[BUG] Thickness result seems to have changed between 1.3.5 and 1.3.6

[markjessell]

Describe your issue

Running a notebook under colab using map2loop 1.3.6@6IAS (branch to fix error in main branch on line 451 of m2l_map_checker.py):

https://colab.research.google.com/github/Loop3D/6IAS/blob/main/map2loop/1a_Building_a_model_from_local_sources.ipynb

gives me only 16 formation thickness estimates, and each one is positionally duplicated with a different thickness estimate.

running v 1.3.5 not in colab gives me 286 points also with duplicates, and I assume the issue with the number of points is the version, not the fact it is colab?

So there are two issues, the duplicates, which is a long-standing weirdness as far as I can tell, and the different number of measuremetns for the same input data between 1.3.5 & 1.3.6

Minimal reproducing code example

see colab link above

Error message

No response

[RoyThomsonMonash]

Hi Mark,

In the changes made in 1.3.6 the way contacts are stored changed from separating multilinestrings to linestrings to keeping them as mutlilinestrings as long as possible. This is due to using geopandas intersections to identify contacts rather than manually looping through each stratigraphic layer's geometry and comparing them to other layers geometry.

This storage change causes the save_contact_vectors function to traverse the MultiLineString segment rather than the standard code segment. There does not appear to be anything different between the code other than separating out the lines from the multilines but the raw_contacts.csv temporary file is much shorter in the 1.3.6 version than the 1.3.5 version. I will investigate further.