Post by iconPost by marcusbelgicus | 2019-06-24 | 18:48:17

Hi Cvetan,
It has been a while! Racing in PDD only...
For short races like the Figaro, the weather forecast for the next couple of hours is crucial. I recall that you inject in the model the first 3 hours of new forecast only after 1 hour, to take into account the progressive phase-in, phase-out of the new forecast in VR. This works nicely for long races, but for short races with rapid weather change, waiting for an hour can be too late to take the best decision. Could it be possible to have a routing model for short races, where all new weather forecast is immediately taken into account ?
Thanks a lot for your tremendous support!

commenticon 3 Comments
Post by iconPost by Badseed | 2019-06-26 | 13:46:23
Hi Cvetan
Marcusbelgicus enlighten me on this weather update protocol on the router, a novelty for me.
I was used to it being progressive, starting at 15:31 UTC and then 15:41; 15: 51 ... and so on and to consider that the progression of the red isochron on the map indicated this state of progress and its immediate consideration in the routing.

Back in the Figaro, I thought it was a bug before I realized that was systematic.
The wind data displayed on the map and the routing remain the same between 15:31 UTC and 16:31 UTC for example.
The change of the display of the data and the new routing takes place 1 hour after the start of the update in one time.
In practice, it is only an hour later that we can get an idea of the changing winds and understand, in retrospect, why some players, who have other sources to exploit the NOAA gribs, have tacked or gybed ¾ or ½ hours before us (recent example on Figaro.leg 3).

For my part, on these races of 2/3 days, I preferred the old method because the corrections of winds and routing were known from 15: 41 or 15: 51 for example. Here, we are in the situation where we know that the new winds are available, but we will only have them one hour later. For an hour, we navigate on old data!

Keep up the good work!

Post by iconPost by zezo | 2019-06-26 | 16:33:03
There is a difference - the router does not interpolate between old and new GFS for the same point of time during the update.

The problem is I am not sure the VR client and server use the same schedule - and they seem to differ.

So in general I can probably sync the router to the client, but can't guarantee that the server sees the same wind, so decided to leave it as it is.
Post by iconPost by marcusbelgicus | 2019-06-26 | 19:21:58
Hi Cvetan, the VR client does not use the same schedule than the server. The new GFS is fed in one shot in the client, this is why we see a big change in the projection line around 11h45 UTC+2 , 17h45, 23h45 and 5h45; while there is a progressive phase-in phase-out on the server. I have been wondering if VR is actually not changing the interpolation parameters between old and new GFS from one race to another to make the routing more complicated. Don't know how much complex it would be to implement a random parameter that would change the interpolation parameters from one race to another.

Topics list
Copyright 2009 by ZEZO.ORG. All Rights Reserved.