Predict Wind: Which model

[Paul L]

Originally Posted by jbinbi

For what I am looking for, which is say on Wed to look what the winds on the bay will be on Sat, to see whether or not it will be a good sailing day, I think should have very high reliability.

The local weather forecast for this 48 hour in advance is pretty darn accurate, they just don't have marine forecast details as when you want to go to the beach, whether or not the winds are at 7 or at 13 makes no difference. But if you are thinking of going sailing, that difference makes ALL the difference!

So with predictwind, I find it odd that 2 of the 4 models agree, and say 1 is quite a bit different.

See attached, which is the next 2 days. First red highlight, 3 are pretty close, so that is probably what is going on, though ecmf is WAY off. Next highlight, on Wed has PWG and GFS saying very little wind, with the other saying OK wind. Which to believe?

The two added "models" on ,Predictwind are just minor tweaks based directly off the GFS and Euro model. They are more marketing hype than useful.
Stick with the base models for useful analysis.

[Dave_S]

Obviously I'm wrong but it baffles me how with all the resources available and the historical data why the predictions aren't more accurate.

I realize the old salts will tell me that it is so much better than it was but it's not much good if it is reliably wrong.

[hellosailor]

Dave-
Very simply because none of the models has up to date information on the monarch butterfly population in Mexico.
[That's referring to mathematical chaos theory, where it is said that the actions of one butterfly flapping its wings can ultimately reverberate thousands of miles away and affect what is happening there days later.]

Can you tell exactly what the cloud cover will be tomorrow? Which affects solar heating on the ground or sea below it, which in turn affects wind formation from the heating of the air above that ground? No. So, you can't be certain of the outcome three or five or seven days later, the best you can guess if often 48 hours.

I've gone to bed with forecasts of "5mph easterly winds" and headed out into 4-8' seas and gusts to 40 knots that had half of a racing fleet cancel before the start. What happened in 12 hours? Damfino.

[chouliha]

Originally Posted by arsenelupiga

here in south pacific, after testing different approaches i follow ecmwf if all models diverge as it hit nail most often.

GFS is immediately after and study above two in tandem.

The other two only use for confirmation - ie if all 4 forecast same then strong probability.

Also if one component predicted similar, say time of arrival of strong winds, but other things diverge, will take time into account for higher probability forecast.

In throughs, forecasts are nearly meaningless and plan B is more important.

Observing sky, clouds, waves, animal behaviour, adds another dimension.

thats what i learned so far.

I second the use ECMWF models in the South Pacific / French Polynesia. Much more accurate than the GFS. We have been using Windty for a number of years and its been great.

[zboss]

Euro model for open ocean weather, GFS for near shore work.