Weather Routing

[rgleason]

That will probably work too because the plugin api is the same for both versions 1.10

Sent from my SCH-I545 using Cruisers Sailing Forum mobile app

[rgleason]

Please see this post regarding Weather_routing_pi updates.
http://www.cruisersforum.com/forums/...ml#post1552098 Post #39

The post following notes having difficulties with the new version 11 Weather_routing. I never got it to work, but uploaded it to Openpn Beta File Thingie
anyway in hope that others would discover how to make it work.

Now today I tried it more completely, trying from a clean slate. Version 11 with Newtons Method and Runge Kutta routing along with several selectible uses of Climatology does not seem to run Climatology or Grib files. I have tried everthing I can think of. Sean do you have any suggestions for settings to start out with, or how to get this version to work? Thanks.

Version 10 works with climatology and grib and the several moving gribs I have tried.

PS: Now copying climatology data today.

[kubek]

Hi,

I tried to work with the new version of weather_routing yesterday.
And it works only in one situation - when I chose "Average".
With all other it failed to compute route.

See attachment

Kubek

[rgleason]

Kubek, Good Job!

[seandepagnier]

I'm in the process of debugging, and many options are broken. In any case, the average one doesn't even work correctly in some obscure cases which I am trying to find the best way to fix. It seems the correct way is slow.

Also keep in mind, all the other options will require the newest version of climatology to work at all.

I'm really not at a point for testing weather routing just yet, it's still being developed. If you want to do some testing, try climatology, or weather fax, as I would like to release updates for these soon.

[rgleason]

Ok but did try it with Clim on average. Ran Newport to Poole 19 days with clim and grib 16 day. There is high cpu that does not subside. Using newest version of clim and grib.

[seandepagnier]

Originally Posted by rgleason

Ok but did try it with Clim on average. Ran Newport to Poole 19 days with clim and grib 16 day. There is high cpu that does not subside. Using newest version of clim and grib.

I made some major fixes in the last few days, unfortunately the more I fix the more I find.

Anyway, the problem is the algorithm is a bit too complex to deal with. Perhaps if I rewrote it, it could work much better, but that won't happen in the near future.

I think It is much more useful for climatology now with the cumulative map, but still in variable areas especially for very short passages, the results using a current grib could be widely different.

[rgleason]

I think its better than that, but perhaps it is time for a rest and some perspective. I have no doubt it is complex.
Hope to send one HD with original climate data plus other stuff, and a Dell win dual processor with a 15 or 16 inch screen today. I hope it does not use too much juice, but you can use it as your win platform.

Sent from my SCH-I545 using Cruisers Sailing Forum mobile app

[seandepagnier]

Originally Posted by rgleason

I think its better than that, but perhaps it is time for a rest and some perspective. I have no doubt it is complex.
Hope to send one HD with original climate data plus other stuff, and a Dell win dual processor with a 15 or 16 inch screen today. I hope it does not use too much juice, but you can use it as your win platform.

Sent from my SCH-I545 using Cruisers Sailing Forum mobile app

I have too much solar power, I have to put tarps over my panels to avoid overcharging the batteries, and this is running a computer all the time and a 15 watt fan.

I made a bunch of fixes, and I'm getting very close to releasing the first version of weather routing (even if it's a bit buggy it's mostly usable, and the bugs just make it run slower, not incorrectly)

I found a few things.. the coastline detection is taking about 90% of the time. This can be fixed/optimized to give a huge speedup but will take a bit of work. The simplest way is to break down the data into smaller than 1x1 degree blocks on the grid. Notice the difference when detect land is off.

I have an optimization which nearly doubles the speed of computation but it has very weird anomalies and so I have to keep it disabled for now...

I believe I can double the execution speed again with basic optimizations and caching of parameters. A second doubling in speed would probably be possible with extensive work.

If all of this is done, I anticipate the routing taking 1 second to do what takes a minute now..

I think it's ready for testing again.

[kubek]

Hi,

It looks like broken logic (reverse/inverse).

As you can see in previous my post routing from Recife to Cape Verde is mostly windward and from all older routing plugins shows 26,5 days to reach Cape Verde. In opposite direction it took 14 days.

But now with new version of weather routing software it is inverse.

Kubek

[seandepagnier]

Originally Posted by kubek

Hi,

It looks like broken logic (reverse/inverse).

As you can see in previous my post routing from Recife to Cape Verde is mostly windward and from all older routing plugins shows 26,5 days to reach Cape Verde. In opposite direction it took 14 days.

But now with new version of weather routing software it is inverse.

Kubek

I should check for the climatology version.. just added the check in git.

The climatology plugin changed as well, I simplified the way the data is handled and as a result, wind direction is reversed.. so upgrade to the latest version (plugin checks now)

. The new one should work with all the options not just "average" which is very poor.

[seandepagnier]

kubek,

I see you are using windows and have the rtlsdr plugin loaded. Does it work? Do you have a dvb-t dongle to test? I would really like to know.. but maybe in the other thread.

[kubek]

Originally Posted by boat_alexandra

I should check for the climatology version.. just added the check in git.

The climatology plugin changed as well, I simplified the way the data is handled and as a result, wind direction is reversed.. so upgrade to the latest version (plugin checks now)

. The new one should work with all the options not just "average" which is very poor.

Hi,

Today, I performed fresh install of all plugins which were provided in "exe" format to have all newest version of it. So Climatology and Weather_Routing are definitely new, data for climatology is version 10. Please try to sail to winward with weather_routing (route from Capo Verde to Las Palmas-Canari will be good example) and you will see what I unveiled.

Thanks for great plugins,

Kubek

[kubek]

Originally Posted by boat_alexandra

kubek,

I see you are using windows and have the rtlsdr plugin loaded. Does it work? Do you have a dvb-t dongle to test? I would really like to know.. but maybe in the other thread.

Hi,

Yes, it is on my wish list. I have already bought a dongle, now I try to find a appropriate antena cable to connect my usb receiver to VHF antena and on 21st of June I will start my next trip from Lizbona to Madera and Las Palmas so I will try how it works.

Kubek

[kubek]

Originally Posted by boat_alexandra

kubek,

I see you are using windows and have the rtlsdr plugin loaded. Does it work? Do you have a dvb-t dongle to test? I would really like to know.. but maybe in the other thread.

Hi,

I have one suggestion, could file "rtlsdr_pi-0.10-win32.exe" installer have bundled rtlsdr_winutilities.zip inside.
Maybe also is good idea to put this "rtlsdr_winutilities" inside plugins directory like: "C:\Program Files (x86)\OpenCPN\plugins\rtlsdr_pi\bin\".

It will be nice If installer can add to the user variable PATH a path to this utilities bin directory.

By doing that, installation process will be as simple as possible.

And there will be no questions on forum how to install all this staff to work together.

Kubek