How can two weather predictions be so different?

[russellsnr]

Hi, so I normally go by the Met office website but just out of curiosity I checked the BBC for the same day.

 

[Gav.]

They both say around 5 deg and 40-50% chance of rain..

 

[russellsnr]

They both say around 5 deg and 40-50% chance of rain..

Actually missed that it was the titles Light Rain and a Gentle Breeze and Cloudy changing to sunny intervals

 

[Brentor]

I thought the Met Office provided the BBC forecasts.

 

[DemiLion]

I thought the Met Office provided the BBC forecasts.

No. They lost the contract to Meteo about five years ago.

 

[hunnymonster]

Which they subsequently regained.

Source

 

[Humphrey]

The BBC one always seems more pessimistic, with their symbols.

 

[lindsay]

A lot can change in 4 days. The expected sea surface temperatures and predicted air pressure, the line of the isobars, a small difference tomorrow can have a substantial difference in forecast for 4 days hence. Even using the same data model, two forecasters can produce different results for the future pattern.

 

[Barney]

BBC can come forward a day when forcast is 4-5 days away, I have planned any outdoor work around their forcast for years. Like clockwork for my area, in anything rain forcast is on the pessimistic side, could be a 1 minute shower for their "percentage chance". If it is two raindrops rather than one in the schematic its more or less always correct.

 

[DemiLion]

Which they subsequently regained.

Source

The contract, yes. But during the handover, the weather still comes from Meteo at the moment.

 

[whyone]

Whilst the summary symbols look different, I'd say the actual forecasts (temperature, chance of rain etc) are pretty similar.

 

[sirch]

I've often thought that someone should create a Weather website comparison site, capture the forecasts and then get people to report actual weather and see how good the forecasts were.

50% chance of rain is really in the "we don't know" territory. What the percentage chance means is that they have run the forecast multiple times with slightly different sets of conditions and the given percentage is the number of times the model predicted rain. So 10% and 90% really means a high degree of confidence. 50% means that half the model runs showed rain and half didn't so it could go either way.

 

[DemiLion]

I've often thought that someone should create a Weather website comparison site, capture the forecasts and then get people to report actual weather and see how good the forecasts were.

50% chance of rain is really in the "we don't know territory". What the percentage chance means is that they have run the forecast multiple times with slightly different sets of conditions and the given percentage is the number of times the model predicted rain. So 10% and 90% really means a high degree of confidence. 50% means that half the model runs showed rain and half didn't so it could go either way.

Didn't there used to be a member on here (Called 'The Weatherman' or similar) that did exactly that?

 

[myotis]

Hi, so I normally go by the Met office website but just out of curiosity I checked the BBC for the same day.
View attachment 468679View attachment 468680

A bit late with this, but the other thing to consider is that the results from all predictive models whether they are for the weather or covid, or the economy etc need professional interpretation before they can become useful. This brings in factors too complex to include in the modelling and relies on the expertise and experience of the people doing the interpretation. Including local knowledge.

Although I have the Meteo app (the people who provide the weather modelling data to the BBC) it hardly ever seems to agree with the BBC forecast.

 

[saltireblue]

I find that yr.no is pretty reliable - although Norwegian, it is available for the UK. Also available as an app.

 

[vintageman]

I've just tried yr.no and realised that somewhere along the way both the BBC and met office have dumbed down to 1, 2, or 3 raindrops. The Norwegians show amounts of precipation in mm. How much more useful! Thanks Malc

 

[DemiLion]

The Norwegians show amounts of precipation in mm. How much more useful! Thanks Malc

The native iOS weather app does that!

 

[F1.2]

I developed a good weather prediction method
I open the front door stick my head out , If it gets wet yep it is raining
Wet my finger and stick it out the door to see how windy it is , also looking at the local trees
Works for me.

 

[sirch]

The most statistically accurate weather forecast is "the weather in the next hour will be the same as the previous hour"

Although that's not a lot of use if you are making plans for a day...

 

[Nod]

Mrs Nod has a couple of Met Office forecasters as students. They look out of the window to see what's happening when asked for a forecast.

 

[Bodian]

I can't find it now, but I had a screenshot once showing a far greater difference between BBC and Met Office. One was for heavy rain, and the other for sunny spells. It was so completely different it was bizarre. The Met Office turned out to be the correct one.

But basically, in the Pennine Shadow at this time of year any old forecast will do - because whatever the forecast the actual weather will be wet, then dry, then windy, with maybe a bit of Helm Wind thrown in - like every other day (I used to live in Kirkoswald).

 

[Bodian]

Mrs Nod has a couple of Met Office forecasters as students. They look out of the window to see what's happening when asked for a forecast.

I went for a bike ride in Yorkshire a while back. We were chatting about the cows and reckoned that about 40% of them were lying down in the fields. We then checked the BBC which said that there was a 40% chance of rain.