On 2010-12-10 11:19:25 +0000, Alastair said:

On Dec 9, 12:40*am, Alastair wrote:
On Dec 8, 11:11*pm, Adam Lea wrote:





On 08/12/10 17:51, prodata wrote:

On Dec 8, 3:19 pm, *wrote:

I've calculated the December mean from 1971 - 2009 as +5.0c, with a
Standard Deviation of 1.5c, so by my calculations the current CET th
at
is running at -2.0c is somewhere between 4& *5 Standard Deviations
.

Is there evidence that these values are normally distributed? Maybe
this has been throughly tested and is well-known? But if not then
personally I'd be happier with a non-parametric analysis.

JGD

Technically speaking they won't be, as the normal distribution is
unbounded, whereas there is a limit on how large a magnitude a
temperature anomaly can be. However for practical purposes it is likely
that temperature anomalies will be close enough to normal that the usua
l
statistical techniques that require normality will be valid.

IMHO, the normal distribution is for random values but weather/climate
is chaotic. Chaos can appear random but is actually deterministic.
The fact that the temperatures do not fit happily into a normal
distribution only goes to prove that they are not random.

The classical case of a chaotic attractor is the Butterfly effect
produced by the meteorologist Edward Lorenz.http://www.viewsfromscience.c
om/documents/webpages/chaos_p3.html
It does not represent a plot of daily temperatures over the years, but
if it did, then you can see that much of the time there it is close to
a cycle just like annual temperatures. However, at times it jumps out
of that cycle into another cycle. What we could be seeing with the
December temperatures being so far from the standard deviation that
the climate is jumping out of the current cycle/state/attractor into
another climate state, i.e. an abrupt climate change. Let's hope not.

OTOH, what we may be seeing is are the effects of an abnormal sum. The
last solar minimum lasted much longer than usual.

Will mentioned in another thread that it may be due to there being
less Arctic sea ice, but the area is not much different from other
recent years. It could be that the ice is thinner allowing more leads
to form and so allowing more water vapour to escape from the sea
beneath the ice. In that case, could that be the reason for low
pressure over the *pole, and equalising high pressure over the North
Atlantic?

Cheers, Alastair.

Cheers, Alastair.- Hide quoted text -

- Show quoted text -

As a follow upt to what I wrote above, can I menton that there as a TV
program on BBC 4 last night entitled "The Secret Life of Chaos" which
can be seen for the next 6 days at:
http://www.bbc.co.uk/programmes/b00pv1c3

It says that weather is Chaotic, and therefore it is detrministic and
not random. In that case the normal distributions does not apply to
weather data.

Nevertheless the normal distribution fits the temperature mean data
extremely well - far better than the power-law distribution that fits
the output of chaotic systems.

A lot also depends on what you mean by "random". In a sense everything
is deterministic. I certainly don't think it's right to say that
because some variable has a deterministic aspect a normal distribution
does not apply.

A ND arises when the value of the variable is determined by many small
variables. Height and IQ are the classic examples, with them being
determined by the interaction of many genes and the environment. The
weather, particularly the mean monthly temperature, is I think very
similar. The temperature is the result of the interaction of a huge
number of determining variables, and I don't see that is in any way
inconsistent with the weather being a chaotic system (which it has to
be, given that very small changes in the starting conditions can lead
to enormous differences later on).

It so happens that my research speciality, when I have the time, are
computational models of mind, and especially dynamic theories of
cognition. I am of course also the world's first and only
psychometeorologist ... :-)

--

Trevor
Erudite in Lundie, near Dundee
http://www.personal.dundee.ac.uk/~taharley/