After “Deep Thought” the Met Office super computer final answer to Newton Rigg is………42…….hopefully not!

Super Computer “Deep Thought” image courtesy Douglas Adams!

Regular readers to my work on the Surface Stations Project will know I have repeatedly pressed the Met Office for the exact details of how “climate averages” data is compiled for existing, no longer existing, and “historic” (extant or not) weather stations. Up to recently I had been unable to have my questions answered but now (and with the assistance of a number of readers) I feel I have made the break though and reached one of two possible answers – either a) they do not know or just as likely b) 42.

I would like to start somewhat oddly with a definition of the “PROCESS” of the simple addition of two whole and positive numbers. I apologise for this simplicity but please bear with me. From AI.

The above is a very easy to follow “PROCESS” known as addition. As everyone agrees with this “PROCESS” we can claim it is “Peer Reviewed”

So taking this PROCESS a step forward if the “SUM” is 42, what are the whole positive numbers that reach that figure by their addition? Obviously you cannot possibly know that – supplying the Sum (output of the PROCESS) and knowing the PROCESS is insufficient information. The calculation for just two whole numbers offers no fewer than 41 ordered pairs and a further 21 unordered combinations.

Extending the calculation to 3 whole positive numbers expands the options dramatically, again as per AI

Now imagine if we can add just 5 positive numbers including those to the first decimal place i.e. from 0.1 to 41.6 The options become multiple orders of magnitude greater.

The point I am possibly over emphasizing here is that simply understanding a peer reviewed “PROCESS” and knowing the output (SUM) of that process is a complete irrelevence if the other critically important factors are not known i.e. the INPUTS and any firm conclusions drawn from the unsubstantiated OUTPUTS are false. Going back to the topic of this post ….

Timeline.

I first queried the 103 non existent sites that were then shown as “Climate Averages” stations with exact coordinates, elevations and site descriptions. I was informed the data was compiled from an internationally recognised “peer reviewed ” process that drew on data from nearby “Well Correlated ” stations. This is an example below of how one of these sites were presented online at the time.

The Met Office forwarded me a pay-walled link to the “process” employed but then went on to change the name of their Climate Averages site to “Location Specific Long Term Averages” and subtly omitted the earlier coordinates, elevation and descriptions.

Turning my attention to their “Historic Stations” data page I discovered that there were ongoing monthly figures being produced for sites that were notably long defunct – I paid particular attention to Lowestoft, closed in 2010. Again I was fobbed off with the notification that average readings were drawn from “well correlated” sites and that the “inputs” were NOT retained. On pressing further and pointing out that not only were there no realistic well correlated stations , all the other sites quoted in the area were similarly non existent, I was then deemed “Vexatious” by the Met Office who refused to answer any further questions nor divulge which sites were used to compile the data. The Met Office even contributed to an attempt to rather ineptly debunk my raising questions which they would not answer. It seems their “external scrutiny” does not extend to mere members of the public asking awkward questions.

“To suggest we’re fabricating data is not true. Everything we do at the Met Office is based on peer reviewed science and subject to external scrutiny. To ensure we [the Met Office] maintain continuity of long-term datasets we use peer reviewed science methods that, where stations have closed, use well-correlated observations from other nearby stations to help inform long-term average figures […] This method is peer reviewed and similar processes are used across the world to ensure datasets are consistent across historic averaging periods. The paper explaining the methodology can be found here………..{pay-walled document link}”  

I was later able to establish the exact paper in question authored by Matthew Perry and Daniel Hollis which is very clearly a major piece of detailed “Climate Science” being published in the International Journal of Climatology for peer review.

This paper however again only describes a “Process” and yet again does NOT specify which stations are used for any specific place or time to provide outputs.

STOP and hold this for thought please………. This is a detailed and highly expensive research work submitted to international journals for peer review…………

Moving again forward I had also pressed for advice on which stations were used in data for compilation of other sites long defunct at Nairn and Paisley but, in now being blocked, no answers were forthcoming. To my astonishment, however, the Met office then opted to DELETE all data for Lowestoft, Nairn and Paisley subsequent to their closure – the house of cards was well and truly collapsing.

Unknown to me at the time, Talkshop reader and close associate Derek Tipp was pursuing my case in the political arena. His constituency MP ( Sir Julian Lewis) took up this issue with the government minister responsible for the Met Office (unelected) Lord Patrick Vallance. I published in full his response which I strongly recommend reading again in full for the specific detail.

Vallance’s Vanishing Act: Official Response to Stationgate

My interpretation is fairly straight forward – the Met Office deleted the data rather than feel the need to justify how it was compiled and then claimed it had NOT been deleted. These were just figures claimed to be for “general interest.” I quote from their current website:

“The purpose of this webpage is to provide a sample of long-term average station data across the UK for general interest. It is not used for formal climate monitoring purposes.”

Really? So senior Met Office scientists (still employed by them to this day) produced a detailed paper for publication in an International Journal specifically for peer review of their “PROCESS” BUT it was only lightweight material for general interest after all. Does anyone who has read so far actually believe that for one second?

Does the Met Office’s financial budget run to providing two separate datasets? Presumably the first for “Light Entertainment” on say Radio 2 and the second heavyweight “real one” for much more serious “Home Service” coverage like Radio 4? I use this broadcast analogy as, after all, a certain Jon Plowman (former Head of Comedy at the BBC) simply must be an expert on all things climate related as he was one of those 28 Anthropogenic Global Warming “experts” who passed judgment on “Settled Science” and restricted/censored any scientific discussion on the causes of climate change thereafter on the BBC.

https://www.theregister.com/offbeat/2012/11/13/secret-28-scientific-experts-who-greened-the-bbc-revealed/522842?page=2

All the objective analysis seen here indicated to me that the Met Office was deliberately trying to obfuscate their data rather than defend their position and reveal the stations actually used to create their climate averages but frustratingly I could not prove that. At this point Talkshop reader “Alex” (requesting anonymity) forwarded me their attempt at getting the Met Office to divulge those stations used for climate averages at Newton Rigg.

Yet again Alex was given the classic Met Office fob off with zero intention to divulge those seemingly “state secret” inputs essential for the production of outputs.

“In order to provide advice and assistance, estimates on this webpage are taken from the gridded data, so provisional versions of HadUK-Grid were used for January 2026 and will be used for February. The four closest grid points are used and combined using bilinear interpolation. The stations used in our gridding vary from day to day and month to month as sites open and close, may have incomplete records for reasons such as maintenance or power cuts, and quality control and quality assurance processes are applied. The gridding process uses whatever is available at a given time. This estimation procedure can be used to provide continuous data series for locations where data may be intermittent. However, we recognise that this could cause confusion when there appears to be a continued flow of data on this webpage from stations that have closed indefinitely. For this reason, we took a decision to remove the estimated data for a small number of sites that are no longer operating for any variables.The purpose of the webpage linked in your request is to provide a sample of historical station data across the UK for general interest. It is not used for formal climate monitoring purposes. We hope this answers your enquiry.“

Alex persisted requesting a review of their answer only to be informed of the following unequivocally stating:

” …..there is no defined list of named observing stations that “contribute” to the estimated values for a specific closed site such as Newton Rigg.

And just re-consider this statement:

“The automated gridding process does not produce or retain a recorded list of which individual stations
contributed to a particular grid value for a specific location and month.”

This is an open admission that the Met Office CANNOT either prove its outputs or offer any transparency of its workings – the answer, apparently, really is no more specific than just claiming 42.

Further Developments.

At this point I felt I should personally try again to challenge the Met Office to answer my original basic questions, however, a separate Talkshop reader (a former Civil Servant with legal training) offered to further the case on my behalf in a very detailed manner. This reader has also requested anonymity and whilst they accept the explicit details will identify themselves to the Met Office they prefer to keep their position not open to the general public – a point I fully understand in the circumstances.

I reproduce in full the response from the Met Office forwarded to me by my reader, which includes the original inquiry.

Request for Information under the Environmental Information Regulations (EIR) 2004

Your correspondence dated 4th May 2026 has been considered to be a request for information in
accordance with the Environmental Information Regulations 2004.
You asked:

I have noticed this online critique of your response to FOI/EIR requests for information regarding
Newton Rigg Weather Station.

Newton Rigg Revisited again. A deceased “Historic Station” producing monthly figures and you are NOT allowed to know how!

This causes me serious concern regarding the scientific transparency of the Met Office. I make the
following request to clarify the exact position:-

Under the Freedom of Information Act 2000 and/or Environmental Information Regulations 2004, I
request the following information concerning HadUK-Grid-derived values published for closed or
inactive observing stations, specifically Newton Rigg.

For the avoidance of doubt, I am NOT requesting a pre-existing published “list of stations used”, if
no such standalone document exists. Instead, I request any information held which would permit identification, reconstruction, or
determination of the observational inputs contributing to the gridded estimate for a specified
location and date.

Please provide:

• A description of the operational workflow by which a monthly temperature value is generated and
  published for a closed station location such as Newton Rigg.
• Identification of all datasets, files, databases, intermediate products, or processing stages used in
  generating the published monthly value for Newton Rigg for January 2026.
• Whether the HadUK-Grid processing system:
  a) temporarily determines which observation stations contribute to each 1 km grid cell;
  b) stores, logs, caches, or serialises this information at any stage;
  c) permits this information to be reconstructed by rerunning the interpolation process.
• If reconstruction is possible, please state:
  a) whether the Met Office could identify the stations contributing to the relevant grid cell for Newton
  Rigg for January 2026;
  b) the estimated staff time required to do so;
  c) whether any software, scripts, or archived inputs required for such reconstruction are retained.
• Copies of any:
• technical specifications,
• algorithm descriptions,
• provenance policies,
• reproducibility procedures,
• audit procedures,
• or internal guidance
  relating to traceability of HadUK-Grid interpolation inputs.
• Clarification of the meaning of “Open” on the public station webpage when the original observing
  station is no longer operational and published values are instead derived from gridded
  interpolation.
• The uncertainty estimate, confidence interval, or error characterization associated with the
  January 2026 Newton Rigg monthly mean temperature value, including any documentation
  describing how uncertainty is quantified for infilled or interpolated station estimates.

And so to the RESPONSE {n.b. bolds are mine for emphasis}

The Met Office holds some of this information.

Please note, EIR gives an applicant the right to access recorded information held by public authorities
at the time the request is made and does not require us to answer questions, provide explanations or
give opinions, unless this is recorded information held. We do not need to create new information for
the request, but can provide advice and assistance on information around the subject that we may
hold.

• A description of the operational workflow by which a monthly temperature value is generated and
  published for a closed station location such as Newton Rigg.
  This is an automated process which completes the following steps:
• Extract daily observations for all open stations
• Calculate monthly stats at each station (average max temp, total rainfall etc)
• Interpolate the monthly station values to a 1km grid (Hollis et al.)
• Extract an estimate for the coordinates of Newton Rigg from the 1km grid using bilinear
  interpolation
  The exception is air frost where we produce grids of tmin for each individual day and then combine
  these to create a monthly grid of the number of days with tmin

• Identification of all datasets, files, databases, intermediate products, or processing stages used in
  generating the published monthly value for Newton Rigg for January 2026.
• MIDAS database stores the original station observations. Publicly available version providing subset of
  station data here: Dataset Collection Record: Met Office MIDAS Open: UK Land Surface Stations Data
  (1853-current)
• Data Components (which implements the business logic to handle different reporting practices,
  multiple rain gauges etc.)
• The ancillary files used by our Climate Grid software i.e. terrain elevation, proximity to the coast,
  proximity to urban areas
• The grids of the 1961-1990 LTA which are used to convert actuals to anomalies and vice versa
• Intermediate product is the pre-QC version of the grid, generated by the auto-QC process – this is
  written to disk.
• Whether the HadUK-Grid processing system:
  a) temporarily determines which observation stations contribute to each 1 km grid cell;
  Yes. Although this is not straightforward. The gridding process involves both a regression step and an
  inverse-distance weighted (IDW) averaging step. The IDW search radius and the distance of each
  selected station from the target grid point determine which stations are influencing (contributing to)
  each grid point. However, the regression step involves using all of the stations in the network to
  construct a model of, for example, the relationship between temperature anomaly and terrain
  elevation. This means that every station in the network has some impact on every grid point. As noted
  above, air frost is determined from the daily tmin grids, which means that for Jan 2026 there are 31
  different models. It is likely that the stations used in each model will not be identical for all 31 days as
  some stations may drop out due to missing data. Lastly, given the final estimates are obtained by
  bilinear interpolation from the four surrounding grid points, it follows that you would need to check the
  calculation for each of the grid points surrounding Newton Rigg to give a definitive answer.
  b) stores, logs, caches, or serialises this information at any stage;
  No.
  c) permits this information to be reconstructed by rerunning the interpolation process.
  Yes. However, re-running the software does not make the information on which stations were used
  any more accessible.
• If reconstruction is possible, please state:
  a) whether the Met Office could identify the stations contributing to the relevant grid cell for Newton
  Rigg for January 2026;
  We do not hold. In order to provide advice and assistance, as noted above, all stations contribute in
  some way, to a greater or lesser extent. Though theoretically possible to identify which stations are
  used in the regression and, separately, which stations are used in the IDW, doing so would require
  adaptation to the automated code which runs regularly to carry out these calculations. This would
  involve the creation of new information rather than the provision of recorded information already held.
  Under the EIR, public authorities are not required to create new information in order to respond to a
  request.

b) the estimated staff time required to do so;
We do not hold. Please see above.
c) whether any software, scripts, or archived inputs required for such reconstruction are retained.
Any code we use will be retained as part of our version control process.

• Copies of any:
• technical specifications,
• algorithm descriptions,
• provenance policies,
• audit procedures,
• reproducibility procedures,
• or internal guidance relating to traceability of HadUK-Grid interpolation inputs.

The complete technical specification and algorithm description for the gridding process can be found in
Hollis et al (2019) at https://doi.org/10.1002/gdj3.78.
The process is highly automated so no internal guidance is needed.

• Clarification of the meaning of “Open” on the public station webpage when the original observing
  station is no longer operational and published values are instead derived from gridded interpolation.
  Open is used when one or more variables are being reported regularly from a station. Newton Rigg is
  still open as a rainfall-only station.
• The uncertainty estimate, confidence interval, or error characterization associated with the January
  2026 Newton Rigg monthly mean temperature value, including any documentation describing how
  uncertainty is quantified for infilled or interpolated station estimates.
  We do not hold this information. In order to provide advice and assistance, general uncertainty
  estimates for a previous 5km gridded product are available in the paper Legg (2014).
  References
  Hollis, D, McCarthy, MP, Kendon, M, Legg, T, Simpson, I., 2019. HadUK‐Grid—A new UK dataset of
  gridded climate observations. Geosci Data J. 6: 151– 159. https://doi.org/10.1002/gdj3.78
  Legg, T.P., 2014. Uncertainties in gridded area average monthly temperature, precipitation and
  sunshine for the United Kingdom. Int. J. Climatol. https://doi.org/10.1002/joc.4062
  We hope this answers your enquiry.

Detailed analysis of this response will be the subject of future posts and I do not wish to in any way to influence or direct readers views – I welcome comments and interpretations from readers on this. However, for my part there are three initial standout issues.

Firstly, for the Met Office to suggest such a sophisticated “process” is to produce data for “general interest” only and is somehow separate from other more detailed data sets seems incredibly implausible. If this were the case it would be an extravagant and unnecessary use of public funds.

Secondly, it would appear that every open weather station is being used in the process regardless of quality or provenance of readings. With such a collection of sites changing over time it would be incredibly easy to “game” the system by selective closure and opening of contributing sites – a point I have suggested on several occasions with a notable example being Dumfriess and Galloway.

Thirdly, the Met Office completely fails any vestige of transparency. Unwillingness, (or probable inability) to justify one month’s, even one day’s readings data formulation fails any scientific scrutiny. There are nearly 400 CIMO rated sites in the UK leading to literally billions of possible combinations to reach almost any end result.

“Black Box” computer modelling at least uses known inputs to produce end results without, controversially, fully disclosing the internal process. The Met Office seems to have this controversial system to hide the inputs and the exact process details. Perhaps opening this Box might relieve us from all these problems and at least leave some hope for a more honest approach in the end.

Source: https://tallbloke.wordpress.com/2026/06/07/after-deep-thought-the-met-office-super-computer-final-answer-to-newton-rigg-is-42-hopefully-not/