Wye, East Malling and Faversham – a discussion piece on compiling a new historic temperature record for Great Britain. (Plus another astonishing discovery)

Original and modern locations of East Malling Weather station on aerial image from 1960

This post is to open up discussion on the process to compile a new historic temperature record for Great Britain defined as England, Scotland and Wales but excluding Northern Ireland and the Channel islands. I also offer some updates on three sites intended to form the basis of one area’s data being the currently operational sites at East Malling and Faversham together with the closed Wye station. I am actively seeking any views, information, guidance, opinion, assistance indeed any useful contributions in any way or form.

In my review of WYE I felt I had identified a site operational from 1924 to 2006 that met all the correct criteria for its data to be used in the construction of a new temperature history for Great Britain. Dave Woolcock then analysed the available data from 1959 to closure and expertly graphically presented in a meaningful way such as this below.

Dave’s “12-month moving average for tmin and tmax using @Average of the previous 365 days.” did not show any dramatic levels of warming in either maximum or minimum with the start and finish end points remarkably similar. Several things arose in discussion that many of the “traditional” school of thought did not consider this a valid way to depict data for reasons that rather struck me as more establishment ritual than straight analysis.

More importantly for me were Dave’s own subsequent questions regarding how to allow for missing data. Wye’s record was very good but not perfect. I personally transcribed from the manual archives the data from 1924 to 1959 and noticed more missing days (broken thermometers for example and the small matter of bombs occasionally dropping on the site!) definitely showing that some form of infilling was required and had to be adressed.

The Central England Temperature Series, in using only a small number of sites, would have encountered the same problem – the solution was to nominate “back up” sites with demonstrated well correlated near neighbouring stations. The whole process used is detailed in this paper from which the indicative extract is below..

For Wye I propose to use the nearby (38km/23 miles distant) East malling site for infilling in two ways. The site was established at almost the same time as Wye. This simultaneous start was not accidental, both sites were originally run by Wye College, they are both agricultural reseach sites on the North Downs in Kent and lay near river valleys at similar elevations. If close and consistent correlation can be established between their readings then both can be suitable to infill for each other if needed to produce 2 independent data sets – i.e. Wye set with occasional East Malling infill and East Malling with occasional Wye infill.

It is important to note that both sites have very good observation records so infills will be rare exceptions not the norm and can be readily marked for transparency. The important issue is that East Malling meets the same quality characteristics that I defined for Wye that are worth repeating.

1. CIMO Class 1 or 2 Site.
2. Not relocated over data reporting period.
3. Identified good maintenance with no significant change of surrounding conditions.
4. Not subject to significant Urban Heat Island effects.
5. No instrumentation changes (i.e. from LIGT to PRT)
6. Consistent good quality observations record. (WYE was excellent)

East Malling is CIMO assessed as Class 1 by the Met Office but in its current iteration I find that highly unlikely as I commented in my original report. The intrusion of thousands of square metres of polytunnels so close to the south east of the screen are serious impediments. however, these are in fact only very recent (though enduring) additions. I did contact the Met Office to query the classification as I felt it was stretching the CIMO regulations to claim this site was still Class 1 with these levels of artificial heat sources so close. The reply

Enquiries	Tue, 11 Feb 2025, 10:18
to me

Dear Mr Sanders

We can confirm that East Malling weather station was assessed as CIMO class 1 for temperature and humidity on its most recent inspection. This is in keeping with the guidance set out in Annex 1.D of the WMO guide to instruments and methods of observation – volume 1. Please see the link below for further details.

https://library.wmo.int/records/item/41650-guide-to-instruments-and-methods-of-observation

Kind regards,

Weather Desk

Very clearly the 100 metre RADIUS circle area iincludes a very large section of artificial heat sources – perhaps another of those Cardiff miscalculations?

Fortunately, for my purposes, this compromising set of polytunnels is a relatively new feature from 2022 onwards and prior to this the site was indeed a very good Class 1 agricultural site. That the Met Office are seemingly at ease with such a potentially large artificial heat source now so close is more a reflection on them, however I do note that their latest CIMO listing demotes the site to Class 2 as all that plastic is still there. (Radius/diameter recalculation again maybe?)

This is how East Malling looked at the time of closure of Wye – strikingly different and very good.

Delving deeper than my inital review of East Malling I was able to establish its original location as per the headline image showing old and new locations. The development of additional labs and offices enforced the move, however, the original location was also perfectly acceptable and again typical of the siting conditions employed by agricultural research centres.

A potential interesting development, if close correlation is revealed, is the automation of East Malling in 2000. Any significant deviation in the correlation between the two sites from automation in 2000 to Wye closure in 2006 should be reasonably noticeable despite only being a 6 year period.

Thus my proposal to overcome missing readings with the absolute minimum of statisitical interference is to ascertain if there is a usable site correlation and transparently demonstrate that point. If (my first passes look very promising) that is established then openly disclosed infills can be used at both Wye and East Malling for paralell sites – a possible imteresting harmony given their related hoistory. Two good quality sites for this section of south east England should work. NOTE As I started this reeport I really welcome advice, opinion, guidance on this as it is critical to get this part of the project correct to roll out around the country. please offer comments below for open discussion.

DATA PRESENTATION

Perhaps the most accessible data presentation by the Met Office of the historical climate trend is their “Time Series” This displays Maximum, Minimum and Mean Daily Temperatures plus Rainfall and Precipitation by country, year, season and month as below in this example of England covering annual maximum temperature.

I am hoping to be able to present data in this same manner by individual station to act as a direct comparison. One significant change though will be the trend line derivation. The Met Office are using a system defined by AI as follows.

“A triangular-weighted kernel smoothing filter, using 14 years on either side of each target year.

In practical terms:

• The smoothing window contains 29 annual observations: the target year, 14 preceding years and 14 following years.
• Observations closest to the target year receive the greatest weight.
• Observations further away receive progressively smaller weights.
• The calculation produces a smoothed temperature for each year.
• The resulting values are joined to form the long-term trend curve.

This is not an ordinary 29-year moving average, because the observations do not all have equal weight.

It is also not a conventional straight-line regression.

2. The actual calculation

A standard triangular weighting scheme for a 29-year window would use the following weights:

Illustration of the 29-year triangular filterIllustrative weights, not a reproduction of Met Office source code.

For example, the smoothed temperature for 1950 would be calculated using observations from 1936 through 1964.

For the standard triangular filter, the central year receives weight 15, the adjacent years weight 14, and so on, down to weight 1 at the ends.

The weighted average is:

The denominator is 225 because that is the sum of all 29 weights.“

…………………………………………………………………………………………………………………………………………………………………

That the trends start from the first year and complete up to the last year on the graphs raises a very significant point. Where do the forward and preceding figures come from? The Met Office via AI explains:

3. What happens at the beginning and end?

This is an especially important detail.

The Met Office paper states that the smoothing kernel is reflected at the ends of the time series, allowing the trend to extend all the way to the first and last observation.

However, the authors warn that this reflection can suppress trends near the endpoints. They specifically advise caution when interpreting the first and last decade of the smoothed series.

That could be significant when discussing recent warming trends.” {my bold}

In other words the first and last 14 years of the graphs are dependent on “predictions” and not real observations.

For this reason I was proposing to use trailing 10 or 15 years trends to avoid any artificial/predicited readings.

Readings Pre and Post site’s data.

Regarding data prior to the instalation of these exampless of East malling and Wye, I was proposing to find further nearby stations which had overlapping data to identify correlation. On this basis an example for Wye could be the Canterbury site from 1866 to 1944. For East Malling the Tunbridge Wells (Bredbury Common) site may suffice subject to furtther provenance checks. This overlapping period observations was also a principle used in the CET to pruduce a continuity set. Though my system is constructed from the central core years and extended backwards and sometimes forward, it follows the same principles.

Many issues relating to instrument cahnges can be overcome (up to 2017) by using only manual sites which continued using LIGT up to that point. The whole question of instrument changes and how to allow for these changes in modern readings will be the subject of further few review particularly with regard to the many findings by Dr Eric Huxter. If anyone has knowledge of longer term sites (starting anythime in the 20th century) which still operate using LIGT (obviously non Met Office or possibly former) these would be extremely useful.

This is a long term open discussion so if anyone has any contributions please feel free to offer them at any time. Rothamsted will obviously be one of the other long term sites repressenting an area just north of London. Other potential contributers around the country to the overall project include the likes of Morpeth Cockle Park, Cawood and Wellesbourne. Further suggestions most welcome for investigation.

For now I intend to complete reviewing the last few current Met Office sites with upcoming “trips” to the Channel Isles and some high altitude Scotttish sites – the roads around Bealach-Na -Ba look very interesting!

CODICIL – Regarding Faversham

Further developing this Wye College theme, in this research I discovered that the College was also originally responsible for establishment of the Faversham weather station. This site is frankly atrocious and I was very surprised that Wye College would be associated with its poor standards. Having dismissed Faversham as Junk throughout the 21st century (under the stewardship of none other than Reading University with its national centre of excellence in …….meteorology) I decided to go further back into its history. This site had in recent time gone from bad to worse but this was not the case in the past. Whilst now massive 5 metre high leylandii hedging blocks winds from every compass point and produces – quite intentionally as it is a fruit farm – a totally calmed “suntrap,” in the past under Wye’s research regime it was a wonderfully open agricultural field.

Wye college closed the site in 1990 when the hedging was planted – on subsequent site take over and 10 years of rapid hedge growth later, Reading University reopened the use of the self same but now compromised screen and got a record and an overheating site ever since. The observations standard typically rapidly declined but when it was read its debateable readings were substantiated against even RMetS criticism. It is still a terrible site as I posted when I visited last year.

Now you might think it could not get worse…..but! – Manually observed sites use a PRT and Data Logger for maximum readings and a LIGT for minimum readings. These latter instruments at “once daily” read sites can, and regularly do, “carry over” readings from the day before, notably minimums when warm fronts warm up the day with the previous day’s low still showing at the base of the rider. This can also carry over maximum readings but generally much less frequently so. Obviously the Met office is aware of this and applies “adjustements” to minimums in its “Climate” models to account for this. The PRT/Data Logger set up also records those minimums and indeed on automation of sites would serve that function but with no adjustement required.

This situation could cause problems if an observer should decide to not use the minimum LIGT and “cheat” by using the data logger – particularly if late for a reading or missing one. If such a case occured unkonwn then adjustment would be applied artificially magnifying the effect rather than compensating for it. Of course this should never happen, but I have often queried the diligence of many observers who miss many days and readings and question their skill levels………………and then I saw this in the archives.

Remarks

Start date	End date	Remark type	Remark
2021-09-22	2021-09-22	SITE INFORMATION	PHOTOS OF INSIDE THE SS IT CAN BE SEEN FROM THIS DATE THERE WERE NO L-I-G THERMOMETERS PRESENT MIN TEMPS FLAGGED NOW
2021-12-31	2021-12-31	SITE INFORMATION	OBSERVER CAN’T BE PRECISE ABOUT EQPT REMOVAL DATES. COVID AND REDUCED CAPACITY HAVE LED TO A REDUCTION IN OBS

That is correct. There were no LIGT thermometers at all in the screen. Nobody knew when they disappeared but blamed it all on Covid 19 anyway – in December 2021. So how were all those readings taken? It is very instructive of the Met Office to say temperatures are now “flagged” but there are no indications in the publicly viewable archives of any “flags”, just readings of unknown validity. “From this date” (22/9/2021) no thermometers. Thence 31/12/2021 over 3 months later no apparent resolution but readings continuing all the time. £Multibillion super computer being fed…………………………………!

Source: https://talkshop.blog/2026/11/12/wye-east-malling-and-faversham-a-discussion-piece-on-compiling-a-new-historic-temperature-record-for-great-britain-plus-another-astonishing-discovery/