On Friday, August 23, 2013 3:06:34 PM UTC+1, General wrote:
"yttiw" wrote in message news:2013082314464610142-cuddles@britpostcom...



On 2013-08-23 13:12:48 +0000, Jim Cannon said:



The secret lies with who they subsequently blame if the forecast goes

wrong - the computer or themselves? Also who they congratulate if the

forecast is correct - the computer or themselves?

========================================



What would give this - perfectly legitimate IMO - approach more credence

would be to place it on an objective basis.



It surely ought to be relatively simple to take the various models'

numerical forecasts (are these the GRIB files or is that something else?)

and devise a methodology to calculate the correlation between them, weighted

presumably for the UK as a specific region and with progressively less

weight further away. You could then condense the correlation down to a

single index, eg 'the Consensus Index (CI)' (TM) and express it on a scale

of eg 0-100 (%, if you like). Then when the CI reached some significant

threshold value like 50% or 60% or some such, you could flag this as a

potentially noteworthy event.



Very possibly this is being done anyway by the professionals. I guess what

I'm curious about is how accessible this is to the amateur community, given

a sufficiently (mathematically) skilled user and access to to the

appropriate data files. Are the forecast files available in a gridded

format, ie something more numerical than chart?



JGD

I'm sure this is used, to some extent, but perhaps the professionals at the MetO just don't have confidence in what they see to issue a percentage forecast at 7, or 10 days (I do wish they would). I would need a much higher probability than 50-60% to make plans based on those forecasts, however. Perhaps the commercial forecasts give that measure of forecast confidence and maybe someone here, on the receiving end of them, knows? I've never seen one.

Try this:

Schools will be going back in early September and many tourists will leave the SW. This weekend could be the time when people without families may be considering "Can we risk catching any early September decent weather in Devon, or Cornwall, or do we fly abroad for a week?". It's a very relevant scenario and it will be being played out in households this weekend. Where do they go for their forecasts and how do they know that the forecast they are seeing has a track record of accuracy?

All they have on the MetO site is that precis that gives next to no real information. Or they go to a site, like Accuweather, which does give a 10-day forecast to the day, but has no indication of the confidence the forecasters have in their own forecast.

Hence, at 10 days, no-one has any confidence in any forecast, if you are a member of the public. I feel that forecasting can be pretty accurate at 10 days **on occasions**. Most of the time, however, my confidence is below 80% and many times, I have very little confidence and forecasting would be no better than a guess.

I think the MetO could be more public with confidence-based forecasts at a longer distance, but I think they are scared stiff of public reaction if they get them wrong.