Incorrect routes through the Northwest Pacific

[pjwilliams11]

This is an awesome and powerful package! One thing I've found: Routes through the Northwest Pacific are erroneous. It seems the base graph lacks sufficient points in the Northwest pacific to properly calculate a minimal distance route, leading to some weird routing. See the image below for a route between Shanghai and Vancouver - this should go north of Japan and generally follow a great circle route across the rest of the Pacific.

[genthalili]

Hi @pjwilliams11, You're right, looks like the network is not dense enough to propose the shortest path as expected:

When I find some free time, I'll look closely to this case and see how to fix it as there must be connection secured between west and east properly with the "segment" file.

Anyways, meanwhile feel free to contribute/send pull request.

[pjwilliams11]

Can you provide a bit broader of a screenshot? I'm surprised that the upper routes aren't being followed for this case, as it looks like the shortest route is probably available by going north of Japan.

Maybe there is a calculation error for shortest route. It seems like the alteration in route (from tracking horizontally to going north) in the original screenshot happens right at the 180th meridian.

[pjwilliams11]

Ah, just looked at the geojson. Yes, it looks like there are no segment connections further north.

[vinisalazar]

Hi @genthalili, I am experiencing a similar problem. I have these two points:

p0, p1 = array([-154.9234, 71.9052], dtype=object), array([-72.46, 72.23], dtype=object)

I am setting restrictions=None to allow routes through the Northwestern passages:

dist_ = sr.searoute(p0, p1, restrictions=None)

This is my resulting route:

plot_world_map_with_lines([dist_["geometry"]["coordinates"],])

I would like instead to have a shorter distance through the Northwestern passages. This superestimation of the distance seems to happen with many of my data points around the poles. Is there a way to maybe obtain a denser network?

Thank you for your assistance and for providing Searoute!

[pjwilliams11]

@genthalili would the update be to segment.geojson or marnet_searoute.geojson?

[genthalili]

Just released version 1.3.1 which should cover both issues. Try it and let me know.

[genthalili]

closing this as no activity, feel free to re-open this issue if needed