The New Zealand temperature vexation

Well, is the country warming or isn’t it?

Here’s an article written by one of the scientists in the NZ Climate Science Coalition (we like to call him Rupert Postlethwaite). Rupert is as much a wordsmith as a boffin and we hope you enjoy his easy-to-read account of the controversy over our national temperature record. He transforms a potentially boring topic into a lively entertainment. Because it was written last year, it does not represent the latest twists and turns in the saga and the numbers are slightly different, but if you forgive those little lapses, he does give a good account of the basic issues. Especially, he makes the reasons for adjusting the temperature readings easy to understand. Feel free to let us know what you think of it.

Air temperature is usually recorded even today with instruments like this one — crude, simple and accurate enough, yet sometimes subject to substantial slip-ups.

So, about that Global Warming…

If someone were to ask you the question “Has New Zealand warmed significantly over the past 100 years?” you’d be excused for looking at them somewhat askance. Askew, even.

“Of course it has,” you’d reply, somewhat taken aback at the audacity of the query. You might at this point make some notes regarding the questioner, just in case. You never know.

“Just look at the data,” you’d respond. “NIWA shows it clearly on their site. You can see a graph showing how we’ve warmed a full degree over the past century. See?”

NIWA’s graph of the NZ annual temperature series.

“And Greenpeace says we’re all gonna die,” you’d add, helpfully.

The person dumb enough to ask the question would hang his or her head in shame and slink away, muttering incoherently about conspiracy theories and invoicing Big Oil. And you’d feel happy and ever so slightly smug, knowing that everything is in good hands. And it is, isn’t it?

Of course it is. NIWA processes and holds the raw temperature data for all of New Zealand, collected from a myriad of weather stations across the country and on some islands too. Like most countries around the world, it sends it all off every now and then to be amalgamated into an enormous Dataset in the Sky, called the Global Historical Climatology Network, or (thankfully) GHCN for short. The good people who manage this for us are the US’s National Climatic Data Center (NCDC), part of NOAA (National Oceanic and Atmospheric Administration). They claim they hold “The World’s Largest Archive of Climate Data”. I suppose somebody has to.

The whole lot gets churned over monthly and when the dust settles it spits out another alarming graph showing sky-rocketing temperatures. The graph is then used to tax us. It’s a simple system and we seem to like it, because we hold rallies and sit-ins to protest that we’re not being taxed enough. “More taxes!” we demand, waving our new graphs.

However, what few people know is that before the data is churned there are some little “inhomogeneity” adjustments that get applied to the raw temperatures, just to sort out minor issues, you understand. They take care of things such as station moves, equipment changes, urbanization, and so forth. Nothing serious, they won’t affect the overall trends at all. Hardly worth doing, just a tweak here and there.

You’ve probably guessed by now where this is going.

“What if,” you say sneakily, “the adjustments are not minor? What if, Heaven forbid, they end up influencing the trend just a little? Or a lot? What would that mean?”

“What about my new graph?” you suddenly blurt out, looking worried. “And my taxes?”

Well, it turns out that the inhomogeneity corrections are not as innocuous as they sound, judging by their effect. Now I know the scientists will have all sorts of arguments as to why they should do them, and point to peer-reviewed papers that discuss in gory detail topics such as “t-tests” and “subannual differences over symmetric intervals”, and so on.

But what really matters, when all is said and done, is: what effect do the “corrections” and “adjustments” actually have on the New Zealand temperature series? On the trends? How much quicker or slower will we appear to have warmed? And how come New Zealand has apparently been warming faster than the rest of the world over the past century? Most global estimates are in the 0.6ºC per century area. We’re apparently almost 1ºC.

Let’s take a look. The NIWA graph I mentioned at the start is made up of data from seven long-term stations scattered around New Zealand. They are Auckland, Wellington, Masterton, Nelson, Hokitika, Christchurch and Dunedin. Auckland and Dunedin have temperature histories that go back to 1853 or thereabouts, the rest kick in around 1864.

Fortunately, we have access to the GHCN datasets online, so we can compare individual stations before and after the “corrections”. We also have access to the NIWA raw station data, so we can look up any station’s history and see if and when any equipment (or even the whole station itself) was moved. So we can see if any “adjustments” make sense. After all, if an adjustment is so large that it affects the temperature trend of the station significantly over a hundred year history, the reason for the adjustment should be obvious in the station history, right? Something like a major meltdown in the station instruments comes to mind here.

So we pull the data and extract little New Zealand from the huge amount of data in there. We’re country 507, did you know that? Well, we are.

Then from the NZ data we pull out our seven long-term stations (well, five actually — it turns out we don’t send Masterton to NCDC at all, and for Nelson NCDC use only a short period from 1951 to 1970) and we plot the results.

At first we don’t believe what we see, so we check that we downloaded the right files. We check and re-check the data, because what we’re seeing is astonishing: four of the five long-term stations (Dunedin is the exception) show severe and unexpected adjustments, and they end up much warmer in recent years relative to the past. How did this happen? Quite simply — the earliest years are almost all reduced drastically.

So what does this do to the trend? Well, if you anchor a graphed line on its right-most point, and then swivel the left-hand side down, what does that do to the trend? The answer is: it increases warming, depending on how much you swivel it. Auckland is pushed down -1.5ºC, while Wellington is reduced -0.5ºC. Christchurch drops by up to -1.3ºC. Hokitika is a shocker – it loses a full -2ºC!

How does all this graph-swivelling affect things? Quite a lot, as it turns out. Auckland has its trend increased 7-fold, from a mild +0.2ºC to +1.5ºC per century. Wellington goes from cooling (that’s right – cooling!) at -0.52ºC per century to warming at +0.35ºC per century. And Hokitika takes the biggest hit: from cooling at -0.15ºC to a whopping +1.7ºC of warming! Christchurch jumps from a negligible 0.08ºC to 0.93ºC per century. The only one to show the opposite trend is Dunedin, which drops from 0.7 to 0.4ºC. Looking at the Dunedin data, I can understand why it was adjusted – the data jumps all over in the early years. But to be honest, it’s the smallest adjustment and frankly gets swamped by the others.

Here are the adjustments, expressed in degrees Celsius per hundred years:

Now I know what you’re thinking. You’re thinking that I’m just cherry-picking the data. Skeptics do that, you know — they can’t help it, they’re deluded. After all, these stations could just coincidentally all be in need of massive reduction of past temperatures. That could happen, couldn’t it? One can easily think up a hundred reasons why people simultaneously, across the country in the late 19th century, would all read their thermometers up to 2 degrees too hot, right? Every day. Year in, year out. Uh huh.

A glance at Auckland. Does any of this make sense?

Auckland temperature series with GHCN adjustments. Click for larger version.

What about Christchurch?

Christchurch temperature series with GHCN adjustments. Click for larger version.

Wellington?

Wellington temperature series with GHCN adjustments. Click for larger version.

Let’s look closely at the Hokitika station. It’s had a quiet semi-rural history, running from 1868 to the present, with a break from 1881 to 1893. The unadjusted data (blue series) varies between 11ºC and 12.5ºC most of the time, and hasn’t changed significantly in 130 years.

Hokitika aerodrome temperature series with GHCN adjustments. Click for larger version.

The red series is the adjusted one. It’s difficult to see how this adjustment can be valid. Have a look at how the adjustments were applied, and when.

GHCN adjustments to the Hokitika aerodrome temperature series. Click for larger version.

The NIWA station record shows nothing untoward, apart from a new screen in 1912. Yet the whole record prior to 1912 is decreased by a huge -2ºC. After 1912 the decrease is a mere -0.6ºC until 1938, when suddenly we decrease by only -0.46ºC from then on. According to the record, nothing unusual happened in 1938 or 1939. The only entry near here is 1937: “Routine inspection. No problems.” The next entry is in 1943 and involves moving a Dines rain gauge. So these adjustments don’t appear, on the surface anyway, to be necessary due to station moves or anything like that.

[UPDATE: We know now there was a station move in 1943; it just wasn’t described very well in the station notes. The entry “instruments to be moved to new site in Aug 43” is now known to be a reference to the Hokitika Township/Hokitika Southside move.]

And then from 1951 the whole charade is over and the data from then is unadjusted until the present. What happened in 1951, you ask? Well, nothing. The nearest entry is 1953: “Routine inspection. No problems.” Notice how the actual temperatures after 1951 show no warming — just a gentle flat trend until the present.

Now it seems to me that something isn’t right here. It seems unlikely that the pre-1912 data had to be corrected because it showed unrealistic temperatures relative to nearby stations, simply because there were no nearby stations back then. Neither the data nor the history shows any reason for the adjustment. If the pre-1912 data is so very suspect that it’s a full 2ºC out, why bother to include it at all?

Based on the laws of chance one would expect the station adjustments to even out. How likely is it that four of our five long-term stations all end up having their early data adjusted downwards, and the only one that doesn’t is coincidentally the smallest adjustment? Even if some of the corrections can be justified technically, shouldn’t the public be made aware that almost all of the warming in New Zealand comes from the adjustments? And shouldn’t they front up with the reasons for the corrections, when so much is at stake?

Why does this matter? Who cares if some little corrections skew the data a bit? Well, normally, almost nobody would care, but you see that NIWA graph we started with uses adjusted data. They don’t publish a graph of the unadjusted data. If they did, it would show a very slight warming trend of about 0.06º per century. That’s right – essentially no warming. Of course, if you measure since 2001, we’ve been declining at -1.9ºC per century! And if the current trend continues, we’ll be heading for an Ice Age!

No, just kidding.

New Zealand aerodrome temperature series with GHCN adjustments. Click for larger version.

And would that graph be useful to a government hell-bent on pushing us into economically disastrous new laws to govern carbon emissions? No, I think not.

– Rupert Postlethwaite

Appendix A: Hokitika Station History

Here are the Hokitika Aero station notes as at November 2009, when we began our investigation into the national temperature record. Our conclusions then still seem justified, but please note that more information has since come to light that gives a better, though partial, explanation for the adjustments here. However, NIWA have still not provided justifications for any adjustments except in the case of Hokitika, where they have totally repudiated their previous explanations (such as highly technical changes on the basis of altitude differences) and simply “compared” the moved station with “neighbours”. Nobody has described how to perform such a comparison or explained why you would do that rather than simply run the old and new stations concurrently for a while.

Hokitika Site Change History
Site Change Date     Description
01-Jan-1866       Station opened at rear of government buildings.                   Good exposure.
02-Jan-1866       Max-Min recorded in tenth degF 1866-1880.
30-Dec-1880       Forms record max on day-of-reading up to Dec1880. (Max                   has been shifted back 1-day in MAX_MIN_TEMP table).
31-Dec-1880       Observations ceased.
01-Jan-1894       Observations recommenced.
02-Jan-1894       Max-Min recorded in half-degF 1894-1899.
03-Jan-1894       Forms record max on day-of-reading up to Dec-1899. (Max                   has been shifted back 1-day in MAX_MIN_TEMP table).
10-Sep-1912       A new screen was installed. The enclosure was too small.                   The raingauge was moved to improve exposure.
19-Jan-1928       Routine inspection. The raingauge was loose in a box and                   probably underestimated rain. Screen needed painting.                   Sunshine recorder on roof and inconvenient for observer                   to change cards. Enclosure still too small at 2.1 m by                   2.7 m.
25-May-1929       Routine inspection. No problems. The enclosure had been                   enlarged on 24 October 1928 to 15 m by 12 m.
10-Jan-1931       Routine inspection. Rusty tacks needed replacing to                   secure louvres in screen.
03-Dec-1934       Routine inspection. The raingauge was moved to improve                   exposure and reset to correct height. Grass minimum                   thermometer giving problems and replaced. Other                   instruments in good order.
14-Feb-1937       Routine inspection. No problems.
24-Jun-1937       Routine inspection. No problems.
01-Aug-1942       Dines gauge transferred to airport.
25-Jul-1943       Routine inspection. Screen in poor shape and repaired.                   Raingauge had small inner can which often overflows. To                   be replaced. Instruments to be moved to new site                   effective August 1943.
13-Jan-1944       Manual raingauge replaced.
22-May-1944       Routine inspection. No problem. First inspection of                   Hokitika Southside.
05-Jul-1944       Routine inspection. No problems.
07-Aug-1945       Routine inspection.
11-Oct-1948       Routine inspection.
03-Mar-1953       Routine inspection. No problems.
09-Aug-1957       Routine inspection. No problems. Very good station.
31-Dec-1964       Station replaced by F20793 which opened in November 1963.
28-Feb-1965       STATS CODES 00,42, continued AFTER station shut on                   19641231
31-Dec-1965       STATS CODES 22,23,24,25,26,28, continued AFTER station                   shut on 19641231

Views: 383