Is Armagh Burning?

Guest Post by Willis Eschenbach

Anthony has highlighted a study by Coughlin and Butler. Their study says that there is little or no urban warming (urban heat island, or UHI) in the temperature record from the Armagh Observatory in Ireland. They say:

It is concluded that temperature observations made at Armagh Observatory have been unaffected by rapid urbanisation over the past three decades.

Why is Armagh important? And is there really no UHI in Armagh?

The Armagh record is very valuable because it is one of the longest well-documented temperature series in existence. Here is the monthly mean temperature record from Armagh. (NOTE: I have replaced the earlier Figures 1 and 3, which only went up to the year 2004, with updated figures which now include 2005-2010. My thanks to those who wrote in with the location of the post-2004 data.)

Figure 1. 209 years of monthly temperatures at Armagh, Ireland. Pale blue is monthly surface air temperatures. Dark blue is Gaussian average of the temperature. Photo is noctilucent clouds over Northern Ireland.

My conclusions from Figure 1?

1. First, one single temperature station says nothing about the temperature of the planet. However, this one says a lot about century-long temperature changes in the North of Ireland.

2. The most striking thing to me is the slow regularity of the two-century-long temperature trend. Yes, there are decadal swings. But they don’t stray far from a simple trendline.

3. The recent warming from ~ 1980 on is not particularly unusual or anomalous compared to earlier periods of warming. From this, however, we can’t tell if there is a heat island signal in the record.

4. The Armagh data shows the same 0.6°C temperature trend over the 20th century that is shown by the global record. It also shows the same features as the global record, warming to the late 1940’s, cooling for thirty years or more, recent warming.

5. There is no sign of any acceleration, and indeed little change at all, in the long slow two centuries of warming.

Oddly, the Armagh Observatory data does not form part of the GHCN dataset that is used by all parties to create global temperature datasets. But I digress. Onward to the UHI.

First, some terminology. “UHI” stands for “Urban Heat Island”. Bad name. There’s lot’s of heat islands that are not urban. Trees, changes in the vegetation of the site, hedges, all of these can cause heat islands. I prefer the term “LHI”, for “Local Heat Island”. I know, I’m swimming uphill, so I call it UHI like everyone else does. But remember it doesn’t have to be urban.

The question of whether Armagh contains a heat island signal is an important one. Casting around for a way to determine the amount (if any) of heat island signal in the Armagh data, I decided to look at the relationship between Armagh temperature and the sea surface temperature (SST) of the North Atlantic and the Irish Sea. I reasoned that for an island on the edge of the North Atlantic, the SST would determine the land temperature. Here are the areas I used to see if my reasoning was correct:

Figure 2. Areas of ocean used for the comparison with the Armagh temperatures. Armagh Observatory is at the center of the yellow house. Left gridsquare is the North Atlantic area. Right gridsquare is the Irish Sea area.

I took the anomalies of the HadISST sea surface temperatures for each of those areas, and of the Armagh temperatures. Here are the results:

Figure 3. Temperature anomalies around Ireland. Monthly averages have been removed. Note that the vertical scale is different from Figure 1. Pale colored lines are actual monthly anomalies, heavy solid color lines are Gaussian averages. Sea surface temperatures (SSTs) are HadISST temperatures from KNMI.

Now, that’s pretty interesting. My observations, in no particular order, are:

1. As I suspected, the ocean temperature around the island of Ireland determines the Armagh temperature. The island is ruled by ocean winds and breezes.

2. The Irish Sea and the North Atlantic temperatures are quite similar. This increases confidence in the precision of the sea surface temperature data.

3. As you would expect, the swings in the land temperature extremes are greater than those of the sea surface temperature.

4. From 1900 to 1986, the averages of all three records are generally all quite close to each other. I always like seeing such a close correspondence of two entirely separate and discrete natural records. It increases the confidence in both datasets. In particular, the wiggle-match between the North Atlantic (heavy red line) and Armagh (heavy blue line) is quite impressive.

5. From 1986 onwards, the Armagh and the ocean datasets diverge in a significant manner.

6. The size of the divergence from 1986 to the end of the record in July 2010 is about a degree.

The Coughlin and Butler paper says:

The grounds surrounding the Observatory and its climate station have remained relatively unchanged over the past 200 years. However, in that time, the town of Armagh has spread in several directions, including to the north and east, past the Observatory site. Much of the development around the site has been in the form of housing built over the past 20-30 years and this development still continues.

Does this mean that Armagh is showing urban or site-specific warming over the last quarter century? I don’t know. But I find it mighty suspicious that after 85 years of running right in sync with both the North Atlantic Ocean and the Irish Sea, the Armagh temperature should suddenly strike out on its own towards new heights, just when the town starts building up around it.

As a result, I’m not prepared to agree with Coughlin and Butler that there is no UHI signal in the Armagh data. They say:

However, recent research into the historical temperature records and comparisons with present day data from rural weather stations indicate that any temperature differences which existed between the Observatory site and the countryside 20-30 years ago have not increased over the intervening years.

Comparison of Armagh with ocean data, however, clearly shows increasing temperature differences in the exact time frame which they have used in their paper to discriminate a valid signal.

My regards to all,

w.

PS – I can’t find any Armagh data after 2004 … does anyone know where it might be available? (Solved, thanks to those who wrote in.)

Advertisements

  Subscribe  
newest oldest most voted
Notify of

Willis,
I like your assumption about sea surface temperatures determining the temperature. The same of course must apply to New Zealand where we have raw temps showing no warming and NIWA’s adjusted” temps showing over 1 deg C warming in the last 100 years.
Does anybody know what the sea surface temps have done over the last 100 years in the Tasman and South Pacific around NZ?

Interesting analysis Willis.
It seems that they missed what you found, does that mean that your analysis invalidates their in part or in whole paper? If so then it seems important to get your analysis published as a peer reviewed paper.
Keep at it.

I suppose it just goes to show how problematic getting genuine, reliable land surface data sets is. If an apparently “unaffected” site is affected by nearby urbanisation rather than direct urbanisation then what hope is there for all the more obviously affested sites at airports etc?

Willis,
Current data at Armagh: http://climate.arm.ac.uk/aws.html

stephen richards

to find all the data you have to pick it from the site bit by bit. It is laborious, cumbersome and time consumming. I emailed Armagh last year for the later data and although we exchanged several very polite emails it seems unlikely for now that they will make it any easier.

LDLAS
richard telford

I think hedges would better be described as a microsite issue than a catch-all LHI (not all microsite issues would give a positive temperature)
Decomposing the temperature into max vs min temperatures, or windy vs calm days would give more insight into any possible UHI at Armagh.
recent data at
http://climate.arm.ac.uk/scan.html

Sue Smith

The Met Office has monthly data for Armagh from 1865 to date.
http://www.metoffice.gov.uk/climate/uk/stationdata/armaghdata.txt
I don’t know how it compares to the data actually held at Armagh.

Nick

Armagh may jump above the SSTs from the mid-1980s,but they both show a similar trend increase from that point. Is that UHI bleeding over the sea?
Need to compare Armagh with nearby genuinely rural sites.

Here is a link to updated monthly temperatures at Armagh:
http://www.metoffice.gov.uk/climate/uk/stationdata/armaghdata.txt
Try this one for daily data:
http://climate.arm.ac.uk/scan.html
I concur that the conclusion by Coughlin and Butler is a little premature.
Having a look at the 360° panorama, just from above the met station, you can see a lot of trees.
How they used to manage that vegetation in the past?
Surely, in Europe there are a lot more trees in these decades then before.
Some old photographs are required!

Phillip Bratby

“I reasoned that for an island on the edge of the North Atlantic, the SST would determine the land temperature. ” Anyone who lives in these fair isles knows that to be the case. And the further west, the more the climate is dominated by the Atlantic. The further east, the more the continental effects become apparent. If it weren’t for the warming effects from the Atlantic, these isles would be pretty well uninhabited.

Peter Dunford

I wonder what caused the spike in the 40’s. Haven’t found much info on google.

Tenuc

We know UHI exists. We know the town of Armagh has spread in several directions, including to the north and east, past the Observatory site. Why wouldn’t this show in the temperature record???
Well done Willis, for an elegantly simple refutation of the Coughlin and Butler study.

David

Armagh already has a significant place in Climate Science because of the Butler and Johnson 1996 paper and its use by David Archibald and others as part of their evidence of a link between solar cycle length and intensity.
Some have suggested that this possible link has become less persuasive in recent years making it doubly important to make sure that its recent record is both available and uncontaminated.
Why on earth isn’t it in the GHCN data set?

Mooloo

Surely the point of showing a lack of UHI at Armagh is to “prove” that the recent warming in CO2 induced, and not an artifact of UHI.
Why anyone interested in proving CO2 warming would go anywhere near Armagh is a mystery to me. I would think they would keep that particular record quiet.

HR

I did my own 25 year smoothing on the armagh data (somebody can tell me if that’s valid or not)
http://i38.tinypic.com/2ewply8.png
And my scale is different to yours. It seems to show a good low frequency oscillation superimposed on a warming trend that seems to stretch back to the start of the record.

John Ballam

Anthony,
Clearly your analysis is different to the original, in that you compare against sea temperature rather than surrounding rural temperature, but it is interesting none the less. My question is this – when sound arguments like this can be made, backed up with hard data, why do they seem to only get published in blogs? If the argument is sound, surely it would be better to turn it into a full-blown paper and submit it to a journal. Wouldn’t that counter the argument that “skeptics never do real research?”
John

HR

The post 2004 data is there but in raw form
http://climate.arm.ac.uk/scans/

Tom Roche

Living in Ireland, one of the factors I feel is relevant over the last 30 years has been the predominance of westerly winds. It has been almost an exclusive feature of our weather, this is an additional factor in warming, try adding data on same to the mix and see if it accounts for the discrepancy.

Reference

Willis,
Does this help?
Data up to July 2010
http://climate.arm.ac.uk/scan.html
See also source page http://climate.arm.ac.uk/

Andrew P.

Another excellent post Willis. The only comment I would make is that rather than the ocean determining the Armagh / Irish land temperature, I would say it heavily has a heavy influence. I say this because while the seas around Ireland (and in my case Scotland) act as a buffer and moderate extremes, it is the wind (and more its direction) which is still the key determinant. (e.g Dec 2009 to mid February 2010 when the North Atlantic was still relatively mild, but north and north-east winds, (and long periods of still conditions) resulted in us having the coldest winter since 1963, (and the coldest December and January since 1914). i.e. it’s the jet streams which determine the weather we get, and the Atlantic which moderate it. (And that’s the key reason why the Met Office were/are so bad at their long range/seasonal forecast – they have invested in models which play around with the symptoms rather than the causes – i.e. the AO and jet stream patterns.
Back to the specifics of Armagh, I agree with one of the commentors on Anthony’s post, that the site will be exposed to noticeable UHI when the wind is coming from the adjacent urban area, at least in the winter months.
One last point (which I think has been made by Tallbloke previously) about the CET and Armagh records; they would be much more useful indicator of the long term temperature record (for the northern Atlantic/Hemisphere), if they had corresponding wind direction and duration data. Who is to say that a year with a temperature anomaly of +3 was actually that warm – it could be that there were far fewer north winds than normal that year, which would have brought down the ‘local’ average.

TinyCO2

Willis E. Says:
I don’t know. But I find it mighty suspicious that after 85 years of running right in sync with both the North Atlantic Ocean and the Irish Sea, the Armagh temperature should suddenly strike out on its own towards new heights, just when the town starts
building up around it.
Henry@Willis
I am puzzled that I see the same or similar upward trend at between 1940 and 1950. What could have been the cause of that?

kadaka (KD Knoebel)

PS – I can’t find any Armagh data after 2004 … does anyone know where it might be available?
http://climate.arm.ac.uk/scan.html

Monthly Records
This section allows you to select, from the databank, an image from the original record book containing the raw meteorological data for any given month between December 1794 and June 2000 but excluding April 1795 to June 1795 and June 1825 to December 1832.
Data from July 2000 until July 2010 is available in digital format rather than as a scanned image.
The data is not corrected for instrumental sensitivity, exposure of the instruments or the time of observation. It is contained in monthly tables of observed data with some general comments on the weather by the observer.
(Calibrated data is available here.)

AJB

Maybe a comparison with Valentia Observatory would be useful.
51° 56′ 23″N, 10° 14′ 40″W.

Christopher Hanley

I don’t understand why the 200 year Armagh temperature record, in conjunction with the 350 year CET record, cannot be treated as a reasonably accurate proxies for the NH or even global temperature, particularly because they are from a predominantly maritime climate.
http://www.junkscience.com/MSU_Temps/CETvsArmagh_long.html

John R. Walker

There are now manual and automated weather stations on the Observatory site
http://climate.arm.ac.uk/aws.html
There is data up to March 2010 in the archives – even though the recent annual folders are all dated 21 April 2004. The data within is more recent.
http://climate.arm.ac.uk/archives.html
http://climate.arm.ac.uk/scans/

Isn’t the small part of Atlantic too small? When comparing annual Armagh record with the whole N Atlantic [0-70N 280-360E], the agreement is much better.
http://i38.tinypic.com/10nx66w.jpg

Vince Causey

Interesting conclusion: “However, in that time, the town of Armagh has spread in several directions,. . . Much of the development . . . housing built over the past 20-30 years”.
But “Comparisons with . . . rural weather stations indicate that any temperature differences which existed between the Observatory site and the countryside 20-30 years ago have not increased over the intervening years.”
So, although there is massive and ongoing development around Amargh, this has not contributed to any warming. This begs the question: are these rural weather stations themselves free from UHI? One would like to Google satellite them at the very least, since some so called rural stations turn out on closer inspection to be not quite as rural as one would have thought.

Leon Brozyna

So, over an eight month period, from Feb to Oct 96, they found the Observatory temps to be warmer than the test sites — Tmax at .1°C warmer and Tmin at .4°C warmer. Slight but real. I wonder what they’d find if they repeated the study now, 14 years later. Especially when you consider that during the period of the study the 6 yr Gaussian avg of Armagh was fairly close to that of the SST values; now there’s more of a spread.

I’d have to agree that Armagh may be UHI affected now. Although the Ovservatory is still ‘connected’ to the farmland the fingers of the town are spreading around it and it is to the Northeast of the town centre. Prevailing winds tend to be Southwesterly so it would be interesting to look at temperatures in relation to windspeed and direction. I believe also that rural areas with regards temperature may be most sensitive to anthropogenic effects in the 2,000+ population band (for which I want to chase up where I read this statement); the population of Armagh is now ~15,000. although it is not a very compact town.
I did a quick comparison using Google Earth – identical images below from 10km altitude, but with the Nasa Earth City Lights overlay (second link) the observatory lies in the midst of the bright area:
http://diggingintheclay.files.wordpress.com/2009/11/armagh-from-10km.jpg?w=632&h=422
http://diggingintheclay.files.wordpress.com/2009/11/armagh-from-earthcityllights-overlay.jpg
Note this overlay is different to that used by GISS for its nightlights adjustments, but I’ve done regular visual comparisons of the Google overlay with the GISS values for sites.

Paul_K

Hi Willis,
Your analysis is suggestive, but leaves open the possibility that the different atmosphere-surface coupling over land and sea is giving rise to the temperature differences. If the recent recorded temperature change was in fact (a) real and (b) due to an increase in downwelling LW due to GHGs, then wouldn’t you expect to see terrestrial temperatures climbing faster than SST? Have you come across any studies of coastal temperature stations to see whether there is a systematic difference in trends between land and sea in such stations? On a separate point, since the original study is within satellite era, is there any chance of persuading Dr Roy Spencer to download the AMSU TLT anomalies in the nearby gridcells to see whether they show an “averaging” of the terrestrial and marine temperature trends over the Armagh region?

Once I compared the MSU trend since 1979 in the 2.5×2.5° Armagh grid with the Observatory record, and while individual monthly anomalies were not exact, the overall trend was exactly the same.

Ed Fix

When I read the Coughlin and Butler paper, I was struck by their conclusion that even though Armagh had grown past the observatory in two directions, the city had left a “finger” of open land reaching from rural countryside to the observatory, and had therefore spared the observatory from UHI effect. Their reasoning seemed to be that it is conceivable that the observatory had been spared from UHI, therefore it was unaffected.
In other words, wishful thinking.

I meant to say also I found figures that population density of Armagh County has increased 20% between 1981-2009 (73.5 to 87.6 persons per sq km). I haven’t been able to find population figures for the town for that period – yet anyway.

Andreas

O, there I learned something new and exciting, thank you for that bit of writing.
My teacher in forest-meteorology always mentions that the temperature series from Uppsala here in Sweden is the longest and best temperature series in the world, maybe he’s a bit patriotic but what do I know. The Uppsala series would maybe show the UHI-development quite well i believe since Uppsala been a “larger” city for many hundreds of years.

Verity Jones says:
August 29, 2010 at 4:28 am
Reply;
Wow the observatory must be more concerned over light pollution, than UHI background.

JohnH

No only will there be the extra heat from the new housing but the % of homes with central heating in the UK has increased over the last 30 years too. Extra commuting too but offset by better home insulation.

BBk

John Ballard:
“If the argument is sound, surely it would be better to turn it into a full-blown paper and submit it to a journal. Wouldn’t that counter the argument that “skeptics never do real research?””
Sure, if you can get past the gatekeepers at the journals. ClimateGate made it pretty obvious that you’re working uphill at best.

Natural trend in the temperature’s anomaly (excluding urban factors) anywhere in the British Islands should not vary to any significant degree, although absolute values may.
There is a good reason to assume that apparent existence of a N. Atlantic precursor may signal a significant temperature drop in the next decade.
http://www.vukcevic.talktalk.net/CETnd.htm

latitude

They found that Armagh was warmer day and night, than the test stations.
What am I missing here?
“The mean difference in daily maxima between the Observatory and the mean of the three rural stations is found to be 0.11oC, whilst the difference in minima is 0.41oC,”
0.11C warmer day time, and 0.41C warmer night time
Wouldn’t that be UHI?

Sandy

It is worth pointing out that the SSTs used as a control may also have some Mann-made warming.

Berger

Some more questions:
what do we know about the station history, the thermometers used (for how long?) and beeing changed (when?), methods (daily time of measurment) etc.

Gary Pearse

Hmm it took Butler et al (The Met Office) to do an elaborate multi-year experiment with new installations to arrive at conclusions that a casual reader of the paper could see were wrong ones. Willis you did an unequivocal experiment with existing data in hours to show what was really going on – a much better approach than simply arguing with the authors about their conclusions. Dr. Butler and colleagues, this is what happens when you “know” the answer before you do the experiment. When I got to their discussion and conclusions I expected they would state the conclusion that their work led me to – yes there is an apparent UHI effect of about half a degree.

Robin Kool

That’s a great article.
You give the facts and you walk us through your reasoning.
Thank you.

BarryW

I wonder, did they run a temperature transect of the area? Additionally, did they check for MHI (Microsite Heat Island)? A few changes, such as a parking lot upwind or AC in the area would be obvious culprits.

Verity Jones says: August 29, 2010 at 4:28 am

…I believe also that rural areas with regards temperature may be most sensitive to anthropogenic effects in the 2,000+ population band (for which I want to chase up where I read this statement)

Try this and this – I still wonder if these Eschenbach-like sensible Russian scientist’s assessments from the Heartland conference have not been equalled.

HR says: August 29, 2010 at 2:55 am

I did my own 25 year smoothing on the armagh data…

I think your graph shows nicely the two factors Akasofu talks about (a) a 60-year cycle (b) overlaid on a steady rise out of the LIA. Note also, if one assumes this to be the case, the trend line would be flatter (start- and end-points of the trendline are not at the maxima or minima).