SUJET :

Hi,
Temp, wind speed and humidity can be found on most grib files available included all the download possibilities provided by XyGrib.
Solar radiation is not commonly used and not even published by most models. It is usually part of non-hydrostatic models (local area models) that simulate local heat exchanges. I suggest trying the documentation for DWD-ICON, DWD-COSMO, Meteo France Arome and other open access models.

Second problem that you might have is to read the solar radiation as most grib viewer do not include it. For this is suggest using NOAA-Panolpy.

Frank écrit: As a long retired professional meteorologist, a sailor and a cynic, I do have to question the real value of the ECMWF forecasts over and above GFS and ICON. I am well aware that ECMWF out-performs the rest, slightly better than the U.K. and a little more so than GFS and ICON. But, when I look at www.weathercast.co.uk/services/ensemble-forecast.html, I see very little consistent differences. Such differences as there are, are well within the natural uncertainty, noise if you like, in the weather. There are limits to predictability and that basic fact is not appreciated by the vast majority of the public, even the more meteorologically aware.

As a sailor, I see no point in using Predictwind to get ECMWF output. You are doing a great job, David.

I disagree.

while it is true that, on average, the global models all perform at nearly the same level.., for any individual weather system, there can be a large difference in the forecasts - especially several days out, and it is pretty common for one or two models to outperform the others by a margin that is meaningful to sailors.

While the ECMWF often scores the highest in verification tests.., and on average might be the best.., that doesn't mean it always outperforms other models, and for any given weather system, it always a good idea to figure out which model is performing the best and try to use that one. In the US, this is easier than it sounds, because the NOAA weather discussions typically include a discussion of how they believe the models are performing, and which models they are using for their forecasting. on those occasions where the EC is doing a better job, i would like to have it.

until a year or two ago, both predict wind, and Squid sold access to the 9km ECMWF - it could be downloaded in grib format and imported into routing programs like expedition and adrena. they no longer do this, because the EC decided that by making the gribs loadable in 3rd party programs, they were violating the license. there was some discussion about value-added - the license required that the re-seller not sell the ECMWF in it's original state, but somehow add value to it. I think both predict wind and squid were slightly modifying/scaling the windspeeds or something like that, but EC apparently decided that wasn't sufficient value added and prohibited them from making the data available to other programs.

as of now, predict wind still sells the 9km ECMWF, and gives away for free the 50km ECMWF - but you can only view it in their software.., you can not use it in adrena or expedition. I think squid has stopped selling ECMWF, except on a custom basis.

Personally, I would be happy to pay for ECMWF in a grib format that could be used in expedition. I am mostly a racing sailor.., and i appreciate that cruising sailors might not see the value

jgf said :
"Personally, I would be happy to pay for ECMWF in a grib format that could be used in expedition. I am mostly a racing sailor.., and i appreciate that cruising sailors might not see the value"

Hello,
I am cruiser and racer, as well as helping my friends (racing or cruising) with weather routing. This means for the later case and to prepare passages that when onshore I have access to multiple sources and I use them, including ECMWF output.
When sailing, sources are more limited for various reasons, including the fact that race organisers allow you only to use the data they are providing. All the numerical prevision models have advantages and disavantages, knowing them is crucial for interpretation as well as local observation (that can be substantially different from the model). Making prevision and taking decisions to be fast and safe can be very complicated sometimes. This also include the fact not having all available sources when taking a decision.

But the most important thing is the basic of our project : "Provide free and open Gribs to the user community that deserves free & open high quality forecast resources. " which exclude by definition non free and/or non open data"

If you (and others) wish to use predictwind (non free and non open and not working on OS linux) we do not have any problem. But it is true that today, it is possible to sail (race and cruise) with excellent free and open tools and data.

DomH écrit: But the most important thing is the basic of our project : "Provide free and open Gribs to the user community that deserves free & open high quality forecast resources. " which exclude by definition non free and/or non open data"

Sorry, I wasn't very clear - i understand and support the goal of opengribs.., and wasn't suggesting that opengribs should think about providing models by paid subscription.