search

rphlo / karttapullautin

A fast and accurate map generator from classified LiDAR data.
GNU General Public License v3.0
32 stars 9 forks source link

Enabling larger depression contours #94

Closed stefankinell closed 6 months ago

stefankinell commented 6 months ago

Pullauta has a limit of how long contours that are being checked for being a contour. This has ment that on one hand - large depressions has been missed, and on the other hand - only a few contours are marked as depressions in hilly negative terrain.

A pull request https://github.com/rphlo/rusty-pullauta/pull/93 has been created to solve this by introducing a parameter for setting limit of when to detect depressions.

Issue discussed more in https://github.com/rphlo/rusty-pullauta/issues/83

stefankinell commented 6 months ago

I have now ran this in two test areas. I thought it could be interesting to see the output/difference for others as well.

  1. Gotska Sandön. Executed with contour_interval=2.5 m and depression_length=10000 https://kartor.gokartor.se/dev/#11/58.3512/19.3128

One can shift between old execution with 181 limit and new with 10000 by shifting layers in top right corner. Version 3 is the new, Version 2 is what I have used for the public map so far.

  1. Säter - an area with some big depressions. Executed with contour_interval=5 m and depression_length=10000 https://kartor.gokartor.se/dev/#13/60.3345/15.8439

This is one of the areas I showed in #83

Conclusion: I am much more happy with this execution. I can see larger depressions, get more depression lines and formlines for depressions are super!

Things I have noticed: Sometimes when there is a border of laz-files - one side of the laz-border is set as contour - the other as depression. One example is here: https://kartor.gokartor.se/dev/#16/58.3685/19.2759

This is another example: https://kartor.gokartor.se/dev/#16/58.3685/19.2759

I have no clue if something can be done about that since I assume it will treat the parts of the contour as any contour. And that depressions really could best be found when there is a circle?

Another example is where a large creek/river canyon is not detected as depressions. By looking at the map https://kartor.gokartor.se/dev/#14/60.3429/15.7796 surely I would understand that the creek runs in the lower end of a canyon, rather than on top of a hill. But from a technical perspective I do not understand why this is not captured as depressions? @Rphlo & @Jagge - do you know why this is happening? I am not sure if it is a big deal, or an issue at all. Just curious about why this happens.

I also tried to merge all laz-files for this area to see if it made any difference - but the canyon still was returned as a hill/normal contours.

Overall I think this is a huge improvement in quality for me. A lot of contours that I have assumed were hills, I now understand are depressions. Just shift between the map images on Gotska Sandön to get a feeling for how much improvement there is.

rphlo commented 6 months ago

The thing with canyon/creeks is that the contour doesn't actually close so it is not considered as depression.

Jagge commented 6 months ago

It is depresson only if it is a ring. canyon is not a depression. Sometimes part of the depression ring is outside the tile. When batch processing there is some buffer taken forn neighboting tiles. It gets detected as depression if it is inside that buffer. If it is not then it's not a ring and gets drawn as normal contour. for example https://kartor.gokartor.se/dev/#15/58.3690/19.2764 east part is less than 100m from the edge and inside buffer, so west tile gets depression, but west part is furher away (140m) and parstly outside the buffer so east tile gets a normal contour.