On Sat, 2 Oct 2010 10:01:37 +0100, Jon O'Rourke wrote:
"Dawlish" wrote in message
news:d54d02f2-3715-457a-9d59-

It's really the kind of thing the MetO should be doing for its
customers.

In terms of the Public Weather Service (PWS) verification relevant to this
thread - extensive verification is carried out for 139 UK locations
throughout the year and form part of the overall Key Performance Targets
(KPTs).

More here http://www.metoffice.gov.uk/corporat...tion/city.html

With the results from 2009/10 here
http://www.metoffice.gov.uk/about-us...ecast-accuracy

Jon.

It's worth bearing in mind that I am not measuring how often any
particular forecast comes true, which seems to be what the links
above are concerned with. The best I can do is see how frequently the
forecasts change - since a forecast for 4 days hence is not worth
much if by the next day it's been changed, and then again for the
day after that ...

As I've said before, the basic problem I see with these forecasts
put on the BBC website is that they try to portray too much detail,
too far ahead. For example: there are 7 different descriptions for
wet weather. So whereas it's reasonable to say with a fair degree
of confidence that the weather today will be "Sunny intervals",
that same statement is used to forecast the weather for 4 days in
advance, where something like "probably dry" would better
convey the degree of confidence[1]. Although forecasting anything
as sunny doesn't work for night time, which for 6 months of the
year is the dominant case.

Now I appreciate that when you have a sub-twitter number of characters
to use, it's difficult to come up with anything better than a
"mostly harmless" style summary - but maybe that's part of the
problem, too. It's a difficult path to walk, between worthlessly
vague and misleadingly over-confident.


[1] it would be refreshing to see a few "we simply don't know"
forecasts when any forecast is below a particular quality
threshold, though I can't ever see that happening.