Climate bombshell — NZ has not warmed for 19 years

The data say one thing

[CORRECTION 1 Nov 2017 1945 NZDT: The original post used annual data. Figures for the standard error at 19 and 20 years have been removed until I can redo them using monthly data. The trends are unaffected. RT]

[CORRECTION 2 Nov 2017 1605 NZDT: There have been numerous changes to align this post with a Coalition submission to the Royal Society. There are minor changes, references to error margins and to the 20-year chart have been binned and the title changed to “… not warmed for 19 years”. RT]

For the national temperature record, the 7SS, NIWA have collected the data, checked it, adjusted it, approved it and published it on their website, so they can scarcely now argue with it. But, on the other hand, it’s totally at odds with what they say in public. Note to MSM: this ought to be front-page news.

NIWA say another

Here is a representative selection of comments by NIWA and other NZ scientists that echo the story we are familiar with: that New Zealand has been warming. The alarmist meme has always been ‘climate change is already here’.

In the Australasian region our climate is changing. There are long-term trends toward higher air and sea surface temperatures; increased frequency of extreme heat events; fewer events of extreme cold; and changes in rainfall patterns.

from Human Health Impacts of Climate Change for New Zealand (Royal Society of NZ, Oct 2017):

[NIWA climate scientist Petra Pearce] says the climate is warming with New Zealand warming about 1°C since 1909 with more heat waves, fewer frosts, more rain and in the south and west, less in the north and east and rise in sea level of about 1.7mm per year since the 1900s [sic].

from NIWA Climate Change Report for Wellington (Aug 2017):

Niwa forecaster Chris Brandolino said last year’s “exceptional warmth” was the result of three factors. Firstly, “as an island nation we are very susceptible to ocean temperatures – and these were higher than usual”, he said. Secondly, sea pressures were higher to the country’s east, and lower in the south and west. That combination caused more northerly and northwesterly winds to blow across our landmass than usual, Brandolino explained. Northerlies are the warmer winds, as southerlies bring cooler air from the Antarctic along with them. Thirdly, climate change. “New Zealand warmed by almost a degree over the last 100 years due to greenhouse gasses,” Brandolino said. He said warming due to greenhouse gasses looked to be a constant for the foreseeable future, while ocean pressure and temperature were variables difficult to forecast more than three months in advance. Temperatures soared especially high in Northland, Auckland, Bay of Plenty, Hawke’s Bay, Whanganui, Manawatu, Kapiti Coast, Wellington, West Coast, Otago and Southland.

from Stuff (Jan 2017):

Many places around the country were warmer than ever before, in some places by up to 2C. Experts say it is the result of a warming climate and consistent with a warming trend globally. It was caused by a combination of climate change, El Nino, and unusually warm ocean temperatures, Niwa forecaster Chris Brandolino said. An increase in greenhouse gases in the atmosphere had warmed the country by about 1C over a century. “You combine this long term warming trend with some natural variability… It adds up to a record-breaking year, temperature-wise, in New Zealand,” he said.

from Stuff (Jan 2017):

Several ski areas delayed their openings this year due to a lack of snow. Coronet Peak closed 10 days after opening; when it reopens on Saturday, most of its snow will have been artificially produced. Treble Cone, Porters, and both skifields at Mt Ruapehu also delayed their openings due to a snow shortage.

from Stuff (Jul 2016):

Climate change is already redefining coastlines and the weather here in New Zealand and around the world. … But the background trend has been upwards for all of the last century and beyond. … Temperatures today are so much higher than they were then [1909] that even a huge volcanic eruption blocking out sunlight for months would not cool the planet back to early 20th century values. … The ocean is warming, especially the Southern Ocean which is melting the edges of the Antarctic ice sheet, contributing to sea-level rise. As the ocean warms it also expands and this has caused half of the sea-level rise observed during the 20th and 21st centuries. … Powerful climate feedbacks, associated with sea-ice, are already expressing themselves in the Arctic, with Arctic ocean summer sea-ice expected to be all gone by as soon as 2050. … The Antarctic Peninsula has warmed 0.5C per decade for the last 50 years — the fastest-warming place on the planet — and ice shelves are collapsing catastrophically.

from the famous “ten things” road trip by Renwick and Naish (Jul 2016):

Now here’s the truth

For 19 years there’s been no local warming — NIWA’s own figures prove it, contradicting their endless public statements. Why don’t they admit this in public? Why do they tell us a different story? Why do the PM’s Chief Science Advisor, Sir Peter Gluckman, and the previous Commissioner for the Environment, Dr Jan Wright, BOTH talk about warming when it hasn’t been happening? Could they have been deceived by NIWA and the Royal Society? I lump them in together because most of the climate panel on the RS are NIWA scientists. They’ve been in bed together on climate, with the IPCC, for decades.

How did I discover this? A few days ago I picked up the Royal Society report, Human Health Impacts of Climate Change for New Zealand, and the very first sentence (highlighted above) rang alarm bells:

In the Australasian region our climate is changing.

Now, most readers know there has been no significant global warming for about 20 years. They also know that the climate has been changing for millions of years, so on the face of it this is a true statement. However, with climate language twisted completely out of shape by environmental propaganda over the last 30 years, “climate change” now means man-made warming. Beware denials, for they are coming, but when they say ‘changing’ they actually mean ‘warming’.

Christopher Monckton was tracking the lack of warming evident in the RSS dataset, which grew to 18 years and 9 months before the El Nino of 2016 set in with higher temperatures. Since the El Nino ended a few months ago, temperatures have been subsiding, and last month returned to previous levels.

But the RS was talking about Australia and New Zealand, not the globe, so I wondered how local temperatures were tracking. NIWA’s website displays the 7SS and on looking closely it was hard to see a trend one way or the other over about the last 20 years.

So I downloaded the 7SS data and, taking it from 1998 to the present because that’s what the IPCC used in discussing the global “pause” in AR5, tried graphing the last 19 years. Here’s the result:

You can see there’s an insignificant cooling trend of 0.0012°C/year (0.12°C/100 yrs). In other words, no warming and no cooling. It reflects quite accurately the widely acknowledged hiatus in the global mean surface temperature.

New Zealand needs to know this.

22 Thoughts on “Climate bombshell — NZ has not warmed for 19 years

  1. Mike Jowsey on November 1, 2017 at 1:56 pm said:

    Good work RT.

  2. Richard Treadgold on November 1, 2017 at 2:03 pm said:

    Thanks, Mike. I hope people see it, including the RS and NIWA but especially the Herald and a few politicians.

  3. Barry Brill on November 1, 2017 at 2:46 pm said:

    Astonishing!!! The NZHerald has stories every week about the harm being caused to New Zealanders by all the warming they thought had occurred in recent years.

    As regards the 1997-2016 record, is the trend 0.0071 ± 0.38 ? If so, no statistician would even think of claiming there has been any warming at all in that period.

  4. Richard Treadgold on November 1, 2017 at 3:13 pm said:

    Hi Barry,

    Quite right, absolutely nobody is asking about the justification for the anthro global warming scare. Except us.

    And no, the standard error of 0.38 applies to the data. I’ve asked Bob to explain how to find the error margins for the trend. Hopefully we’ll have an answer soon.

  5. If you had chosen 2000 instead of 1998 as your start-point, you would have derived a warming estimate of +2.38°C/century. You can’t meaningfully derive trends from such short time series. Take a look at your R² and p-values. They are not significant. Your standard error is meaningless.

  6. Richard Treadgold on November 1, 2017 at 4:36 pm said:

    Hi Simon,

    Well, you’re basically right. But the start date isn’t cherry-picked. As the narrative says, it simply mirrors Monckton’s presentation of the RSS data, where he went back as far as possible without finding a trend. it needs no apology. The fact that I found a pronounced hiatus for the same period is an amazing coincidence, but it’s true. However, the small number of data points gives a misleading impression, and I’ll be downloading the monthly data in the next few days to see what that looks like. I’ve been on a steep statistical learning curve, so please be patient.

    How would you answer my questions for NIWA and the RS?

  7. Richard Treadgold on November 1, 2017 at 6:06 pm said:


    Bob has helped me understand more about the trend and its error margins. Of the 1998-2016 chart, he says: “The trend (at the 95% confidence level) is 0.0071±0.032 °C/year. In other words, 0.71 ±3.2 °C/century.”

    He goes on to explain that, for the 7SS, which was also annual data, there were 100 data points, so the confidence interval was way down at 0.3 °C/century for both NIWA and our R&S. We should get a similar figure for these charts when we use monthly data.

  8. Barry Brill on November 1, 2017 at 6:30 pm said:

    In WG1 of AR5, the IPCC discussed the “pause” in warming trends. They defined the period as commencing with the 1998 giant El Nino. [Technical Summary, p. 37 and pp. 61–63.]

    Since then, there have been scores of papers and commentaries on this phenomenon. Even Wikipedia has an extensive (slanted) article on it, with no less than 95 references. (Just look at those names! Trenberth, Karl, Rahmsdorf, Foster, Santer, Hansen, etc – where is Mann?) They all talk about a “pause”, “slowdown”, “hiatus” or “standstill” that commenced with the 1998 El Nino.

    Nobody has any doubt that the ‘hiatus’ refers to the period from 1998 to the present. This commencement date was identified by the late Bob Carter when he first discovered the pause back in 2006.

    The IPCC felt that 15 years was not long enough to be wholly conclusive because of the unquantified temperature impacts caused by natural variability. Santer et al later found that climate models might be inaccurate for continuous periods as long as 17 years. I know of nobody who thought a 19-year trend was too short to be interesting.

    The IPCC also felt that –”The exceptionally warm El Niño year of 1998 was an outlier from the continuing temperature trend, and so subsequent annual temperatures gave the appearance of a hiatus.”

    This is fair comment but the criticism could be overcome by taking the global average surface temperature anomaly (GASTA) from one large El Nino in 1998 to an even largerEl Nino in 2016.

    That’s what your graph has done for New Zealand average temperatures. The peaks at the beginning and end equalise out and we are left with a very convincing 19-year trend of zero change.

  9. Gary Kerkin on November 2, 2017 at 10:32 am said:

    Simon, I venture to suggest that the correlation coefficient is not worth considering in this context because no one else is concerned or surprised that it is very low. Putting a trend line through a data series that has much variation can only indicate the, well, err, trend. It cannot be used for prediction. Moreover the correlation coefficient and standard error do not relate to the data in the series: they relate to the accuracy and confidence we can have in the trend line parameters—the slope and the constant. Surely you understand that Simon? Why then would you criticise Richard for using the same techniques, on the same data, as used by NIWA and other “experts”?

    In producing a trend line there are two error components which can be calculated: the standard error of the slope and the standard error of the constant. A little bit of ordinary algebra will indicate that the error of the estimate increases with time. For example if the trend line is t = ay + b where y is years and t is temperature, and if the standard errors of a and b are a’ and b’ then it follows that the estimate of temperature is
    t’ = (a±a’)y + (b±b’) which implies that the error in the estimate of temperature for the series is ±a’y ± b’.

    It is worth noting that the distributions of temperature data series in New Zealand are not normal i.e. a regular, bell-shaped, gaussian curve with one mode (peak) at the mean. A quick survey of the data for all of the 7 stations in the NIWA series shows that all are multimodal. It was suggested to me some time ago that they should be, at least, bimodal because of diurnal variation. I’m not so sure about that—the smoothing of the short-term data to get annual means must wipe out information relating to daily swings in temperature. The last time I looked at the stations data the distributions were at least trimodal. In most cases the width of the bulk of the distributions was quite wide and often they were skewed to one side or the other. That does not, though, imply any warming or cooling over time. Many of those outliers occurred quite early in the 20th century.

  10. Gary Kerkin on November 2, 2017 at 10:41 am said:

    Anthony Watts concurs with you Richard, using HadCRUT data.

  11. Choosing a start-point based upon the absence of trend is a particularly cynical form of cherry-picking. It’s like trying to go down an up escalator.
    Your standard error is an underestimate even at 0.0071±0.032°C/year, because you are not randomly sampling from a normal distribution. Temperatures are a non-stationary time series with auto-correlation and a mean that is increasing over time. You need a longer time series to minimise the bias. Using monthly or daily data will not help you, because short-term temperatures are even more auto-correlated.

  12. Richard Treadgold on November 2, 2017 at 4:45 pm said:

    Simon, sorry, but references to standard error have been expunged from this post. But I strongly object to your calling the topic ‘particularly cynical cherry-picking’. It can only be cherry-picking if I misrepresent the results, and I do not, in fact I tell you exactly what I did and what I found. I followed the IPCC, so are you accusing them, too, of cherry-picking? I simply looked at the same period, and whaddya know — a great chunk of no warming, coincident with activists crying wolf about how we’re marching to oblivion and the destruction of the environment. The strident claims of dangerous warming by self-claimed qualified scientists and their admirers (who frequently took matters far beyond sensible bounds) were, by the facts as now known, quite unjustified. There is simply no defending that. Neither the global nor the local temperatures have increased for 19 years!!

  13. Simon,

    Since 1950 it has only warmed from approximately 1979-1998 – 19 yrs out of the last 67 (28%), and that was 19 yrs ago. To try to claim that it has been warming during the entire time from 1950 or even for the last 19 yrs due to some warming 19 yrs ago is about as disingenuous as it gets. The ‘hiatus’ (IPCC) over the past 19 yrs is at least the same length of time as any actual warming there was during 1979-1998 – if 19 yrs is too short a period to draw any conclusions then that applies to 1979-1998 also.

  14. Gary Kerkin on November 4, 2017 at 10:40 am said:

    Good response, Richard, and I notice that Simon has picked up on my point about the nature of the distribution in a temperature-time series. However I am not convinced about his reference to auto correlation in the temperature series, and I don’t understand his comment about a longer term time series. Is he referring to a longer term of annualised data? Is he saying that monthly or daily data won’t provide the information no matter how long the data goes back? Personally I would prefer to deal with 100 years of hourly information than 100 years of annualised information. Calculating a daily mean from hourly data loses the all important diurnal variation. Calculating an annual mean from from the same data loses the seasonal variations which are equally important. Averaging in any situation will lose information about outliers in the series. Auto correlation methods applied to a long term series of hourly information will reveal the relative importance of diurnal and seasonal variations and coupled with Fourier analysis will identify the dominant frequencies associated with those variations. Please don’t ask me to show you how—it is 40 years or more since I used it. However, modern spreadsheets make the use of such analyses relatively easy.

  15. Mike Jowsey on November 4, 2017 at 12:52 pm said:

    Great point Magoo!

  16. This article gets cited by Ian Wishart in the following piece:

    Royal Society making stuff up on climate, says group

  17. Richard Treadgold on November 5, 2017 at 1:30 pm said:

    Isn’t that helpful? Thanks for the tip, Andy.

  18. Brett Keane on November 5, 2017 at 1:36 pm said:

    It always was retrospective. If the endpoint back then was cooler, the pause would be longer. Waste of time expecting Simon, whom I know of old as a chronic dissembler, to be honest about the proper start point…. Sad sack.

  19. I have estimated the trend to present day for each year in turn, i.e. a cherry picker. There is only one year where the trend to present day is negative, and it is 1998. The error around that estimate is ±1.61ºC/century. Here are the results.

  20. Gary Kerkin on November 6, 2017 at 9:32 am said:

    Not overly helpful Simon. It would have been clearer if you had picked out a few, example, starting points and then plotted the trend lines on a graph of anomalies vs time. When you say “current day” to what are you referring? Is it the the monthly average for October 2017, or is it the annual average to October 2017? Or is it something different again? The important point is that it should be well clear of the 2016 El Niño which will show up in any annual averages to date. Oh no! That would be cherry picking, I hear you say? But aren’t you cherry picking—selecting an end point which suits your arguments.

  21. More traffic from WhaleOil, this time

  22. Richard Treadgold on November 6, 2017 at 10:22 am said:

    Wonderful! And Cameron took it from Ian Wishart. Brilliant. I’ll send a link to the GWPF and Not many people know that.

Leave a Reply

Your email address will not be published. Required fields are marked *

Post Navigation