Version française : LIEN
Thanks to Alain Cadre, a very experienced windsurfer and meteorologist, to share his great experience.
Forecasts come from numerical forecast models. so it’s important when you go to a weather forecast site to know which model has been used. no need to go to another site that uses the same models.
The results are displayed as value tables or maps. always prefer the maps, the tables are extractions of the cards at a accurate point and don’t show the tendencies. Indeed it is interesting to know what is happening around the place where you are. for example for the passage of a front, if it passes earlier or later than expected you will not have the same conditions. the value chart will not give you the weather situation and will not allow you to anticipate if there is a time lag compared to the forecast.
There are many predictive models these are the ones I use: for the long term, 10 days, the European model ECMWF. for the medium and short term the French models arpège, 4 days, and aroma, 48 hours.
you will find several sites displaying these models in the form of dynamic maps in a more or less pretty way. these are the ones I use: for ECMWF http://www.windy.com, for arpeggio and aroma http://www.meteociel.fr/modeles/arpegezoom.php and http://www.meteociel.fr/modeles /aromezoom.php. (The display is not very nice and only in km / h, but I have not found better for the moment).
you have to look at the average wind and the gusting wind to get an idea. the average wind is over 10 minutes, unlike, for example, wind-up readings (2 minutes).
Do not look at one model, for example for the next day, look at what gives these 3 models. do not take wind values to the letter, but adapt them according to what you have seen in recent days compared to what was planned. Ask for the advice of the locals. In other words, the more information you have from different sources, the better you can get an idea of what lies ahead.
Tracking the evolution over several days before a deadline will also anticipate changes not foreseen by models such as a front-run faster or slower than expected.