Find Intersections of a self-intersecting polyline feature

[KylePurdon]

Hello,

Dataset:
- Shapefile with multiple polyline features representing flight paths. Features intersect each other and some intersect themselves as well.

Task:
- Find and create a point at every line cross point (intersection)

Current Solution:
- Use the "Intersect (Analysis)" tool to find the intersections and save a point at each intersection.

Problem:
- The intersections of a feature with another feature are found but the intersections of a feature with itself are not found.
- See the attached image for an example. The selected item is a single polyline feature in the shapefile.
- The tool finds the points shown, but not the self-intersections.

Ideas? Many Thanks!

[KylePurdon]

A new development.

I can split the line at it's vertices. It was converted from points so there are tons of vertices. Now there is no self-intersection and the Intersect Tool finds all the crossover points I want, but....

It also finds the intersections that happen at the vertices where each line was split. This results in tons of points, not just the crossovers.

Though this might help somebody spark an idea.

Also, here is a response from a professor of mine.

"The Intersect tool won�??t find intersection for a polyline itself for sure. If you convert the shapefiles to the Coverage format, you can try the Clean and Build tool under Coverage Tools\Data Management\Topology.
The other possibility is to create a geodatabase with the topology of not allowing line intersect with itself and then try to validate the topology." -Xingong Li (University of Kansas Geography)

I will try the above solutions but please post your ideas!

Thanks!

[KylePurdon]

Also here is a link to the original point data:

https://data.cresis.ku.edu/data/rds/2011_Greenland_P3/csv_good/Browse_2011_Greenland_P3.csv

1) Add the XY Data in WGS84
2) Export to SHP
3) Point to Line by FRAME and sort with TIME
4) Find line crossovers!

Thanks!

[SteveLynch]

Have you thought of using the RepairGeometry tool ?

[EricRice]

Hi Kyle,

Take your lines, and input them into the Feature to Lines tool. Take that output and intersect it with itself. The result is much better than splitting the lines by vertices. Feature to Line will essentially planarize the lines and remove overlapping lines. Because of this you will still have some undesirable points of intersection as noted in the screenshot (the points left of the grid pattern). It should be a lot easier to clean these up compared to your previous result.

Regards,
Eric

[KylePurdon]

Eric,

That result does look promising, however I need a solution with no manual cleanup. This process will be integrated into a model that uses the "crossover" points to create a buffer at each point and compare the value of "Ice Thickness" of all the original data points within the buffer. This is a measure of accuracy in the data processing that goes into these points creation.

I'll keep working on it and if anyone else has an idea keep it coming!

[EricRice]

If you fix your .csv file then you will not encounter this at all after you planarize the lines.

[KylePurdon]

If you fix your .csv file

What do you mean "fix" the CSV file? What is wrong with it?

Note: here is a more complete (Not decimated) file
https://data.cresis.ku.edu/data/rds/2011_Greenland_P3/csv_good/2011_Greenland_P3.csv

[EricRice]

The issue is with the construction of the lines or data - not sure. Maybe it would be best if we spoke? Some of the lines that look like single lines are indeed 2 or more lines that are only millimeters apart. Intersect and Feature to Line have an XY tolerance settings. I may need to play some more with those settings. Are you supposed to have nearly coincident line work only mm's apart? I inspected my intersection points (created from Feature to Line result) for one area and found that the points do not align with the original line work from Points to Line result. Its almost 4cm shifted indicating the tolerance settings are key.

Note, Intersect is giving the expected output given what I've thrown at it so far.

Regards,
Eric

[KylePurdon]

Eric,

That is the data I believe. The "FRAME" variable represents segments of data collected via airborne platform and they actually have some overlap along the lines. This may be the reason for duplicate lines very close to each other. There is also some lines that are actually flown right next to each other, but those are handled fine by the process you have described.

If you have some other ideas or need more information feel free to send me an email at kylepurdon@gmail.com and we can exchange contact information other than email if needed.

Thank you very much for the help.