Hello Col.
Although your questions have been answered some time ago, I'd like to answer
once again, perhaps with a little more hind-sight on my part.

You are right when you say current thinking suggests it is impossible
...........! But I have to ask you - who makes up current thinking?
Obviously the scientific approach to forecasting is paramount and that means
computer forecasting.

And how does computer forecasting work? - well, you take the latest analysis
of the actual weather and project it by some means in a series of successive
charts until you reach the required forecast date. Naturally the further you
go from the real-weather- analysis the worse your forecast will be.
Furthermore, the method is sensitive to error-inputs because these explode
into larger errors the further you calculate.

Now a body of experts exists who says there is only one method to forecast
the weather and that is per computer. Therefore success is limited to only
five or six days - ten days at the most - with ensembles perhaps a few days
more. Anything beyond that cannot be done - therefore it is not done. And
it is not easy to break down the walls of that way of thinking.

The Countryfile forecasts you mention are a product of what I have been
talking about. I marvel at the fact that computers are capable of producing
forecasts for five or six days ahead when so little input is available to do
the job. And of course its inherent effect of the analysis chart ends
there, the moment it is entered into the computer.

So there we have it Col, the people who determine current thinking have no
other possibility but to quote from their own experience.

The method I use is entirely different from the one which requires
successive forecasts to be made from an actual weather situation. My
method starts with finding out which factors are responsible for the weather
situation in hand. These I calculate from knowledge of the date and the time
and finish up with an index figure which is universally applicable for
determining the pressure situation for that date and time. I don't need to
attach the forecast to an analysis of real weather, I'm free to go where I
want. I pick out the date and time, and the forecast I end up with is for
that date. This means that there is only one possible forecast chart for that
date - It doesn't alter the forecast chart if I make the forecast a year
before.

This leads to the question - is weather random or organized? According to
science there is a deep feeling that weather is random and chaotic because at
present the scientific approach hasn't been successful and at the moment the
best improvement to the situation is merely ensemble forecasting.

From my own point of view weather is finely organized. If I'm 0.5 of an
index figure out, I could pick out the wrong forecast chart. It's interesting
to watch movements of pressure centres caused by changes in index figures.
If an index figure shows that a depression should remain still or be
replenished by other depressions holding it back, then the depression will
remain on the spot in a real-weather situation. In other words, as the
index figures are (you could say) almost a product of nature, and the
figures indicate that a depression must be in a certain position at a certain
time, then the depression has to be there at that time - and it usually is.

There is one point that I have to make and that is that the detail of the
forecast charts is not as good as the ones derived from a computer. My
charts could be improved by graphical addition of a number of weighted
charts. But it's a big job and I haven't the tools to do it - and It would
be a trial and error story in the end, anyway. So I'm leaving things as they
are and I'm gathering more experience of where the limits are.

I'm almost embarrassed to say that retro forecasting for most of the
important historical events has been one of my main interests for a number of
years but I've only posted one or two because of the immense opposition to
single-day long range weather forecasting.

The success of the forecasting method has to be good. If any method can
produce a rough copy of the actual pressure situation for any place, on
any date mentioned - then the method must be good. If pressure charts can be
drawn up to cover a single time, a two day period, a five day period, a 15day
period and a monthly period - then the method must be quite good. If the
pressure chart can be changed into a 500mb top chart, show areas of strong
wind, show areas of lift or show areas of rainfall - then the method must be
even better.

All these things the method can do. It requires time, and weather is endless.

Col, I'm sorry to have made it so long - and I apologize for any errors in my
English. It's a bit late I suppose

Check the November forecast for yourself and form your own judgement.

Cheers, Keith

Col schrieb:

"Keith Darlington" wrote in message
...
During the period 24th December 2003 till 1st January 2004 the weather
will be determined by the mild SW winds - turning to light
anticyclonic winds with night frost - an finishing with a small
depression near SE England which deserves our attention if we are a snow
fan.

snip

Current thinking suggests it is quite impossible to forecast the weather
in this much detail so far ahead.
I've seen Countryfile forecasts for 5 days ahead that forecast to this
much detail go badly wrong, how can you hope to do this for over
2 month ahead?

What methods do you use and what sucess have you had?

Col
--
Bolton, Lancashire.
160m asl.
http://www.reddwarfer.btinternet.co.uk