A Century Of UK/NI Rainfall (part 7)

A butcher's at drought and deluge over the last 100 years. Are things getting worse, and what does ‘worse’ even mean?

In part 6 of this series I finally got to producing a time series for mean monthly rainfall for the UK for the period 1883 - 2021 using records held by Armagh, Durham, Oxford, Sheffield and Stornoway Airport observatories, these being dubbed the ‘big five’ a.k.a. UK5 in graphs and tables. We observed that the rainfall record over 139 years has remained remarkably consistent, with the possibility of a positive trend from 1950 onward. Counts of drought months revealed things were certainly not getting drier as many alarmists claim and, if anything, things are getting wetter.

Rainfall on a small island caught between a warm ocean and cold northerly sea is pretty darn variable, which is why there is both sunscreen and wellington boots in the back of my old Land Rover. Not just any old wellies either - herewith a selfie of my summer and winter wellies that prove I take rain seriously:

But Five Is Not A Lot, Is It?

No it’s not. The trouble is long series weather stations are pretty thin on the ground and, as far as I am aware, these are the only stations that hold continuous records dating back to 1900 that are still in operation today. But there’s another issue here, and this is best illustrated by a slide:

Here we see the UK mean rainfall derived by averaging monthly data for the big five (dark blue line) together with rainfall for four stations that came online in recent years. It’s pretty wet up at Dunstaffnage (as we noted previously) and if we add these records to our national mean it will look as though the climate changed dramatically in 1972. By adding in Ballypatrick Forest, Camborne and Cardiff Bute Park we can ensure we seriously bias the rainfall record good and proper, swap drought alarmism for deluge alarmism and go full-on climate activist.

A Solution

There is an elegant solution to this problem and this is to derive a series of rainfall anomalies. We do this by defining a climatological normal, being a 30-year reference period from which deviations are calculated. The period 1961-1990 is the favoured normal for the big players despite 1991-2020 being available for revision. The trouble is four of the stations in my sample weren’t operating in 1961 so we can’t really calculate their 1961-1990 mean value unless we bend the rules and permit truncated normalisation (slap on wrists!).

At this point I shall sip my lemon tea and allow the concept of bias introduced through truncated normalisation to seep into the collective consciousness. I shall then pose a simple question: is this what they’ve been doing with the temperature record?

Being brought up proper, I shall use the period 1991-2020 as my climatological normal in the derivation of rainfall anomalies for the 34 stations in my sample that are still operative. To illustrate normals in action here’s a plot of monthly rainfall as recorded at Cambridge NIAB and Dunstaffenage followed by the average of both series:

As we can clearly see, if I try to create a mean series from these two station records we end up with an artefactual hike in rainfall from Jun 1971 onward when a much wetter Dunstaffnage got going. This isn’t climate change in action - it’s undiluted idiocy!

Here’s the same data but with station records normalised to mean values obtained over the reference period of 1991-2020:

The two series now coincide and we can calculate their combined mean anomaly without introducing bias caused by a much wetter Dunstaffnage coming online in 1971. This method of normalisation allows us to combine wet stations with dry yet preserves variability in the rainfall record, for not only is Dunstaffnage wetter than Cambridge but it is also wilder!

The Big 5 Become the Big 34

After calculating the 1991-2020 individual anomaly series for all 34 stations we can go about calculating their grand mean anomaly. Before we look at this we better take a look at the sample size over time:

There we go! We started out in 1853 with just Armagh and Oxford putting their bucket out and ended-up with 34 stations putting their bucket out, with very few missing months along the way.

There were concerted efforts to get stations operative as WWI and WWII loomed for pretty obvious reasons but what’s not so obvious is the sizeable surge during 1959 when 6 stations were opened (Braemar, Cambridge NIAB, Eastbourne, Newton Rigg, Paisley & Sutton Bonington). This profile serves as a proxy for the certainty that we may attach to the grand mean UK34 anomaly, which shall now be revealed:

Umpteen hours and a lot of biscuits later we finally arrive at something we can have a reasonable amount of faith in!

The red line is another of those ordinary least squares regressions (OLSR) that climate folk seem to slam through everything. This is as flat as a pancake (possibly flatter than mine, especially if I miss the toss) and the coefficient of +0.002mm per year is utterly insignificant (p=0.889). This means we are back where we started some 169 years later!

I’m not keen on straight line thinking and so I’ve crayoned some locally estimated scatterplot smoothing (LOESS) in green. This gives a kinky flavour to things, and we now appear to have a slight decline in UK/NI monthly rainfall from 1870 down to 1965 after which there is a noticeable upturn that was mentioned in part 6. This is most pleasing because it means my ‘big 5’ were doing a jolly good job of tracking UK/NI rainfall all along!

So where is that upturn taking us next?

My money is on yet another of those cycles that alarmists prefer not to talk about. There are all sorts of contenders for a driver such as oceanic oscillations, solar activity, and cosmic ray flux and I shall be looking at a few of these now we have a decent time series for rainfall.

Kettle On!