Is UK precipitation really becoming more extreme?

In recent posts here and here we looked into the question of whether the UK’s recent wet weather was unprecedented and concluded that it wasn’t. But is it symptomatic of an upward trend in the intensity and frequency of extreme rainfall events caused by climate change, as some have claimed?

This post addresses this question in two sections. The first summarizes how different organizations define “extreme weather” (or “climate”). The second applies four of these definitions to the UK Met Office’s monthly England & Wales rainfall record to determine a) which months qualify as “extreme” and b) whether there has been an increase in the frequency and intensity of extreme rainfall months since 1766, when the record begins. The conclusions are:

• There is no accepted definition of “extreme weather”. Extreme weather is whatever one wants it to be.

• There has been no statistically significant increase in extreme monthly rainfall in England and Wales for at least the last 250 years regardless of how “extreme” is defined.

• Since England and Wales make up 62% of the surface area of the UK it is likely that this conclusion applies to the UK as a whole, although this can’t be confirmed.

1. Definitions of “Extreme”

We begin with the BBC, which puts the problem in perspective:

Extreme weather is when a weather event is significantly different from the average or usual weather pattern. This may take place over one day or a period of time.

Here the BBC identifies three issues: a) how long should the period of time be, b) how to define “significantly different” and c) how to define “average”. In this review the time period (one month) is fixed by the data and the question of how to define “average” is discussed later. The main question is how to define “significant”, and while there is general agreement that the way to do this is by setting a percentile threshold based on the historical data distribution there are differences of opinion of over an order of magnitude as to where the threshold should be set, as exemplified below:

The IPCC one-in-ten definition :

An extreme weather event is an event that is rare within its statistical reference distribution at a particular place. Definitions of rare vary, but an extreme weather event would normally be as rare as or rarer than the 10th or 90th percentile. By definition, the characteristics of what is called extreme weather may vary from place to place. An extreme climate event is an average of a number of weather events over a certain period of time, an average which is itself extreme (e.g. rainfall over a season).

According to this definition any month above the 90th percentile is rare enough to be considered extreme. I don’t think one month in ten comes close to qualifying as extreme, but I use the IPCC definition anyway. (Note also the uncertainty as to the difference between an extreme “weather” and an extreme “climate” event. Monthly rainfall would appear to fall somewhere in between the two.)

The NOAA one-in-twenty definition

Extreme weather includes unusual, severe or unseasonal weather; weather at the extremes of the historical distribution—the range that has been seen in the past. Often, extreme events are based on a location’s recorded weather history and defined as lying in the most unusual ten percent.

The “most unusual” ten percent includes the dry months at the bottom of the distribution as well as the wet months at the top, so here we are looking at the data above the 95th percentile, not the 90th percentile. But defining one month in twenty as “extreme” maybe still overdoes it a little.

The UK Met Office one-in-a-hundred definition

The Met Office rarely refers to “extreme” weather, preferring the term “severe”, which it fails to define. But after some searching I came across this:

The meteorological or statistical definition of extreme weather events is events at the extremes (or edges) of the complete range of weather experienced in the past. Defined in this way, extreme weather events include, but are not limited to, severe events like heatwaves or intense rainfall. UKCP09 (UK Climate Projection 2009) provides probabilistic projections of changes in climate extremes that are based on the 99th percentile, or the value that is exceeded on only 1 in 100 days.

The 99th percentile, or one-event-in-a-hundred, is certainly a more robust definition of “extreme” than the one-in-ten or one-in twenty definitions of the IPCC and NOAA.

The one-in-three-hundred “extreme heat” definition

The recent decade has seen an exceptional number of extreme heat waves around the world that caused severe damage to society and ecosystems. These events were highly unusual with temperatures typically three standard deviations warmer than the local climatology lasting for several weeks.

This quote is from the paper by Coumou & Robertson linked to in the heading, but more than three standard deviations above the mean, otherwise known as the “three-sigma” limit, seems to be the accepted yardstick for defining extreme heat waves. It defines the top 0.3% of the data, or in this case one extreme month in every 333, which makes it over thirty times more restrictive than the IPCC definition. It’s not clear why it should be applied to extreme heat and not to extreme rainfall, so I have incorporated it in the review.

2. Identifying extreme events in the UK rainfall record

The monthly UKMO rainfall record for England and Wales, which begins in 1766, is plotted in Figure 1 (the UKMO Northern Ireland and Scotland records don’t begin until 1931): The yellow trend line shows only a very minor increase in average precipitation over the last 250 years, but that doesn’t necessarily mean there has been no increase in extreme rainfall events:

Figure 1: UKMO England & Wales monthly rainfall record

Two points should be noted before proceeding:

• Monthly rainfall does not tell us how many floods occurred in the month, although arguably the worst UK flood since 1766 occurred in March 1947 , the month which this review identifies as having had the most extreme precipitation. There are no “monthly number of flood” records for the UK that I am aware of.

• On average the UK receives about 60% more rain in October than in April, so percentile thresholds set relative to mean annual rainfall will tend to make a wet October seem more anomalous than it really is and a wet April less so. Separate “extreme” thresholds are therefore defined for each month using the mean monthly rainfall and standard deviations shown in Figure 2:

Figure 2: Monthly mean rainfall and standard deviation from Figure 1 data.

Now to the results. Figure 3 shows the extreme rainfall months defined by the four thresholds (click to enlarge). The values plotted are the differences in millimeters between observed monthly rainfall and the monthly extreme threshold, i.e. they are a measure of how extreme the monthly rainfall was. Months when rainfall is below the threshold are ignored:

Figure 3: Extreme monthly rainfall events

None of the four plots provides any good evidence for an increase in the intensity of extreme monthly rainfall events since 1766. The trend lines do show shallow positive gradients, but because of the large amount of scatter none of them comes close to being statistically significant. The R squared values of zero or close to it in fact suggest that there has been little if any change in the intensity of extreme monthly rainfall in England and Wales over the last 250 years.

Figure 4, which plots the number of extreme events per decade for the 90th, 95th and 99th percentile thresholds (the three-sigma case does not have enough points to plot), provides information on whether extreme rainfall events are becoming more frequent. The trend lines for the 90th and 95th percentile cases show shallow upward gradients but again at very low R squared values, and the trend line for the 99th percentile case shows a shallow downward gradient. In summary, there is also no good evidence to suggest that extreme rainfall events in England and Wales are any more frequent now than they were in 1766.

Figure 4: Number of extreme monthly rainfall events per decade

Concluding comments:

In a recent Financial Times article the following prominent scientists are reported as having made the following statements:

Professor Alan Jenkins, Deputy Director of the Centre for Ecology and Hydrology : “We are absolutely convinced that there is weighty scientific evidence that the recent extreme rainfall has been impacted by climate change.”

All climate experts who have analysed this December’s record-breaking temperatures and rainfall in Britain see a significant role for global warming in addition to natural variability in the weather, according to Professor Adam Scaife, head of long-range prediction at the UK Met Office.

Professor Myles Allen, a climate scientist at Oxford university, said predictions of heavier winter rainfall in mid-latitudes as the world warms go back 25 years. “So none of this should really come as a surprise to anyone,” he added. “The hopefully diminishing band of diehards who continue to dismiss climate change concerns as ‘just green crap’ have mud on their hands.”

These gentlemen are invited to point out the flaws in my analysis.

This entry was posted in Climate change and tagged , , , , , . Bookmark the permalink.

28 Responses to Is UK precipitation really becoming more extreme?

  1. Tom Bates says:

    Wet and dry are not unusual. Great Britain does have a problem. A lot more people living in flood plains and a government which refuses to build water storage dams or kick the people out of the flood plain. Here is a study of wet and dry in europe

    • oldfossil says:

      Thanks for the link. Not just the flood plain problem Tom, it’s the river valley management too. A bit of dredging and damming would not go amiss, but it’s not going to happen. Some of those Defra people are abnormally and perversely thick. Cameron et al are actually encouraging building in flood plains. The insurance on flood-prone homes is subsidised by a levy on low risk homes built by sensible people well above flood lines.

  2. Don’t expect a response from the three so-called “climate experts”.

    • I wouldn’t expect one. These guys are sailing into a stiff scientific headwind. They don’t even get any support from the IPCC:

      “Overall, the most robust global changes in climate extremes are seen in measures of daily temperature, including to some extent, heat waves. There is limited evidence of changes in extremes associated with other climate variables since the mid-20th century. Precipitation extremes also appear to be increasing, but there is large spatial variability ….. there continues to be a lack of evidence and thus low confidence regarding the sign of trend in the magnitude and/or frequency of floods on a global scale” (From the AR5).

  3. Doug Brodie says:

    In April 2014, Channel 4 TV screened a series entitled “Britain’s Most Extreme Weather” by weatherman Alex Beresford. It was a very balanced series, discounting its obligatory reverential nods towards “climate change” and “global warming”. The four episodes covered storms, cold, floods and heat. Each episode asked if recent examples of severe weather in the UK were unprecedented and in each case the answer was a clear no. See

    • steve says:

      Last week the BBC had John Hammond, their leading weather presenter explaining that the jet stream travels were probably a result of reduced arctic ice and temperature gradients from a warming pole and, presumably less warming parts to the south. He chose a lady expert, whose name I have forgotten. Anyway, the propaganda was effective.

  4. Jim Brough says:

    Thanks for the analyses.
    There is not a sign of a hockey stick in the graphs in spite of increasing atmospheric CO2 levels.

    Looking at the mandatory windscreen sticker for a new car to describe its CO2 emissions it says that CO2 emissions are the main cause of climate change. I speak from Australia, does this apply in UK or EU ?

  5. Euan Mearns says:

    One thing I can say from own experience is that if you are affected by or are proximal to storm or flood damage one becomes more receptive to ideas about “extreme” and wanting to blame this on something. In my case the quiet Sun.

    The Dee flood of 30 Dec 2015 and its impact were down to a number of factors.

    1. Meandering jet stream that brought cold snowy weather and then very warm weather succeeded by continual rain and then more snow. The rain in Aberdeen was not exceptionally heavy, it just went on for days/weeks.

    2. It turns out that Scotland had record rain in January since 1931, but probably exceeded this in centuries gone by.

    3. The flood on the River Dee was also a record in that time frame, but nor for all time.

    4. One aspect of this is that rainfall is NEVER evenly spread. It is inevitable that whenever one approaches a record high for a country like Scotland that some parts of the country will experience record busting highs and extreme events experienced by a minority of the people. This is the statistical view that I would expect government scientists to be expressing. Instead we get Red Top alarmist BS from individuals who don’t deserve the title scientist at all.

    5. The impact of extreme rainfall by way of flooding is highly variable. Take fisherman’s bothies along the river Dee as an example. If a bothie was more than 5M above normal level it would likely be untouched. Below that level about 50% chance that it was simply swept away.

    6. Rivers are active geological features that like to change course all the time. Its what they do. But Man has built infrastructure along their banks and since doing so has done all He can to prevent the rivers from going where they want to go. They can become like a coiled spring, building and holding stress, waiting for an opportunity to return to their normal state.

    • Roger Andrews says:

      One thing I can say from own experience is that if you are affected by or are proximal to storm or flood damage one becomes more receptive to ideas about “extreme” and wanting to blame this on something.

      You’re not alone, Euan. A lot of people have been looking for someone or something to blame for the floods, and the UK’s “climate scientists” have jumped on this golden opportunity to blame them on man-made climate change. And according to this article they’ve succeeded in convincing 37% of the public that man-made climate change was indeed the cause:

  6. john Pitman says:

    The magnitude, duration, frequency and distribution of UK rainfall events is a aspect our understanding and prediction of flood events, and you will find a useful overview in this 2009 paper by Rodda et al:
    Hope that this helps.

  7. Yvan Dutil says:

    I would like to point out that your analysis assume a gaussian distribution. True rainfall follow is much more asymmetric Therefore, extreme are more extreme and the difference are far for trivial to show up. .

    • Roger Andrews says:

      My analysis assumes a gaussian distribution only for the 3-sigma case. I picked the 90, 95% and 99% thresholds off monthly data lists.

    • john Pitman says:

      As a retired hydrogeologist, we used extreme value distributions when analysing extreme rainfall and flood events. T
      The most widely used distribution after a Gaussian log-normal.-where we used the logs of the variable of interest- are the Gumbel , Fréchet or Weibull Distributions. There is a nice discussion these at which you might enjoy.

      • John: Here are two graphs showing the distribution of monthly E&W rainfall since 1766 (3,000 months) as a histogram and a log-transformed cumulative frequency plot. Where does “extreme” start?

        • Peter Lang says:

          Where does extreme start? Depends on the context. 1:10,000 year flood for spillways for large dams. 1:100 year flood common for other public facilities, e.g. roads..

        • john Pitman says:

          The reality is that the distribution of data versus time is continuous, from the lowest recorded value to the highest value. The objective of this kind of data analysis is to determine the return period, and therefore the probability of its likely occurrence, for the analysed data.
          Of course. using extremal analysis one can calculate the probabilities and return periods well outside the actual data range, although that can de dangerous!

          In the UK, one of the current approaches to extreme rainfall events and flood risk uses SUDS (Sustainable urban drainage) analysis package


          which uses the M5-60 Rainfall Depth (mm) which is the rainfall depth for the 60 minutes 5 years return period event. M5-60 value is a parameter used in deriving rainfall depths for other return periods and durations. Flood Studies Report (NERC, 1975).
          The other parameter that is also used is the ‘r’ Ratio M5-60/M5-2 day Variable “r” represents the ratio of the rainfall depth of the 60 minute to the 2 day, 5 year rainfall event..

          In other word, it all depends on which value the engineer/hydrologist/planner define as important!
          The SUDS site above gives fuller details, and it is possible using the M5-60 and the M5-2 day rainfall amounts to determine the recurrence interval e.g. 1 in 50 years, 1 in 100 years etc. It all depends on what you are designing, their cost and size and expected life-time.

          If you are simply interested in the largest value recoded, as the earlier parts of this discussion implied, then we can use extremal analysis determine its probably and return period.

        • Roger Andrews says:


          Early in the post I stated “There is no accepted definition of “extreme weather”. Extreme weather is whatever one wants it to be.”

          The two graphs demonstrate this. The histogram shows a “fat right tail” distribution similar to the Bayesian-transformed distributions of climate sensitivity estimates that statisticians continue to argue about. The log-transformed plot shows a lognormal distribution up to maybe the 60th percentile but progressively departs from lognormality above that threshold. Ideally what we would like to see is an abrupt inflection in the plot that defines what we in mining call a “high-grade outlier” population, and if there was one we could use it to define “extreme” or whatever other adjective we might deem appropriate. But there isn’t one. We are at liberty to use any threshold above the 60th percentile to define “rare”, “severe”, “extreme”, “exceptional” or even “unprecedented” events, as the whim takes us. The process is totally subjective.

          What’s the solution? Well, if we are going to continue to talk about extreme events then everyone has to agree on an extreme event threshold even though it may have no firm scientific basis (the UNFCCC has already done this with its 2C global warming target, which is a number picked out of the air). The 90th and 95th percentiles define too many events to be considered extreme and the three-sigma threshold too few, so I would be inclined to go with the 99th percentile, or one event in 100. And may the Gods of Statistics forgive me. 😉

        • Euan Mearns says:

          My Wag on this would be that “extreme” would require a new bump to the right of the existing distribution that would indicate the creation of a new regime as opposed to repetition of events that have happened before. But we could expect that for all those months 150 or over that owing to uneven distribution of precipitation, some areas would experience flooding and the sensation of extreme events, especially if this is reinforced by the BBC and Dame Julia.

          The other interesting thing is at the low ppt end of the distribution, the country and areas within it experience no rain fall. That is also an “extreme” event by “their” definition. Drought! In the UK this can normally be fixed by irrigation – farmers pump water from rivers.

          But the main problem Roger is that Holocene climate has been completely non-varying everywhere and Man and all other creatures great and small have lost their ability to adapt to any regime that does not fall exactly on the mean. I watched a nature program the other night called “Wild Canada” narrated by David (past his sell by date) Attenborough. It was about polar bears. One passage showed a perfectly healthy polar bear with two cubs eating seaweed and berries augmented by migrating Char (the size of salmon). Looked pretty like a white grizzly to me. But the narrative went along the lines…. polar bears have taken thousands of years to evolve to survive in a stable Arctic climate (I kid you not) and are now struggling to survive on berries and fish. Given the choice, what would you prefer. Blackberry crumble and smoked Char in the comfort of warm dry land, or, wandering about on pack ice looking for a seal?

          • As I understand it the polar bear evolved from the grizzly bear and the two are so genetically similar they can still mate, which I guess would result in the birth of light brown grolar or pizzly bears, take your pick. 😉

          • Roger Andrews says:

            But the main problem Roger is that Holocene climate has been completely non-varying everywhere and Man and all other creatures great and small have lost their ability to adapt to any regime that does not fall exactly on the mean.


          • Euan Mearns says:

            Satire my friend 😉

          • john Pitman says:

            I really don’t want to bring it up, but we also need to be aware of the sampling problem, and the fact that most natural phenomena are non-stationary, so that in reality the population mean and standard deviation are continually changing even if we abide to the moving 10 year 30 years mean for our observations.

          • No reason why you shouldn’t bring it up, but in this case non-stationarity doesn’t make much difference. Means and standard deviations for the E&W monthly rainfall record calculated over moving 30-year intervals don’t change that much. The two-sigma (~95%) threshold ranges only from 140mm to 155mm over the entire period of record and is only a few mm higher now than it was in 1780.

          • Grant says:

            I have often wondered why seals breeding in the Isles of the UK are considered nice, cuddly and protection worthy when those in the Arctic are evidently there only as food for polar bears.

            What have people got against polar seals? Why do the “charities” not set up campaigns to protect the majority of the Arctic wildlife from rampaging polar bears?

  8. garethbeer says:

    Gridwatch UK & FR – demand flat out close to 52gw (wind dropped to 0.4 Gw)
    uk, 84gw FR.

  9. Owen says:

    1947 was one of the worst flood years in Ireland and UK in living memory, some photos here :

    What made it worse than this winter’s floods was the amount of snow that lay on the ground when the heavy rains came.

  10. Peter Lang says:

    Off Topic:

    “On the sad passing of Bob Carter, a great scientist and friend”

    My comment:

    Bob Carter did an enormous amount to help save us all from stupidity. He was one of four non-climate scientists who convinced Senator Steve Fielding (the only engineer in Australia’s Parliament at the time) that the climate scientists’ projections of catastrophic human caused global warming were not based on sound objective, analysis of the relevant evidence and were highly suspect. This was the beginnings of the undoing of Prime Minister Kevin Rudd, Environment Minister Penny Wong and the advocates for the ‘Carbon Pollution Reduction Scheme’ and later the Gillard Carbon Tax.

    Thank you Bob Carter for helping to bring some rational, objective analysis to the debate. Thank you for the long view you brought to the debate that only geologists can bring. Thank you for so much!

  11. Greg says:

    Yes, RIP Bob Carter!

Comments are closed.