Windguru: inconsistencies?

Greenheart

Greenheart

Well-Known Member
Joined
29 Dec 2010
Messages
10,384
Visit site
I'm a little confused by the Windguru pages (this one's for Calshot): http://www.windguru.cz/int/index.php?sc=28765

...in theory it's great, offering localised forecasts that seem much more reliable than general 'weather' reports.

But scrolling down to the lower half of the page, there a 'delayed' forecast, showing yesterday's wind...except that it also continues into the rest of the week, which is well into the future, and which seems to show much more wind than the 'current' forecast does, as well as indicating far more than yesterday's forecast showed.

Where is the 'delayed' detail of yesterday's forecast for the future coming from, when it wasn't actually part of yesterday's forecast?

I can understand updates, but these figures seem to have been adjusted retrospectively without any reason that I can see. Yesterday's forecast for Friday showed very light breezes. Now, yesterday's delayed forecast for Friday a.m. shows winds nudging force 4 until mid-morning, despite the current forecast still indicating F2.

And...what's the reason for a second, 'delayed' forecast for the same period, directly below the first set of 'delayed' figures, but showing different readings?

Thanks for any clarification. :encouragement:
 
No idea but guess if you were keeping a log you could compare how near to forecast the weather was is .Guess there are periods of accurate prediction and periods of unpredictability, which may help you judge how far of the prediction the next few days might be. Read into that what gobbeldy gook you may
 
I don't see your "delayed" one either, but I assume it's because I'm a registered user of the paid-for version (a few Czech blatts per year).

I suspect that the two forecasts are using different models, one GFS and one something else, maybe WRF, and the "delayed" one is a kind of trailer to show what you get (without the delay) when you pay for it.

Pete
 
dancrane said:
I'm a little confused by the Windguru pages (this one's for Calshot): http://www.windguru.cz/int/index.php?sc=28765

...in theory it's great, offering localised forecasts that seem much more reliable than general 'weather' reports.

But scrolling down to the lower half of the page, there a 'delayed' forecast, showing yesterday's wind...except that it also continues into the rest of the week, which is well into the future, and which seems to show much more wind than the 'current' forecast does, as well as indicating far more than yesterday's forecast showed.

Where is the 'delayed' detail of yesterday's forecast for the future coming from, when it wasn't actually part of yesterday's forecast?

I can understand updates, but these figures seem to have been adjusted retrospectively without any reason that I can see. Yesterday's forecast for Friday showed very light breezes. Now, yesterday's delayed forecast for Friday a.m. shows winds nudging force 4 until mid-morning, despite the current forecast still indicating F2.

And...what's the reason for a second, 'delayed' forecast for the same period, directly below the first set of 'delayed' figures, but showing different readings?

Thanks for any clarification. :encouragement:
Click to expand...

Windguru uses free US models for its "forecast". The delayed forecast is using the more detailed WRF model and is delayed because they want you to subscribe to get the up to date forecast.

It's all here: http://www.windguru.cz/int/help_index.php?sec=models

Be aware that the "localised forecasts that seem much more reliable than general 'weather' reports" are nothing of the sort. They are just interpolations. A bit like me looking at a weather model that says that Basingstoke will be 20 knots and Aylesbury 10 knots and then selling you a detailed forecast for your house halfway between the two which just splits the difference to give you a wind forecast of 15 knots. You will be impressed because I'll give you slightly different figures for your front and back garden.
 
Last edited:
I think I see, thanks.

Regarding the accuracy of Windguru's forecasts, they seem to me always to be impressively specific and never far wrong, as opposed to the terrible, vague and grossly inaccurate forecasts Google automatically provides on my PC, if I just request local weather. I'm disappointed to learn that Windguru isn't based locally...but it seems as good as if it was.
 
Don't confuse resolution with accuracy, they are two different things. It's a little bit like your GPS that gives you a position to the resolution of a few inches, but this is an average over time of a position that is only accurate to a few yards.

On the other hand, I subscribe to Wind Guru and find it frighteningly accurate. If anything, I find it slightly over estimates the wind speed by a knot or two. Well worth the cost. It makes the inshore waters forecast look as accurate as sea weed!!!
 
Alpha22 said:
I subscribe to Wind Guru and find it frighteningly accurate.
Click to expand...

I can't say I agree with you as I find the GFS model tends to underestimate coastal windspeed by about 10 to 20 percent, but more importantly Windguru is just a packaging exercise. The same data is used by all the other commercial websites - XCweather, Weather Underground etc. Only the display is their work. The Windguru locations (originally selected as useful for windsurfers) are just interpolations but they give it a spurious impression of accuracy, like when you use a calculator to calculate some value based on measurements taken with an old school ruler.
 
Nico said:
I can't say I agree with you as I find the GFS model tends to underestimate coastal windspeed by about 10 to 20 percent, but more importantly Windguru is just a packaging exercise.
Click to expand...

They are only packaging other data, but I find them a convenient source of a variety of models to compare. I do not use the GFS except for long-range planning. The majority of basic weather sites are GFS only.

Pete
 
prv said:
They are only packaging other data, but I find them a convenient source of a variety of models to compare. I do not use the GFS except for long-range planning. The majority of basic weather sites are GFS only.

Pete
Click to expand...

Yes, that's true that many only use GFS - which is a bit unsophisticated and the main reason I don't rate Windguru. I don't have a subscription. Try Theyr. They have their own models and are a genuinely different player to add to what is really only a handful of options.
 
I am deeply cynical about “detailed” forecasts, especially those provided by a number of private firms. There are two incontrovertible facts.

• A grid can only define weather patterns and topography on a scale of 4 to 5 grid lengths.

• Small weather features have short lifetimes. This limits their theoretical predictability. Weather on a scale of about 30 km has a lifetime of around 6 hours.

As far as I know all the commercial models start with the US GFS data analysis only available on a 0.5 degree grid, ie about50 km. They will not know about any weather detail at T=0 smaller than 200 to 250 km.. Further they can only update values around the boundaries of their forecast areas using predictions on the same grid ie 50 km.

They will be able to model topographic effects such as large scale sea breezes but success in doing so will depend upon getting such detail as cloud amount general wind speed and detection correct. That is a tall order.

Several National Met services start with input from models with shorter grid lengths – 0.125 degree (about 15 km) in the case of Norway Sweden, Denmark Spain and some others. They are also able to use data on higher resolutions obtained from satellite and even weather radar. Further they update their boundaries using the data from the same source (ECMWF) on the 0.125 degree grid. They have a substantial edge over the commercial models.

Nobody can yet predict detail of a few tens of km (around 20 to 40)for more than a few hours ahead – except in a probabilistic sense. Those models which start with detailed analyses and use the higher resolution boundary information will fare the best to about 24 to 48 hours ahead.

Detailed model data produced by Iceland Norway, Sweden, Denmark and Spain should be the most reliable over this length of time. The UK detailed 4 km model should be at least as good as any – pity that we cannot see the output unless you are in the Southern Baltic using the Polish met service website!

In the Med I would use http://212.175.180.126/DTS/sea.php issued by Turkey but in fact ECMWF output on a 0.125 degree grid. West of Gibraltar, that is denied to us. I use the GFS modified in the short term by GMDSS broadcast information. Add some commonsense and experience and you will not go far wrong.

As has been said the GFS – available in many guises and disguises – often under predicts the strongest winds by around one beaufort – 20% is a good figure. Whatever you use, keep an eye on the relevant GMDSS forecasts. They will be broad brush and very general but they have human input.
 
I've woken this thread up because I reckon you gents who've replied previously will know the answer to my present query.

Below is a screenshot I took of the Windguru detail for this coming Monday, June 1st. For those that aren't familiar with the layout, the green numbers represent wind-speed, the figure below it is the predicted gust-speed, the arrow is wind direction and the yellowish figure below that, is temperature...pretty miserable for June...

Screenshot_2015-05-30-10-13-24_zpsx4dig60s.png


...now, below that come three rows of figures, which are described as "Cloud cover %, high/medium/low"...and below that, the likely weight of rainfall, here 0.3...not much...

...but what about the cloud cover? Surely "high" should be higher than "medium", which in turn should be higher than "low"? In the pic above, at 13:00hrs, the "medium" figure is lower than the high or the low, above it and below, whether you regard the top and bottom figures as being high/low or low/high...and this is commonly the case.

Is it just sloppy keyboard skills at the office, or am I seriously misunderstanding the layout of the tables?
 
Last edited:
Thanks Frank. The 200-250km scale will explain why the weather forecasts from Windguru and the others are often very wrong around my cruising area - which is where the Pyrenees come down to the Mediterranean sea.

50km one way we have a huge expanse of pretty flat land and beaches for 200km or more, we have about 50km of crinkly bits with the mountains with some valleys in between before there is another long stretch of pretty flat stuff.

I guess in the modelling that 50km of mountains does not really show up well.

Just do a quick google for "Knock Down Cap Creus" for some reports about others who got surprised.
 
dancrane said:
I've woken this thread up because I reckon you gents who've replied previously will know the answer to my present query.

Below is a screenshot I took of the Windguru detail for this coming Monday, June 1st. For those that aren't familiar with the layout, the green numbers represent wind-speed, the figure below it is the predicted gust-speed, the arrow is wind direction and the yellowish figure below that, is temperature...pretty miserable for June...

Screenshot_2015-05-30-10-13-24_zpsx4dig60s.png


...now, below that come three rows of figures, which are described as "Cloud cover %, high/medium/low"...and below that, the likely weight of rainfall, here 0.3...not much...

...but what about the cloud cover? Surely "high" should be higher than "medium", which in turn should be higher than "low"? In the pic above, at 13:00hrs, the "medium" figure is lower than the high or the low, above it and below, whether you regard the top and bottom figures as being high/low or low/high...and this is commonly the case.

Is it just sloppy keyboard skills at the office, or am I seriously misunderstanding the layout of the tables?
Click to expand...

Think altitude ... http://www.metoffice.gov.uk/learning/clouds/cloud-names-classifications
 
Ah! Thank you, that seems very logical...

...but don't winds also vary at altitude? So, if the service provided is altitude-focused for paragliders or mountaineers as well as sailors, why is the wind speed limited to sea level?
 
dancrane said:
I'm a little confused by the Windguru pages (this one's for Calshot): http://www.windguru.cz/int/index.php?sc=28765

...in theory it's great, offering localised forecasts that seem much more reliable than general 'weather' reports.

But scrolling down to the lower half of the page, there a 'delayed' forecast, showing yesterday's wind...except that it also continues into the rest of the week, which is well into the future, and which seems to show much more wind than the 'current' forecast does, as well as indicating far more than yesterday's forecast showed.

Where is the 'delayed' detail of yesterday's forecast for the future coming from, when it wasn't actually part of yesterday's forecast?

I can understand updates, but these figures seem to have been adjusted retrospectively without any reason that I can see. Yesterday's forecast for Friday showed very light breezes. Now, yesterday's delayed forecast for Friday a.m. shows winds nudging force 4 until mid-morning, despite the current forecast still indicating F2.

And...what's the reason for a second, 'delayed' forecast for the same period, directly below the first set of 'delayed' figures, but showing different readings?

Thanks for any clarification. :encouragement:
Click to expand...

Right now the forecast I get was generated by the computer at 0600 hrs today. It covers 5 days ahead. The second part below is the forecast that the computer generated yeasterday at 1200 hrs again for 4 days but hourly instead of every 3 hours. Below that is another forecast generated at 1800 hrs yesterday covering much the same time period as the second one.

I suppose they give you an idea of how the forecasts are changing, if indeed you want that idea.
 
Frankly I think that Windguru (and everyone else) just gives you an idea about what the weather might be like, and nothing more.
I live on the N.Kent coast and maybe our weather is influenced by continental weather, but more often than not the forecasts hereabouts are significantly out, and sometimes laughably so.
Most of my sailing is in and around the Thames Estuary and I've found that out of all of them, the forecast most likely to be something like right is the BBC Coastal Forecast. Why it should be any less worse than any other, I have no idea.
 
You must log in or register to reply here.

Share this page

Top