The National Institute for Water and Atmospheric Research (NIWA), accused last week of fiddling the long term New Zealand temperature record to create spurious warming, has released information showing that the attack mounted by the NZ Climate “Science” Coalition and Climate Conversation Group has no merit.
The NIWA announcement shows that the warming trend in the long term record is also found when weather stations with long term records that require no corrections are used. From the release:
Dr Jim Salinger has identified from the NIWA climate archive a set of 11 stations with long records where there have been no significant site changes. When the annual temperatures from all of these sites are averaged to form a temperature series for New Zealand, the best-fit linear trend is a warming of 1°C from 1931 to 2008. We will be placing more information about this on the web later this week.
I’ll have more detail on that series when it’s made available. So the warming in the record is robust, found in sites all round New Zealand, and doesn’t depend on mysterious adjustments. But the Treadgold/CSC report also made claims about data being hidden:
Requests for this information from Dr Salinger himself over the years, by different scientists, have long gone unanswered, but now we might discover the truth.
That’s an outright lie, as the NIWA release shows.
For more than two years, New Zealand Climate Science Coalition members have known of the need to adjust the “seven station” data. They have had access to:
- the raw data
- the adjusted data (anomalies)
- information needed to identify the adjustments made by Dr Salinger
- information needed to develop their own adjustments.
The NIWA release cites emails to CSC members Vincent Grey and Warwick Hughes in July 2006, which provided all the references required to calculate the necessary adjustments themselves. In particular, all the information about the station site changes has been publicly available since 1992 and details of the methodology since 1993!
So where does this leave Treadgold and the CSC? They have published a report, issued press releases and made blog posts that misrepresent the facts, and have shown themselves incapable of conducting good science. They have proven themselves morally and ethically bankrupt, and should — if they had any decency — withdraw and apologise. But I won’t be holding my breath.
Had a quick read through the article “Trends and Variations in South Pacific Island and Ocean Surface Temperature”
Reconciles fairly well with the NIWA land based records.
Concludes:
“The results also extend previous work showing that annual and seasonal surface ocean and island air temperatures have generally increased by 0.68–1.08C since near 1910 throughout a large part of the South Pacific southwest of the SPCZ. This is consistent with warming
observed in Australia and the southern oceans to the south of that continent (Salinger et al. 1995, 1996). To the northeast of the SPCZ, decadal increases in annual temperature are only widely seen since 1970, by 0.38– 0.58C with some cooling before that time since 1940,
the beginning of the record discussed here.”
The main difference is a hump during the 30’s that is kinda missing from the NIWA records, but all in all a bit of an own goal for NZCSC really.
Still in the dark about the methodology for the Hokitika adjustment but this other study clearly shows NZ warming since 1850, so the overall conclusion is reaffirmed.
Thanks R2. “A bit of an own goal” is the message that I hoped would come through… 😉
Excellent. Hopefully some egg sticks to face.
So the crackpots were wrong. What a shock.
And the data fiddlers got caught with their trousers down! You will not bluff and bullshit your way out of this.
CSC – liars and morons.
The wheels have come off your dung cart sonny!
Roger – I assume that you really meant that comment to apply to the CSC themselves, where it actually fits.
No Billy. You have been dropped right in it by your mates Mann, Jones, Briffa and others at CRU. There is even talk of criminal charges pending!
Oh no! Not… Talk.
There is talk that Roger Dewhurst is nothing but talk…
It would be nice if Rodney Hide…
http://hot-topic.co.nz/who-writes-rodneys-rubbish/
…would admit that he’s been a dick about it.
hi there!
on topic, would there be any chance whatsoever from enzed records or Salinger@NIWA of coming up with cloud influences at Kelburn? I ask because evidence is growing that Lindzen had things the wrong way around and that clouds are in fact positive feedback.. ie warming. Which may modify the ‘theoretical’ conclusion of Kelburn’s height above sea-level as cooling..
off topic but I hope you’ll bear with me as my first input here. Quite probably the blog yesterday was attempting fun, satire, what have you. It sure worked insofar as “readme” exercise goes. My beef is with your choice of once funny-man Jim Hopkins. In many subjects he is witty, but for folks like myself who have read his distinctly sceptic-aligned views in hard print there would be some risk to yourself of being compromised. Unwittingly I’d guess. Hope you don’t mind and welcome the criticism as constructive..
Best
Hi tomfarmer
I don’t think the line of argument in your first paragraph is very promising. As I understand it you are suggesting that positive cloud feedbacks change the ‘theoretical’ (your quotes) cooling with height.
The drop-off of temperature with height is a pretty robust observed feature of the atmosphere. The factors that determine the exact value of the mean lapse rate are many and varied (and clouds are definitely involved) but the basic phenomenon is not in question. (Like a lot of features of the atmosphere, really.)
“The drop-off of temperature with height is a pretty robust observed feature of the atmosphere.”
Mark H obviously doesn’t live in Christchurch where the inversion layer is an easily observed climate phenomenon. I have often observed a 2 degree drop descending from Cashmere to the City. Assumptions, projections and adjustments to data are what make this whole debate so difficult.
The huge economic consequences of the current political decisions on global warming really do need stronger evidence to justify them than data trends that rely so heavily on debatable adjustments to make their case.
Rod – that’s why it is called an ‘inversion layer’ – the temperature change goes against the normal lapse rate. There’s been a lot of hot air on that topic from some quarters recently.
As for your comment on huge economic consequences of the current political decisions on global warming really do need stronger evidence to justify them than data trends that rely so heavily on debatable adjustments – there are multiple lines of evidence ranging from physics, ecology, historical weather records, ice cores, GCMs, etc which are all consistent in pointing to AGW. Of course, the huge economic consequences you mention is the major reason why this issue is being drummed up into the biggest fight since tobacco.
Hi Rod,
Inversions are an interesting phenomenon, but even in Christchurch they don’t happen all the time. The cooling of the lower atmosphere with altitude is hardly controversial, nor are the station adjustments made when building long data series. Look again at the NIWA release: stations that have never been adjusted still show the same warming trend… (More on that in a later post, I hope).
I’ve measured more inversions than you’ve had hot dinners, sonny!
On average the temperature drops with height at something like 0.6 degC per 100 m, plus or minus. Even in Christchurch.
Three points:
It always amazes me that some people think that climatologists and meteorologists aren’t aware of the effect of local conditions on local climate, given it is what many of them do for a job.
What debatable adjustments are we talking about? What effect have they had on the temperature record?
The assumptions used in economic analysis are far more suspect than anything in used climate science.
Mark H,
my suggesting(your word) is a tad strong. Asking was where I was at. Still, I’m happy to accept that my theoretical was inappropriate and that the Kelburn claim in – was it Salinger’s writing or the blogger’s opinion? – is robust. Though I assume this robustness is based upon prevailing evidence.
Which in another way adds point to my ask. E.g has the Kelburn locality experienced greater cloud cover in recent times. If so, is this likely to last there.
Picking up a little on Rod’s point – I do not agree with what appears a call to greater uncertainty from this fellow – Canterbury this year (season) has had a good deal more cloud.. this week alone 4 days continuously.. and.. should the observation of positive feedback hold then air and ground temperatures (agricultural values) may be holding up.. to thus explain my brassicas phenomenal growth.
tom – what I understand you as saying is that if there was more cloud cover in recent times, then the measured temperature would be higher than it would otherwise have been.
I imagine that answer to that is not simple – it might depend for example on whether the greater cloudiness is at night (when cloud keeps things warmer) or in the daytime (when it tends to cool).
In any case, the cloudiness doesn’t mean there is anything wrong with the temperature record – if it is warmer, then it’s warmer. You might be hinting that “it’s not all CO2”, and you’d be right…
The difference between Kelburn and sea level sites like the airport is straightforward and consistent on a daily, monthly and annual basis and variables like cloudiness have a negligible effect on those outcomes. tomfarmer is trying to find complications that do not exist.
In the case of Dunedin, the latest site is a comparatively warm one at sea level (Musselburgh).
Why is it I feel like a pork chop that has strayed into a synagogue?
Whatever, it happens I fall into neither warmist or skeptic camps, but have a little knowledge of matters statistical. What has caught my eye is the consistent pattern in the differences between the Treadgold and NIWA graphs. If the difference can be explained just by past recorder site relocations and suchlike adjustments, I would have expected some randomness in the differences – some adjustments up, some down. In fact, I will go so far as to suggest I would have expected the effects to cancel each other out. In my experience with time series they usually do if the sample is big enough. My statistical alarm bells are ringing because the differences follow a pattern that coincidentally seem to serve one or other point of view. Statistically this is very odd, and points to something systematic going on. I am in no doubt whatsoever that there needs to be a better explanation than just a few recording sites being relocated over the years. For example, has there been a consistent policy to move recording stations always to warmer (or colder) places when they are moved? No, that’s not credible. What, then?
Data from sites that were not relocated and sea temperature data are interesting in the grand picture, but divert attention from the statistical question of why there is such a clear pattern in the differences over time between the original two series that gave rise to this debate.
As an average bloke who is going to be asked to help fund some pretty big steps being planned to counter climate change, I believe someone in the know needs to give a more convincing explanation.
Rod: Consider that one of the principal things that gets corrected for is station altitude. Given that a lot of stations are already close to sea level, there cannot be an “equal expectation” that site corrections will average out to nothing, becuase stations will not (in general) move to negative altitude — below sea level.
Rod, you’d be best to ask CSC about that – they presumably chose to highlight those particular locations.
Rod: Comparing the Treadgold and NIWA graphs shows quite a number of station changes, some up and some down, but you are quite right – the overall pattern is for most of the early stations to have their temperatures adjusted down relative to the modern data. This implies that the modern sites tend to be at relatively colder locations, compared to what would be measured today if there were still gauges at (say) Auckland’s Albert Park or in the Wellington CBD.
Well, the conspiracy theorists and CSC prefer the other implication of fiddling the data. But if you have confidence in the integrity of the professionals, as I do, how do you explain the trend to relatively colder locations over time?
My suggestion is that early settlements in New Zealand were located to take advantage of the local climate – eg, the warm side of the valley instead of the cold, the sheltered inland area instead of the exposed coast, the harbour-side instead of the top of a hill – and the Met instruments located nearby. Then as the towns expanded and the land was required for other purposes, the Met stations had to be relocated to less hospitable (and often colder!) environments where land was available.
I can’t really prove my argument, but I would point out that the one exception in the site changes is Dunedin, where the site adjustments go upwards not downwards with time.
Didn’t the Scots settle there?
RW,
Gee if what I’m trying to find is complicated then let’s call the whole thing off. Not ask any further questions, can’t have people even so much as monitor themselves and their knowledge or evidence to date..humn.
Except that climate change does mean change.. that things change.. how quickly and consequentially being more important today than in the past.. That’s only complicated Rupert for folks who set themselves and others to ‘mate in 12’ (yeah I took a peek at your w/site) board problems. Described as not subtle etc..
Meantime you can redeem yourself by wishing my brassicas free of bugs that their growth this season suggests to me at anyrate has arisen from higher temperatures. Aforementioned.
With best wishes
On the raw data “the best-fit linear trend is a warming of 1°C from 1931 to 2008”. Isn’t that rather a lot?
Looking at IPCC’s AR4, the global movement from the 1930s (a warm period) to 2005 seemed to be about 0.4deg or 0.5deg at most. Did we all know that New Zealand has been warming at about twice the global average rate for the last century?
This blog seems to use words like “liars” “bullshit” “dick” and “morons” quite a lot. Do you guys feel this scientific jargon helps to clarify your reasoning – or just your spleens?
So, on the Rodney Hide thread, I made some suggestions around urban heat island. A number of people suggested (politely) that I a) educate myself, and b) get some data and do something.
Thanks to some assistance on Kiwiblog on how to actually log on to NIWA, I downloaded the Kelburn temperature record, and wind speed record.
The statement someone made to me was that Wellington wasn’t subject to urban heat island effects, as it had little urban density, and the usual high winds meant that the heat was blown away / new air off the sea came in. (I’m paraphrasing and elaborating a little on the argument there, but you get the gist).
My hypothesis was that there was, in fact, some urban heat island effect. If this were true, then I would expect more heat island effect on days with little wind than days with lots of wind.
To test that hypothesis, I correlated wind speed and max temp for each day 1961-2004. (Kelburn wind data only available back to 1961). I then looked at the wind speed at 9am, and split the data set into high wind days (> 5m/s) and low wind days (< 5 m/s). I then separately graphed the average max temperature for the high wind days in a month, and the low wind days in the month. For each graph I added a linear trend line for each month.
Note that even a high wind day under my coarse definition would/could have some periods of no wind. So even the high wind days could have some urban heat island effect – to prove my hypothesis it isn't necessary that high wind days show no warming at all, only that they show less warming than low wind days.
For my hypothesis to be true, it should be the case that high wind days show less warming than low wind days. This would indicate that, in fact, the data could be contaminated by urban heat island effect, or perhaps some other effect that is correlated with wind.
So, drumroll please….
Low wind days show an average monthly warming over the period of 0.56 degrees celsius. High wind days show an average monthly warming of 0.06 degrees celsius – or almost no warming at all.
Does this tell us anything? Well, no, because my methods are pretty rough. Caveats are:
– I used Excel's linear regression function on the graphs, and then read the gradient off the graph by eye. So not hugely accurate
– I'm not sure linear regression is the right regression to use – but then it looked to me like the original graph had a linear regression too…..
– Using my method, it should have been the case that every trendline crossed the x-axis midway through. Some didn't, so something is wrong in my analysis
– I arbitrarily selected 5 m/s as a split between high and low wind. It looks to split the days roughly 50/50, but I didn't check exactly. Some months may have very few high wind or low wind days
– There aren't all that many data points in a single weather station. The result could be random luck.
– Conversely, this is the example weather station that NIWA plotted and showed warming in, so I could be successfully calling into question that single station
If I have time some point in the future, I may look for a weather station that has more wind data. But in the mean-time, you'd have to say that was an interesting result, no?
Argh. I think this blog has limits on comment length. 🙁 I’ve posted this comment bit by bit using the edit feature, I presume the other comment will exit moderation at some point, then we’ll have a duplicate. Ah well.
On the Rodney Hide thread, I made some suggestions around urban heat island. A number of people suggested (politely) that I a) educate myself, and b) get some data and do something.
I downloaded the Kelburn temperature record, and wind speed record.
The statement someone made to me was that Wellington wasn’t subject to urban heat island effects, as it had little urban density, and the usual high winds meant that the heat was blown away / new air off the sea came in. (I’m paraphrasing and elaborating a little on the argument there, but you get the gist).
My hypothesis was that there was, in fact, some urban heat island effect. If this were true, then I would expect more heat island effect on days with little wind than days with lots of wind.
To test that hypothesis, I correlated wind speed and max temp for each day 1961-2004. (Kelburn wind data only available back to 1961). I then looked at the wind speed at 9am, and split the data set into high wind days (> 5m/s) and low wind days (< 5 m/s). I then separately graphed the average max temperature for the high wind days in a month, and the low wind days in the month. For each graph I added a linear trend line for each month.
Note that even a high wind day under my coarse definition would/could have some periods of no wind. So even the high wind days could have some urban heat island effect – to prove my hypothesis it isn’t necessary that high wind days show no warming at all, only that they show less warming than low wind days.
For my hypothesis to be true, it should be the case that high wind days show less warming than low wind days. This would indicate that, in fact, the data could be contaminated by urban heat island effect, or perhaps some other effect that is correlated with wind.
So, drumroll please….
Low wind days show an average monthly warming over the period of 0.56 degrees celsius. High wind days show an average monthly warming of 0.06 degrees celsius – or almost no warming at all.
Does this tell us anything? Well, no, because my methods are pretty rough. Caveats are:
– I used Excel’s linear regression function on the graphs, and then read the gradient off the graph by eye. So not hugely accurate
– I’m not sure linear regression is the right regression to use – but then it looked to me like the original graph had a linear regression too…..
– Using my method, it should have been the case that every trendline crossed the x-axis midway through. Some didn’t, so something is wrong in my analysis
– I arbitrarily selected 5 m/s as a split between high and low wind. It looks to split the days roughly 50/50, but I didn’t check exactly. Some months may have very few high wind or low wind days
– There aren’t all that many data points in a single weather station. The result could be random luck.
– Conversely, this is the example weather station that NIWA plotted and showed warming in, so I could be successfully calling into question that single station
If I have time some point in the future, I may look for a weather station that has more wind data. But in the mean-time, you’d have to say that was an interesting result, no?
Laurence, give him credit for critical self evaluation. I think this is a good start to evaluating a genuine issue. If you would rather ignore it thats your prerogative but I think your criticism is unwarranted. Perhaps constructive feedback next time would be a better idea, if you can do better suggest improvements.
Interesting. Perhaps a guest post is in order? Would help to see images and more detail to allow proper critique by others.
One question, (or 3) did the number of windy days per year remain constant? Did the windy days stay in the same months? Were there any interesting trends in general in the windy days that may have caused a sample bias?
I’ve just had a look at the “windy” days.
There is variation over time – min 148, max 256. From eyeballing it I’d say windiness on average decreased over time, but was also low at the very start of the series.
There is variation by month – min 602, max 824. The higher months are sept/oct/nov/dec/jan, which is unexpected – I’d have expected the equinoxes. If vegetation is a correlation, then I would have thought there would be fewer windy days Dec/Jan/Feb, and more windy days in Jun/Jul/Aug.
It is possible there are other correlations with windiness. If, for example, windy days were getting more extreme, and windier days were cooler, then that could also explain the correlation. But there is no particular reason to believe that is true, just noting that there is at least one possible explanation that doesn’t involve UHI.
There are problems in trying to use the Kelburn (1928-1997) wind record (ie daily windrun) at all. A NIWA scientist examined environmental issues affecting the windspeeds (variable vegetation heights in surrounding zones) and found the timing of considerable “jumps” in the records (monthly averages etc) matched sudden changes in these heights. The composite record can’t safely be used for anything. Maximum gust speeds however have been separately measured and recorded at a suitable site from 1972 (some paper records 1967-1971), and windrun is now also being meaured away from tree-growth effects as well. If you’re using data from 1961, then the early part of that data cannot be relied on – and indeed if the 9am windspeed was measured on the affected equipment rather than on the “gust-site” equipment, then none of it can be regarded as consistent.
To get back to basics: Salinger cites 11 NZ stations, of which Kelburn is just one, that show an averaged gradient of 1.0C over 1931-2008.
PaulL – interesting analysis but I’m not sure if you’ve tested the UHI effect. You have just shown that the wind in Wellington can be cool.
Who knew?
If temperatures are warming you would expect windy days to warm as well.
If it is only the UHI effect that is causing the warming then only non-windy days will warm.
AndrewH – no. I’ve maybe shown that windy days aren’t getting warmer, and non-windy days are. I’d hope I’d notice if I were just comparing temperature between the two data sets – I’m comparing warming between the two data sets.
On vegetation – maybe. But surely vegetation would be constant for a given time of year (at least on average) – if there’s less wind in summer because of vegetation, that doesn’t really impact the fact that windy days aren’t warming and non-windy days are.
Happy to share the data and spreadsheet. It came from NIWA, and my analysis, as I said, is pretty rudimentary. I need to learn how to drive the regression function directly in Excel. And I’ve been thinking a bit about why I averaged by month and regressed each month separately, instead of just regressing the whole series. Anyway, if anybody wants it, just shout (and give me somewhere to send it).
NIWA may be interested, they may be able to explain it, or it might be a puzzle in a larger picture. Or it may turn out not to be a robust result.
I’d suggest a good faith email to NIWA. They may simply take note of it, but I’m sure they’re interested in improving the quality of their analysis and maintaining reliable records. This could help with that.
Last attempt – there have been several sudden changes in the environment in that 1928-1997 record, with a highly significant effect on measured winds. You can contact Steve Reid at NIWA (though he may have left, I’m not sure). However, since the intenation of a number of people in this whole silly debate is to try and counter the notion that Kelburn has got warmer without the “benefit” of UHI , just like the other sites, I’m sure the red herrings will continue to be placed on the counter.
Have any of you read my last post (11:21), or has my text just become invisible? This wind research has established NOTHING.
Your text is visible, please politely point to me to the published NIWA paper about the Kelburn ( and others? ) environmental factors on the wind data.
Whilst you’re about it, could you clarify the selection processes used to demonstrate that environmental factors can compromise some data, but don’t affect all measurements from the site.
You will have to look at the archive of NIWA papers. Notes on the viability of a particular raw dataset aren’t all going to make it into published papers. The issues with this particular data have been known for many years however. Again, contact the relevant person I mentioned. Incidentally, as an example of the unreliability, separate gust data indicated that October 1990 was an average month as far as windiness was concerned – yet the windrun record showed the monthly average to be the lowest in the entire record for October! (Incidentally, this was also quite a warm month).
Even if your base data were reliable, you’ve given no indication of differentiating wind directions in the “windy” days.
On the last point, I suggest you ask NIWA directly, but would suggest that in the case of wind measurements the “fetch” and consequent flow speed can be seriously affected by obstacles mcuh further away than those which could shelter other measuring devices. In Australia the wind measurements in the capitals are all taken atop very tall poles (as have been the Kelburn gust readings I referred to).
OK, I’ve redone the analysis, and corrected some of the things I didn’t like about my first analysis.
1. I’ve moved away from grouping by month, then doing a trendline for each month, then averaging those. I think that is statistically shonky, although my statistics is rusty enough that I cannot explain why. I didn’t think this would make a material difference, and looking at the results I don’t think it has.
2. I’ve moved to using Excel’s “SLOPE” function for the regression, rather than reading a trendline off a graph. I’ve not used this function before, but I think I have it right. I’d welcome someone to review what I did there, and whether it is statistically appropriate.
I’m therefore no longer analysing the average monthly temperature, I’m just doing two regressions – one for all the high wind days 1961-2005, another for all the low wind days 1961-2005.
The high wind days show a warming of -0.14 degrees, the low wind days show a warming of 0.5 degrees. Again, happy to share these results with anyone who’d like to check. paul at planar dot id dot au
Next, to some of the comments.
On vegetation, I understand the concern, but:
a) I don’t have any way to correct it. If this site has concerns around the accuracy of measurement, then I’m not sure why it was used as part of the overall combined temperature record. I’m happy to either:
(i) look at another site that is part of that temperature record
(ii) consider another way to proxy wind speed that doesn’t use the suspect measures
(iii) consider a correction that someone might suggest
(b) On vegetation, I’m not sure where the argument is that any distortion would be anything other than randomly distributed. I guess there could be an argument that there is more vegetation in warmer months, and therefore less windy months are warmer, but this
(i) should be consistent year to year, so wouldn’t create a trend
(ii) isn’t borne out by the data, which shows that the windiest months are the warmer months Nov-Feb
(iii) the number of windy days has reduced more recently, which is counter intuitive if the vegetation is now being controlled – surely it should have gotten windier?
As for e-mailing to NIWA, I will do so once/if I think I have something that is worth their attention. At the moment I’m just some guy with a spreadsheet looking at one weather station. If you think that climate scientists are nervous about opening their results to scrutiny by amateurs, you’d have to expect that amateurs might want a little time to refine their results before sending them for scrutiny to scientists. I’d rather not have a glaring hole in my analysis when I do that. My later spreadsheet is a bit cleaner, but I’d still like to stew on it for a coupla days, and ideally have someone else look at it and give some tips.
Thinking a little further, if I’m looking at localised UHI effects, then I probably want the contaminated data. If the hypothesis is that the warming is due to UHI, and that wind reduces the UHI, then surely I’m interested in the wind speed at the temperature measurement site, not the wind speed many metres above. If the vegetation was reducing wind speed, then it would also have been increasing the UHI effect?
Good on you for looking into this.
I don’t see the few metres between where the wind speed and temperature measurements are made as an issue.
You could look at other sites, especially rural sites, as a control.
I think you’re probably right and are detecting UHI effects.
“..a) I don’t have any way to correct it. If this site has concerns around the accuracy of measurement, then I’m not sure why it was used as part of the overall combined temperature record.”
This sitel, to the best of my knowledge, does not have concerns about the other variables. Feel free to ask NIWA.
On vegetation changes: you haven’t understood the point. Regardless of the time of year – THIS IS NOT A SEASON-RELATED issue – the pattern for both monthly and annual values has been one of periods of a little decline in windiness, followed by substantial leaps in windindess after vegetation was cut back. Then another period of decline, etc. Most notably the record shows there is a decline after about 1972 which steepens at about 1984 and then after restabilising does not recover to older levels – I can’t verify it, but there was almost certainly no retrimming of major vegetation in this period. Reid has the details. The seasonal relativities of the months did NOT change over the 70-year record, as ALL were affected by the ups and downs.
Note: going by the gust data from the “tall pole”, there has been a small decline in the average values of top gust speeds – about 8-9% over a 36-year timespan.
But, talk about reinventing wheels – why don’t all you amateur doubters challenge Jim Salinger to a debate on the NZ temperature record? I don’t think you’d come out of it very well. But you’d probably prefer to argue about angels dancing on pinheads by going through one NZ site at a time. BOM in Australia has bigger resources than NIWA – why don’t try taking them on while you’re at it?
RW: I understand from what you’re saying that this is not season related, and not consistent from year to year.
What I don’t understand is why that would impact my result. If I’m looking at whether the place we measured the temperature was windy, then I don’t really care whether it is windy because the vegetation was cut back, or windy because the day was overall windy.
To put it another way, the reason for lack of wind is irrelevant. What matters is that there is a correlation between lack of wind and warming. The hypothesis is that correlation might be caused by urban heat island.
Rather than me trawling the site looking for another location to do the same analysis on, do you happen to be aware of another location that was included in the overall series, that is urban-ish (so Hokitika, which is the other one I know, isn’t a great candidate), and that has more reliable wind and temperature record for the majority of the time period in question?
Exactly, if it was not windy on this location, but the rest of the city was windy, you don’t want to match the wind speed in the rest of the city with the temperature at this location. If it was sheltered due to vegetation, and if the steel and concrete in the area was radiating heat, that will create a heat island an artificial warming.
RW, strange that you are so hostile to this analysis. It appears you are not trying to hide your bias. Paul seems to be making a genuine attempt to analyse the data and has not shown any bias thus far. He is open to critique and suggests. To attack him is absurd. Offer suggestions for improvement and if your hypothesis that there is no heat island contamination is correct the data will show that.
What is with all these citizen climatologists all of a sudden?
Do you guys do your own dentistry as well?
haha yeah its so sad when people think for themselves isn’t it? How did you tackle the climate change question when it entered your life Carol?
Probably an astroturf operation in defence of NZC”S”C, Carol.
Give the guy a break eh – he’s at least got the data and some some analysis for himself, which is more than most people have ever done.
Paul, my suggestion would be to second the earlier suggestion to make an enquiry direct to NIWA (or Jim Salinger?) – they’d hopefully be able to help clarify what you have found in your analysis – whether it is an indication of ‘UHI’ or if there is anything else you might need to consider.
I’d also suggest that to really show it is actually UHI that is causing the difference and not something else like ‘land’ vs ‘sea’ (ie windy days effectively give temperature of cook strait) or ‘wind’ vs ‘non-windy’ (ie windy days the air is more mixed up – see Pielke) then you really do need to do the same analysis on a comparable set of rural data.
Carol, Rob, RW: Interesting circling of the wagons there.
The reason I spent some time on it is that I commented on a previous thread here. It was suggested there was nothing to see in the NIWA data, and that the CSC paper was a crock. That is entirely possible.
I pointed out that pretty much all the corrections went in one direction – to reinforce the warming. That seemed a red flag to me, I would have thought on the law of averages the corrections would work in both directions. I noted that on one thread.
It was suggested to me that was because the stations mostly moved inland from coastal sites to non-coastal sites, that would explain the corrections. My question to that was whether all the corrections were for location, with no other corrections, for example for UHI, which would work in the other direction.
The answer was that most NZ locations don’t have UHI as they are too windy / insufficiently urban. A couple of people were quite pointed that I should “educate myself”, and that if I thought there was something in it, I should download the data that was freely available and do some analysis.
I have stated my hypothesis, downloaded the data, and done some analysis. I’m happy to make available my workings.
What I now find is that my work is considered worthless by some because I am an amateur. So we’ve moved from “this is science, the facts stand for themselves, go get yourself some facts” to “you’re an amateur, your facts have no relevance because you don’t have the right credentials.”
I fully understand that different people have made the different comments, and that therefore it would be unfair to suggest any one person was attempting to hold those conflicting views. But it is a little frustrating.
Other than your belief that I am somehow trying to disprove warming (which I’m not), or that I’m insufficiently qualified to drive a spreadsheet, is there any critique of what I actually did? Do any of you want to spend your own time to do that same analysis and show it as wrong?
Fair comment, PaulL.
I’m not qualified to critique your analysis of the instrumental record, but I’ve emailed Jim Salinger advising him of this thread. Can’t guarantee he’ll have time to look at it, though.
Yes, what Rob said.
Is there not a certain mistrust providing the motivation for your efforts? (maybe not, maybe it’s just curiosity).
Carol, not mistrust as such.
It does seem a little unusual to me that NZ has more warming than the rest of the world, when logically as a long thin island nation we should have a bit less warming than the rest of the world.
In terms of the political end of my views (v’s the science end), my views are pretty simple:
1. I think we’re warming, there is a lot of evidence saying that and I would say that most of that evidence is correct. In some cases I think that some of that warming is overstated because there is a lot of pressure to find warming everywhere. This particular example was one that looked that way to me – most of the corrections going in one direction sounds a little like people trying to find warming. It may be legitimate, but it was a red flag.
2. I think that warming is largely human caused, CO2 and methane, with consequent water vapour increases.
3. I am sceptical about the models that show significant positive feedbacks, and therefore predict that the warming will increase substantially beyond the current straight line correlation between CO2 level and rate of warming. In some ways it seems counter-intuitive to me, and this isn’t something that has been measured, it is something that has been extrapolated using models.
4. I think we should do something about both reducing warming, and mitigating. In the reducing warming category, I very much dislike trading schemes, and strongly believe a carbon tax would be far more efficient and effective. In the mitigating warming category, I believe that it is unlikely that we will prevent further warming, so we may as well start planning for how we’ll live with a somewhat warmer climate.
Does all that make me a denier? (I know I just put words in your mouth there). Dunno.
In terms of this particular instance, my main motivation is that I had some suggestions and someone told me I was an idiot. (again paraphrasing). That’s always going to have the result of making me look harder.
Hi Paul
I also suggest a polite email to NIWA might help shed some light on your questions/data.
Regarding the 1 degree of temp rise, to me this doesn’t seem a big deal, for starters although global temp rise since 1930 is around 0.7 degrees (NASA GISTEMP record) there is significant regional variation in rates of warming.
Also the NIWA press release made it clear there has been a 0.7 degree rise in air temperatures over the ocean around NZ, since air temp over land increases faster than over oceans, this also suggests to me that 1 degree of rise sounds reasonable.
Also its not just models which show positive feedbacks (especially from water vapour). This is the area of climate sensitivity and there have been a large number of studies looking at the paleoclimate record to empirically calculate climate sensitivity.
The consensus is that positive feedbacks are real and the best estimate of a 3 degree rise for doubling of C02 remains unchanged.
If you can you could check out Annan & Hargreaves 2006, or Knutti & Hegerl 2008 for some recent papers on this.
Note to Richard: this got trapped by the spam filter, so it’s a bit late arriving on the site. Bryan
Paul
Advice on analysis of kelburn wind records/different rates of warming.
(which will reiterate partly what RW has been saying)
You will need to carfeully examine the site record for changes in the height of the anenometer, type of anenometer, and the method of recording the speeds. The automated logging method introduced around 1994 resulted in, I believe, a reduction wind speeds due to a sinplified way in which the instrument response curve was treated. So you will likely be seeing a higher frequency of days with winds < 5/ms over the past 15 years. The site also moved in 1966 and went from a Dines to a Munro at the same time and was raised from 13 m AGL to 20 m AGL. So you might be seeing more light wind days in the first 4to 5 years of your record.
Also, how much real development urbanization has there been near Kelburn since 1961? Most expansion in Wellington has been in the Northern suburbs past Johnsonville. There won't have been much to the south or in Thorndon. Also, it might pay to split your high wind cases into Northerly and southerly cases and reconcile with SST's (over at least some of the period – these should be there since 1993).
How about also finding out in which seasons Wellington is supposed to warm up the most, and is the warming more because of higher night time minumums or higher daytime maximums. Check to see whether your analysis is consitent with the predictions made there.
OK, some interesting suggestions in there, some of which will take more time than I easily have available.
On wind speed, I’m trying to construct a scenario where the change in instruments results in a decrease in the measured warming only on the windy days. How about this hypothesis:
1. Windy days are colder than non-windy days
2. New measurement approach reduces the measured wind speed, bringing some days previously regarded as windy into the not-windy category
3. Since windy days are colder than average, bringing some windy days into the not-windy category results in cooling in the not-windy category
4. And since windier days are colder, those remaining in the windy category are colder than they were before
Problem for me here is that both the windy and not-windy categories should now show cooling, and only the windy category does.
I note Whaleoil’s latest post (I’m guessing not many of you frequent his blog 🙂 ) with a photo of the Kelburn measurement site right next to the asphalt carpark. http://whaleoil.gotcha.co.nz/2009/12/07/climategate-niwa-got-some-splaining-to-do/
So whilst I give some credence to your hypothesis that changes in measurement approach have led to data that looks like UHI but isn’t, surely you’d equivalently have to concede that the big asphalt carpark next to the thermometer might give some UHI on windless days?
Want data?
plot your own temp graph
http://woodfortrees.org/plot/
Further to Richard’s remarks – there has been negligible development in the Kelburn area over the 40-odd year timespan. Regarding the 1994 changes, nonetheless the maximum gust records show little change after about 1988.
PaulL, you can analyse away for a few years – then come back with the breathtaking discovery that Wellington, like everywhere else in NZ , has indeed warmed.
Meanwhile some small issues with your hypothesis: mean temperatures are “day-night” ie max/min averages, not daytime ones. The statement that windy “days” are colder than non-windy ones is a gross oversimplification and downright wrong in some contexts. It depends on the wind direction and the season as well, and of course on cloud cover/insolation. October 1988 was by far the windiest month in the recent Wellington records, with gusts to 63kph or more on 29 of the 31 days. It was also a warm month. The period from about 24 June – 17 July 2009 was notable for low windspeeds and very cold temperatures. For your further edification, windspeeds at Kelburn have been shown to be higher during daytime than at night, and higher in the warmer months than in cooler ones. Northerly flow is by far the dominant direction, at a ratio of about 2:1 over southerly.
I suppose any observed warming at Wellington Aero, that well-known wind tunnel, is also influenced by UHI? Not to mention Chateau Tongariro, Campbell Is etc?
RW: no idea about Wellington Aero or Chateau Tongariro, I haven’t looked at their data.
I started with a hypothesis, identified a way to test it, and the results from that test did not disprove the hypothesis. They didn’t prove it either, but I’m a bit stuck. I still think it is reasonable to expect a correlation between UHI and windspeed – that UHI would be less visible on windy days.
I don’t understand why we think wind direction, vegetation or any of the other concerns people have raised would be likely to affect the result, as I haven’t identified a plausible hypothesis for how those things would impact the result. I thought the scientific method required more than just a correlation, it also required a plausible method of causation. For the UHI effect, I have both a correlation and a plausible causation. For the other confounding factors that people have suggested, I don’t yet have any plausible way that they could be causing it. I’m not saying there isn’t a plausible way, just that nobody’s explained it. I’m tempted to believe that it is people just believing what they want to believe, and retrofitting the data to it.
Read this PaulL.
http://www.skepticalscience.com/urban-heat-island-effect.htm
My final comment on this NZ temperature record stuff: even if you could establish some significant UHI effect at Kelburn (unlikely), nothing changes the fact that New Zealand and the surrounding ocean areas have warmed, by the amount claimed. Still angels on a pinhead stuff, I’m afraid. Goodbye.
I guess it’s nice to have an open mind.
I didn’t ever claim that NZ hadn’t warmed, and I think it likely that NZ has warmed.
Having said that, if all the 7 stations used in that data set did have problems with their temperature record, then that would change the fact that NZ had warmed. Because this is measurement of fact – if you can’t measure it then it didn’t happen. I am open to changing my opinion based on the facts, you appear not to be. I’m not sure that is science, I think that is religion.
Paul, I think the central point is that there is a lot of “prior art” in the examination of the sorts of things you’re looking at. Professional meteorologists (you know, the guys who do this for a living, who are required by their work to have a thorough understanding of the scientific literature on the subject) have been looking at this sort of issue for a long time, and in detail.
To discuss the issues at their level, you need to put in the work to build your understanding, so that your questions make sense. To ask them layman questions is a different process: it depends on them being willing to spend the time to impart their knowledge, and your being willing to take that knowledge on board.
That doesn’t mean that you can’t play with the numbers yourself — it might be fun, even educational — but doing so without first building your understanding of the subject means that any conclusions you may draw are unlikely to be robust. That’s also true, it has to be said, of the likes of Treadgold and the NZ CSC, who seem unwilling to do the learning required to make reasonable criticism possible. They prefer to make unreasonable criticisms, which are the subject of this whole affair…
Putting my cards on the table first (although, in scientific debate, it should not be relevant): I’m inclined towards NIWA’s analysis, and certainly convinced by AGW.
PaulL is adhering to good scientific approach here, folk knocking him for that do little other than display ignorance, and ‘faith-based’ argumentation, in cases introducing additional potential factors solely because his findings don’t fit with their ‘world view’ – I think that’s tragic – it certainly does nothing to promote their cause.
I look forward to further analysis, and perhaps rebuttal, from a scientific perspective, rather than what has been posted by his nay-sayers so far. I would also encourage PaulL to ask the questions of NIWA and|or Salinger.
My gut feeling is he’s on to something relevant – but that the effects will be ‘washed out’ by the averaging across sites. But my gut feeling is no more relevant than those who are dismissive of him.
In short, I’m somewhat disgusted by the faith-based dismissal of his findings, by those who would (presumably) argue that science is at the heart of the question. I am supremely disappointed that PaulL felt he had to set out his overarching opinions in order to ‘defend’ his data based research – and that I also felt that that would be ‘helpful’ at the outset – data is data, method is method, and the moment we start to analyse those based on an understanding or inference of an ‘agenda’, we do science a great disservice.