What to do when the `offset` is too large for the last lithology layer

[johnjarmitage]

Hi Matt et al. I can propose a solution to the issue I encountered with the sediment log I was testing.

issue Error when offsetis to large causing the function tops_from_loglike in utils.py to look for values in the array a that are out of range.

fix I propose to create a while loop that will chop of the last lithology unit if it is too thin for the chosen offset (from line 478 in utils.py)

try:
        values = a[tops + offset]
    except:
        lasttop = len(a) - tops[-1]
        i = 1
        while lasttop < offset:
            lasttop += len(a) - tops[-i]
            i += 1
        tops = tops[:-i]
        values = a[tops + offset]  # If the offset is too big relative to the last layer remove last top

I can make a pull request for this fix if you want, but I am not sure how generic this solution is.

[kwinkunks]

Thanks for this John. I had a fix in the develop branch, but I have not tested it rigorously. It was a different approach to you. If you get a chance to try it out, I'd be interested in knowing if it fixes the problem you were experiencing.

You'd install with python -m pip install --upgrade https://github.com/agile-geoscience/striplog/archive/develop.zip

[johnjarmitage]

Hi Matt, it is possible your fix does not work as I get the out of bounds error message. Is there a way to check that I am using the develop branch after running the pip command? (Sorry that I have a slow return around time on this issue)

[mtb-za]

Hi John, you can check the version with the following in a python interpreter:

import striplog
striplog.__version__

for me that is 0.8.8, which is the most recent development version. I am not sure exactly when the fix Matt mentions above went in, since the last commit is dated to January 12th.

[johnjarmitage]

Yep that is the version number I have. I am testing the modifications on the develop branch, so before replying that it does not work I would like to be sure. I had a look at the repo and it says that for the develop branch: "This branch is even with master." Does this perhaps mean that Matt's modifications have not been pushed to the repo?

[kwinkunks]

Oops, my bad, I think I was in master, sorry about that. Anyway, develop is now synced and both are pushed.

[johnjarmitage]

I have not had success:

---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
<ipython-input-6-771446b87056> in <module>
      1 from striplog import Striplog
      2 
----> 3 strip = Striplog.from_image(infile,
      4                             start=0, stop=2631,
      5                             col_offset=5,     # How far across the image to extract pixels.

~/miniconda3/envs/sedlogs/lib/python3.9/site-packages/striplog/striplog.py in from_image(cls, filename, start, stop, legend, source, col_offset, row_offset, tolerance, background)
    850 
    851         # Get the pixels and colour values at 'tops' (i.e. changes).
--> 852         tops, hexes = utils.tops_from_loglike(loglike, offset=row_offset)
    853 
    854         # If there are consecutive tops, we assume it's because there is a

~/miniconda3/envs/sedlogs/lib/python3.9/site-packages/striplog/utils.py in tops_from_loglike(a, offset, null)
    476     tops = np.append(0, tops)
    477 
--> 478     values = a[tops + offset]
    479 
    480     if contains_nans and transformed:

IndexError: index 12413 is out of bounds for axis 0 with size 12413

I attach an image of the log I am trying to digitalise so you can test it at your end to be sure it is not my fault for not managing to update to the most recent version of striplog.

[johnjarmitage]

sorry, the attached image comes out huge...

[mtb-za]

Hi @johnjarmitage I was just taking a look to try and fix this. The code in your error message does not look like the current code that I am seeing in the repo. The current version looks like this for the lines giving the error you posted above:

offs = tops + offset
values = a[offs[offs < a.size]]

This snippet should prevent the IndexError that you are getting. I am not sure which version you are using, but it might not be happening with the newer version. Do you have some shareable data that I can test this with to try and replicate it? I am not sure if the snippet that I am seeing has had a release, but it looks like the change was here: https://github.com/agile-geoscience/striplog/commit/9ef74bf5622397c76750a8081831f6a7029a75e8

[johnjarmitage]

Hi @mtb-za I ran python -m pip install --upgrade https://github.com/agile-geoscience/striplog/archive/develop.zip and then relaunched my notebook. There is now no error message, so it is possible that this bug has been fixed during the SWUNG meeting. You can close the issue I think.

[mtb-za]

I am glad that it seems to have been resolved. Let us know if there is anything more that you come across.