Favourite Weather sites - opinion?

franksingleton

Well-known member
Joined
27 Oct 2002
Messages
3,499
Location
UK when not sailing
weather.mailasail.com
Which website?

Interesting. I always see something new from such a thread. Some comments of my own -

First, most GRIB or GRIB derived products, and that applies to all well based forecasts that are automatically generated, are derived from the US GFS. There is little systematic difference between this and other NWP models eg the UKMO, CMC.

ECMWF does a little better at the longer times ahead ie their 96 hour forecast is as good as the GFS 84 but is available several hours later. This is because ECWF concentrates on the medium range and does not start its forecast until all available data have been received and incorporated. The US GFS, UKMO, Météo France, DWD etc have more immediate operational constraints and have to use data cut-off times a few hours earlier than ECMWF.

Secondly, I know of no detailed NWP forecast, available and useful to us, that starts with a good detailed data analysis using all the masses of data available - see http://weather.mailasail.com/Franks-Weather/Weather-Observations-Nwp. The models used are fine. However, the necessary data handling to get the best analysis possible is, itself, a huge task and involves some very heavy mathematics. A quick look at http://www.metoffice.gov.uk/research/areas/data-assimilation-and-ensembles/4d-var-research will give an idea of what is involved. That is not to say that these “detailed” forecasts are no good. Compared to the GFS, the UKMO, the UK NAE and COAMPS they are likely to give more data but no more real information.

Remember that there is nothing truly certain about any non-trivial forecast except for the date. Even then, keep the theologians...........................

In summary, there is little to choose between the various weather sites mentioned. Use that which you find most convenient. Do not expect any greater accuracy from any individual site. Use forecasts, especially the GFS, UK Met Office forecast synoptic charts and ECMWF for planning ahead up to about 5 days. For day to day localized forecasts, use whatever forecast you prefer but with your own input borne of experience.
 

jfm

Well-known member
Joined
16 May 2001
Messages
23,711
Location
Jersey/Antibes
Visit site
I use www.theyr.com for the Med winds and like it, and I use the mueller models linked to from www.greatweather.co.uk that Tim mentions above for synoptic charts. In the med you have to be careful using synoptic charts becuase much wind (eg Mistral) is not gradient wind so a synoptic chart isn't gonna help you much. I agree with you that you have to dig a bit to find which model they're using
 

Hurricane

Well-known member
Joined
11 Nov 2005
Messages
9,405
Location
Sant Carles de la Ràpita
Visit site
Interesting. I always see something new from such a thread. Some comments of my own -

First, most GRIB or GRIB derived products, and that applies to all well based forecasts that are automatically generated, are derived from the US GFS. There is little systematic difference between this and other NWP models eg the UKMO, CMC.

ECMWF does a little better at the longer times ahead ie their 96 hour forecast is as good as the GFS 84 but is available several hours later. This is because ECWF concentrates on the medium range and does not start its forecast until all available data have been received and incorporated. The US GFS, UKMO, Météo France, DWD etc have more immediate operational constraints and have to use data cut-off times a few hours earlier than ECMWF.

Secondly, I know of no detailed NWP forecast, available and useful to us, that starts with a good detailed data analysis using all the masses of data available - see http://weather.mailasail.com/Franks-Weather/Weather-Observations-Nwp. The models used are fine. However, the necessary data handling to get the best analysis possible is, itself, a huge task and involves some very heavy mathematics. A quick look at http://www.metoffice.gov.uk/research/areas/data-assimilation-and-ensembles/4d-var-research will give an idea of what is involved. That is not to say that these “detailed” forecasts are no good. Compared to the GFS, the UKMO, the UK NAE and COAMPS they are likely to give more data but no more real information.

Remember that there is nothing truly certain about any non-trivial forecast except for the date. Even then, keep the theologians...........................

In summary, there is little to choose between the various weather sites mentioned. Use that which you find most convenient. Do not expect any greater accuracy from any individual site. Use forecasts, especially the GFS, UK Met Office forecast synoptic charts and ECMWF for planning ahead up to about 5 days. For day to day localized forecasts, use whatever forecast you prefer but with your own input borne of experience.

I have spent very happy :confused: evenings reading your website.
Personally, I use Windguru as it has two models displayed side by side but the big thing I picked up from your site was that local knowledge (headlands etc) play a huge part of the actual conditions you get.

I hope I read it correctly and thanks for providing such fascinating information.
 

franksingleton

Well-known member
Joined
27 Oct 2002
Messages
3,499
Location
UK when not sailing
weather.mailasail.com
Hurricane

Thanks for that. Like I said, if you find it useful use it. However, be aware of the limitations; but, yes, you are correct. Whichever fine scale model is being used, local knowledge, experience and nous are important tools in the thinking mariner’s armoury.

Windguru starts from the GFS, interpolates, uses no more data (as far as I am aware) and runs a model on a 27 km grid. However, if you go to Weatheronline.co.uk, you can see the GFS which now runs on a 27 km grid globally using all the available data. I know which has my vote.

Windguru and others interpolate further to smaller grid lengths, without more data – again as far as I know – and compute on a 9 km grid. They will calculate local effects well but it will only need a small error in the large scale pattern to affect adversely the outcome. The UK NAE uses a 12 km grid but, again, uses all data available. It should be better. The same should apply to the USN COAMPS, although the last time that I checked they were not using quite as sophisticated analysis scheme as the GFS or the UKMO.

PS If Windguru pr any other (non-National Met Service) producer of meso-scale forecasts really does use the data sources listed at http://www.ecmwf.int/products/forec...ynop-ship!00!pop!od!mixed!w_coverage!latest!/ with a 4D Var or 3D Var data assimilation scheme then I will be more than happy to say so.http://www.metoffice.gov.uk/research/areas/data-assimilation-and-ensembles/4d-var-research is a "simple" guide to 4D Var.
 
Top