Guest essay by Tom Barr
Central England Temperature stations, long considered a benchmark, have been affected by land use change and urbanisation
The answer is blindingly obvious, from above: they are all subject to considerable modern local urbanisation immediately to the North, including heated greenhouses designed to replicate a Mediterranean climate.
The Met Office relies upon just three weather stations to record the Central England Temperature: Stonyhurst (Lancashire), Pershore (Worcestershire) and Rothamsted (Hertfordshire). http://www.metoffice.gov.uk/climate/uk/about/archives
The Met Office averages these three temperatures and makes a 0.2 degrees C negative adjustment to compensate for the UHI effect. http://hadobs.metoffice.com/hadcet/
The Met Office has made this adjustment since 1974. Bizarrely, they don’t seem to know exactly why 1974 was chosen other than being able to speculate that “It was probably around then that the effects of urbanisation began to be noticed, although this date as chosen may be slightly arbitrary”[1].
But is this 0.2 degrees negative adjustment adequate for the three CET weather stations? I suggest it may not be:
All three weather stations have material modern urbanisation starting within 100 to 200 metres in their Northern quadrant, this development being associated with the rapidly expanding research and educational institutions which control the land on which they are sited. These institutions are often built around old country houses with formal gardens sited to their South in the UK, so their commonality of orientation is unsurprising.
Rothamsted Research (f/k/a Rothamsted Experimental Station) now shelters its weather station from the North winds and warms it up by modern buildings rapidly thrown up to the NNE:
Stoneyhurst College has expanded in the NNW, and now hugely shelters its weather station from the North winds and warms it up with new building and nearby tarmacadam ball sport courts:
Pershore College has gone a step further and covered the land immediately to the NNW with heated greenhouses https://www.warwickshire.ac.uk/colleges/pershore_college/facilities/pershore_development.aspx , car parks and new building, having a similar effect on the Pershore weather station as its two sister sites. Surely a greenhouse designed to enable “plants to flourish in a semi-Mediterranean climate” sited up wind of a weather station may have some effect?
It is not beyond the wit of man to postulate that these localised, specific effects might artificially inflate the recorded temperature. Increasing the data to a temperature not properly compensated for by the small 0.2 degrees C adjustment applied by the Met Office, thereby inflating the CET to a number more amenable to the Met Office and the IPCC.
This would particularly warm data recorded in colder temperatures, where winds are more likely to be from the North.
The Met Office plans to
“overhaul” its CET datasets during mid-2017. Included in this “overhaul” is a review as “to what adjustments have been made (e.g. urbanisation adjustments, and changes to allow for differing climatological characteristics between the different sites used over the years so as to avoid introducing any inhomogeneities)”.[2]
Not before time. The CET problem, from a birds eye view, is staring the Met Office in the face.
[1] Met Office EIR Response dated 23 January, 2017.
[2] Met Office EIR Response dated 23 January, 2017.
Letters from Met Offce: 0052166-barr (PDF)
Addendum: Here’s the 1945 urbanisation at Pershore: virtually nil:
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Why not just use the Climate Reference Network for temperature records instead of the old, ad-hoc weather stations?
Because the CNR is a US weather station network, not UK.
Re the top post, who considers CET a ‘benchmark’? And a benchmark for what?
Experiment is still the most robust way to validate something. In this case an experiment could be conducted to establish any urban influence with reasonable accuracy. Mount (say) 10 accurate thermometers within a logical perimeter of the devise being tested where urban influence is less obvious. They are establishing regional mean temperature plus highs and lows. So, spacing could be as far as 20 km. One year’s records may be enough to establish any difference. Three-five years would be better.
It would also be possible to study the site history to try to duplicate the pre-development conditions .i e. set up the test apparatus in sites similar to that which dominated the historical record.
This ain’t rocket science and need not be that expensive.
Can you tell me why my comments are blockeed
They aren’t blocked, they are held for approval as a first time commenter. If we didn’t have such a safeguard, we’d have all people abusing the privilege to comment.
It makes me wonder what the scientists from the 1700’s and 1800’s that were diligently recording these temperatures would think about ‘climate science’ and its political fallout.
Bud,
Thanks.
The old-timers were doing their best – “diligently recording these temperatures”.
Technology has moved on.
Auto
You start by installing a weather station north of each site and seeing if the temperature difference is significant. If so, that is your compensation value.
I LOVE VALENTIA remote SW Ireland since 1880 – 0.42 degs C a century
v’
As someone who lives close to Stonyhurst it made me smile to see Tom refer to “…modern urbanisation…” when referring to the College. The majority of the College buildings have been around since the early 1800’s and are possibly some of the most striking school buildings in England. Although there has been some building works recently at the College (the Refectory and Weld House notably) they are on the North and North West side of the College. I assume that, if the College buildings have an effect on temperature readings, this would have been constant since the 19th century.
The College also sits in the shadow of Longridge Fell to the North which shelters it, to an extent, from northerly winds. There isn’t much development close to the College with most of the surrounding area being farmland.
You are right, Stoneyhurst college is virtually unchanged since 1894 – I checked on a site called “Old-maps” which has large scale Ordnance Survey maps going back to the late 19th Century.
https://www.old-maps.co.uk/#/Map/369182/438862/12/101394
The weather station is in a Stevenson Screen in a sheltered garden location surrounded by trees and walls. Tom Barr’s pin is spot on. In fact on Google maps aerial photo if you zoom in far enough you can see the screen as a white dot. A picture of the screen and surroundings is here:
https://www.flickr.com/photos/swilliams2001/4802168811/
About the only thing that would have changed in the last 120 years or so is the size of the trees you can see in the picture.
Just a small point, I thought one point of a Stevenson Screen was to protect the thermometer from the wind, so it eliminated wind chill from the reading. In that case this seems like a very good location for the weather station. Sheltered, rural garden location unchanged for probably 200 years or so.
Apparently the readings are taken by the school’s Classics master in his spare time. Nice to see a continuation of the “Gentleman amateur” tradition.
I bet they heat the school buildings better now. Even catholics are no longer so keen on mortification of the flesh.
(This is one of the top private schools in UK, founded by Jesuits)
The school is 580 ft roughly downwind (prevailing from SW) of the observatory and the thermometer is behind a Stevenson Screen. The pictures (and map contours) I have seen suggest that the observatory is behind a 6-9 foot rise in the ground between it and the school buildings, so there may not even be line of sight to the school. In which case there can be no convective or radiative impact from a slightly warmer school on the thermometer. Not much to worry about here.
That is, of course, provided they are using the full temperature history from the site in compiling the CET. Hopefully they are not stitching the recent figures on to the historic figures from the site it replaced.
The Pershore site seems to have much bigger issues.
As per Ralph above, Stoneyhurst College is a recent replacement for two egregiously terrible sites.
As I said, there is no problem provided it is a full substitution and they are using the all the records from the Stonyhurst site and putting the records from the “egregiously terrible” sites in the trash.
Roy Spencer did some research to find out how population an heat are related
http://www.drroyspencer.com/2010/03/the-global-average-urban-heat-island-effect-in-2000-estimated-from-station-temperatures-and-population-density-data/
Thanks.
So the Met’s 1974 adjustment is way too small.
A better adjustment could be calculated for each station.
Tiny Lebanon, KS, closest town to the geographic center of the contiguous 48 United States, has experienced no global warming during the interval of rising CO2 since the end of WWII.
https://temperature.weatherdb.com/l/14320/Lebanon-Kansas
Record high of 117.0° on July 24, 1936. August was also warmest that year. Only record highs for other months during the supposed late 20th and early 21st century warming interval were in Apr 1989, Jan 1990 and June 2012.
Record low of -32.1° on February 12, 1899, but also monthly low records for six other months since 1948, to include 1979, 1984, 1989 and 1997.
This is a Met office paper on urban heat, temps and CET related issues.
http://centres.exeter.ac.uk/cee/prometheus/uhi_paper_preprint.pdf
Seems to me they are well aware of any influence on their data.
We also know that the Berkley Earth project examined thosands of surface temp measurements to see if urban heat was biasing results. They concluded that no, it wasn’t.
http://berkeleyearth.org/faq/#question-15
“The Urban Heat Island effect is real. Berkeley’s analysis focused on the question of whether this effect biases the global land average. Our UHI paper analyzing this indicates that the urban heat island effect on our global estimate of land temperatures is indistinguishable from zero.”
(original contains link to detail)
Griff,
If you seriously imagine that urbanization has had no effect on central England since 1850, then you’ve never been there or even studied the history of industrialization. The population of Birmingham in 1851 was 232,638. In 2011, it was 1,074,300, who used a lot more energy each.
But then a commenter so paleoclimatologically ill-informed as to imagine that Arctic sea ice has never been lower than now since the Eemian will fall for any lie.
Gloateus
“But then a commenter so paleoclimatologically ill-informed as to imagine that Arctic sea ice has never been lower than now since the Eemian will fall for any lie.”
But then a commenter so paleoclimatologically ill-informed as to imagine that Arctic sea ice has never been lower than now since the Eemian will fall for any old carp spewed from a watermelon believer’s tonsils.
Does that clarify and help?
I hope so.
Auto.
Between 1971 and 2014 Philip Eden maintained a separate record of CET with the aim of more faithfully representing and continuing the record developed by Gordon Manley from 1659 up to 1971. Instead of the 3 sites used by the Met Office , Philip Eden used 2 sites in Oxfordshire and Lancashire to be analagous to that developed by Manley. There are small differences between the Met Office and Philip Eden versions of occasionally up to 0.4C on an annual basis but an average of 0.03C annually.
The fact that these two independent series agree so closely seems to me to be a vindication of the fact that neither series has been unduly distorted by the UHI effect
Between 1971 and 2014 Philip Eden maintained a separate record of CET with the aim of more faithfully representing and continuing the record developed by Gordon Manley from 1659 up to 1971. Instead of the 3 sites used by the Met Office , Philip Eden used 2 sites in Oxfordshire and Lancashire to be analagous to that developed by Manley. There are small differences between the Met Office and Philip Eden versions of occasionally up to 0.4C on an annual basis but an average of 0.03C annually.
The fact that these two independent series agree so closely seems to me to be a vindication of the fact that neither series has been unduly distorted by the UHI effect
At 1200 km radius per site (4.5 million sq km @ur momisugly), only 113 standardized stations would be required to cover the whole globe (510 sq km). But ideally a range of elevations and distances from the sea would also be required for a meaningful average of the absolute temperate of the planet.
Maybe someday such an automated network will exist. The North Pole is over water 4000 m deep.
It beats me how you can ever hope to get a reliable “average” surface temperature for the globe given the huge range of conditions at the surface (elevation, prevailing winds, vegetation, humidity etc etc. That is why the only meaningful / reliable readings for the recent past and future are from the satellites. Let’s hope Trump does not throw the satellite baby out with the NASA GISS bathwater.
UAH relies on state and other federal grants rather than NASA.
Yes, but who funds the actual satellites? UAH and RSS are just academic bodies interpreting the data that comes from them. I think NASA launches and manages the satellites themselves.
…and by average temperature I mean, of course, average temperature anomaly. I doubt any sane person would try and average absolute temperatures between the Gobi Desert and Antarctica!
IPCC has disappeared the previous heat record of 136 F from Libya in the 1930s, and doesn’t recognize the new Antarctica record low of -136 F. But they average to zero F.
Quite! Perfect illustration of why that is nonsense.
The alternative is to work out the change in the temperature over time (aka anomaly). And, as we all know, averaging that out over the whole globe is fraught with problems and probably has very little meaning.
Is a change of 1c in the average annual temperature on Greenland from. say, -10c to -9c going to have the same impact as +15c to +16c in a temperate zone. Who knows? At the moment these changes get a geographical weighting but no weighting that reflects their impact on the world’s climate.
The only exercise of any real value is to study different countries / regions / climate zones, see if temperatures are changing and what the effect is locally. Here in the UK we have warmed a bit, but not much and very difficult to discern any real impact on the way we live. Less snowy winters in the south, possibly, but that is no bad thing!
In the Pacific NW of the US, the winters are not quite as cold as during the cold interval (despite rising CO2) 1940-77, but the summers aren’t any hotter. A human footprint outside of local land use changes is not detectable.
The confounding of UHI and homogenization presents a sort of can-o-worms that people can argue about until the next ice age sneaks up on us all. Some years ago I read a PDF at a NOAA site file that summarized how USHCN data were adjusted for various affects, and besides being left with an impression of an inadequate treatment of UHI, I’d also swear that the process is one of homogenizing before correcting for UHI. I don’t think I have to say much about why this would lead to problems in final data. Now I suppose if a person believes that UHI has no effect on temperatures, as Griff claims above at 6:51 am, for the Berkeley Earth effort, then the order of corrections does not matter. But if UHI effects are that small there isn’t much point in correcting for them at all.
Lately I have read several people on this site suggesting that a good measure of non-warming comes by way of a comparison of dates for record low and record high temperatures. I looked into this back on January 6th when the Laramie Airport Station (KLAR) reported -40F, but for days at NCEP the record low for the U.S. was listed as -13F at Wisdom, MT. I was puzzled and sent a request to NCEP (or NOAA) for information about how record temperatures were determined. I never got a reply, but many days later I saw that the record for the 6th had been revised to -48F at Walden, Co. There is a small note on the NCEP page that implies the record is selected from reports received as of 6 UTC on the date in question which implies in turn several problems with record temperatures: they have to be actively submitted, they are often from cooperative observers (you can read records that report location as one mile NW of Gunnison, or something like that); and, they must have time-of-day biases.
There seems to be no surface records truly clean enough to speak credibly of record hot or cold, or to measure average earth temperature to the precision people claim.
Hey – go argue with Berkley Earth.
I am pretty sure I linked to a page which links to their findings on that.
To Tom Barr, Philip Eden, Paul Homewood and any whom it may concern:
For the last two years I have been recording 3 CET-related temperature series from the Met Office website (to thenearest 0.1 degree), to see if I can replicate their figures and also detect their UHI adjustment. But Stonyhurst data are not available daily, so I have to use a nearby proxy. I expect my proxy to be slightly warmer, because it is south of Stonyhurst and also more likely to suffer from UHI. Therefore after they have subtracted 0.2 the official CET minus mine should be less than -0.2.
Here are the mean monthly differences (CET minus mine) starting in November 2014 and ending in December 2016:
bias = c( 251, -18,
4, 27, -46, -314, -131, -188, 54, -50, 82, 89, 73, -155,
74, -97, 106, -18, 41, -121, 102, -117, 115, 263, 114, 240)/1000
The mean is 0.015 degrees (amazingly small really).
This is not less than -0.2.
I expect you may be as shocked as I am…
Rich.
K. Kilty on January 30, 2017 at 9:47 am
As is explained by Berkeley Earth, UHI exists but is such a tiny phenomenon that it has no really perceptible effect on temperature, especially worlwide.
A simple way to get convinced of that is to separate the GHCN unadjusted record into two subsets:
– the one produced by all GHCN very rural stations (type “R” AND with least nightlight factor „A“);
– the rest.
Even for the world’s region one might expect to be most affected by UHI (the CONtiguous US), the differences
http://fs5.directupload.net/images/170130/d95pto4b.jpg
(1) probably won‘t so really look as expected by many people between 1880 and 1979,
and
http://fs5.directupload.net/images/170130/wv47lgi2.jpg
(2) are indeed relatively small since 1979 (in blue you see in addition UAH6.0 for USA48 aka CONUS).
And while the GHCN CONUS rural (349 stations) has a slightly negative linear trend (-0.04) when starting by 1880, its trend since 1979 (0.15 °C / decade) is higher than that for the nonrural rest (1493 stations, 0.11 °C / decade).
Source: ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v3/
Done a little data strangling of my own just now. CET change since 1659 is roughly +0.0027c a year on average (a rate of 0.27c a century). 1950 to the present, covering the period when CO2 emissions have taken off, it has been rising at a rate of +0.0157c a year (+1.57c a century).
However, most surprisingly, during the first 16 years of the 21st Century the CET has been falling at a rate of -0.0209c a year (-2.09c a century) yep, national cooling! So the “pause” has definitely been a real event in Central England.
TLM Useful post – thank you
TLM on January 30, 2017 at 1:55 pm
So the “pause” has definitely been a real event in Central England.
Certainly TLM… and by far no only there.
Recently I computed the trends for all 7,280 GHCN stations from their respective beginning to their respective end. Your message motivates me to perform a global selection therein for 2000-2016.
Here’s a terrible thought: Suppose a world war broke out that caused lots of bombs to go off in the fighting zones of the world, or worse, some really big weapons of mass destruction came into play. How would this affect our ability to trust global average temperature at all ?
A radical idea might be to admit that a number simply does NOT exist to gauge any such measure. The only sure way we know that the globe was heating abnormally would be if most of its surface experienced a simultaneous huge increase in temperature.
nuclear weapons?
The effect of letting off a lot of those has been modelled/extrapolated from the tests…
I often wonder what effect the many burning cities 1939-45 might have had on climate -much smoke and soot. Got to be something like a volcano in terms of scale of effect????
Anyone know if anybody ever looked at that?
Of course it has been studied, and results misused by “nuclear winter” polemicists like the astronomical Dr. Sagan.
Re the top post, who considers CET a ‘benchmark’? And a benchmark for what?
Hundreds of weather enthusiasts all over the country
[snip]