In article ,
Martin wrote:

Back in the days when they concentrated on describing the likely
movements and properties of air masses, the presentation was
better (though the data available were MUCH worse). At least
the Met. Office's web page has improved by being less definite
in giving the probability of precipitation, but a reliable
forecast would need to be highly probabilistic and conditional.
I would dearly love to see them provide such a page for the
general public.

The Dutch Met Office does but they try to cover the whole country with
just one forecast.

That's reasonable, given its size and lack of terrain, but they
really should have some adjustments for the times and probabilities
that a front is passing through or hovering there.

From long ago, in the time before their one computer, a Ferranti
Mercury, or any other computers were used by the Met Office for short
term forecast, I remember that short term forecasts for specific
locations like airfields were very accurate compared to general
forecast for the area in which the airfield was located. Perhaps
partly because forecasts were funded by the Air Ministry in those days
and the forecasters had vast experience of forecasting for airfields.

Sort of. It was because Old Fred had many decades of experience for
that particular location, and used his subconscious. The general
forecasts were based on the science, which was basic, and the data,
which was almost non-existent. They were little better than crude
guesswork.

My first job, in 1966, was physically within the Met. Office, and I
was programming a Mercury. It was completely inadequate for anything
as tricky as a forecast, though that was the sort of programming I
was trying to do.


Regards,
Nick Maclaren.