Re: Why 1961-1990?

On Oct 3, 10:55 pm, cumulu...@xxxxxxxxxxxx wrote:
On 3 Oct, 20:32, Graham Easterling <GEasterl...@xxxxxxxxxxxxxx> wrote:

Is there a reason why the MetO still use 1961-90 for 'normal' weather?
(www.metoffice.gov.uk/climate/uk/2008/september.htmlasan example )

As I understand it the 30 year period was brought into use as it is
long enough to produce an average which can be considered 'normal',
without being unduly outdated by climatic change. Surely the most
recent 30 years should be used?

In fact, bearing in mind the difference between the 60's & 70's
compared with more recent years, there is an argument for using a
slightly shorter period (Say 1980-2005)

Graham, WMO defined various 'standard' 30 year periods over 40 years
ago, the first being 1931-60, the others defined as following on from
this but specified as being non-overlapping - hence 1961-90. According
to this arcane ruling, the next standard period averages will not be
updated until the period 1991-2020 is complete. (Even then no doubt
it'll be many years before the Met Office updates its averages.)

I don't see what is so arcane about establishing work-bases.

If you just use random dates, the danger of finding ones to suit is
pernicious. Having data sets at least stops the range of statistics
overwhelming.

One might even be able to picture event sequences.

Unlikely but possible.

No doubt this made sense in the days when every national met service
had its own rules for averaging periods, and probably from an era when
everything had to be laboriously calculated by hand rather than in a
millisecond in a spread*** or database, but it certainly makes
little sense these days, particularly with rather rapid warming taking
place resulting in almost every month being well above the 1961-90
normal for years on end.

WMO really should look at changing the recommendation to 'the most
recent 30 year period ending in a decade' (which now would mandate
1971-2000 of course, which is now pretty much standard anyway - but
the Met Office still uses a bizarre mix of 1961-90 and 1971-2000
averages, for no very obvious reason). This would at least mandate a
reasonably topical averaging period. At least the next averaging
period under this method isn't very far away - 1981-2010 is fast
approaching.

The reason it's a 30 year period comes from statistical theory - from
a theoretically infinitely large dataset with a stable mean (dubious
with temperatures now of course), the mean of 30 samples should give
an average practically indistinguishable from the true population
mean. Victorian meteorologists worked this out empirically (although
they used 35 years - 1881-1915 being one such period), and it was
later proved on theoretical grounds that 30 was sufficient. And the
reason it ends in a decade is just tidiness, I guess ... !

Contemplating one's navel is a sign of peace and plenty. One needs
plenty of food and beer and no aggro in order to do it well.

Contemplating one's anus merely requires something to base statistics
on. And the urge to lose ones' self in a singularity.

Having said that, the 30 year period suits lunar cycles quite well.

.