Weather Routing

[hasse_A]

Originally Posted by rgleason

Is this happening all the time, with any configuration and any grib file or is it a specific setup?
If it is only a specific setup, please compress the offending files and upload or send to me at my email. Thanks.

This happens once in while.
It's a bit treacherous. It comes and goes and I have not been able to correlate the problem with a special configuration. I first experienced it 2 years ago and it did at the time exist on 2 different computer both running Win 10. At the time I think I was running OpenCPN 5.2.0.
Anyhow the problem shows itself on the revisions specified above.
I will salvage input files and configuration and send them over.
Maybe not until tomorrow.
Thanks Rick

/Hans

[LeaseOnLife]

Originally Posted by seandepagnier

Did you check the settings? Maybe make the line thickness more?

Originally Posted by kevinvh

The iso chrons no longer display when computing a weather route using WR PI 1.13.39.0 and OpenCPN 5.6.1+1842.1427f287c. Works with OpenCPN 5.6.1+6ab733c73.

I have logged a Bug Report with Flyspray.


MacBook Pro (15-inch, 2018)
2.6 GHz 6-Core Intel Core i7
16 GB 2400 MHz DDR4
Intel UHD Graphics 630 1536 MB
MacOS Monterey V12.3

same here on a linux amd64 laptop with latest from github. the routing completes fine, the statistics show number of isochrones calculated but the isochrones do not show up on the chart, even with thickness =10 (max).

[seandepagnier]

Originally Posted by LeaseOnLife

same here on a linux amd64 laptop with latest from github. the routing completes fine, the statistics show number of isochrones calculated but the isochrones do not show up on the chart, even with thickness =10 (max).

Is this with opengl enabled or not?

Maybe changing this setting? Does everything else display?

[LeaseOnLife]

Originally Posted by seandepagnier

Is this with opengl enabled or not?

Maybe changing this setting? Does everything else display?

opengl was enabled and everything else I use is showing (grib / climatology / tactics / weatherfax / radar)


with opengl off, the isochrones are being displayed!

[wernertoonk]

Just a brief note to confirm the above findings: With OpenGL de-activated (via Settings > Display > Advanced : Use Accelerated Graphics (OpenGL)) the isochrones are shown. (OSx, OpenCPN 5.6.1+1427f287c). Though, without OpenGL the Climatology overlay map is unsupported.

[rgleason]

Werner Thankyou. I can confirm on Win10 v5.6.1+1427f287c
That with opengl= Off the isochrones show. For me, Climatology shows with opengl on or off. In WR think I twisted some settings in cmake for opengl. I am going to check.
This also confirms Leaseonlife's report (thank you)



Werner, also I agree with you. This is something new to me I hadn't routed in the southern hemisphere on the other side of the time zones. Routing Whangaroa Bay to Whangamumu Harbor time period set to 5 min.

1. Turning off opengl allows the isochrones to work. (I must have twisted something in the builds. I think I know what it is.)
2. WR works with Climatology and a fresh Grib, as long as land avoidance is not turned on.
3. When Land Avoidance is checked, it fails.
4. Is this related to https://github.com/rgleason/weather_...g_pi/issues/79 and https://github.com/seandepagnier/wea..._pi/issues/256 ?

First two are set to 5minute interval. Without Land avoidance checked and with land avoidance checked.

[rgleason]

Here is a fairly tough routing around Nova Scotia in our part of the world, and it works.
Time interval set to 1.5 hr. Why is routing with Land Avoidance = ON working here, but not in NZ???

[seandepagnier]

it certainly used to work in the southern hemisphere...

Can you confirm landfall avoidance is really a southern hemisphere issue?

[rgleason]

I can confirm that landfall avoidance does not work in and around NZ with WR 1.13.30.4 and the current version 1.13.40.0 with Grib 4.1 However it does work with Climatology 1.4.44


So can Werner Toonk

[rgleason]

Now with a newly compiled version with some minor changes WR 1.3.40.2
and trying to route around Cape Reinga, it seems to work!!!


I am not really understanding this just jet.

[rgleason]

Ok, I have something more definitive now. I can route from the West side of Cape Reinga to Bream Bay or Bream Bay 2 further out, or Bay of Islands but not Matiuwhi Bay, Waikare Inlet, with these settings.
Max Diverted Course = 160
Time Step = 30 min
Reset all Advanced Parameters
Safety Margin from Land = 0.1 nm <--to avoid land,
However I notice that there are some big discrepancies in the High Res Land configuration as compared to CM93 Charts!

I haven't been able to get the NZ S-63 yet because they have not sent me a Permit.txt file for 4 weeks, after signing up and requesting same.

I also have some observations bugs about weather_routing.


1. WR does not check to see if the Start or Destination points are greater than the setting for "Safety Margin from Land" and provide an appropriate error message. It just calculates endlessly and finally comes up with "no data".

2. WR is pretty dumb when it comes to continuing calculations way past the destination point. If the best route has come pretty close to the destination and then successive calculations result in longer routes that are further away from the destination, shouldn't the program just stop? It is even worse when you have checked "Last valid if data deficient". Generally it just calculates until it has run out to the grib file extent.

3. WR does not have a dedicated "Stop Calculation" button, which would be very nice because, it would leave the calculations and isochrones available for inspection rather than wiping them out when hitting the "X".

4. Also hitting the "X" to stop the calculations causes the program to fail and OpenCPN to crash.

5. When there are many many calculations, the program seems to reach a limit on memory and just crashes, rather than stopping and giving us a message that it has exceeded the memory limits of Opencpn or something like that.

6. If it is calculating, and I try to Close the "Weather Routing Configuration" menu, that causes the Plugin to freeze and then fail.

7. If it is calculating, without the "Weather Routing Configuration" Menu open and I try to move the screen to get it centered better, the PI freezes and fails Opencpn.

8. When the PI fails, all my settings are lost and I have to repeat and do it all over again! Why can these settings be saved earlier in the process, so that if there is a failure, I can go back in and just change one parameter?

9. The "Safety Margin from Land" is a excellent feature, but it caused me a lot of time and trouble because I had it set on 2 miles without being aware of that and I kept having routes that failed because my destination was too close to land. There should be a warning about this.

10. When "Reset Advanced" is selected the "Safety Margin from Land"is set to "0". This is good and is the best setting for newbies.

11. I see that the plugin does declare success sometimes even if the actual routing at the end has not fully completed. I would like to know what is the built in logic around this, perhaps it can be improved?

12. I still often wonder if a routing has come quite close to completion, say within 1.5 of the time interval being used, but it does not succeed, why can't the time interval be halved for subsequent calcs, also why can't the "Safety Margin from Land" be halved if it is not Zero? Additionally Max Diverted Course could also be changed to 180 degrees at that time.

13. I would like to see the entry for "Safety Margin from Land" show decimal point and smaller numbers better. The period does not show at all on my screen. (I used 0.1 nm and it showed as "1" which is kind of confusing!

[rgleason]

To get it to compute from Bream Bay to West Side of Cape Reinga I had to set max diverted angle to 180 because it would not get out of Bream Bay.

This is an example where the start point Parameters should also be dynamically improved until it succeeds, and then the parameters should be set back to the original settings.

Perhaps there should be another Tab for Configuration, in addition to "Advanced" which would be labeled "Slow start/arrive" and would be the dynamic settings to be used when the routing is close but about to fail.

Time Interval = (some setting, perhaps 10 min)
Max Diverted Angle = 180
Safety Margin for Land Avoidance = 0
any other critical parameter.

See Issues

Slow Start https://github.com/seandepagnier/wea...ng_pi/pull/205

RFE: Dynamic adjustment of start and end isochrone time interval to complete https://github.com/seandepagnier/wea..._pi/issues/202


qtVlm and Sailgrib has weather_routing that is more user friendly and seamless. I wish OpenCPN's weather_routing was just as user friendly.

[rgleason]

With Opengl turned on, using this WR version WR 1.3.40.2 the isochrones do not show. I have gone as far back as WR 1.3.30 (including others in between) and find that this is the way Weather_routing has been working.


I do not remember if WR every had the isochrones showing when opengl was turned on, perhaps Sean remembers. It would be nice if they worked in both states.


I will push this newer version WR 1.3.40.2 up to my github and it will build in cloudsmith opencpn/weather-routing/alpha or beta. It will also be automatically pushed to a PR on Sean's repository for his review.

[rgleason]

I've commited and pushed WR 1.13.41.0 + TP 1.0.196 to github.com/rgleason/weather_routing_pi and that is now finishing the OS builds (except Android) and has been deployed to Cloudsmith opencpn/weather-routing-prod

This version has also been pushed to sean's repository and is found under Pull Requests.

We would like testing and reports to confirm what I have found. Use the "Import Plugin" feature to download your OS tarball from Cloudsmith.

[seandepagnier]

Originally Posted by rgleason

Max Diverted Course = 160

Why not 180?

1. WR does not check to see if the Start or Destination points are greater than the setting for "Safety Margin from Land" and provide an appropriate error message. It just calculates endlessly and finally comes up with "no data".


2. WR is pretty dumb when it comes to continuing calculations way past the destination point. If the best route has come pretty close to the destination and then successive calculations result in longer routes that are further away from the destination, shouldn't the program just stop? It is even worse when you have checked "Last valid if data deficient". Generally it just calculates until it has run out to the grib file extent.
[/quote]
It's dumb, well it's hard to be "smart". Don't use lat valid if data deficient if you want to deal with realistic results.

3. WR does not have a dedicated "Stop Calculation" button, which would be very nice because, it would leave the calculations and isochrones available for inspection rather than wiping them out when hitting the "X".

4. Also hitting the "X" to stop the calculations causes the program to fail and OpenCPN to crash.

5. When there are many many calculations, the program seems to reach a limit on memory and just crashes, rather than stopping and giving us a message that it has exceeded the memory limits of Opencpn or something like that.

6. If it is calculating, and I try to Close the "Weather Routing Configuration" menu, that causes the Plugin to freeze and then fail.

7. If it is calculating, without the "Weather Routing Configuration" Menu open and I try to move the screen to get it centered better, the PI freezes and fails Opencpn.

8. When the PI fails, all my settings are lost and I have to repeat and do it all over again! Why can these settings be saved earlier in the process, so that if there is a failure, I can go back in and just change one parameter?

Why dont you start up and make your settings then exit cleanly saving them and avoid this?

Obviously the plugin shouldnt be crashing and should allow better control over it's processes. It may need to check and make sure it isn't using too much memory or something, though I generally found it to be cpu bound.

9. The "Safety Margin from Land" is a excellent feature, but it caused me a lot of time and trouble because I had it set on 2 miles without being aware of that and I kept having routes that failed because my destination was too close to land. There should be a warning about this.

10. When "Reset Advanced" is selected the "Safety Margin from Land"is set to "0". This is good and is the best setting for newbies.

11. I see that the plugin does declare success sometimes even if the actual routing at the end has not fully completed. I would like to know what is the built in logic around this, perhaps it can be improved?

12. I still often wonder if a routing has come quite close to completion, say within 1.5 of the time interval being used, but it does not succeed, why can't the time interval be halved for subsequent calcs, also why can't the "Safety Margin from Land" be halved if it is not Zero? Additionally Max Diverted Course could also be changed to 180 degrees at that time.

There are too many possible cases and parameters generally. You could try to code this "logic" but for other cases it would fail. It would make the resulting isochron ma difficult to read unless you are careful with the rendering.

13. I would like to see the entry for "Safety Margin from Land" show decimal point and smaller numbers better. The period does not show at all on my screen. (I used 0.1 nm and it showed as "1" which is kind of confusing!

it would be better to somehow get vector chart data and use that. That way it could use safety margin from grounding depth, and the resolution would be high

at .1nm with gshhs you will have problems