red and green lines ahead of isochrones
Post by iconPost by jfloviou | 2020-12-15 | 18:36:06

Anyone able to explain why red lines and green lines dots are ahead of blue isochrons since a few hours (or maybe yesterday ?)
This is weird.

commenticon 12 Comments
Post by iconPost by BooBill | 2020-12-16 | 03:20:39
Are you possibly talking about the track lines? Those are normally turned on by default. If you've just discovered them, then you've just found one of Zezo's most powerful features.

The redline is your predicted track at a constant bearing.
The green line is your predicted track at a constant TWA.
The blue line is your orthodromic path if you were to say set a WP.

The angles for each are in that box that floats above your cursor.

In the future, to turn on this feature, it's the button in the top row of the controls just to the right of the zoom level.
Post by iconPost by NumeroDixTV | 2020-12-16 | 11:37:30
I think he meant the green line is going further than the iso
Post by iconPost by zezo | 2020-12-16 | 12:14:28
That's not unusual close to the VMG angles (150-155 in the Vendee now). The router does not get those angles right for the current wind speed, that's why you should always check the polars.

Another possibility is during the wind update process when the router and track tool can use different wind data for a while. That's usually just few minutes, up to one 10-minute refresh interval, but can be more if NOAA has issues and the entire forecast is delayed.
Post by iconPost by MidnightFoiler | 2020-12-16 | 12:38:53
As I said on the other thread, I also suspect the isochrones are factoring in sail changes between Staysail and HG which is perhaps more likely close to the max VMG twa of 155. You would see saw teeth shapes in the isochrones where it should be smooth curve (if it did no sail changes) and if you switch off the pro winches the saw teeth got bigger - could be based on adding gybes too though but I don't think so as I think it would require a sub 10 minute gybe for the size of the saw teeth.

I imagine the router does not factor in autosail change tolerance but simply picks whichever is faster and applies sail change penalties if it flip flops for different wind strengths.

Post by iconPost by zezo | 2020-12-16 | 12:52:14
That's very good observation. You've mastered the tool ;-)

The router tends to exaggerate the polar crossover points. The sail changes penalties mess with the algorithm a bit.

Because it uses dynamic programming and has no history of sail changes it will try blindly to sail in the current best direction, which creates worse isochrone than actually sailing two angles for a whle with one sail change. Also, with the penalties it can't guarantee a global optimum, because the problem space will grow exponentially.

But that's mostly academic issue. In reality the new weather forecast will bring more error than the one caused by the penalties.

We also have the green line tool to check for that condition (but it does not account for sail changes. Maybe it should)


Post by iconPost by jfloviou | 2020-12-16 | 14:38:16
Thanks for you answers Cvetan and MidnightFoiler. I think it explains this weird situation.
@Boobill: thx, i knew the tool, and how to use it, but i was surprised to see it not function properly.
Post by iconPost by Inicio | 2020-12-18 | 19:35:51
If it were possible to introduce the penalties in the calculation of the green line, that would be great. I almost always follow the green line. But with the autosail it can be a disaster!

Does it take a lot of calculation to enter them?
Post by iconPost by zezo | 2020-12-18 | 19:46:59
It's easy. But the sail changes are not exact. And one minute on the hour may not be very visible anyway.
Post by iconPost by Inicio | 2020-12-18 | 20:15:03
Nothing is exact in this business!

One minute two or three times a day means three or four hours right now in the race! That's many thousands of places in the ranking!

It would make a fantastic Christmas gift!

When you have time...
Post by iconPost by zezo | 2020-12-18 | 20:18:46
It's significant for sure, but will only be 1-pixel difference on the chart. And then the router runs with different step beyond 6 hours, so the track might use slightly different winds, etc. But I'll take a look.
Post by iconPost by Inicio | 2020-12-18 | 20:32:12
Thanks!

Post by iconPost by MidnightFoiler | 2020-12-21 | 14:21:12
That 1 minute might get stretched in a rich get richer situation. It always amazes me how much different pro winches makes when you compare etas with them on and off - often much more than the actual penalties.

Having said that I wouldn't want it added to the cursors if it ends up with the same problem as the isochrones. Don't think it would as I have not seen the sail change squares when the isochrones seem short of where they should be.
border
Topics list
Posts
border
5
border
border
Copyright 2009 by ZEZO.ORG. All Rights Reserved.