Feedback on new source GRIB files and their accuracy - new favourite

[tudorsailor]

Rather than continuing an old thread which had drifted I thought that I would feedback on the new GRIB files that a fellow sailor has made available for Med sailors. The website is http://openskiron.org/en/ There are two types of GRIB files on the site, but the WRF ones are meant to be more accurate.

On a PC one can view them with Zygrib. On an ANdroid device they are easily downloaded into SailGrib WR.

I have just returned from 3 weeks in Greece going from Pelopennese to Ionian. I used to like Predict Wind but did find that it was inaccurate even with the choice of models.

I have to say that SailGrib with the WRF files quickly became my preferred choice for weather. It was surprisingly accurate. It now includes CAPE data (for thunderstorms) which was missing in PredictWind.

The Skiron website could do with tidying up, but once you get used to choosing the correct file, it is easily brought into SailGrib

The author of the Skiron site is quick to respond to messages and is altruistic in providing open source information. I would suggest others try the files in other parts of the Med and report back.

For people with iPads, I think that the process is not so streamlined. You download the grib into the download folder and then use PocketGrib to open.

I have no connection to either Skiron or SailGrib. Just a happy use.

TudorSailor

 

[lpdsn]

I've been really impressed with the GFS since it went to 0.25 deg cells. Both this weekend and last I was following it from ten days out or more and it was pretty consistent all the way in to the actual weather we got - certainly for NI and the Irish Sea.

 

[franksingleton]

If they are willing to reply to you, could you please ask them what initial weather data do they use. My best guess is that they use the GFS 0.25ndegree data and nothing else.

Their model is the WRF which is as good as any limited area model. The Predictwind model was developed by CSIRO. I have little doubt that it will be similar to the WRF. More depends on the data input than on the models per se. At one time PW used to use the GFS and the Canadian models as input but at a lower resolution – when the GFS was issuing 0.5 degree data, PW were running a global model with a one degree grid. They then ran limited area models with no more weather data. It seemed crazy then to run models with 8 and 1 km grid lengths. I do not know what they do now.

The PW idea of running their model with two different data inputs is not very intelligent. It is a two-member ensemble; ensembles should be far larger – 50 or so. Two members tells you nothing.

As long as you are prepared to use a little common sense over the first 36 hours or so, the GFS will be as good as anything over that period. It will outperform all limited area models beyond that.