On Sun, 4 Jan 2004 15:00:19 +0000, John Hall wrote in


In article ,
Waghorn writes:
OK, so the WMO are a bit slow on this - but why use out of date
averages when talking to the UK public about the UK? Seems plain daft
to me.

Stoopid question-why use 30 yr averages.Why not extend the reference period decadally as new data
becomes available?

Because our climate isn't constant. There therefore has to be a
trade-off between having a period long enough to iron out most of the
year on year random variations but one short enough to be representative
of the current climate. Presumably it has been decided that 30 years is
a reasonable compromise.

This thread had me thinking back to my student days, as I was sure the
number 30 had statistical significance. It plays a part in the concepts of
sample and population means, and standard deviations of same, IIRC.
Statistical work is best undertaken on samples of 30 numbers or more. In
the case of climate - more than 30 would hide the continuous changes we
know exist. So 30 is, as John writes, a compromise - allowing rigorous
analysis but not too long as to hide change.

The confidence in a mean and SD when n30 (in relation to the population
values) is quite low compared with n=30.

--
Mike 55.13°N 6.69°W Coleraine posted to uk.sci.weather 04/01/2004 18:20:19 UTC
My aurora images here http://www.mtullett.plus.com/29a-oct and
http://www.mtullett.plus.com/20-nov/