One of the very first articles that I read in the Weather Magazine as a young outstation assistant was entitled “A simple summer index with an illustration for summer 1971” by R. Murray which was published in April 1972. Now over forty years later as a retired programmer with the Met Office, I have decided to revisit the summer index and update his record.
I have a number of advantages that Murray could only dream of, and they are a powerful personal computer, up to date freely accessible climate data, and of course the Internet to access that data from. The Met Office provide the data in the form of monthly regional and national gridded climate data back to 1910. This provides you with all the temperature, rainfall and sunshine that you require to calculate a summer index, and the advantage of this data is that you can generate a summer index not only for the UK, but for England, Wales, Scotland, Northern Ireland as well as any of the other twelve other regional areas.
All that was required to access the latest climate data from the Met Office website was internet access. The program converted the text files I downloaded into a data structure to hold each month’s mean temperature, total rainfall and sunshine values necessary to compute each year’s summer index. The slightly tricky bit was to calculate the quintiles of each month’s mean maximum temperature, and the terciles of the total sunshine and rainfall that the formula requires.
The summer index [SI]
SI = 3T + 5S – 5R – 9m
- m = number of months
- T = sum over m months of quintiles of monthly mean temperature
- S = sum over m months of terciles of monthly sunshine
- R = sum over m months of terciles of monthly rainfall
Quintiles and terciles are statistical terms used with any series of data arranged in order of magnitude. Rainfall is conventionally divided into three equal classes; the driest third being tercile 1, and the wettest tercile 3. With temperature the data is divided into five equal classes; quintile 1 refers to the coldest and quintile 5 to the warmest. There is a drawback in using the climate data series from the Met Office, although the temperature and rainfall series extend back to 1910, the sunshine series is only available from 1929, so I was unable to reach back quite to 1881 as Murray did originally. Using Murray’s formula the absolute best ‘meteorological’ summer can score a maximum SI of 48, and the absolute worst a SI of -48.
A simple summer index
The summer index was first proposed by Davis in 1968, its beauty lies in its simplicity, but a good summer can be ruined by a wet last week in Autumn, so the index is far from perfect. You could dream up a summer index that looked more closely at daily values of temperature, rainfall and sunshine, but at the moment the Met Office do not make daily regional climate data available, so for now monthly data will have to suffice.
How do you define what constitutes a ‘good’ summer? It is very subjective, and as we grow older, it may have less to do with weather, and more to do with other things that are going on in our lives. Keeping it strictly meteorological, and if you’re older than 70, you probably look back at the summer of 1959 as being the best, older than 50 and it’s highly likely that 1976 will be your perfect summer, younger still and it may well be the summer of 1995 or 2003. The worst summer in contrast is not so easy to quantify, and many people if asked will struggle to name the worst summer that they have experienced in their lifetime.
The ‘best’ summers
As you can see from the table of best summers (fig 1), 1976 tops the Summer index back to 1929 for the UK which probably comes as no great to surprise to many. In fact its score is the perfect maximum of 48.
The extended summer index
The beauty of the algorithm is that you can also calculate an extended Summer index (May through to September), which gives an entirely different slant on what was the best summer. The table below (fig 2) shows that 1959 has the highest extended summer index of 62 (out of a possible 80), and that 1976 is only eleventh in the rankings, with an index of 28. So why was the extended summer of 1976 so much worse? If you compare the various quintiles and terciles for 1976 and 1959, you will see that 1976 was in fact duller and wetter than 1959 in May and September so the extended index score was reduced.
Because the data is also split into regional as well as national values, it’s easy to compare what kind of summer other parts of the UK experienced. As you can see in the breakdown of the extended summer of 1959 (fig 3), the northeast of England and the Midlands score a very high 72, whilst somewhat lower down the rankings came the north and west of Scotland.
The ‘worst’ summer
The summer of 1954 has the lowest summer index -48 of all summers in the UK since 1929 (fig 4), you just can’t get a summer index lower than -48. 1954 was the very antithesis of 1976, it was not only wet, it was cold and dull. Even if you compare 1954 using the extended summer index, it’s still has the lowest index of -64 for the UK. Just to show you how poor that summer was, here are the headlines for each month of the extended summer of 1954 that I’ve copied from the Monthly Weather Report.
- May 1954 mainly dull and wet, with frequent thunderstorms; large variations of temperature.
- June 1954 mainly dull and cool; periods of rain, heavy at times.
- July 1954 notably cool and dull; wet in some areas.
- August 1954 cool and dull, mainly wet in England, Wales and southern Scotland.
- September 1954 cool and unsettled; wet in most areas; sunny on the whole.
What about summer 2016?
To a lot of people, especially those in the south and east, summer 2016 was very good, and the eagle-eyed amongst you will have noticed that it ranked joint 19th in the table of meteorological summers, and joint 11th in the table of extended summers, just behind 1976 and on a par with 1947. But how did the extended summer of 2016 look nationally and regionally? According to the extended summer index East Anglia fared best in 2016 with a summer index of 42 (fig 5), closely followed by southern England and the Midlands.
Finally, here is a graphical way of looking at the summer’s since 1929 as a whole by means of two scatter graphs. The first graph plots temperature against rainfall (fig 6), whilst the second graph plots temperature against sunshine (fig 7). They show at a glance just how each summer compares with each other, for instance although 1995 was very slightly drier than 1976, it was slightly less sunny and not as warm.
Are summers getting any better?
The one remaining question is – are summers getting better? Well with the help of another chart (fig 8) I’ve plotted the summer index and overlaid it with a five-year centred moving average (dashed line with a yellow outline). I’ve also added a simple linear trend (dashed black line), although climatologically this may be frowned upon (because any trend certainly wouldn’t be linear), it does help to highlight the increase in summer index that there has been since 1929. So the short answer to that question is yes, the summer index has increased over the last 87 years, whether that equates to better summers, I will leave that for you to decide.
Davis, N. E.1968. An optimum summer index. Weather 23: 305-317.
Murray, R.1972.A simple summer index with an illustration for summer 1971.Weather 45:161-169.
Many of my regular readers may remember that I wrote an article about the ‘Summer Index’ last July. I fully intended to get a fuller version of it published in the Weather Magazine of the Royal Met Society. I lost heart in the end, I don’t think I write in the way that they like, and I’m not good at writing in any other way. I went to a bit of trouble in putting the article together, so instead of it just languishing in a folder named ‘Weather Magazine’ in my Google drive account, I thought that I’d publish it in my blog just for posterity. The other thing that I like about a blog is that you can always fix typo’s or mistakes which you can’t do in a printed magazine, hopefully there are not too many of those.