Wight:St Catherines Point WMO03866 – a stalled attempt at an area climate history reconstruction for quite bizarre reasons.

50.575543 -1.296948 Met Office Assessed CIMO Class 4 Installed 1/12/1924

Another of the Met Office’s centenarians and one of the many weather stations that the island has graced over the years. The very earliest on Wight that I could find details of dated from 1839 and was operated by the Ventnor doctor George Anne Martin. There will almost certainly have been earlier sites for naval purposes but, in common with most of these types of sites, records are no longer retained or have yet to be “unearthed”. St Catherines Point (the lighthouse) is now a fully equipped synoptic site but this was not always the case – a slightly strange history unfolds

There are currently three sites on the island recording temperatures with the often quoted “Needles” site principally recording wind velocity and direction but not temperatures. The other current island sites at Ventnor and Shanklin are basic climate reporting sites of notably low grade standard whilst St Catherines records over a wider spectrum of observations especially wind with variously different placed wind masts over time – 2 are visible in the image below by the wall and on the lighthouse itself. Sailing “conditions” are a key element to be known around the island.

The “remarks” section of the CEDA archives indicates this site was relocated in 2000 from a site further from the coastline at an elevation of 46 metres amsl rather than the current 20 metres. The coordinates given are not accurate enough to pinpoint the site, however, the elevation indicates the site was to the north of the lighthouse and just within the perimeter walling shown at the top of the above ground level view in the distance. From this point of view it is unlikely the move would have amended the climatology and there is no reunumbering involved to confirm that.

My only real point to note with this site is the very common situation of being unnecessarily close to a wall. I have lost count of the number of sites that are similarly located and I can only assume that there is a reason but that I am not yet aware of it. The bush to the northeast though really should be removed, the image is dated 2017 and more recent aerial imagery shows significant growth that is probably the reason for the low CIMO Class 4 rating. Surely the Met Office should be able to have this latter problem removed to avoid compromising data. The image below shows just how much this growth is affecting the 10 metres area.

Despite the few issues with this site and those I identified with the 2 other currently operational sites on the Island, I did feel that with so many other former long term sites available that a small area historic climate record could potentially be produced.- the climate history of the Isle of Wight. I started looking at the earliest records for St Catherines Point and came across the some very strange archiving processes.

Most traditional climate reporting sites report once daily at 09:00 both maximum and minimum. These are then archived in the same way by the Met Office and this is what those archives look like for this site.N.B. column “D” indicate 24 for once only “ob-hour” with column B showing 09:00.

Handling once daily data like this is prone to the carry over effect of previous days highs/lows potentially recording twice as I notably highlighted in my reports on Sutton Bonington amongst many others. There is nothing at all unusual about any of this so far. However, I decided to clarify the conversion from Fahrenheit to Celsius used as I have noted in the past this is variable and looked into the manuscript readings to compare with the digital transcriptions. This was quite a revelation! Readings were actually taken three times daily – 07:00 (Yes SEVEN A.M.) 13:00 and 18:00.

This regime is well out of my experience range and I had to consult an expert reader for advice on day attribution protocols and what effect the earlier a.m. reading may have. It transpires the 07:00 and 18:00 readings are the dominant figures with a one day drift on readings. These multiple day readings though do largely overcome the carry over effect in the same way as modern day automatic readings do.

Quite why the Met office has given the false timing appearance of 09:00 for 07:00 readings smacks of form filling rather than accurate data recording. For a large part of the year these early readings would have been taken in the dark and the boundary effect and Time of Obs bias would have been assumed to have been greater BUT in reality they would not have occured due to the “invisible in the archives” later readings time. The Met Office claims to make (almost inevitably) “adjustments” to historic once daily readings in their “climate models”. Exactly how these adjustments are made is again one of those mysteries that the public is not party to. I asked AI to locate the Climate Model adjustments but despite multiple attempts through different AI models could not find a definitive answer. The is the best AI could manage:

“I found relevant references, but not a published Met Office formula confirming the particular adjustment you describe. The documents distinguish standardising observation periods from statistically correcting time-of-observation bias.

The most useful references are:

1. Met Office Surface Data Users Guide, sections 4.5–4.6. This directly documents the different observing practices: once-daily climate stations report maximum and minimum temperatures over 0900–0900 UTC, while other stations report separate 0900–2100 and 2100–0900 periods. It establishes the comparison problem, but does not give a statistical correction formula. artefacts.ceda.ac.uk
2. HadUK-Grid FAQ, “Why are the daily temperature maxima and minima for different 24-hour periods?” This explains how observations are assigned to a calendar date:Value assigned to day DObservation periodMaximum temperature0900 on D to 0900 on D+1Minimum temperature0900 on D−1 to 0900 on DThe rationale is that maxima usually occur in the afternoon and minima around dawn. This is a date-assignment convention; it does not itself remove every bias caused by the 0900 boundary. The same FAQ defines daily mean temperature as ((T_max+T_min)/2). Met Office
3. Hollis et al. (2019), “HadUK-Grid—A new UK dataset of gridded climate observations”, section 2.1.1. This is the closest reference to the processing you may mean. The authors use a preprocessed MIDAS database with “a series of business rules” that create daily/monthly values and handle differences in observing practice and instrumentation. However, the paper does not enumerate those rules or specify a once-daily versus twice-daily temperature correction. Section 2.1.2 says the full processed station inputs were available on request from the authors. Wiley Online Library
4. Met Office UK climate extremes—temperature notes. For official daily temperature records, the Met Office explicitly uses 0900–0900 UTC for comparability with manual stations, although automatic stations also supply daytime and nighttime extremes. Here, comparability is achieved by using a common observation window. metoffice.gov.uk

A qualification: these references concern observational climate datasets, which can be used to assess or calibrate climate models. The HadUK-Grid FAQ explicitly says that its input data were not homogenised, so it would be too strong to infer a general time-of-observation correction from these sources. Met Office

The precise missing reference to request from the Met Office would be: the MIDAS preprocessing “business rules” for deriving daily maximum and minimum temperatures from 12-hour and 24-hour reports, including any adjustment for thermometer reset time or observation-boundary bias.“

Thus apart from the fact that the process of adjustment is certainly not transparent the obvious next consideration is whether or not any form of adjustment is made to St Catherines Point’s data which has the appearance of once daily 09:00 reporting in the official archived data – but in reality had multiple timed readings. Was, or was not, unnecessary adjustment made to the frankly botched archiving when input into thate peer reviewed process of deriving area climate averages?

To answer my own conjecture………….I doubt even those responsible at the Met Office even know the answer or had even ever considered it.

Very helpfully my expert reader has offered to transcribe the data from original source when time allows and hopefully I may be able to continue with my local area temperature reconstruction in the near future. Experience though suggests there are probably lots more other quirks in this older data to iron out. A good trial run for a larger database national project.

Source: https://talkshop.blog/2026/10/04/wightst-catherines-point-wmo03866-a-stalled-attempt-at-an-area-climate-history-reconstruction-for-quite-bizarre-reasons/