Roger Andrews: So How Much Warming Was There In New Zealand Anyway?

Posted: November 1, 2012 by tallbloke in Analysis, Dataset, Incompetence, Legal, methodology, Politics, propaganda, Surfacestation

So How Much Warming Was There In New Zealand Anyway?
by Roger Andrews

While discussions in the recent NIWA post concentrated on the legal aspects of the NIWA vs. New Zealand Climate Science Coalition (NZCSC for short) proceedings I was looking into this question. Was it indeed as much as 0.91C/century, as NIWA claims?

Or only 0.34C/century, as NZCSC claims?

Or was it somewhere in between the two?

To investigate the issue I constructed two surface air temperature series from scratch using twelve New Zealand GHCN v2 records. These series act as robust checks on the NZCSC and NIWA series for the following reasons:

The data sets are different. The raw data used by NZCSC and NIWA come from over 30 separate records that NZCSC and NIWA composite into single records at seven stations. But only five of these stations (Auckland, Wellington, Hokitika, Dunedin and Nelson) are listed in GHCN and therefore included in the twelve-station GHCN v2 data set I use, and seven of the GHCN stations I use (Christchurch, Invercargill, New Plymouth, Kaitaia, Gisborne, Napier and Whenuapai) are omitted from the NIWA data set. As a result there is only an approximate one-third overlap between the two data sets and the data are not always the same where they do overlap.

The analytical approaches are different: NZCSC and NIWA apply stair-step adjustments to the 30 separate  NIWA records – NZCSC using the Rhoades-Salinger method and NIWA a method apparently of its own devising – before averaging the seven composite records to obtain time series. The twelve GHCN v2 records I use are already merged (by GHCN), and other than reduction to a common baseline and rejection of bad or suspect data I apply no adjustments to them.

GHCN v2 lists a total of eighteen records in New Zealand. Five of them (Auckland, Albert Park, Auckland Albert Park, Christchurch New Zealand and Dunedin Aerodrome) are too short to be of any use and the two records at Whenuapai can be combined into one, leaving twelve records containing a total of 846 years of readings since 1881, and these are the records I used. The locations of the corresponding twelve stations are shown below:

I first constructed a baseline-adjusted series from the twelve GHCN v2 records with temperatures expressed as anomalies relative to 1950-80 averages. Before constructing it I deleted the following bad or suspect record segments:

  • Wellington before 1929 (large stair-step discontinuity)
  • Hokitika before 1913 (large stair-step discontinuity)
  • Dunedin before 1914 (record implausible)
  • Gisborne after 1995 (anomalous warming)

Here is how the twelve records compare after the deletions:

The generally good overall match indicates that none of the records is seriously distorted over the long-term by artificial discontinuities, UHI gradients etc. and it also shows that the records don’t need homogenization (they are in fact already more homogeneous than the GISS “homogeneity-adjusted” versions). I therefore used the annual averages (red line, two-station minimum) as my baseline-adjusted series.

Next I constructed a first-difference series using the same GHCN v2 data. The graph below compares this series with the baseline-adjusted series. There are only minor differences:

So how do these two series compare with the NIWA and NZCSC series? The next graph plots all four of them together between 1909 and 2009, the period covered by the NIWA and NZCSC series –

– and it’s difficult to see what’s going on because the short-term temperature fluctuations obscure the long-term trends. So at the risk of incurring the wrath of statistical purists here are the data replotted as ten-year running means instead of annual means:

Now we can see that the NZCSC, baseline-adjusted and first-difference series show a respectable match but that the NIWA series is way out in left field. This result effectively validates the NZCSC series (within limits) and falsifies the NIWA series.

Having established that we can now address the question of how much warming there has been in New Zealand over the 100-year period between 1909 and 2009. Making an exact estimate is problematic because of the large short-term fluctuations in annual mean temperatures, but we can come up with an approximate estimate based on regression line gradients, which for the ten-year means plotted on the graph above are NZCSC = 0.45C, baseline-adjusted = 0.52C and first-difference = 0.49C, and for the annual means shown on the graph before that are NZCSC = 0.34C (the value quoted by NZCSC in its report), baseline-adjusted = 0.44C and first-difference = 0.43C. The mean of these values is 0.45C and the standard deviation is 0.06C, so I’m going to say that 0.45 +/- 0.12C of total warming over the last 100 years is as good an estimate for New Zealand as we are going to get.

The NIWA series shows about twice as much warming as the other three series and it’s now obvious that the extra warming is manufactured. And to see how it’s manufactured we need only look at the warming adjustments NIWA applies to its raw records (all estimated to the nearest 0.01C, incidentally):

An argument can in fact be made that if adjustments this large are needed to make the raw records “correct” then the raw records were far too heavily distorted to have been used to begin with.

  1. Ray Tomes says:

    Rog, I am very interested in this whole subject, especially in regard to NZ where I live. The most telling thing to me is the last graph. the average adjustment to raw data is close to 0.7 C which explains the entire extra warming that NIWA comes up with. I find it hard to believe that all the site changes etc work in the one direction like this. Also of course no allowance is made for Urban Heat island which will affect all the sites to some extent and some to a large extent. I doubt that any real warming is happening at all. Regards, Ray

  2. Roger Andrews says:


    I don’t think UHI impacts are significant. The Auckland record, where we might expect to see the largest UHI effect, doesn’t show an obvious urban warming gradient. The NZ surface records show no more warming than the surrounding island records (Lord Howe, Norfolk, Raoul, Chatham, Campbell, Macquarie) and in fact show less warming than UAH TLT.

  3. John Snow says:

    Roger: Congratulations on a very succinct and clear analysis. Once again we see a few things that are common to the AGW discussions.
    1. There is some warming but it is fairly small. We have had much greater warming since the end of the previous ice age. Therefore it seems that even if we attribute most of this warming to manmade processes it is still miniscule. Certainly not worthy of ruining the world economy to limit carbon emissions.
    2. The lines are straight. Extrapolation easy. No hockey stick! Infact the trend for the last ten years is flat.
    3. Those that want to show a significant and sudden uptick in the warming profile have to fudge (excuse me, adjust) the data to get the answer they want.
    4. The amount of warming – or climate change – is within that which can be explained by natural phenomena such as solar activty.

    A prediction: Evidence such as this very nice accomplished piece of work will be dismissed and derided.

  4. Roger T, a point you probably realised is that all the sites you use are on the coast and influenced by sea temperature. Of the seven NIWA sites in the last graph only Masterton could be catorised as country (being 100km north east of Wellington). Lincoln is in the country but only 22km south of Christchurch and close to the coast. Do not know NZ very well but when I was there on visits I found Wellington, Nelson, Hokitika, Invercargill, and Dunedin were getting cold winds from the cold ocean- glad it was summer, I sure would not want to be there in winter except maybe the ski fields of Queenstown.

  5. Roger Andrews says:

    John Snow:

    Thank you for your kind words.

    One of the intriguing things about the NZ warming is that everyone – including NIWA – agrees that most of it occurred between 1950 and 1960, which certainly isn’t compatible with CO2 forcing. In fact I’m not sure what it is compatible with. None of the ocean indices did anything in particular over this period and it doesn’t seem to have had anything to do with solar activity either.

  6. Roger Andrews says:


    Right, they’re all coastal stations, but nowhere in NZ is very far from the sea (Masterton is only 40km inland). And if the cooling sea breezes blow all the time there shouldn’t be any impact on long-term temperature trends.

    According to the data I’ve been able to dig up the ocean around NZ is also about 3C warmer than the land, although I guess that doesn’t matter because cooling sea breezes work by evaporation 🙂

  7. tchannon says:

    keep in mind the 0.4C step circa 1951I have pointed at but been generally dismissed.

    How many of the stations is there genuine instrument readings?

    You are doing good stuff indeed.

  8. tallbloke says:

    Roger Andrews says:
    November 2, 2012 at 3:04 am
    John Snow:

    Thank you for your kind words.

    One of the intriguing things about the NZ warming is that everyone – including NIWA – agrees that most of it occurred between 1950 and 1960, which certainly isn’t compatible with CO2 forcing. In fact I’m not sure what it is compatible with. None of the ocean indices did anything in particular over this period and it doesn’t seem to have had anything to do with solar activity either.

    It seems to be a feature of the Pacific ocean that it is pretty efficient at shifting absorbed heat elsewhere (See Bob Tisdale’s graphs). So the ‘unprecedented’ strength of solar cycle 19 may have had a more noticeable effect on land than ocean. The recent cloud studies show a drop in cloud cover from mid C20th. That would affect land surface temperatures.

  9. Roger Andrews says:


    The NZ surface warming is entirely compatible with heat releases from the oceans around NZ:


    “How many of the stations is there genuine instrument readings?” I’m not sure what you mean by genuine, but if you mean read off a thermometer as opposed to made up in a pub I would guess all of them.

    And you do good stuff too 🙂

  10. tallbloke says:

    Roger Andrews says:
    “In fact I’m not sure what it is compatible with. None of the ocean indices did anything in particular over this period ”

    Then Roger Andrews says:
    “The NZ surface warming is entirely compatible with heat releases from the oceans around NZ”

    Make your mind up mate!

  11. Roger Andrews says:

    Hah! I present you with a major breakthrough and you fail to recognize it.

    What the graph shows (and I hadn’t looked at SST when I wrote the post, hence my uncertainty) is that the surface warming in NZ – however much it was – was likely caused mostly if not entirely by heat released from the surrounding ocean. And if this is the case then the culprit is surely the sun, which put the heat in the oceans in the first place during solar cycles 17, 18 and 19. 🙂

  12. tallbloke says:

    Well, John Christy emailed me couple of SST datasets for the lat and longs either side of NZ and the Hadsst2 shows a 0.95C warming over the C20th. Reynolds doesn’t. I suspect your ICOADS doesn’t either. So we have another issue to resolve.

    Perhaps you could drop a comment on the NZcsc site to invite them over to give us their insight.

  13. Roger Andrews says:

    TB: It doesn’t make much difference which NZ SST series you use.

    I tried dropping a comment on the NZCSC site but it hasn’t showed up yet. Either it’s being moderated or it went directly to the great hard drive in the sky. Maybe you could give it a try. I assume they know you over there.

  14. tallbloke says:

    Hmm, I had the same problem with trying to place a comment. Strange. I couldn’t get a comment onto Jo Nova’s site either. I’ve emailed Bob D.

    What are the lats and longs of the ocean areas you used?

  15. Roger Andrews says:

    38 to 48S, 165 to 179E.

  16. Roger Andrews says:

    OK, I got it.

    Which series is this?

    Incidentally, you just won the award for the longest post ever on a climate blog. I measured it at fourteen feet nine inches. 🙂

  17. Roger Andrews says:

    I do you unjust. On remeasurement I get fifty five feet four inches.

  18. Doug Proctor says:

    The graphs you provide are simple and simply read, and intuitively they make you wonder how such a progressive, unidirectional correction could be realistic. They remind me of Watts’ graphs of GHCN corrections; again, only the non-intuitive would think the pattern of corrections over time didn’t look funny. But the warmist community do not think these things look odd.

    The gorilla in the room is not invisible, from what I can see, but no one outside of the skeptics is commenting on the gorilla being present. This is what I find peculiar. The dead-pan silence. So many of the warmist’s “observations” or facts are clearly disputable, but no one seems able to bring the subjects to a point of open review. The passive-aggressive response by the warmist mob to complaint by the skeptics is truly amazing.

    The Australian BOM reviewed the NIWA work prior to the ‘suit. The BOM report was supposed to be released at trial but NIWA refused. The judge did not decide that this was some sort of contempt, but decided that NIWA didn’t have to justify what they did/how they did to anyone but themselves. He also ruled that the Coalition did not have competent witnesses, so there was no one capable of interpreting the data and report outside of NIWA at trial, so he had to take NIWA’s word that the BOM said their work was fine. The judge’s decisions look like a simple extension of a very clear non-response response.

    There have been various bizarre beliefs promoted widely and accepted by otherwise practical and smart people over the centuries. Even Harold Camping, who said the world was going to end (twice) in 2011 got a lot of followers before the end did not come. With cults, we explain away the apparent obvious disconnect between belief and reality as a function of brainwashing, groupthink and the like. But the success of CAGW narrative worldwide, like the former narrative of witchcraft, makes me think more of a viral infection. More of a pathology than a philosophy.

    Hysterical blindness, the Freudians used to say. Anorexics look in the mirror and see a fat person. Is it possible that the pathologization of climate “denial” several warmists are promoting is true – in the sense that all of us are pathologized in our beliefs if the social environment is supportive enough?

    I’ve read that the Danes in Greenland starved when the climate changed and beef and sheep did not survive (evidenced by middens, essentially their garbage dumps). They refused to eat fish. I’ve heard the same about the Newfoundlanders of 100 years ago or more who refused to eat lobster: one term I heard when visiting, was that lobster was the poor man’s turkey. Better to go hungry than be seen as poor. You would think that both cases were examples of pathologies – at least a pragmatic man would think so.

    I work with professionals and executives in the petroleum and financial industries who believe amazingly silly things that routinely are proven false by insolvency and bankruptcy. Is the problem at the genetic level, that Man (not just Mann) is physically incapable of accepting information that runs counter to certain levels of social and personal need?

    Perhaps hysteria and its attendant blindness is not about shrieking and rushing about in a panic. Maybe it is about a mental breakdown of critical thinking. Something that can happen to anyone succeptible, be they a streetsweeper or a peer-reviewed climate-based holder of a certificate explaining that the holder did something among the crowd whose joint effect was to impress the Nobel prize committee enough to hand a prize to the representatives of the group, neither of which did any work on the subject at hand.

  19. tallbloke says:

    Roger A: Did you get both of them? There were two separate comments.

  20. tchannon says:

    “In 1952, Hokitika’s airport was moved from the Southside Aerodrome on the south bank of the
    Hokitika River to its present site on the flat terrace overlooking the town.”

  21. tchannon says:

    This might be the Hokitika Aerodrome met enclosure

    Google images are very different 2006 and 2009. I’ve pinned what might be a double Stevenson screen 2009, is an AWS. Aligned east/west
    Previous image 2006 might have a single but if so is not aligned.

    How many moves? (kt x q chk)

    Just about all these stations have gone AWS and probably still kept changing. Before that needed daily servicing at least. Be various moves.

    In every case full bridging data is needed. Betcha it does not exist.

  22. Roger A, good graph with SST. I had thought earlier to mention winds. Your latitude 38 to 48S is mostly in the “roaring 40’s wind band. Further, NZ is effected by the cold Antarctic Circumpolar current. The capital Wellington is known as “Windy” Wellington. The prevailing winds from the west and south west are cold. I doubt your point that the oceans are 3C higher than the land. I would suggest that the ocean temperature on the north east coast of the North Island is at least 10C higher than the ocean temperature around the South Island. Talking about an average temperature of oceans makes no sense when there are warm and cold currents. In the ocean off the Australian east coast there is a warm current that gets down to about Jervis Bay in summer (I have seen coral growing there) latitude 35S. In winter that current ends at around the Gold Coast -latitude 28S. I have not been to the Bay of Islands (latitude 36S) on the north east coast of NZ (maybe next trip) but I have heard the water is warm and the weather pleasant.

  23. Roger Andrews says:

    TB: Yup. Got them both

    What are they please?

  24. tallbloke says:

    Roger A: Hadcrut4 according to John.

  25. Roger Andrews says:

    Tim C: One way of identifying shifts resulting from station moves etc. is to subtract the mean of all the records in the area from the record. Here’s the plot for Hokitika. The shift caused by the 1911-1912 station move is obvious but all we see after that is a bulge in the 1950s that may or may not be related to the station moves in the mid-1960s.

    The NIWA corrections are added for comparison purposes. To homogenize the Hokitika record to the NZ mean they should broadly track the NZ mean-minus-Hokitika line and the shape is indeed about right. However, the total warming correction is about 0.5C too high.

    cementafriend: Between 1950 and 1980 mean sea surface temperature around NZ was14.5C and mean surface air temperature at NZ coastal stations was 13.0C, a difference of 1.5C. There aren’t enough SST records around NZ to go into much more detail than that.

    TB: I don’t know what your and John Christy’s plans were but HadCRUT4 can be downloaded from KNMI if you want it. But with mostly ocean blocks in the area I suspect it would look pretty much the same as HadSST2 or 3.

  26. Roger A, I think your figures for sea temperature are high see this Have not done the sums but I would guess that the long term average temperature is in the range 13 to 13.5C. I note the sea temperature difference between Balclutha (on southern part of South Island -been there and the water was freezing) and Bay of Islands north east of the North Islands is 8C -a little less than my estimate.of 10C but it is not summer yet, maximum sea temperatures (and air temperatures) off north east coast of Australia occur in February.

  27. I should add my guess would be based on land area or straight line coast rather than an arithmetic average of measurement points which seems to be the case for the average on that web page. The South Island is bigger in area. I am very surprised at the relatively high temperature in Milford Sound but maybe that is because of still water not affected by currents. I now recall that when I was on a boat trip there that some crazies went swimming and they were told not to dive down as the temperature about 1m down was at least 5C lower and they could have a heart attack (most did not last more than 5 minutes in the water)..Also, there was a noticeable difference in air and seaspray temperature when the boat went out into the open sea. Maybe the same applies to most of the measurement points (eg in the Greymouth and Golden Bay harbours). I feel that the average temperatures at Balclutha, Dunedin and Milton (around 10.5C) are more typical. of the whole of the South Island.

  28. […] Andrews has investigated the warming in New Zealand over the last 100 years and is published at Tall Bloke. He happily confirms the NZCSC audit of NIWA’s […]

  29. Roger,

    Congratulations on your painstaking research. Your conclusion in your graph comparing the upward trends for NIWA (1degC per century) with the trends for NZCSC, Baseline Adjusted, and First Difference data (0.34 to 0.5degC per century) is very clear: NIWA have obviously developed a “method” that does not work, either through incompetence, or negligence, or worse.

    Now warmists might conclude that maybe NIWA has got it right and that NZCSC has got it wrong. But the killer argument in favour of NZCSC’s figure is that it conforms almost exactly to the 0.4degC per century trend in the University of East Anglia’s HadCRUT3 world mean temperature data series since 1850. See:

    It irks me that even hard line skeptics often write stuff like “warming has only been 0.8degC per century” when even the clear long-term trend from the University of East Anglia (that fount of non-alarmist data accuracy, not!) is only 0.4degC per century.

    So congratulations on publicising yet another example confirming that the true long term world temperature trend is likely well under half a degree C per century, which of course puts it fairly and squarely within the bounds of natural variability.

    And I wouldn’t be surprised as time goes by if even that figure turns out to be too high. For example, the Central England temperature since 1660 shows a long term trend of just 0.26degC per century. See:

    Only regional data, of course, but when data from many regions all shows the same un-alarming trend it can slowly add up to a convincing conclusion.

  30. Barry B says:


    Great analysis. Interesting that the whole century’s warming occurred in a short burst during the 1950s.

    You say: “I don’t think UHI impacts are significant. The Auckland record, where we might expect to see the largest UHI effect, doesn’t show an obvious urban warming gradient.”

    Albert Park, which makes up the Auckland record 1909-76, was subject to continually-changing exposures as a result of tree growth. The dramatic sheltering effects are shown by photos at

    Hessell (1980) found that the sheltering disguised the impact of UHI, when the urban population increased by 60% between 1936 and 1966.

    10 of the 12 stations used are cities (Hokitika and Kaitaia being the exceptions) and it seems highly unlikely that they would all be immune to UHI.

  31. Roger Andrews says:


    Thanks for the info on Auckland. That may well explain why Auckland shows no UHI gradient.

    But on the other hand, it may not.

    I’ve looked for UHI impacts in thousands of temperature records from all over the world, and – well, let me tell you a few of the things I’ve found and you can tell me what’s going on.

    Urban areas in the southwest USA (Phoenix, Tucson, LA, Las Vegas) all show strong UHI gradients but few cities anywhere else in the USA do.

    Major cities in South America such as Buenos Aires, Rio and Sao Paulo show strong UHI gradients but other major cities, such as Lima, Santiago, Cali and Bogota don’t.

    Sydney, Melbourne, Brisbane and Perth show UHI gradients but Adelaide, Auckland and Cape Town don’t.

    Of all the cities in Asia only two (Sverdlovsk and Lanzhou, China) show obvious UHI gradients.

    Of all the cities in Europe none shows an obvious UHI gradient.

    Every town in Japan with a population of greater than a few thousand shows a UHI gradient.

    Rural stations often show stronger warming than urban stations.

    Usually, however, there’s nothing to choose between urban and rural stations. Here’s an example from Europe (the warming gradients are effectively identical).

  32. Richard C (NZ) says:

    Nice work. Especially good to see GHCN used this way instead of how BEST adulterates GHCN around NZ in the early years.

    BTW, keep an eye out at Niche Modeling over the coming weeks. David Stockwell informs me that he is constructing a series too although not specifically NZ I don’t think (perhaps it’s Australasia). It will be similar to BEST but different methodology (don’t ask me, wait for David’s write up). That will probably be his input rather than enter the NZCSET v NIWA fray.

  33. Barry B says:


    That’s astonishing! Obviously there are other factors at work that disguise the UHI gradient in many instances – or possibly create a false UHI gradient in others.

    In regard to New Zealand, have a look at Hessell (1980), a peer-reviewed paper available from At page 6 Table 6, he compares nine city sites with nearby rural counterparts finding that the city temps are higher – by an average 0.57°C – in every case but one.

    In Auckland, ten years of Albert Park data were compared with Auckland Airport at Mangere. The mean of the former was found to be warmer by 0.6°C.

  34. Roger Andrews says:

    Richard C:

    BEST actually sticks a knife into NIWA too, although it doesn’t twist it:


    The late John Daly found a similar vegetation problem at Low Head, Tasmania.

    And because of it Low Head – supposedly a rural station – shows as much warming as Melbourne.

    I’m beginning to suspect that temperatures are affected much more strongly by conditions in the immediate vicinity of the station than they are by UHI impacts.

  35. tchannon says:

    The Low Head instance is very important. I go further than Daly, the effect also prevents free radiation.

  36. Roger Andrews says:

    Whatever the effect is, it’s stronger than urban warming:

    Warming gradients 1895-1992:

    SE Australia: 0.4C
    Melbourne: 1.0C
    Low Head: 1.3C

  37. Roger you mentioned little UHI for Melbourne how about this “Quantification of the Influences of Wind and Cloud on the Nocturnal Urban Heat Island of a Large City” by C. J. G. MORRIS AND I. SIMMONDS in Feb 2001 American Meteorological Society. In the concluding marks “However, even
    under conditions of strong winds and 8 octas of cloud cover, Melbourne exhibits a UHI”.I have a graph from a traverse across Melbourne in 1992 which shows the CBD 10C higher than the rural surrounds. In the paper it shows winds are on average higher over urban areas than rural areas and this reduces the UHI effect at some measurement points. Think about Chicago and Toronto CBD’s for windy miserable places in winter.
    As one of the comments above indicates you have to consider local conditions. The whole temperature thing is nonsense Beside standard surrounds, measurement accuracy, minute by minute variations, it is necessary to factor in wind velocity, cloud cover and humidity. to get some indication of local changes but this can not be smeared over a greater area than 5kms. Too many getting paid to analyse useless data which can be cherry-picked by alarmists.

  38. Roger Andrews says:

    The Melbourne metro area comes out to be about 1C warmer than the surrounding areas when temperatures are averaged over several years. This number seems to be typical of most larger cities. (The biggest UHI I’ve found anywhere is Los Angeles, with +/-2C.)

    Since 1895 Melbourne has warmed by about 1C and maybe 0.4C of that is natural, suggesting that the UHI effect in the Melbourne metro area has increased by about 0.6C since 1895.

    A UHI effect, however, will generate an urban warming gradient only if it increases with time. Many cities don’t show urban warming gradients because the UHI hasn’t changed significantly.

  39. Ray Tomes says:

    Roger Andrews wrote (November 3, 2012 at 4:18 am):
    “One way of identifying shifts resulting from station moves etc. is to subtract the mean of all the records in the area from the record. Here’s the plot for Hokitika. The shift caused by the 1911-1912 station move is obvious but all we see after that is a bulge in the 1950s that may or may not be related to the station moves in the mid-1960s.”

    This makes it very clear thanks Roger. Nearly all the longer term movements of Hokatika from NZ average were put in by the “adjustments”. Perhaps we should rename these to what they really are, “fiddles”.

    I have a proposed statistical test that can show adjustments for what they really are. Take all the stations annual changes for every year and divide them into two sets, one where there are two measurements and no adjustment, and ones affected by adjustments. A simple F test will then show whether the adjusted data means and standard deviations could possibly have come from the real data in nature. My money is on a ridiculously small probability that they could.

    Is anyone able to do this test, or to email me the data so that I can. ray(at)tomes(dot)biz

  40. Ray Tomes says:

    Roger Andrews says (November 1, 2012 at 3:34 pm):

    I don’t think UHI impacts are significant. The Auckland record, where we might expect to see the largest UHI effect, doesn’t show an obvious urban warming gradient. The NZ surface records show no more warming than the surrounding island records (Lord Howe, Norfolk, Raoul, Chatham, Campbell, Macquarie) and in fact show less warming than UAH TLT.

    Roger, this is contradicted by appendix 5 which states “The 100-year trend in the Auckland composite temperature series is substantially
    higher than the warming trends found at the other six locations of the “7-station”

  41. Roger Andrews says:


    “The 100-year trend in the Auckland composite temperature series is substantially higher than the warming trends found at the other six locations of the “7-station” series.”

    Right. After NIWA adjusts it.

    The GHCN v2 Auckland record doesn’t show an urban warming gradient.

  42. Gary Kerkin says:

    If the raw data for New Zealand are considered from the earliest records (around 1860 – 1870) they show cooling for some 40 or more years and then varying increases until the last decade and a half. NIWA dropped the early parts of the records, prior to 1911, because the data of the several stations was sparse and, they imply, it is not possible to construct proper anomalies. So they eliminated what appeared to be reasonably substantial cooling in the latter part of the 19th century. There is a far wider problem, though. So far they have refused to offer estimates of absolute errors in the several measurements. The early measurements were probably subject to at least ±0.5ºC let alone errors occasioned by the situation of the station. More recent measurements are probably subject to errors of at least ±0.1ºC plus the situational errors which, based on the investigations of Anthony Watts and some particular comments he made about the Wellington station at Kelburn, could be as high as ±2ºC. If we are generous and suggest station errors to be in the vicinity of ±1ºC, which would not be unreasonable, then it hardly matters whether the increase is 0.5ºC per century or 0.9ºC. The increase is within the “noise”! We believe this is the reason NIWA will not release any estimate of absolute errors but instead relies on statistical nonsense to obfuscate the truth. If they were to reveal them their reports and arguments to Parliament could not be sustained and a government which based policy on them would find itself in a parlous situation. And no doubt the funds available to NIWA might be curtailed.

  43. ian says:

    Fwiw…I live rurally outside Auckland…and routinely measure early morning temps against those carried in radio newscasts…as a rule country temps are about 2 degrees lower than urban temp.

  44. Bob D says:

    Roger Andrews:
    “The GHCN v2 Auckland record doesn’t show an urban warming gradient.

    The Auckland site with the most obvious UHI/sheltering was Albert Park. Hessell found a lot of excess warming in his 1980 paper. NIWA themselves admitted it was contaminated, but then they did nothing about it – no adjustments for UHI or sheltering were carried out.

    When we checked, we found that Albert Park warmed by an additional 0.09°C/decade relative to the nearest reasonable-quality rural site (Te Aroha).

    The unadjusted GHCN chart you show is fine, but bear in mind it is made up of several sites spliced together. You’ll notice that in the early part of the record (say 1920s) Auckland is below most of the other stations. By the time 1965 is reached, it’s at the top end of the others, due to the extra warming at Albert Park.

    It could well be that GHCN switched to another site in 1965 (notice the gap). For example, Mangere was open from 1959. I’d need to look at the site combinations that make up the Auckland record before I could comment more fully though.

  45. Bob D says:

    NIWA said this about Albert Park:
    “This result would suggest a sheltering influence could be affecting the Albert Park record through at least the period 1928-1960. If the Te Aroha differential is taken as an approximate measure of the sheltering effect, then the Albert Park record of mean temperature shows warming by about 0.3 °C more than it ‘should’ over 1928-1960 (and maximum temperature by twice the amount).”
    -Auckland Review document, Appendix 5.

  46. Roger Andrews says:

    A UHI effect doesn’t automatically generate an urban warming gradient, and if it happened the way it’s described the Albert Park warming isn’t urban warming either. It’s analogous to the vegetational warming at Low Head:

    I was in the process of writing something about UHIs and urban warming when I got rudely interrupted by paying work, which is why I haven’t been posting comments recently, but as soon as I can regain my amateur status I will be back with a post explaining how temperatures are affected much more strongly by conditions in the immediate vicinity of the station than they are by urban heat islands.

  47. tchannon says:

    Consider fleshing that out sometime since it might make a good article as part of the surface stations project.

    Conceptually I agree with you.
    Class 1 sites in cities might start to show what is going on, perfectly possible but doesn’t happen.

    Changes in the management of vegetation are one of the hallmarks of urbanisation, part of which is no longer caring, scrub, unkempt all around. This applies to the UK where I’ve seen the changes, angry about it because we have lost views, sight of the real country.

    Does anyone have video of a met station? Preferably with a usable copyright. I’m up to something.

  48. Barry B says:

    John Daly’s comment on Low Head temp station is “we have an isolated lighthouse station with some bushes causing a daytime anomaly to it’s long-term data, resulting in the illusion of a strong warming trend.”

    So vegetational warming raises daytime temp trends while UHI has the same effect on night-time trends. Hessell (1980) noted that the UHI effect masked the sheltering effect at Albert Park. This probably referred to shading, which reduces daytime warmth – possibly at different rates throughout the year.

    All this means that stations which have known site effects should be omitted entirely from trend studies. The artificialities can’t be identified with enough precision to correct them.

    Probably the best one can do is contrast the long-term trend of a suspect station with the correlation-weighted average of a group of neighbouring ‘pristine’ stations. If the difference is significant, the station has to go.