Find 2D intersection in lat lon between curve (rivers) and satellite traces (2D polygon) on a map

Hello, this is an expanded question to the original 1D problem in https://es.mathworks.com/matlabcentral/answers/1740820-find-curve-intersection-points-in-lat-lon-between-rivers-and-satellite-traces-on-a-map

I have a set of (lon,lat) points defining the course of a river. Then I also have a set of (lon, lat) points defining the traces of satellite measurements on-ground. The trace is not 1D, there is a 'left' point and a 'right' point, denoting what woud be in reality a continuous 2D band. In fact both the river and the trace are continuous. I want to find the intersections between the river (a curve) and also the 2D trace. I managed in 1D with the answer of the link above, but not evident in 2D at all. Thank you

PS: I include now included exemplary mat data, with the [lon lat] values of the river, and the [lon lat] values of the satellite trace, defined by the left and right points that define the band/belt extremes. On of the problems I see as well is that close to the poles the bands mix because of the projection effect...

[KSSV]

Attach your data.

[Albert Zurita]

Done! Thanks

[Jeffrey Clark]

It seems that all riv points are within trace_1 and trace_2 is that what you expected?

[Albert Zurita]

yes, this is intentional. I want to detect those river points that fall within trace_1 and trace_2 as you mention

[Jeffrey Clark]

@Albert Zurita, yes the closer to the poles the more problematic lat-lon. Before realizing you created the case where all riv points were in the trace_1/2 I included a step in my proposed solution to convert lat-lon to a cartesian coordinate system in three dimensions, in this case Earth-Centered Rotated (right) as compared to the lon-lat data (I augmented and reversed columns of the lon-lat of all three files with altitude of zero to do the transform from Lat-Lon-Alt to ECR in the figure on the right):

[Albert Zurita]

Hi, what is ECR, like ECI? And what coordinates do you plot there, because I guess x,y,z will matter. Thanks

Answer 1

[Bjorn Gustavsson]

If you have a single-pass of the satellite you could check what "river-points" are inside the ground-track with the inpolygon function. Check the help and documentation for illustrating examples. My approach would be to split up the satellite-track into single-pass-segments and search for points inside the ground-track pass-by-pass and collect what I need from each.

HTH

[Albert Zurita]

That is not a bad option indeed! Splitting by orbits or semi-orbits seems to work, but need to be clever when splitting the orbits, especially due to the circularity of the longitude...

[Bjorn Gustavsson]

If you know where your rivers (assuming you want to generalize this to arbitrary rivers) are located it should help you with the wrap-around/branch-cut type problems. You might even get use for the unwrap function, or possibly convert the ground-tracks and river-positions to some ECEF-type coordinates.

[Albert Zurita]

But I think it makes more sense to organize by orbit because the wrap-around would make sense to fasten the code, which at the moment it's not a problem.

On the coordinates the problem with ECEF is that they are 3D coordinates, and it seems a lot easier to work on projected coordinates, with of course the caveats we could encounter...

[Bjorn Gustavsson]

Oh, yes, by orbit first. The idea to work with unwrap and ecef would be if you have rivers close to your +180/-180 (or 0-360) cuts. For ecef you could possibly work with the x-y coordinates if you are safely on the northern or southern hemisphere - I guess there the cut-problem would be the equator. But now I'm mainly throwing completely unbaked ideas at you for the possibility that I by stupid luck manage to randomly combine a couple of phrases in the right order...