In article ,
Granity writes:
|
| While I appreciate it's a difficult art surely they, with all the super
| computers etc they have, they could do better than this.

I was trying to educate some of my colleagues about this only last
week. There are three problems with weather forecasting:

1) Lack of data. Up until the 1960s, there was ONE weather ship
in the North Atlantic, and the patterns were deduced by guesswork
primarily from shore station data. That has been largely resolved
by satellite data.

2) Lack of computer power. Up until the 1970s, a reliable
forecast for 24 hours ahead took over 24 hours to calculate :-)
That has been largely resolved by modern microprocessors.

3) Insolubility of the problem. This is traditionally called
numerical instability, but has been renamed as chaotic behaviour
by the popular press. That is not possible to resolve, as it is a
mathematical restriction!

The weather forecasters said that the one consequence of global
warming that they were certain of was increased instability; that
was not published in the popular press, on the grounds that it was
too hard to explain. They were right. In the past two years, the
weather forecast for the UK has been changing faster than the weather.

Note that, in this respect, even journals like Nature may count as
"the popular press" - I haven't read it regularly in years, but it
was (at that stage) very weak on advanced statistics, mathematics and
computational theory. It probably still is.


Regards,
Nick Maclaren.