What is a river without rain?

This is a guest post by Ben Gillespie, a doctoral researcher at University of Leeds

You would have to live in bubble not to realise that rivers in South West England have been flooding. These recent events have bought untold stress to the people who live locally. But, from a management perspective, it is important not just to focus on the water within two banks (and without as the case may be).


Met Office, Hadley Centre regions.
© Met Office (http://www.metoffice.gov.uk/hadobs/hadukp/)

Media headlines have suggested that the recent rainfall is record breaking and that large rainfall events are becoming more frequent in some areas of the UK. It is therefore a good time to take a closer look at rainfall data for South West England to get an idea for what’s happening there.

The Hadley Centre hold records of daily rainfall by region which you can view and download here. The data for the South West start on the 1st January 1931 and (currently) run until the end of January 2014.

An overview since 1931…

So, lets have a first look at the data (units: mm/ day)…

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##    0.00    0.04    0.90    2.85    3.90   57.90

Lets see which day the maximum amount of rainfall was…

##              day day.num rain       week    month year
## 14089 1969-07-28   14089 57.9 1969-07-24 Jul 1969 1969

OK, the summer of ’69! To see if there are any obvious trends, here’s a plot all the data…1

Hmmm, nothing sticking out to me. So, what does a linear model fitted to all the data tell us?

Well… it turns out there is a very slight, statistically significant (at p<0.05) decrease over time (estimate = 6.4225 × 10-6 p = 0.0245) (i.e. an increase of 1 day will, on average result in 6.4225 × 10-6 less mm of rain). However, this tiny decrease is within the limitations of the equipment used to measure rainfall so we can disregard it.

How about just looking at days where ‘heavy’ rainfall occurred (arbitrarily defined as days where rainfall is above 20mm)?

2A linear model suggests a non-significant, very slight reduction in rainfall over time (estimate =  2.3572 × 10-5 p = 0.4934). Has there been any change in the number of days per year where ‘heavy’ rainfall occurs?

3It looks as if there may have been a general increase (certainly, 1989, 2000 and 2012 have had their fair share of heavy rainfall days). However, a linear model suggests this increase is non-significant (estimate = 0.0173 p = 0.0569).

Periodic summaries

Recent floods have been attributed to prolonged periods of heavy rainfall – it’s therefore a good idea to look at aggregates of rainfall over periods longer than 1 day.

Weekly rainfall

Let’s take a look at total rainfall per week since 1931…

4A linear model applied to total weekly rain suggests a non-significant, very slight decrease (estimate = 4.4286 × 10-5 p = 0.1467).

Monthly rainfall

5Now we have a non-significant increase in mean monthly rainfall (estimate = 0.0704 p = 0.2583).

Annual rainfall

6OK, so at an annual scale we can see a non-significant increase (estimate = 0.6778 p = 0.277).

But what about variation?

Climate change predictions suggest that we will experience more extremes (i.e. more periods of high and low rainfall, more variation).

So, assuming standard deviation is a good measure of variation, let’s look at it calculated based on daily rainfall per year…

7Well, since 1990 we’ve had two years with record high variation (cf pre 1990) and, by eye it looks like variation might be increasing, however, there is no evidence for a statistically significant increase (estimate = 0.0032 p = 0.1221)

We can also look at weekly and monthly variation…

…and neither of these show statistically significant changes over time (estimates = 5.6116 × 10-6 & 0.0034, p = 0.1739 & 0.0844 respectively).

To summarise

Flooding is currently a hot topic. The link between flooding and rainfall is clear, however, the link between rainfall and climate change is less clear as commented on by CEH.

This very rudimentary look at daily rainfall records for South West England since 1931 provides a little insight into ‘long-term’ mean and variation.

Mean daily rainfall since the start of the Hadley Centre dataset is 2.85mm. The highest volume of rainfall recorded was in 1969 where almost 60cm fell on the 28th July.

Overall, this assessment provided no evidence for significant change when viewed at daily, weekly, monthly and annual scales. There was also no evidence for significant change in variation in daily rainfall when viewed at weekly, monthly and annual scales.


Rainfall forecast for 15/02/2014.
© Environment Agency, Dave Throup (twitter: @davethroupea)

These points raise important questions:

  1. How does February 2014 fit into the picture? We’ve seen yet more rain – this will be the subject of a following post.
  2. If we are seeing more flooding in South West England (as suggested here), why is this when there doesn’t appear to be evidence for increased rainfall?
  3. How/ when will we know whether climate change is occurring with regards to rainfall records? Should we rely on statistically significant trends? Do we have enough baseline data to compare with?
  4. How can we prepare for the future with regards to predicted patterns in rainfall and river management?


The analysis carried out here is simple and could be criticised in the following ways:

i) The use of linear models may not be the most appropriate in this situation;

ii) The data represent the whole of South West England – they do not represent areas smaller than this where localised rainfall trends may differ. You can read more about how the data are calculated by the Hadley Centre here

iii) I didn’t look at the data split by month or season – other authors have done so and have found interesting results: see here and here for more.

The analysis and post was done in R using R Markdown and Knitr – you can download the script here. Best to open it in RStudio.

I’m not a climate change denier!

About Ben Gillespie

Researching the impact of reservoirs & environmental flows on downstream hydrology, physical-chemistry, sediment transport & benthic macroinvertebrates. Twitter: @RiversBenG Web: http://www.geog.leeds.ac.uk/people/b.gillespie LinkedIn: http://goo.gl/0ObRnc
This entry was posted in Flooding, Hydrology, Meteorology. Bookmark the permalink.

2 Responses to What is a river without rain?

  1. Jake says:

    Surely that would 60mm fell on the 20th July 1969 (summary note)?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s