13.4 centimetres of sea

[Waves crashing on the seafront at Blackpool. Sea levels have risen by 13.4cm in the UK since 1993 © Peter Byrne/PA]

I was rather alarmed by the observation reported at reference 1 that sea levels around the UK have risen 13.4cm since 1993, a little over thirty years ago. Which sounds like an awful lot of water. And a big rise relative to the big, flat beaches like that at Ryde on the Isle of Wight.

Reading on, I was perplexed by the graphic snapped above. Was it another case of the FT getting carried away with data visualisation; of a picture not being worth a thousand words at all. Maybe they would have done better with a table of raw data supplemented by a few well chosen words?

I thought that it might help if I went to the source at reference 2. Which led me to an Appendix A on datasets, which started on page 57 (of 68). Where I read that most of the data - more than three quarters - in the key HalUK dataset (described at reference 3) is about rainfall, with the rest covering things like temperature, hours of sunshine, air pressure, humidity and windspeed. And, as it happened, the summary Table 1 at reference 3 did not include sea level. Not something that can be recorded at your average weather station.

But I got stuck on the table snapped above. Weather jargon overload. 

After which I thought I would ask Gemini and he proved pretty good as a study assistant - and a dialogue with him was very helpful. He might not have been word perfect, but in this context, that did not really matter. He had enough grasp of the general principles underlying data of this sort to help me along.

Which led to the summary diagram of some of the concepts involved above. With the trickiest link perhaps being that between the location of a recording station and that of the far more numerous grid squares: grid squares might be good for users of weather data, but they are a long way from the source.

The following day, the second day, aka Wednesday, I have another go at the orange figure above, which has been taken from section 1.3 - Daily Temperature Extremes - of reference 2, which starts at page 8 and where I find lots of tricky charts and tables before I get to the original - Figure 10 - of the orange figure. Figure 10 does up anomalies and is paired with Figure 11 which does down anomalies, with the sums of corresponding figures in the first column on the left approximating to the number of days in the year.

Further discussions with Gemini, who does pretty well, but who is rather dependent on what I tell him about a table which he cannot actually see. The key to it all seems to be the level at which you count anomalous days, the level which strikes the right balance between too much noise and too much smoothing. And for the purposes of Figure 10, the authors appear to have chosen county.

Gemini is also good at flattery. He has clearly been told that laying it on with a trowel works, even with people who claim that it doesn't. And it does work, even when you know that the chap with a trowel is a machine! From where I associate to reports of the elderly in Japan getting attached to their robotic pets. Much cheaper and more convenient than real ones.

It so happens that we have just watched the 1970 De Laurentiis film about the battle of Waterloo, about which I think I know a bit. A good film which has worn well and which I thought caught the atmosphere of the whole bloody business pretty well. But you would have had a job to say how the battle unfolded afterwards, if the film was all you had to go on. Never mind the detail, catch the atmosphere. Which is perhaps where the fancy graphics which the FT likes, like the orange figure above, are at.

Oddly, for a report which contains a lot of statistics, there is none of the heavy going statistical stuff which you get in other contexts. Or, at least, I have not found it yet.

All of which prompts two sub-digressions.

First, suppose we have a height and a location for each of the 70 million or so people in the UK. Suppose further we have a hierarchy of areas, perhaps starting at what the census people used to call enumeration districts and ending with the constituent countries of the UK. Perhaps starting with full postcodes, working up through post towns (usually the first two letters of the postcode) up to the constituent countries. It is easy enough to calculate average height for all of those areas. Then is it always true that there is more variation in average height among the small areas, there are more extremes in the small areas, than there is among the big areas - or is that a matter of statistics, of the law of large numbers?

Second, suppose we have a function A, where A(X) = number of days for X for which P is true. Where P is some proposition defined on the area X and daily data is available. Then if the area Y contains the area X, what can we say about A(Y), given A(X)? Or if Y is the sum of a number of Xs. My guess is nothing, unless P is constrained in some way.

But beach calls and there the matter will have to rest for the moment.

PS 1: a bit later on the first day, I got an email from Imperial, asking me to take the cognitive test which came with the questionnaire reported at reference 5. A test which must have taken around half an hour and a lot of which I found quite difficult. Also quite tiring, not helped by various minor glitches with design and IT. And I did not notice any advice about a good time of day to take the test or about not taking coffee beforehand. Noting that the computer will know what time of day you did take the test, but not about about the coffee.

At the end, they told me how I had done, with three good scores balanced by three average scores and  one zero score. Which last I might say I fail to understand: I had thought that I had done badly on delayed memory for shapes, but not that badly! Maybe they will give me another go at some point in the future.

PS 2: I had thought that tens of thousands of people were killed at Waterloo, perhaps as many as 50,000. But the story at Wikipedia is that I had got this quite wrong too. Out of around 200,000 soldiers engaged, perhaps 10,000 were killed. 5%. So a lot of deaths, but the figure I had remembered was grossly exaggerated. Clearly important to check this sort of stuff from time to time.

References

Reference 1: UK sea level rising faster than global average, study finds: Britain is getting wetter and warmer as new temperature records become norm - Attracta Mooney, Janina Conboye, Financial Times - 2025.

Reference 2: International Journal of Climatology: State of the UK Climate in 2024 - Mike Kendon, Amy Doherty, Dan Hollis, Emily Carlisle, Stephen Packman, Svetlana Jevrejeva, Andrew Matthews, Joanne Williams, Judith Garforth, Tim Sparks - 2025. Also mixed up with the Royal Meteorological Society and their Journal of Climate Science.

Reference 3: HadUK‐Grid: A new UK dataset of gridded climate observations - Dan Hollis, Mark McCarthy, Michael Kendon, Tim Legg, Ian Simpson - 2019.

Reference 4: https://peterbyrne.co.uk/. Might take a while to respond, but it does.

Reference 5: https://psmv5.blogspot.com/2025/07/imperial-health.html.

Reference 6: https://en.wikipedia.org/wiki/Battle_of_Waterloo.