GFS model is currently only correct 25 percent of the time

[Triumphant]

If anyone has anyone been looking at the weather in the last few days for the NW Caribbean., they will have noticed that the GFS model tried to forecast a fairly aggressive low developing just north of the Bay Islands, Honduras producing 30 knot E to NE winds to the north of the low, they have since withdrawn from this forecast and fallen in line with ESE winds along with everyone else

Both Chris Parker and Crown weather did not believe it and ignored it in their forecasts

Then I received an email from Crown weather - See extract below

Crown Weather Services
Issued: Monday, May 22, 2017 12:05 pm EDT/11:05 am CDT

Verification statistics reveal that the GFS model currently has a 25 percent verification rate, which means that the model is currently only correct 25 percent of the time.

On the other hand, the European model guidance has a current 75 percent verification rate, which means the European model has more credence and believability right now.

Keep this in mind when looking at the GFS model guidance.

As far as I know, a lot of the internet weather package use the GFS model
i.e Passage Weather, Wind Guru, Windyty, Zygrib to name but a few

Question

If the GFS model is so inaccurate at the moment, can anyone suggest other internet weather products which either

Allow you to select the model

or

Are based on models other than the GFS?

[CarlF]

Predictwind offers GFS, ECMWF (European) and two of their own design based on these - PWG and PWE.

Santusky.com - GFS, Icon (German - Max Planck) and GEM (Canadian?)

Interesting piece from the NYT last October. The GFS is behind due to inadequate funding of the US Weather Service. So much so that the US Air Force pays the Brits $100,000 a year to use their software.

https://www.nytimes.com/2016/10/23/m...ther.html?_r=0

[wsmurdoch]

For us in the Bahamas it is really simple. In the morning we listen and record Chris Parker on 4045 and the same with BASRA on 4003 or WWRC on 7268. Now, here is the extra bit: we send an email to query-reply@saildocs.com with the subject "Spot Request" and with "send spot(NDFD):24.8N,76.2W" (not the quotes) in the body of the email. Five minutes later we have the NDFD forecast for that location in hand. It is a nice low, low bandwith process for HF email.

You might take a look at Available Saildocs Grib Data

Give it a try.

[Triumphant]

Thanks for the replies

re Predictwind , very interesting although when I tried it in W10 Chrome it did not run right told me I needed to Use Chrome!!!! but very nice on Edge


Santusky.co I think there may be a TYPO?

re the Saildocs option, I assume you need to have subscribed to get this service?

[barnakiel]

Originally Posted by Triumphant

Question

If the GFS model is so inaccurate at the moment, can anyone suggest other internet weather products which either

Allow you to select the model

or

Are based on models other than the GFS?

Unisys Weather

Have your pick. I never found ECMWF better than any other model but there are people who claim it is superior. Even if it is actually based on GFS.

Alas, I am not sure EU sages deliver US coverage. They gladly use US gps system but being Europeans they are not willing to do anything that would be of benefit to everybody. They also tend to see Americans as bad, greedy people.

I say, wait for gfs to smooth out the kink, and keep on rolling.

;-)
b.

[StuM]

Windy.com offer either GFS 22km or ECMWS 9km.

And it's easy to click between the two and see the difference at your chosen date/time.

[wsmurdoch]

Originally Posted by Triumphant

re the Saildocs option, I assume you need to have subscribed to get this service?

Saildocs is free. It is part of Sailmail. Home Page They have a catalogue of infomation that they can go and get for you, reduce to an email, and send.