosts-eco-warriors/3961947/
He (Salinger) said the effects of climate change can be seen by the changing weather worldwide, such as a European heat wave.
“Extremes like this are expected to become more frequent,” said Dr Salinger.
He is not a scientist. Scientists don’t lie.
Pingback: Climate Change and Science | New Zealand Climate Change
Pingback: How NIWA added lots of warming in New Zealand – and got away with it – so far « Tallbloke's Talkshop