Looking at numerous sites for three to five days ahead there are some
significant differences between the forecasts.
Has anyone looked back at these sites and compared the actual weather to the
forecast for the same period?
I would be interested in any comments or recommendations for sites with a
greater degree of accuracy or those that, for the most part, are not so
accurate. I once read with interest that forecasters (can't remember if it
was the BBC or Met Office) received a bonus depending on their accuracy.

Your recommendations or comments!

Martyn
www.calvertoncam.co.uk/weather

Nytram wrote:
Looking at numerous sites for three to five days ahead there are some
significant differences between the forecasts.
Has anyone looked back at these sites and compared the actual weather to the
forecast for the same period?
I would be interested in any comments or recommendations for sites with a
greater degree of accuracy or those that, for the most part, are not so
accurate. I once read with interest that forecasters (can't remember if it
was the BBC or Met Office) received a bonus depending on their accuracy.

Your recommendations or comments!

Comparing actual with a forecast is of course easy, but deciding if the forecast
was accurate or not is very difficult.

For example, on the BBC weather web site, they give 8 factors relating to the
weather for a given day. This is updated regularly, which poses another problem.

Which forecast does one compare? The forecast for today they gave this morning,
or the one given last night, or yesterday midday, etc.. At least there is only
one actual (-:

If one selects say midday the previous day as being 'the' forecast, then how
many of the 8 factors have to be right or wrong for the forecast to be right or
wrong. They never get all 8 exactly correct so one could say the forecast is
always wrong ... but that is a little harsh.

I had been planning to make a comparison based on the noon forecast for the next
day, but concluded that there is no point. Whatever I decide is the threshold
for accuracy (6 of 8 or whatever) will just be my view, and therefore of no
value to anyone but myself.

In reality, the categories would have to be weighted, which makes it even more
personalized. After all, if they get the maximum temperature wrong by one
degree, that is 'less' of an error than if they say it will be dry and it rains
all day ... or is it?


Regards,
Gianna


--
* * * * * * *
The information contained in this message is confidential and access by anyone
other than the intended recipient is unauthorised. This message has been
scanned by Norton Anti-Virus Professional and is free of all known viruses.
* * * * * * *