Of Myths and Met – “Gang aft agley”

Image from “How to Lie With Statistics” Darrell Huff.

This represents my 500th posting since restarting the Surface Stations Project in August 2024. Here I will highlight some of the very strange and often inexplicable findings concerning how observations are recorded ranging from absurd errors to the seemingly contrived and the simply plain wrong.

RECORDS

The Met Office provides a guide to “Climate Extremes” which solely lists specific examples of “Weather”. Semantic interpretation on my part? Possibly, but how accurate are the details? This example for Scotland:

UK climate extremes

Temperature by country 

Highest daily maximum temperature record

Country	Temperature	Date	Location

Scotland

34.8°C

19 July 2022

Charterhall (Scottish Borders)

This should be fairly straight forward – what could possibly be controversial about a maximum reading? Well rather a lot in reality. Floors Castle official Met Office weather station did record higher that same day (35.1°C) but its reading was discounted on the grounds of the site somehow not meeting “standards” for record attribution. But was this defined as inaccurate reading “expunged”? No it was duly archived and is still there to this day – not considered good enough to be a “record” but good enough to be recorded.

As I regularly point out, it is only through a very “deep-dive” (a popular Met Office term) that I uncover these errors – a modern day student is unlikely to ever question such things and would accept authority statements. The absurdity here is that when I pointed out this error of inclusion to the Met Office and queried why it was archived (under Freedom of Information request IMS0292131 18/5/2023), they refused to respond deeming me “Vexatious“. The known and confirmed as inaccurate reading has NEVER been corrected. False data, which is what it must be considered, is proven to knowingly remain on file and this practise appears to be deemed acceptable.

The Charterhall reading itself is even more debatable. A station out of action with no readings for the preceding 17 days suddenly produces a record from its very first reading back online? A site which normally reads 12 hourly drops down to only 24 hourly readings on the 29th June followed by no readings at all from the 2nd July and then hey presto the record.

This Met Office MYTH of expunged readings is by no means unusual. Continuing with that same page of weather extremes.

Highest daily maximum temperature record

Country	Temperature	Date	Location
England	40.3°C	19 July 2022	Coningsby (Lincolnshire)

The Coningsby reading has always been regarded as controversial given its location and events noted at the time – so I shall add a bit more. These are the archived readings at nearby RAF Waddington for the same day.

So two sites recorded that “record”? Well no because as I was advised by email in Frebruary 2025.

“At that time Waddington was having its temperature data marked as suspect as the grass under the screen had been treated with weed killer.

Kind regards,

Weather Desk”

Thus we have an open admission (whether the Met office likes it or not) that “suspect readings” routinely enter the national archives – known false data. However, consider that “grass treated with weed killer” reasoning – below is an image of the Coningsby Stevenson screen taken the day after the record on 20/7/2022 from the official report. Temperature readings should be taken over grass but there is none underneath the screen. The area bears a remarkable resemblance to having been affected by weed killer and is nothing remotely similar to the surrounding grass. A further reason, that is demonstrably stated by Met Office to warrant suspicion of the record.

And then more topically there is the truly bizarre admission on the part of the Met office that they no longer operate on averaged readings over a 60 second period as stated in their own operating manual.

At the Class 5 atrociously sited Lingwood station (which if it were a private weather station claiming a record would be totally ignored as unsuitable by the Met Office) the Met Office simply accepted a single pulse reading as the all time June high of 38°C.

“The original reading of 37.7C was taken when the HMT display was in T-trend
mode. T-max/min mode gave a recorded maximum of 38.0C.“

Whatever next? Will pointing one of “Covid specials” at the inside of a Stevenson Screen soon be deemed an acceptable reading in future?

For all those readings hailed as records for the day at Cardiff:Bute Park, and Wiggonholt and Eastbourne in 2022 – are these now to be removed as “suspect” given the admissions of being unsatisfactory. I will certainly be checking whether or not they are ever archived and given the veneer of accuracy – good enough to record.

Rejected figures

Figures are first reported as ‘provisional’ to enable this considered verification process to take place, often utilising further data not available at the time of original recording. If this process cannot affirm the expected levels of confidence during the analysis the figures are rejected.

Such is the rigour of this verification process, some sites have been discounted from providing new records on this occasion. A broken slat on the Stevenson Screen at Wiggonholt invalidated the readings from that site and remedial work is now underway.

The site at Eastbourne was also discounted as the grass was too long and nearby interference at the site didn’t meet usual high standard and so this data was excluded from the daily extremes.

In addition, the recent temperature figures reported from Cardiff Bute Park have also been rejected following a detailed review of the site, which found the station no longer meets the standards required for national temperature records.

And as for all those proclamations of new all time “National Records” at Kew?

These readings were taken from a site that Dr Eric Huxter finally got them to admit was really Class 5 unreliable Junk from which records should not be verified.

Kew Gardens CIMO Site Class Update

Which brings me to issue of:-

Quality Control

Whether or not Met office temperature readings are accurate when taken, they are all, allegedly, subject to “quality control” before (and after) archiving. Any readings that look dubious should be excluded but I shall further demonstrate how this is often an optional exercise. I can produce enough examples of frankly bizarre readings being archived to fill tens of reviews but for this exercise just a few to demonstrate the point.

Firstly a quick recap of how temperature readings are archived from different types of instrumentation/observation protocols.

The Meteorological Day runs from 09:00 one day to 09:00 the next. Manually observed sites take daily readings at 09:00. Automatic stations take multiple readings often now continuously throughout the day – potentially 1,440 minute by minute. These multiple readings are archived twice daily with the highest and lowest of these readings being noted at 21:00 and similarly at 09:00.

As the Meteorological day is different to the calendar day, the daily attribution is as follows (this table generated by GROK AI, DLY3208 are Manual readings, AWSDLY are automatic ones)

Attribution used

Reading	Maximum belongs to	Minimum belongs to
DLY3208 at 09:00, 24-hour	Previous calendar day	Observation date
AWSDLY at 09:00, 12-hour	Previous calendar day	Observation date
AWSDLY at 21:00, 12-hour	Same calendar day	Following calendar day

Consequently:

• A calendar day’s AWS maximum comprises the 21:00 reading that day and the 09:00 reading the following morning.

• A calendar day’s AWS minimum comprises the 21:00 reading the preceding evening and the 09:00 reading that morning.

• The DLY3208 value is already the corresponding 24-hour result: maximum taken from the following morning’s report, minimum from that morning’s report.

So here is an extract from when a station (Dalwhinnie No 2) was in the process of conversion from Manual to Automatic readings and both systems were simultaneously operating. These should record the same readings being in the same place at the same time. First the archived readings for anyone to check and confirm – column “I” is maximums, column “J” is minimums.

And this is how both I and AI interpreted just the minimum readings for these days.:

Using the same attribution rule, the AWS minimum for a calendar date is the lower of:

• the AWSDLY value reported at 21:00 on the preceding date; and
• the AWSDLY value reported at 09:00 on the date concerned.

The difference below is DLY3208 minus reconstructed AWS. A negative difference means the manual DLY3208 minimum was colder.

Assigned date	Previous 21:00 AWS	Current 09:00 AWS	Reconstructed AWS minimum	DLY3208 minimum	DLY − AWS
24 Sep 2003	Not visible	1.7°C	Incomplete	1.5°C	Incomplete
25 Sep 2003	9.7°C	9.4°C	9.4°C	2.5°C	−6.9°C
26 Sep 2003	10.0°C	5.5°C	5.5°C	5.5°C	0.0°C
27 Sep 2003	6.1°C	0.7°C	0.7°C	0.9°C	+0.2°C
28 Sep 2003	5.8°C	2.0°C	2.0°C	2.0°C	0.0°C
29 Sep 2003	6.5°C	8.1°C	6.5°C	3.8°C	−2.7°C
30 Sep 2003	7.6°C	−0.3°C	−0.3°C	−1.9°C	−1.6°C
1 Oct 2003	2.9°C	0.1°C	0.1°C	0.4°C	+0.3°C
2 Oct 2003	2.5°C	−0.7°C	−0.7°C	−0.6°C	+0.1°C
3 Oct 2003	7.1°C	7.8°C	7.1°C	−2.9°C	−10.0°C
4 Oct 2003	6.3°C	2.7°C	2.7°C	3.0°C	+0.3°C
5 Oct 2003	0.8°C	0.6°C	0.6°C	0.7°C	+0.1°C

The conspicuous cases are:

• 25 September: DLY3208 is 6.9°C colder.
• 29 September: DLY3208 is 2.7°C colder.
• 30 September: DLY3208 is 1.6°C colder.
• 3 October: DLY3208 is an extraordinary 10.0°C colder.

The other dates agree within 0.0–0.3°C, which is consistent with ordinary sensor, exposure and rounding differences.

The 3 October result is particularly striking: both AWS half-periods were mild —7.1°C and 7.8°C— yet DLY3208 reported −2.9°C. The minimum therefore cannot have occurred within either AWS interval as represented in this table. It is a strong candidate for the carry-forward or observation-boundary effect under discussion*, although checking the precise DLY3208 reset and attribution rules would still be necessary before treating that as proven.

I further asked AI to verify my own calculations of the effect this would have on mean minimum temperattures for the examined period.

For the 11 days from 25 September to 5 October 2003 inclusive:

Dataset	Sum of daily minima	Days	Average minimum
DLY3208	13.4°C	11	1.22°C
Reconstructed AWSDLY	33.6°C	11	3.05°C

I must emphasise that the above figures have been subject to Met Office Quality Control procedures both before and after archiving. This clearly demonstrates how changing the observations /archiving protocols can give remarkable discrepancies that nobody seems to notice.

So am I just cherry picking occasional dates and exposing freak errors? Certainly not – I could demonstrate many of these and I genuinely do not spend that much time scouring millions of readings just in case I find an occasional mistake. Perhaps, though, the best way to demonstrate this point is where my comparisons have been raised with the Met Office themselves and their own responses.

In my report on the Cambridge Convolution I noted significant differences in minimum readings between the two sites in close proximity. The example I chose was randomly picked out of many as 1/2/2019 almost 6 years prior to my post. Subsequently, in March this year a then new Talkshop reader (and Cambridge resident) raised their concerns directly with the Met Office. They wish to remain anonymous but have forwarded me their complete reply from which I quote this closing extract. n.b. the specific subject of the enquiry was minimum readings with no reference to maximums nor averages.

“There is no evidence that any readings were knowingly recorded incorrectly or misrepresented. The slightly higher Botanic Garden daily maximum for 31 January (5.3°C) reflects the timing of manual data entry: the 1 February 09Z observation was submitted later in the morning, and the 31 January 09Z observation was uploaded a few days afterwards. The identical values in both submissions point to a simple manual input or timing error rather than any deliberate action. Although this was not flagged by quality control at the time, the daily minima used for climate records remain accurate and consistent with established observing procedures.

Kind regards,

Service Desk

Met Office  FitzRoy Road  Exeter  Devon  EX1 3PB  United Kingdom“

I have emboldened the points for a good reason. The manual system of daily reset of the minimum thermometers in winter at 09:00 occurs around the coldest time of day. Should the weather warm up over the course of the day and remain warmer overnight (which other PWS noted it did) , the minimum reading from the day before will inevitably still remain until the next reset point and thus recorded again the following day as I demonstrated and proven here below.

The -5.8°C minimum reading shown for two consecutive days was nothing that “Quality Control” could be at fault for, it is simply a facet of this type of readings archive protocol which displays this effect. The modern automated system overcomes this deficiency of the traditional system. It is the change in protocols that is artificially contributing to an exaggerated appearance of warming especially in night time minimums as so emphatically demonstrated in the Dalwhiinie example above and here in Cambridge.

A further point here that I find exceptionally dubious is the way that when questioned, Met Office staff can immediately find reasons/excuses to justify readings of six years prior at Cambridge but cannot/will not confirm dubious readings if a 24 hour “time window” has passed. When another reader questioned on 23/1/2026 the accuracy of a reading posted the day before they were notified

“Thank you for your email. We cannot retrieve these details as the system we use to view the times of daily extremes resets the following morning. We’ve had a large backlog of emails to work through so I apologise that your email was not picked up in a more timely manner.”

However, the Met Office does rather like to confuse these issues no end. Leading up to my “Stationgate” report in which government Minister, Lord Vallance, denied the deletion of data that the Met Office had deleted, there was a suggestion that “Climate Averages” and “Historic Stations” data where just for “General Interest” which presumably means some form of light entertainment. This is not the case as in previous queries these matters were clearly described as “Peer Reveiewed”processes to the highest international standards. Who “Peer Reviews” light entrertainment?

It follows then that when the Met Office posts this below regarding “Weather Extremes“, the public is simply required to be mildly interested and not actually genuinely informed.

About this extremes data

All times are in UTC.

Highest maximum temperature – 09:00 to 21:00 on the date shown.

Lowest maximum temperature – 09:00 to 21:00 on the date shown.

Lowest minimum temperature – 21:00 on the previous day to 09:00 on the date shown.

So to emphasize, for its “extremes” daily page the Met Office now alters its calendar day attribution system and makes minimums only appropriate to the overnight readings DESPITE being for the previous day.Thus whilst they are perfectly capable of providing the correct minimum by their conventional archiving principle they choose not to. Confusion appears to be part of their best lain plans in this as in so many other areas.

The issue of Scale Conversions continually both amazes and frustrates me. – it also introduces surprisingly large distortions to readings and makes historical comparisons often unrealistic.

To briefly recap, the Met Office archives readings in Celsius to the nearest tenth of a degree. Traditionally temperature readings were recorded in degrees Fahrenheit with various differing levels of accuracy dependant on the “Order” rating of the site. Lower orders recorded to nearest whole degree, some to nearest half degree and the best “First Order” sites to the nearest tenth of a degree. As one degree unit of Fahrenheit equates to 0.55°C (thus a tenth is 0.055 °C) this requires a rounding principle to revert to the modern recording standard.

The conversion of thermometers started in the very late 1950s with all readings archived from 1961 in Celsius. However, a surprising number of sites continued observations in Fahrenheit until the 1970s – most surprisingly Reading University:White Knights (the home of Meteorology undergraduate courses) was one such late converter. The problem of rounding leads to only specific points being possible on the celsius scale creating a very “blocky” feel of only certain points being possible such as demonstrated here from Heathrow. All the highly trained observers accuracy skills to the nearest 0.055°C wiped out into a few botched unrepresentative number points.

This sorry state of affairs then gets compounded when rounding of very accurate Fahrenheit readings occurs prior to subsequent rounding of the celsius conversion – a double whammy. As I pointed out in my report into my personal transcriptions of readings at Wye here is an example:

34.4°F Recorded gets rounded to 34.0°F…………………………………………………………………….. Converts to 1.1°C

34.5°F Recorded gets rounded to 35°F converts to 1.65 °C………………………………………………….Rounds to 1.7°C.

A difference in reading of just 0.1°F (0.05°C) converts to a more than an elevenfold difference of 0.6°C in the archives as a regular occurrence. Despite this the Met Office and “Climate Science” in general claims accuracy in “Climate Averages” and uplifts to the second decimal place…….but then reading accuracy moved into the surreal world of 21st Century instrumentation.

When I first received this claim of readings to the fifth decimal place I simply assumed this was just a very naive junior member of Met Office staff, I was completely wrong and this is the standard readout level in regular use. So to put this 100,000th of a degree into perspective, it is akin to measuring how rapidly World 100 metre record holder Usain Bolt took to travel just 1 millimetre in his record breaking run as that is 100,000th of 100 metres. It certainly can be averaged as a calculation but not measured by the instrumentation in use at the track.

It is physically impossible to measure temperature anywhere near this level of accuracy in the ambient environment. To assume this time-average figure represents anything other than a meaningless mathematical conjuring trick is beyond absurd. As retired research physicist John Marlow put it in his paper on metrology:

“Many people will read and record all the digits on an instrument whereas, in fact, the last two or three are of no consequence whatsoever.   A perfect example of this was the Met Office site where temperatures were recorded to 4 or 5 decimal places, giving the readings the sense of extreme accuracy, whereas they just showed the stupidity of the observer”

A point from all this I regularly notice is the Met Office (and other associates) preoccupation with the “accuracy” of their instrumentation which is, in reality, unlikely to be a problem but rather the real issue being the actual environment in which the reading is being taken. Despite having written papers on the problems associated with overheating within Stevenson Screens in still air conditions, Dr Stephen Burt., in his report of the 2022 Coningsby reading actually claimed it may be an under record due to the slower response time of the notably still conditions. A really almost perverse interpretation of his other work but assists the all important narrative.

So to conclude the many problems with Met Office climate data and statistics are not just a matter of

poor original siting,

deterioration of sites over time,

selection of sites and change of those used over time,

changes of instrumetation

changes of archiving protocols

unreliability/infrequency of readings

skill levels of observers

But also include

changes to archiving protocols

reliability of transcribing data

Conversion of scales

alterations to averaging periods

In reality the system is so incredibly opaque it is almost impossible to actually verify what is going on and the veil of secrecy surrounding data, that should be openly and freely available, is only getting worse. The narrative of catastrophic anthropogenic climate catastrophe has to be maintained – scientific rigour is expendable.

Source: https://talkshop.blog/2026/09/03/of-myths-and-met-gang-aft-agley/