http://news.bbc.co.uk/1/hi/sci/tech/8462890.stm

Thoughts?

Maybe a better explanation of their processes and remit may be in order, though you have to wonder when they tell you to expect snowy conditions the next day in your part of the country and you only get light showers at most... Maybe something is lost in translation between the facts and the weather presenters

I think that the public's grasp of stats is so low that they can't really grasp the lack of certainty.

I demand accurate long term weather forecasts. I don't believe it's difficult at all, they're just being lazy and stupid. I know nothing about it at all, so I'm perfectly qualified to write critical reviews.


+1

I just hope the rumours about selling it off are bogus


That's the truth.

I have a beef with the BBC's weather forecasters (who are Met Office employees, I believe). The past two or three years have seen a move to style over substance, with little in the way of actual forecasting going on. The BBC's current map system is obscure and full of fancy-schmancy graphic nonsense which tells me nothing about what the weather will be. I much preferred the old fashioned pictograms. You could tell a rain cloud from a snow cloud because it had a little drop under it. Now it's all shades of blue and brown, with white highlights. It's a mess, but you can complain till you're blue in the face and no-one listens...

Leaving that aside, when predictions of a "barbecue summer" and a "warmer winter" turn out to be polar opposites of what actually happens, you have to wonder what the Met Office is actually trying to achieve. The forecasts up to five days are getting so much more accurate now, I wonder why they bother trying to guess what's going to happen six months on.

I also couldn't tell by the on-screen graphics recently if it was to rain or snow...

I couldn't figure out if they meant snow or ice either. The fact we can't even agree on our misinterpretations says it all.

Its the long term forecasts that make the money. There is a huge demand for 1 - 6 month forecasts from industry.

Agreed.

The trouble is that by the Met office's own admission their regional 48 hour advanced forecast is accurate 6 days out of 7.
If you extend this forecast to 14 days then it is below 50% accuracy (i.e. little better than a guess).
As part of my job I am sent a 14 day advanced forecast every Monday and the predicted accuracy for the second week is always low because of this.

Andrew Neil recently asked some similar questions on the BBC as well.

Clicky.

Not going to disagree with you.

For 9 out of the past 10 years they've predicted that annual temperatures would be higher than they actually have been.

I find it difficult to rely on forecasts of any type that are based on statistics that are only correct 10% of the time.

Huh? I'm actually quite impressed by the accuracy:



So the predictions have been to within the stated margin of error, and to my mind 0.05C really is a small amount. You're hardly going to say:

"I was going out today, but it's only 18.95C. Those damned fools told me it would be 19C, and I was totally relying on it".

Error bounds FTW!

"Barbecue Summer" and "Mild Winter". That is all.

I'm sighing right now JJ.

Well, err, yes.





I still find the weather map annoying. If it’s not the swirling around the country, it’s the juddering that the name places seem to develop at times. It is unnecessarily over-engineered.



That’e because so far climate models are very unsatisfactory. They give vague ideas about what may happen, but really have the accuracy of a pile of goat’s bones and sheep entrails.