[FrancesHinden]

BBC Weather Centre, 5 day forecast for The Hague, Netherlands:
Thurs: Sun, high 30
Fri: Sun, high 29
Sat: Sun, hgih 27
Sun: Sun, high 23
Mon: Sun, high 22

Yahoo weather, 5 day forecast:
Thurs: partly cloudy, high 27
Fri: Rain, high 22
Sat: Showers, high 22
Sun: Rain, high 22
Mon: Showers, high 20

Telegraaf 5 day forecast
Thur am: cloudy, 24; afternoon/evening high 28, rain.
Fri: sun/thunderstorms, high 25 60% chance of rain
Sat: sun/showers high 25, 40% chance of rain
Sun: sun/showers high 26, 40% chance of rain
Mon: Sun/showers, high 24, 30% chance of rain

Thurs am observation (9am): heavy rain

So, beach or museum at the weekend? Who can tell?

[Walddk]

Beach?? I hope that's a typo. You obviously mean CRICKET, and if you're in the Netherlands, the least you can do is to tune in to BBC Test Special Live.

You may argue that you can bring a radio to the beach. However, if TV is not an option for you (that may well not be), you should spend some of your money on getting Sky.

The heat on the beach is unbearable. The heat on the cricket pitch is heavenly

On a more serious note, I think many of us are suffering from the extreme heat at the moment, not least in countries where a/c is not standard. That applies to most places in northern Europe.

Roland

[FrancesHinden]

There are two legal ways to watch the cricket here in the Netherlands (getting a UK Sky subscription is not really legal outside the UK, though lots of people do it), and one of them would also work for you, Roland.

- You can subscribe to ECBTV and watch it - live or highlights - on streamed video over the internet. Video quality isn't bad.

- We have invested in a strange piece of hardware called a Slingbox that plugs into our TV and router in England, allowing me to watch (and control) our TV anywhere I have a PC and broadband internet. Video quality isn't great, but it's just about watchable.

[paulg]

Over the past twenty years I've been involved in the sale of computer systems to many of the top European weather forecasting bodies. For the suppliers this is a lucrative business as these folks buy the most expensive systems available and have very demanding storage (and retrieval) requirements. In Europe I think only CERN has more storage than each of these customers though its retrieval demands are far simpler.

So spending $30-40M per annum (per country) enables them to have reasonable accurate weather forecasts and do climate research.

Of course, you can get 90+% accuracy by predicting that today's weather will be the same as yesterday's. Aside from the military requirements (and the Metrological Office in the UK has always been part of the Ministry of Defence) it would be interesting to see the business case for the increased accuracy

Holland can get better than 90% accuracy by predicting today's weather will be the same as the previous day in south-east England.

Cheers

Paul

[Walddk]

Since we almost always have a westerly wind blowing across Denmark, it's a pretty good bet to predict that we get the Midlands weather the day after. If it had been straight westerly, we would get Scotland's.

I'm glad we don't. We would run out of brollies and wellingtons very soon. Some wicked souls claim that the word "sun" is non-existent in a genuine Scottish accent

Roland

[FrancesHinden]

cardsharp, on Jul 27 2006, 11:11 AM, said:

Holland can get better than 90% accuracy by predicting today's weather will be the same as the previous day in south-east England.

Cheers

Paul

90%? That's quite good going. I have been watching the BBC weather forecast most mornings; I'll just have to try and remember what yesterday's was.

It hasn't rained in SE England yet, however this week - we seem to be ahead of things.

[hotShot]

cardsharp, on Jul 27 2006, 12:11 PM, said:

Of course, you can get 90+% accuracy by predicting that today's weather will be the same as yesterday's.

@Paul:
I think you are wrong, it's just about 65% accuracy.

@Frances
The main problem with the weather forcast is that it is averaged over a wide area.
Yesterday e.g. we could hear some thunder and some aereas less than 10 miles away were flooded with rain. We here haven't had a bit of rain for the last 4 weeks.
So what would you say? Do we count rain or sunshine forecasts as accurate?

[Gerben42]

In many places "same as yesterday" works well, but usually in the Netherlands this won't work. Better % will be achieved by "same as yesterday in England" since the wind and therefore the weather almost always comes from that direction...

For The Hague just be sure to get the official KNMI forecast as shown on teletext.

Several-day forecast:
http://teletekst.nos.nl/?704-01

Today:
http://teletekst.nos.nl/?702-01

[helene_t]

What does an accuracy of 90% mean? 90% probability that the temperature prediction if off by less than 10 degrees? Or less than 1 degree?

[FrancesHinden]

Gerben42, on Jul 27 2006, 01:07 PM, said:

For The Hague just be sure to get the official KNMI forecast as shown on teletext.

Today:
http://teletekst.nos.nl/?702-01

Ah yes, I looked at that on teletext last night. It said it would be sunny, 27 today. Already wrong.

[paulg]

BBC Weather is based principally on the UK Met Office data and provides detailed local weather information (as Frances has shown).

I find it is most accurate at 7pm when it appears that "today's forecast" is changed to reflect the actual weather. The 5-day forecast is generally unreliable in my locale, but I guess that Roland would say that 'rain' is the only forecast I need

Paul

[Rain]

Sure, I believe weather forecasts. From weather.com . Erm, how else to know what to wear to go out, or even if stepping outside of your house is a good bet?

Obviously weather forecasts are not 100% believable, you take a peek outside and decide how much to trust your favourite weather sites.

When in Singapore I never had to care btw. Its always hot and damp, sometimes wetter when there's another flash thunderstorm, but the heat is constant. I've found that knowing the weather is very important in non tropical areas though.

[hotShot]

Weather forecasts are best for a small area and a short time ahead.
Sometimes the weather situation is stable for a few days, but often it is not. So predictions for the same day are usually correct. Predicions for tomorrow are quite good. But from the 3rd day on you should treat it more like a trend and not like a prediction. At stable weather situations predictions of 7 days are possible.

[1eyedjack]

I am actually quite surprised at the disparity in the original post (regardless of the absolute accuracy of any particular source). I would have expected all "presenters" to get their info from the same ultimate source. They can't all afford their own satellites. If even if they could there would be no incentive to incur the cost.

[slothy]

I predicted there was a 100% chance of Rain in this thread....

[paulg]

1eyedjack, on Jul 27 2006, 07:18 PM, said:

I am actually quite surprised at the disparity in the original post (regardless of the absolute accuracy of any particular source).  I would have expected all "presenters" to get their info from the same ultimate source.  They can't all afford their own satellites.  If even if they could there would be no incentive to incur the cost.

I guess the meteorologists might believe there is little purpose in playing bridge ... after all, you all start with the same hands

The weather folks have the same intense rivalry as many disciplines. The prediction models are hugely complex, have been developed over decades, and each centre has its own. In the UK there are two large organisations, the Met Office and the European Centre for Medium-Range Forecasting, who have very different models and typically, completely different computing infrastructure. They may share the input data from satellites and observation stations, but it's what you then do with it that counts.

Clearly weather is not easy to model due to large number of known (and unknown) variables. Each centre tends to believe it has the best model to meet its commercial demands.

Paul

[1eyedjack]

cardsharp, on Jul 27 2006, 08:30 PM, said:

1eyedjack, on Jul 27 2006, 07:18 PM, said:

I am actually quite surprised at the disparity in the original post (regardless of the absolute accuracy of any particular source).  I would have expected all "presenters" to get their info from the same ultimate source.  They can't all afford their own satellites.  If even if they could there would be no incentive to incur the cost.

I guess the meteorologists might believe there is little purpose in playing bridge ... after all, you all start with the same hands

The weather folks have the same intense rivalry as many disciplines. The prediction models are hugely complex, have been developed over decades, and each centre has its own. In the UK there are two large organisations, the Met Office and the European Centre for Medium-Range Forecasting, who have very different models and typically, completely different computing infrastructure. They may share the input data from satellites and observation stations, but it's what you then do with it that counts.

Clearly weather is not easy to model due to large number of known (and unknown) variables. Each centre tends to believe it has the best model to meet its commercial demands.

Paul

Fascinating. Does anyone "in the know" keep any kind of statistical track of the competing agencies' respective success rates? I would hazard a guess that each outperforms the other on occasion, and it would be interesting to know if there is a pattern to the occasions when one model is on balance more reliable than the other. Then I might tune in to accuweather.com on (say) chaotic summer days and to the BBC on (say) chaotic winter days. Non-chaotic days should I hope contain close correlation when it won't matter.

Anyway, I had more or less worked out that there were competing processing models from the source data. It is the only explanation for the different forecasts. I was only expressing surprise that competing models survive commercially in competition with one another over a significant period. It doesn't sound like the sort of market that behaves like most other markets.