Posts Tagged ‘Met Office’

Methodology Is Important

Monday, December 21st, 2009

I need to make an admission: I’m not really THAT bothered about climate change. There, I’ve said it. Berate me as much as you like. I know I should care, but I just don’t. Maybe it’s because it doesn’t affect me enough. In fact, that’s exactly what it is. But anyway, I’m a leftie who doesn’t really care about climate change.

That said, knowing I really should care, it’s probably about time I had a good look for myself. Conveniently, the Met Office have released raw data from their weather stations. A perfect opportunity to have a look for myself. More on that later.

However, other people have also had a look. Iain Dale has posted up some findings found by a reader of Iain Dale’s Diary which shows that Oxford is getting cooler. The methodology? To compare the highest temperature recorded in set time periods and compare them. The result? Well, to be honest, the commentary is rather garbled and unclear. It seems to suggest that Oxford is cooler now than in 1938, but that it was warmer last year. So, on balance, it would seem to argue there’s been little difference in temperature over the last 70 years – but that’s just a guess. Like I said, it’s pretty hard to work out exactly what the argument is.

There is one major problem with this method – there is no allowance for extreme values. Extreme values skew data. You can prove anything with them – especially when looking at a complicated issue like climate.

So, here is an alternative analysis of the raw data from the Oxford weather station. It is by no means perfect. It has been done very quickly. I have not subjected it to complicated scientific tests. But it has a method. Which is as follows:

I have taken the last decade (the Noughties) and taken the maximum temperature in June of every year in that decade. I would take the mean temperature, but that data is not available. I have then totalled those ten numbers (this year’s is still provisional, but is unlikely to change significantly) and divided them by ten. This gives us – wait for it – a mean maximum temperature. So, if you take the last ten years, you have an average of what that would be in any given year. Not a fantastic measurement, I accept, but it’s the best I can do with the data available. I have then repeated the process for the first decade of the 20th Century. I have picked these BEFORE looking at the data, so you can’t say I’ve taken convenient data. I have taken it randomly. I have then taken an intermediate value between the two – the 1950s.

On the other end of the scale, I have taken the lowest recorded temperature in each December and applied the same method to achieve a mean minimum temperature. This should provide a (superficial) indication of whether winters and/or summers have got warmer over the last century. By balancing the figures over a decade each time, extraneous values in either direction should have minimal effect. So, for anyone still reading, here are the figures:

Table 1.1 – Absolute Temperatures.

1900s

1950s

2000s

Absolute Minimum Temperature

0.6

-0.7

0.2

Absolute Maximum Temperature

24.3

23.9

27.1

Table 1.2 – Mean Temperatures

1900s

1950s

2000s

Mean Minimum Temperature

2.09

2.45

2.98

Mean Maximum Temperature

21.83

21.59

22.54

So, what does this show? Firstly, comparing the 1900s with the 2000s shows a clear rise in the mean maximum and minimum temperatures. The 2000s, according to this (very loose) data, were warmer. However, the lower minimum temperature in the 2000s suggests there could potentially be more fluctuation in temperature now than 100 years ago. Or it could be an anomaly.

Interestingly, the 1950s featured cooler absolute values for both June and December, the mean temperature in June was the coolest of the three, and the mean for December is halfway between the two.

Verdict: Further investigation required into the 1950s. Investigate possible reasons for anomalies. Clear 2000s clearly warmer than 1900s according to this data. Further comparison of the other decades in between needed to verify this data is not an anomaly itself.

Most of all, though, we need much more detailed data – day by day data is required much more than month by month if any meaningful data is to be extracted from this exercise.

Feel free to comment – healthy (and polite) debate welcomed, but leave essays for your own sites, please. I have no agenda. I just like looking at numbers. Thanks for reading.