windguru and the met. office

[fergie_mac66]

Sat on the boat in the IOS with 3 anchors out at 22.00 hr based on a prediction by windguru of winds over 32 knots for tonight. They are consistently accurate in there predictions we had F8 at22.30 ..a GF 8 imminent arrived overthe VHF at 23.45 ? I am confused if windguru.cz can predict winds accurately and consistently. why can't the met office.

 

[grumpy_o_g]

Sat on the boat in the IOS with 3 anchors out at 22.00 hr based on a prediction by windguru of winds over 32 knots for tonight. They are consistently accurate in there predictions we had F8 at22.30 ..a GF 8 imminent arrived overthe VHF at 23.45 ? I am confused if windguru.cz can predict winds accurately and consistently. why can't the met office.

Ah, but Windguru can't predict the weather decades ahead the way the Met Office can...

Actually that's really the answer - they've lost focus on their primary job of forecasting the weather and, instead of employing competent meteorologists to do this, they are busy trying to model the climate decades head. They used to have access to weather data that most others didn't but this has largely been lost, partly because of the wind-down of the Armed Forces and especially air bases and partly because the Internet and cloud-sourcing means that everyone can get good data pretty much in real time.

Where's the IOS?

 

[fergie_mac66]

Ah that sounds about right. I O S. Isles of Scilly

 

[fergie_mac66]

Swmbo has just said its because the met office dont have William Foggitt to tell them any more (died 2004)

 

[longjohnsilver]

We're also in IoS in New Grimsby on a buoy. Interesting evening, strongest wind I saw was 41kts, but all arrived and went pretty swiftly. Hope all around you we're ok.

 

[Spyro]

I've always found windguru very accurate, even within hours of forecasting changes. Weathoronline is leaving much to be desired now as well i'm afraid to say.

 

[prv]

Where's the IOS?

I guessed Isles of Scilly, but I can't help reading it as "Imps of Satan", as used by a lady on a forum I used to read to affectionately refer to her children

Pete

 

[Bosun Higgs]

I am confused if windguru.cz can predict winds accurately and consistently. .

But they dont. Gale force 8 is mean wind speeds of 37kn varying 34 to 40. So 32kn isnt gale force 8.

Where is the IOS? Assuming it is the isles of scilly, the met office were right with the winds reaching 37 knots at 00:00 hrs.

In general terms I find the met office just a little bit more reliable than the automatic forecasts from windguru / widfinder / etc

 

[LadyInBed]

I've always found windguru very accurate, even within hours of forecasting changes. Weathoronline is leaving much to be desired now as well i'm afraid to say.

I have also found that Weathoronline isn't what it used to be, I've started using passageweather.
My current favourite is the android app for Meteo Consult Marine, easier to use than WindGuru and good presentation.

 

[GHA]

As all the free sites use the same gfs data how come there's any difference? Has anyone actually spent a week or so comparing a selection of forecasts with actuals? Sounds like a load of work but would be interesting.

 

[Boreades]

Sat on the boat in the IOS with 3 anchors out at 22.00 hr based on a prediction by windguru of winds over 32 knots for tonight. They are consistently accurate in there predictions we had F8 at22.30 ..a GF 8 imminent arrived overthe VHF at 23.45 ? I am confused if windguru.cz can predict winds accurately and consistently. why can't the met office.

The Met Office is heavily constrained by it's own Standard Operating Procedures. All observational data received has to be fed into their supercomputer (how else can you justify having monstrous computer processing power?), and churned through their climate modelling software, then validated in various time-consuming ways, then published via partners (like the BBC) in various time-consuming ways. By which time it is sometimes hindcasting instead of forecasting. See how much of the BBC weather is telling us what did happen, not will happen.

So no surprise that smaller nimble outfits like WindGuru can get the same source observational data and produce a result quicker. Even better if you get the prediction before the event.

 

[parito]

Not only GFS

Windguru provides GFS, WRF27, WRF9 models.

I find WRF9 model very accurate. it's 9 km grid. Much more refined than metoffice's "north foreland to selsey bill"

There is also predictwind.com - they take couple of public models (GFS and Canadian ), interpret them and create their own 1km resolution weathercast, which takes landmasses etc into effect. Very accurate as well.

When people check different sources they often check the same model and get the same results..

Different models is what matters, not different websites..

Metoffice does not use either GFS or WRF as far as I am aware..

 

[GHA]

I find WRF9 model very accurate. it's 9 km grid.

Ta, didn't know that.

Re the accuracy, sorry but I'm a bit dubious when reading about x being a more accurate forecast than y, it must take an awful lot of work to log all that data and compare it against actual events to know for sure. Humans tend to see what they want to see quite often.
Bet Frank would know more.

 

[parito]

Well, smaller grid (27km, 9km, 1km) is obviously better than larger grid - just because it smaller grid takes into account things like local topography.. At least that is how I see it.

and yes there will always be differences where models could not predict how localized winds would act

and yes, I am not a meteorologist

 

[grumpy_o_g]

Ta, didn't know that.

Re the accuracy, sorry but I'm a bit dubious when reading about x being a more accurate forecast than y, it must take an awful lot of work to log all that data and compare it against actual events to know for sure. Humans tend to see what they want to see quite often.
Bet Frank would know more.

Agree with that - resolution and accuracy are very different things. I'd rather have accurate forecasts 50km apart than inaccurate ones 9km apart. Having said that it could easily be both more accurate and a higher resolution of course.

I think that most of us simply learn to interpret a particular forecast so consistency is just as important, perhaps more important than accuracy or resolution.

 

[GHA]

and yes, I am not a meteorologist

Me neither

Unfortunately continually surprised by how often I'm proved wrong about things which seemed obvious....

 

[PaulMcC]

Interestingly one of the things the modellers do with lower resolution models is to lots of model runs and generate probabalistic forecasts - 'ensemble forecasting'. This tends to be a lot more accurate than single runs at higher resolution - unless the local situation means that high resolution is needed, for example to capture steep slopes. I expect a lot of the difference you see between met office and windguru and others is to do with their approach to resolution vs ensemble size and whether high resolution is particularly important where you happen to be.

(I may be biased as I used to do ensemble forecasting!)

 

[dslittle]

Re the accuracy, sorry but I'm a bit dubious when reading about x being a more accurate forecast than y, it must take an awful lot of work to log all that data and compare it against actual events to know for sure.

We have been on the boat for three years now and I use Windguru all the time. I have found their predictions very accurate and have often been amazed as rain starts almost on the hour that they predict. On one occasion we sailed back having checked the weather (had to motor last two hours due to lack of wind) and were surprised not to see any boats out on a lovely Sunday afternoon when we got back. Later I saw the Met Office forecast in the Marina showing a Gale Warning!!! That was the most extreme difference but the Met Office forecasts frequently 'over egg the pudding' - I do understand that they cover a much wider area.