Tuesday, 23 October 2012

Engineering – and earthquakes

Silence for over a month – I’ve been catching up after a couple of trips so here we go again. There’s been some discussion recently about the need and opportunities for engineering projects in the UK. I say ‘some discussion’ because part of the problem is that engineering has a very low profile here.  I was going to research some exciting, interesting, leading-edge British engineering achievements or plans (not HS2!) and then came the news about the Italian earthquake convictions. While not strictly an engineering story, I think it epitomises the lack of understanding by one part of the establishment of another. Do we have another “two cultures” here?

The chance – probability – of an earthquake is a statistical thing. Have a look at the work done in California – another quake region – on the Southern California Earthquake Center web site here. To quote one section: “According to the new forecast, California has a 99.7% chance of having a magnitude 6.7 or larger earthquake during the next 30 years (see Figure 1). The likelihood of an even more powerful quake of magnitude 7.5 or greater in the next 30 years is 46%. Such a quake is more likely to occur in the southern half of the state (37% chance in 30 years) than in the northern half (15% chance in 30 years)” Now, this forecast was made in 2007 so what’s the chance of an earthquake there tomorrow?

While a little simplistic for the chance of an earthquake, one of the basic statistical models is the normal distribution. This is the distribution of probabilities around an average. Lots of things follow such a distribution – heights of a group of people, number of items in supermarket trolleys, in fact anything that is ‘normal’ The average is the most likely but some measurements will vary from the average – the further away, the fewer there are. The Standard Deviation of the distribution measures how spread out are the figures: in a normal distribution 68% of measurements will be within one standard deviation either side of the average. Go to two standard deviations and 95% are within this and at three, 99.7%. Standard Deviation is abbreviated to the Greek letter sigma (r) There’s a wonderful explanation of Standard Deviation on the Maths is fun web site here. There’s also a great model on the web which generates a normal distribution in front of your eyes. I left it running while drafting this entry:

If you want a really full explanation of Standard Deviation, have a look at the Wikipedia entry, but it dies get very dense.

Who was it who said “Lies, damned lies and statistics”?

No comments: