Weather Routing

[rgleason]

I just checked the default routings, and they are not using the power polar, so I must have changed it earlier.

[rgleason]

Would a Tabular formated screen like this help at all after routing?
https://github.com/seandepagnier/wea...g_pi/issues/80

[rgleason]

Here is yet another Windows Release, the result of both Stelian and Sean's collaborative efforts.

Weather_Routing_1.11
https://github.com/rgleason/weather_...eases/tag/1.11

[Gilletarom]

Hello Sean, hello All,

I wanted to update the French tutorial page for WeatherRouting. I found here the latest version and I did some tests.

1 ° A small detail, though:
Here is a screenshot. the blue route has been exported.
Another road appears almost confused with the blue route.
in both cases, I wanted to create a route from the starting point located at the small green flag, bottom left, to the point marked "3", at the top of the Cotentin.
We note that during the export, the last part of the road disappears. This always happens. It is therefore not possible to keep the expected arrival time.

2 ° For this test, I used a grib file with only the wind data. I did not use the Climatology data. I also used a very complete polar (Figaro2). On the screenshot, I placed various points and in particular points 3 and 4.

Whether with 1h of time step or 30mn of time step, with 5 ° angle or 2 ° angle or 10 ° angle, the rest being identical to the content available during the installation of the plugin, it is impossible to obtain the creation of a route from the starting point (green flag) to point 4, down east of Cotentin. In this case, the search stops because the route comes up against the west coast of Cotentin, about halfway up the Cotentin. Obviously, as with older versions of the plugin, the error announced is "polar error".

It is quite curious, because with the same settings, it is easy to get a route that starts from the green flag and goes to point 3.

It is really difficult to extol the merits of this plugin in a tutorial. I see that it's difficult to improve since during my last test campaign on this plugin, there is a little over a year, I also encountered abnormal situations more or less of this kind.


Nota bene : Cotentin, French coast, channel sea.

Here join grib file and WeatherRouting points. Put out ".doc" on each file.

[stelian]

Gilettarom,

Try 'max deviation from COG' at 120 (or more), it will complete the routing to your 4th waypoint.

[Gilletarom]

Originally Posted by stelian

Gilettarom,

Try 'max deviation from COG' at 120 (or more), it will complete the routing to your 4th waypoint.

Indeed, Stelian, you're right, in this case, with 120 ° it works. the road is built.

That said, how do you want the alpha user to know how to use such a tool, given its complexity? Myself, I am on this subject the, an alpha user. And it will take me a long time to find a solution.

Didier wrote in another post, that we had to find a way to simplify the way to run this plugin, and I understand that it's difficult, but in the current state of things, it's Didier who has reason.

[stelian]

I wonder what is the reason behind this setting (max deviation from cog). Is it only in order to simplify computation ?

Many times I've found that routings fail because of this setting being too low. Maybe we should just rethink its default value.

[seandepagnier]

if the max deviation is high enough, say 180, then it can explore all directions, but takes a very long time. With a more direct route possible the smaller this can be, and the routing completes quickly.

Now maybe if the optimal route is stopped by this, we could suggest the user to increase it. Also, if an optimal route is found, but the route touches the edge of the route map, this value should be increased.

[rgleason]

These are good ideas for improvements.

Currently, the examples are set at Max diverted course of 90 degrees, perhaps we should adjust that to 120 for all of those examples?

Max Diverted Course 90 deg [120 deg?]
Max True Wind 100 knts [25 knts?]
Max Apparent Wind 100 knts [30 knts?]
Max Swell 20 meters [5 meters?]

Max latitude 90 [ good]
Tacking Time 0 [ 120 seconds?] Is this seconds?
Wind vs Current 0 [What is this value? What does it mean? What does it do? Units?]
Max Course Angle 180 degrees [How is this different than [Diverted course?]
Max Search Angle 120 degrees [Can you explain what this is for specifically? How is it different than diverted course?]

Sean if you can take a moment to explain these, I'll promise to get them into the manual! Also let me know if I should change any of the configuration settings please. Thank you.

Originally Posted by boat_alexandra

if the max deviation is high enough, say 180, then it can explore all directions, but takes a very long time. With a more direct route possible the smaller this can be, and the routing completes quickly.

Now maybe if the optimal route is stopped by this, we could suggest the user to increase it. Also, if an optimal route is found, but the route touches the edge of the route map, this value should be increased.

[SoggyPaws]

Originally Posted by rgleason

Would a Tabular formated screen like this help at all after routing?
https://github.com/seandepagnier/wea...g_pi/issues/80

YES, absolutely. From the currently supplied reports, I have a hard time "validating" the reasonableness of the routing. A tabular form that lets you run down the route points, providing wind speed and angle, and boat speed and angle, makes it very easy to see if your Polar is set up properly (and whether the routing algorithm is working as expected).

I know I can get that info from the graph, but it is much more difficult.

It might also help in figuring out what actually stopped the routing process. I found I had had inadvertently set the "Max Latitude" setting in a route, and the routing couldn't reach the point I was trying to reach. The error was "Polar Failed", but it should have been something like "Constrained by Latitude". I did eventually figure it out, but looking at the Polar would not have fixed the problem!!

I use the tabular form also for validating in another way.. I have exported to an Excel spreadsheet, and then DURING the passage, entered a new line under each routing point what my actual conditions were and actual boat speed and ground track were. Again, trying to get the polar tweaked for real conditions on MY boat vs a theoretical polar by some boat designer under ideal condition. (yes, I know about the Polar plugin, have recently used to record, but not taken the time to analyze results yet--problem is our NMEA speed is not accurate, and when sailing in currents, neither is the GPS for this purpose).

Sherry

[seandepagnier]

Originally Posted by SoggyPaws

YES, absolutely. From the currently supplied reports, I have a hard time "validating" the reasonableness of the routing. A tabular form that lets you run down the route points, providing wind speed and angle, and boat speed and angle, makes it very easy to see if your Polar is set up properly (and whether the routing algorithm is working as expected).

What you described is different from the screenshot rick posted on github issue #80

The screenshot shoes the various results broken by route which is exactly what is in the weather routing window, except rows and columns are swapped.

What you describe is a grid, which is only for a single route, and the rows being for each route point. This is correct?

I know I can get that info from the graph, but it is much more difficult.

It might also help in figuring out what actually stopped the routing process. I found I had had inadvertently set the "Max Latitude" setting in a route, and the routing couldn't reach the point I was trying to reach. The error was "Polar Failed", but it should have been something like "Constrained by Latitude". I did eventually figure it out, but looking at the Polar would not have fixed the problem!!

Again, it can't know that the latitude is the reason it failed unless, maybe that the destination is greather than the max latitude (I will add a warning for that) There are a multitude of reasons that it can fail or not, and constraining the latitude may yield a different route that succeeds, or it may fail, but changing other constraints may allow it to succeed again.

I use the tabular form also for validating in another way.. I have exported to an Excel spreadsheet, and then DURING the passage, entered a new line under each routing

point what my actual conditions were and actual boat speed and ground track were. Again, trying to get the polar tweaked for real conditions on MY boat vs a theoretical polar by some boat designer under ideal condition. (yes, I know about the Polar plugin, have recently used to record, but not taken the time to analyze results yet--problem is our NMEA speed is not accurate, and when sailing in currents, neither is the GPS for this purpose).

I think you need to log a lot of data to make a polar and average it down. It is not really possible to create an accurate polar that can be reused again anywhere because seastate changes everything.

[SoggyPaws]

Originally Posted by rgleason

Here is yet another Windows Release, the result of both Stelian and Sean's collaborative efforts.

Weather_Routing_1.11
https://github.com/rgleason/weather_...eases/tag/1.11

Thank you for continuing to build a Windows version for beta testing. It makes it possible for me to test with the latest and greatest version!

A couple of comments, after uninstalling my previous version, cleaning my data folders thoroughly, and installing the new version.

The exe is still named 1.10? (Just making sure I really got the right one...).
(weather_routing_pi-1.10-Climatology-Setp-win32.exe)

All routings only use one Boat file... Boat-Climatology. Might be more interesting to provide at least one routing with a different Boat file. (Maybe this was set up to just test the Climatology Only routings, but did you mean to include it in the "release" that way)?

Also, in the default data that the default routings point to, the TWS-6-20 pol file is not really 6-20 knots?? (this has been that way for several of your releases). (see screenshot) This would be really confusing to a Noob trying to figure out the Boat vs Polar file issue.

twa/tws;1;2;3;4;5
52;4.3;5.1;5.7;5.9;5.9
60;4.6;5.5;6;6.2;6.2
75;4.9;5.8;6.2;6.4;6.4
90;5.1;6;6.3;6.5;6.5
110;5;5.9;6.4;6.7;6.7
120;4.7;5.8;6.3;6.6;6.6
135;4.2;5.2;6;6.4;6.4
150;3.6;4.5;5.4;6;6

IMHO, there should be at least one "Boat" example installed with as realistic an example as possible, including changing sail configurations, in several polar files. You are close there with the 3 polar files, except there is no overlap in wind speeds (I think from Sean's comments that it is possible and intended to be able to set up polar files where the wind parameters overlap slightly, rather than 0-6 and 6-20, more like 0-8 and 6-25 and 20-60...) I would also like an example that includes a realistic "Crossover Percentage". Does Sean have a full set of polars and a boat file that is fully featured??

Maybe put it in as "Sean's Boat" or "Advanced Example"

Originally Posted by rgleason

Max Diverted Course 90 deg [120 deg?]
Max True Wind 100 knts [25 knts?]
Max Apparent Wind 100 knts [30 knts?]
Max Swell 20 meters [5 meters?]

Be careful setting the constraints too low in the default, at least until we can figure out how to tell the user that it is a constraint, and which constraint, that stopped the routing calculation. Specifically the Max and Max Swell I think are a little low for the default setting, as long as the "failed to route" message only says "Polar fail" or "failed" with no hint as to why. I just ran the Provincetown to Lizard Point sample routing, with those settings, and got "Polar: Failed: No Data" for a December 15 start date. Changing only the Max True Wind from 25 to 30, the routing completed successfully.

One of the goals you are aiming for, IMHO, is that first timers, who just install the plugin and download any random grib file for their area and put in any 2 waypoints in their area, get a successful routing on their first try, or a clear reason why the routing failed.

Finally, I mentioned earlier something about the display settings not being remembered. To be more specific on this--the column widths of the columns in the "Configurations" portion of the routing results screen, in Windows, do not get remembered from one startup to the next of OpenCPN. They DO get remembered if I just close the window, and then re-open Weather Routing again. But if I shut down OpenCPN and start it back up, and open Weather Routing, all the columns are set back to their original near-zero widths. Is this how it is supposed to work?



It does remember the column settings (which columns to be visible). And it does remember the window width, and the location of the split between the Positions panel and the Configurations panel, just not the individual column width in the Configurations panel. It's a minor thing, but I thought I would clarify my earlier mention of it. And make sure it is not something that works OK in Linux and not in Windows.

And one more bug report, I think. In trying to understand the details of the chosen routing, the Weather Route Plot window is unreadable on my Windows computer. I have played with resizing the window and moving the sliders back and forth, and nothing will get the numbers on the left from overlaying the graph. It is a little better when I expand the window down, I just realized, but my screenshot is from the default window size after a new install of Weather Routing.



Also, I am confused at the numbers for the Apparent Wind Angle plot. I would expect the numbers to be ranging between +180 to 0 to -180 (or something like that), but they range from +5976 to -8433. I can't figure out how to interpret that.

Again, thanks again for all your amazing work. (all of you).

Finally getting routings to work and making slow progress in understanding how it all works and why it could be so complicated to get one routing for one boat to work. rgleason's continuing work on documenting all the configuration and result parameters are (and which ones can get you into trouble with an incomplete routing), and a starter dataset as an example, will be useful!

Sherry

[SoggyPaws]

Originally Posted by boat_alexandra

What you describe is a grid, which is only for a single route, and the rows being for each route point. This is correct?

Yes, actually. There are at least two possible grids... one being one route being calculated over a number of time periods and/or other constraints, to determine best time to leave. Lets call that one "Departure Planning".

Second being a grid that is only for a single route, and the rows being for each route point.

Both would be useful. What I was describing was the second one.

Truthfully, I have never tried to do a "Departure Planning" routing scenario with OpenCPN. (have only just gotten it working enough with these latest tweaks to even bother messing with it). I am still learning.

Update: I just tried the "Batch" mode for departure planning on the Provincetown - Lizard Pt route that installs as part of the default data. Amazing! 4 of the 5 days failed due to the Max True Wind > 25 problem, and one completed OK, meaning the max wind is probably better on that departure date. Funny though, using Climatology data... Anyway, one new feature I learned today. No wonder it is so hard to get started with just a simple boat with one polar and one route.

So, you export the route from the plugin, then save it from the route manager?

No. Sorry I wasn't clear. I have done this previously with Maxsea 10.3, and with Fastseas.com. Each of these routing programs had the capability of viewing a table/grid listing each route point and the position/true and apparent wind/current info/and selected course/speed for that routing point...only for the final (best) route, not for all the possible alternatives.

Have not with OpenCPN Weather Routing.

[NahanniV]

@rgleason

One of the supplied polars seems to have a problem:

[seandepagnier]

Originally Posted by SoggyPaws

:Tha
IMHO, there should be at least one "Boat" example installed with as realistic an example as possible, including changing sail configurations, in several polar files. You are close there

Ok, but this might also be confusing to have as the default simple case users are first confronted with.

with the 3 polar files, except there is no overlap in wind speeds (I think from Sean's comments that it is possible and intended to be able to set up polar files where the wind parameters overlap slightly, rather than 0-6 and 6-20, more like 0-8 and 6-25 and 20-60...) I would also like an example that includes a realistic "Crossover Percentage". Does Sean have a full set of polars and a boat file that is fully featured??

In the "alexandra" directory in the polars directory shipped with the plugin, you should be able to load all the polars into one boat.

Maybe put it in as "Sean's Boat" or "Advanced Example"


Be careful setting the constraints too low in the default, at least until we can figure out how

Again.. it's not something I can "figure out" unless I can read the mind of the user. Think about it. The constraints are intrinsically intertwined. If you can show me with a case-by-case basis (For example, setting start/end locations outside of the max latitude mentioned above) I can try to improve it, but otherwise I'm not sure what I can do.

Further.. the "polar failed" is nearly always reported. This is because by default the degree step is from 0, where very few polars define a valid speed at 0 angle. For this reason, the error message is always reported to the user. If you change the angle steps, to lie within the definition of the polar used, then "polar failed" will likely only occur if the wind speed exceeds what the polar defines, and this error will not be reported.

to tell the user that it is a constraint, and which constraint, that stopped the routing calculation. Specifically the Max and Max Swell I think are a little low for the

The max swell should default to 20 meters which is virtually never triggered...

default setting, as long as the "failed to route" message only says "Polar fail" or "failed" with no hint as to why. I just ran the Provincetown to Lizard Point sample routing, with those settings, and got "Polar: Failed: No Data" for a December 15 start date. Changing only the Max True Wind from 25 to 30, the routing completed successfully.

Sure. And if you used a different grib, with a max wind of 25, it would also work, and might give a different slower route (avoiding higher winds) than if it is allowed to sail in 30 knots.

Now, it could report to the user that the max wind was triggered somewhere on the map (maybe only one place that ends up being irrelevant anyway because it is never anywhere near a reasonable route) if the routing failed. Should it do that?

Changing even only one other parameter, possibly time step for example, or diverted course, could again allow the routing to succeed, rather than changing the max wind.

One of the goals you are aiming for, IMHO, is that first timers, who just install the plugin and download any random grib file for their area and put in any 2 waypoints in their area, get a successful routing on their first try, or a clear reason why the routing failed.

I agree. To do this, it needs relaxed constraints, and a polar that defines boat speeds all the way to 60 knots.

Finally, I mentioned earlier something about the display settings not being remembered. To be more specific on this--the column widths of the columns in the "Configurations" portion of the routing results screen, in Windows, do not get remembered from one startup to the next of OpenCPN. They DO get remembered if I just close the window, and then re-open Weather Routing again. But if I shut down OpenCPN and start it back up, and open Weather Routing, all the columns are set back to their original near-zero widths. Is this how it is supposed to work?

Yes. This is exactly how I have written it. Would you like the column widths to be remembered? Do you manually size them? They are meant to be automatically sized when data is populated into them.

And one more bug report, I think. In trying to understand the details of the chosen routing, the Weather Route Plot window is unreadable on my Windows computer. I have played with resizing the window and moving the sliders back and forth, and nothing will get the numbers on the left from overlaying the graph. It is a little better when I expand the window down, I just realized, but my screenshot is from the default window size after a new install of Weather Routing.



Also, I am confused at the numbers for the Apparent Wind Angle plot. I would expect the numbers to be ranging between +180 to 0 to -180 (or something like that), but they range from +5976 to -8433. I can't figure out how to interpret that.

Wow, I couldn't either. The plot isn't great, and maybe should be replaced with something else, or I should work on it more...

Again, thanks again for all your amazing work. (all of you).

Finally getting routings to work and making slow progress in understanding how it all works and why it could be so complicated to get one routing for one boat to work. rgleason's continuing work on documenting all the configuration and result parameters are (and which ones can get you into trouble with an incomplete routing), and a starter dataset as an example, will be useful!

Sherry

Originally Posted by SoggyPaws

Yes, actually. There are at least two possible grids... one being one route being calculated over a number of time periods and/or other constraints, to determine best time to leave. Lets call that one "Departure Planning".

Yes, this is the main weather routing window.

Second being a grid that is only for a single route, and the rows being for each route point.

This doesn't exist, but can you export the route once it is computed and open the track in the route manager and see the properties? Maybe it doesn't include all the parameters you wish to see? I think I can add this, as an additional dialog... It would be similar to the plot window, but instead of a graph, it would have a table.

Both would be useful. What I was describing was the second one.

Trurthfully, I have never tried to do a "Departure Planning" routing scenario with OpenCPN. (have only just gotten it working enough with these latest tweaks to even bother messing with it). I am still learning.

Update: I just tried the "Batch" mode for departure planning on the Provincetown - Lizard Pt route that installs as part of the default data. Amazing! 4 of the 5 days failed due to the Max True Wind > 25 problem, and one completed OK, meaning the max wind is probably better on that departure date. Funny though, using Climatology data... Anyway, one new feature I learned today. No wonder it is so hard to get started with just a simple boat with one polar and one route.

Also, you can select all the routes at the same time and change all of their max winds at the same time (multiple selection editing configurations fully supported for all constraints) I hope this makes it easier.

No. Sorry I wasn't clear. I have done this previously with Maxsea 10.3, and with Fastseas.com. Each of these routing programs had the capability of viewing a table/grid listing each route point and the position/true and apparent wind/current info/and selected course/speed for that routing point...only for the final (best) route, not for all the possible alternatives.

Have not with OpenCPN Weather Routing.