The Central England Temperature [CET] record is a meteorological dataset originally published by Professor Gordon Manley in 1953 and subsequently extended and updated in 1974, following many decades of painstaking work. The monthly mean surface air temperatures, for the Midlands region of England, are given in degrees Celsius from the year 1659 to the present.

This record represents the longest series of monthly temperature observations in existence. It is a valuable dataset for meteorologists and climate scientists. It is monthly from 1659, and a daily version has been produced from 1772. The monthly means from November 1722 onwards are given to a precision of 0.1 °C. The earliest years of the series, from 1659 to October 1722 inclusive, for the most part only have monthly means given to the nearest degree or half a degree, though there is a small ‘window’ of 0.1 degree precision from 1699 to 1706 inclusive. This reflects the number, accuracy, reliability and geographical spread of the temperature records that were available for the years in question (intro courtesy of Wikipedia).

Very mild February in Central England despite Buchan’s cold spell

Despite a relatively cold spell from the 5th to the 12th (fig 1), which almost coincided with Alexander Buchan’s first cold spell singularity, February ended up a very mild month, with a mean temperatures in Central England of 6.1°C, which was +2.34°C above the 1961-1990 long-term average. There was an all time maximum set on the 20th for 14.1°C, which was the warmest day in Central England since at least 1878, I did cause a bit of a stir on the day when the temperature reached 18.3°C at Northolt and Kew in London.

Figure 1 – Data courtesy of the Met Office

I make February 2017 was the joint 30th mildest since 1659 (fig 2) in the CET data that I download from the Met Office, hopefully this month I’ve downloaded the latest finalised data unlike last month!

Figure 2 – Data courtesy of the Met Office

Because so many recent February’s have been mild in recent years, February 2017 was only the mildest since 2014 (fig 3), cold February’s these days, are getting few and far between I’m afraid.

Figure 3 – Data courtesy of the Met Office

Expect Spring a little early again this year

With my tried and trusted ‘First day of Spring’ algorithm, I estimate that Spring 2017 will be around eight days early in 2017, probably occurring on the 13th of March in Central England (fig 1). As usual I estimate it by calculating the degree days from the Winter solstice, and fix the first day of Spring when that count exceeds the mean number of degrees days between Winter Solstice and the Vernal equinox (December 21st and March 20th). And yes, I do know the time of both these can vary by day or so each year. The estimate is based on using the long-term average temperature for future temperatures, so if it’s warmer than usual in the next couple of weeks it may come even sooner, or later if the next six weeks or so are colder than average. That makes this Spring just over three weeks later than it was last year.

Figure 1

Cold Winters mean cold springs, so the Spring of 1963, the third latest since 1772, didn’t happen till the 4th of May that year, likewise a very mild Winter means a very early Spring as in 1989 , the earliest Spring since 1772, which occurred on St Hilary’s day of that year (fig 1). Of course, Central England has warmed since 1772, and the first day of Spring is occurring much earlier, in fact I estimate that Spring is now 20 days earlier than it was in 1772.

Figure 2

A new way of looking at Central England Temperatures

Figure 1 – Data courtesy of the Met Office

Ever wondered what looking at the CET for every week in your life would look like? Well, the heat map above (fig 1) does just that, it looks very formidable at first glance, but if you take a closer look it’s just full of information about temperatures of England since 1958. Each cell represents the mean temperature anomaly for each of the fifty-two weeks in a year, so each row represents a single year. I think splitting the year into weeks means is an excellent way of perusing the CET series, you can easily pick out the exceptionally cold or warm period since the daily CET started back in 1772, if you used a longer period such as a month, these shorter spells wouldn’t be visible. The application can display heat maps of anomalies (fig 1) or mean temperatures (fig 2), as well as extreme high maximums or low minimums (fig 3).

Figure 2 – Data courtesy of the Met Office

Figure 3 – Data courtesy of the Met Office

It was easy enough to add some extra code to find the extreme warmest and coldest week from all years in the series and highlight them in the grid (fig 4).

Figure 4 – Data courtesy of the Met Office

It also added functionality to display terciles, quintiles and centiles (fig 5), but they don’t like quite yet, because I don’t think I’ve perfected the colours I use in displaying them. Anyway, I’m sure someone will let me know what I’m doing wrong. The only trouble is with the size of these heat maps, is that I can only display them back to 1958, so until I get that 27″ monitor I keep promising myself, an even bigger heat map will have to wait.

Figure 5 – Data courtesy of the Met Office


Is St Hilary’s day the coldest day of the year?

Figure 1 – Courtesy of Wikipedia (By Chordboard)

In weather folklore there is a widespread belief that St Hilary’s day, which falls on the 13th of January, is the coldest day of the year. But is this true? We sent our meteorological reporter xmetman out to check the CET series, in an attempt to find the truth. And what did he conclude after analysing the daily CET series back to 1878? Well, the short answer to the question is no, the 13th of January is not statistically the coldest day of the year in the CET series, it’s close, I make the mean minimum for that day is 0.96°C, but another Saint’s day, St Valentines on the 14th of February is, with a mean minimum of 0.67°C. The chance of frost on the 13th of January in the 137 year record since 1879 is 37.1%, whilst the chance of frost on the 14th of February is 40.3% the highest frequency of frost in the whole year. It’s also interesting to see that the dip in temperature in the lower graph (fig 2), ties in quite nicely with the first of Alexander Buchan’s cold spell s (February 7th to 14th) that he found in 1867. I think with the weather turning milder, a frost this year, might be pushing it a bit in Central England tonight.

Figure 2 – Data courtesy of the Met Office

Rather mild January 2017 in Central England

Figure 1 – Data courtesy of the Met Office

A bit of a late look back at the CET values for January 2017, which was not a particular exciting month in Central England, with a mean temperature of 3.95°C, and mean anomaly that was +0.13°C higher than the 1961-1990 long-term average. Although the first half of the month was very mild, a cold anticyclonic spell from the 18th to the 30th, brought the mean temperature for the month back closer to average. Because the Central England region lies further north than the southeast of England, it escaped the worst of the frosts that occurred there. Just five of the last twenty January’s have been colder than average.

Figure 2 – Data courtesy of the Met Office

Figure 3 – Data courtesy of the Met Office


If there are any Guardian comment readers still here at this point, I must confess that I did screw up with the comment saying that the January CET value was wrong, it was correct, well I may have rounded it down to 3.9°C, but it wasn’t the 4.45°C that I said it was but actually 3.95°C. I hadn’t downloaded the latest verified daily values for the month, and so the value that I calculated was still based on the estimates from the Met Office. The reason why I say the January value is +0.13°C above average and the article in the Guardian says it’s -0.2°C below the average, is that I used the 1861-1990 long-term average and the article quotes the 1981-2010 long-term average. 

I’m still sticking to my guns about the other comments I made about the January sunshine though, but as I said the statistics used to produce the graphic may well have been calculated from individual station rather than gridded data. If you want to download the data and work out your own anomalies please feel free to do so.

Figure 4 – Data courtesy of the Met Office

I can’t add to the comment (or delete it for that matter) so this will have to suffice as an apology for the moment.

Split personality Winters

Yes, ‘split personality Winters’ is a bit of a weird title for a blog, but I couldn’t think of a better one at the time of writing this piece. What started me investigating, was just how the Winter of 1946-1947 suddenly transformed from being so mild in its first half, to be being so cold in its second. I wondered just how many more ‘split personality’ Winters (or seasons come to that) there have been down through the years. So I enlisted the support of the trusty daily CET series, which started in 1772, to see just how unique 1946-47 actually was as ‘split personality’ Winters go.

Here is a table of the meteorological winters [DJF] with second halves colder than their first halves (fig 1), and surprise, surprise 1946-47 is only joint eighth in the list. Top of the list is the Winter of 1854-55 (fig 4), the second half of which was 6.5°C colder than the first. The Winter of 1946-47 only managed a 3.9°C difference in temperature (fig 3), but thanks in part to the coldest February in the CET series back to 1659, and to the prodigious amounts of snow that came with it, it is burnt into the psyche of anyone interested in the weather and climate of the UK.

The temperature difference between the first and second half of Winter 1985-86 and 1941-42 was also greater than that of 1946-47, but neither were memorable. It’s understandable how we forget about weather events that occurred outside our own lifetimes and never directly experienced, I’m sure that the Winter of 1854-55 was talked about for the rest of the 19th century as being the winter of two halves, much in the same way as we talk about 1946-47.

Figure 1 – Data courtesy of the Met Office

I’ve overlaid, as usual, a  moving average and a linear trend on the series of meteorological winters from 1772 to 2015 (fig 2), and it’s interesting to see how gradually the second half of winter has become colder than the first half over the past 244 years.

Figure 2 – Data courtesy of the Met Office

Figure 3 – Data courtesy of the Met Office

Figure 4 – Data courtesy of the Met Office

Lies, damned lies, and CET

I borrowed my title of course from the great Mark Twain and gave it a twist that only a blogger trying to attract new readers can come up with. But it’s basically true, you can look at any set of data and view it in such a light statically, that you can use it to prove that global warming is a reality even in Central England, or the complete opposite and prove that we are entering a new ice age. I’ve been playing around CET ever since I copied the entire monthly catalog into a program that I had written for my BBC B Computer in 1983, and I reckon one of the best ways of visualising it is by means of a 365 day centred moving average (in red), and just to reinforce that and smooth it a little I add a 14 day moving average on top of that (thick black line). I havent the faintest idea of what that’s called in statistical parlance, probably bollocks, but I like it nonetheless. So here are a few graphs to illustrate just what I’m talking about.

The no axe to grind graph

As you can see this graph (fig 1) shows the entire daily CET series since 1878. I could have extended it back to 1772, but it does get a little cramped in the x-axis department. To save any debate about what long-term average to use, I decided to generate one for the entire 1878-2015 period. Over those graphs I’ve added a simple linear trend for the whole series. The warming trend is a very modest +0.09°C per decade, a value that even our most skeptical reader shouldnt balk at.

Figure 1

The skeptics graph

Let’s use the comparatively cold 1961-1990 long-term average for the skeptics graph (fig 2), and restrict the x-axis to show only the last 20 years, which of course changes the linear trend to show a negative anomaly of -0.20°C per decade, but there’s still an awful lot of red that gives the impression that the CET series is warming.

Figure 2

But if we chose another long-term average, say for the years 1986-2015, you’ll notice (fig 3) that things get a lot bluer, even though the linear trend is showing exactly the same -0.20°C per decade cooling it just looks like there is very little going on. Wonderful, at this rate there’ll be frost fairs on the Thames gain within 50 or 60 years.

Figure 3

The true believers graph

To get the most out of the data we’ll use the 1986-2015 long-term, and using the same data restrict the x-axis to show only the recent years since 2010, which of course changes the linear trend again, and we now have a massive warming trend of +1.61°C per decade – lovely we’re now out doing GISS.

Figure 4

Personally I like the first graph. It makes best use of the maximum amount of ‘accurate’ CET data that we have. Things are warming, you don’t need a linear trend to show you that, and in the last 30 years or so that warming looks quite pronounced. I think the above graphs should please most people, be they skeptics, believers, or those just like me, sitting on the fence in the firm belief that climate is always changing just like the CET series.

18th warmest year in central England

2016 in the CET series looks to have been the 18th warmest year since the monthly series began in 1659. It was according to my statistics just +0.02°C warmer than last year. That makes it that 12 of the last 20 years have occurred in the last 20 years, if my eyesight is to be trusted. The mean temperature of 10.33°C was +0.85°C above the 1961-1990 long-term average.

Figure 1 – raw data courtesy of the Met Office

Quite a number of new day maximum records (the eleven red stars in fig 2) were broken in 2016, including the one for Christmas day. Nine new high minimum records were also set (blue diamonds (todo: change colour to red)). Only one new minimum record was set during the year for the 28th of April (blue star). The two lower charts (fig 2) show you the warm and cold spells that occurred during the year, and as you can see the whole year was dominated by warm spells, some of them occasionally very long although it’s difficult to see when viewing an entire year.

Figure 2 – raw data courtesy of the Met Office

This annual heat map (fig 3) makes it very easy to visualise the annual CET anomalies of the last 357 years at a glance.

Figure 3 – raw data courtesy of the Met Office

December 2016 in Central England

Figure 1

December 2016 can be summed up as a generally mild month where the daily CET value alternated between distinct periods of very mild and brief colder spells, especially at the start and end of the month, with a record warm Christmas day thrown in to boot just for good measure. Overall the final anomaly of +1.3°C for the month, will be forever dwarfed by the massive +5.0°C of the previous year!

Figure 2

Charts of winters past, present and yet to come


Courtesy of the Met Office

I know this is a bit late for those of you that hold to ‘meteorological’ seasons, but for the traditionalists amongst us, publishing it on the 24th of December seems just about right. Tomorrow may bring the warmest Christmas day on record or it may not, and early in the New Year we might see a cold northerly outbreak, not that the North Pole is that cold at the moment and there is a distinct lack of ice in Barents sea. But despite all this there is everything still to play for in Winter 2016/17 – well that’s what I keep telling myself – and just to cheer you all up here is what happened in January 1947!

The blue stars are extreme low minimums and the red diamonds (todo: make the red diamond blue – Bah! humbug) are extreme low maximums that are still standing.