extract data from mat file, reproject the facies and plot the virtual cores by xyl96 · Pull Request #4 · MindTheGap-ERC/TrackingHabitat

xyl96 · 2025-12-16T13:04:43Z

Extracting the data from MAT file from STACKER, reproject the facies and plotting.

xyl96 · 2025-12-16T15:23:44Z

Here is some discrepancies:
The figure I have obtained is listed below.

While the three cores (mid three) directly from STACKER is listed below here:

I do not why this is happenning.

xyl96 · 2025-12-17T08:53:34Z

Here is some discrepancies: The figure I have obtained is listed below. While the three cores (mid three) directly from STACKER is listed below here: I do not why this is happenning.

I did something erong in the code! I will check it today!

xyl96 · 2025-12-17T12:38:11Z

The plotting issue has been solved

EmiliaJarochowska · 2025-12-17T19:34:51Z

As a general feedback on this PR: it is not clear what the problem is. From the issue description #3 it seems to me that plotting the columns is the problem, but from the PR title it seems that extracting data from mat files is the problem. Which one is it?

If it is the plotting that is needed, is there anything about these cores that requires writing completely new code for plotting columns? There are a bunch of packages for doing that, including our home one https://mindthegap-erc.github.io/stratcols/ by @NiklasHohmann

So my general feedback:

Specify what is the feature you are developing (extracting, plotting, comparing?) and make sure the PR matches the problem declared
The thicknesses of the cores don't match the figure from Stacker (e.g. there is no core with a thickness of 0.6 m)
Store data in data not in src
Would using stratcols resolve the problem (but see 1 - what is the problem exactly?)

EmiliaJarochowska · 2025-12-17T19:43:16Z

+    for idx in eachindex(core_layers_norm)[2:end]
+        top = core_layers_norm[idx]
+        bottom = core_layers_norm[idx-1]
+        if core_facies[idx] != 100


why assign properties to hiatus (color and transparency) if it's never plotted?

I added it specifically to check whether the codes are following what I want - it happens when I was writing the code.

xyl96 · 2025-12-18T10:13:30Z

As a general feedback on this PR: it is not clear what the problem is. From the issue description #3 it seems to me that plotting the columns is the problem, but from the PR title it seems that extracting data from mat files is the problem. Which one is it?

If it is the plotting that is needed, is there anything about these cores that requires writing completely new code for plotting columns? There are a bunch of packages for doing that, including our home one https://mindthegap-erc.github.io/stratcols/ by @NiklasHohmann

So my general feedback:

Specify what is the feature you are developing (extracting, plotting, comparing?) and make sure the PR matches the problem declared

The thicknesses of the cores don't match the figure from Stacker (e.g. there is no core with a thickness of 0.6 m)

Store data in data not in src

Would using stratcols resolve the problem (but see 1 - what is the problem exactly?)

reply:

The initiative and aims have been updated. the aim is 3-fold, including extracting, reprojection of facies and plotting. Comparison is on the way not done yet. Should be in another issue. So it is not pure 'plotting'.
That's the thing I am thinking. I checked the original STACKER code yesterday and I think I followed the way he did.
no data was stored in src - they were stored in results. Except the results from STACKER. What should I do with the MAT file?

works, but not elegant

EmiliaJarochowska

Not related to this PR specifically but more generally: please add README (don't hesitate to ask for help)

I think the main problem is storing all the mat files in the repo. Binary files cannot be tracked with git and they take a lot of space. It would probably make more sense to put them in Zenodo (separate repo) and download using datahugger? But you can ask @NiklasHohmann for a second opinion

In rastering.jl I get:

ERROR: KeyError: key "trackID" not found

Consider not tracking files that are generated in the code, just to keep the repo lean

NiklasHohmann · 2026-02-26T11:08:04Z

Not related to this PR specifically but more generally: please add README (don't hesitate to ask for help)

I think the main problem is storing all the mat files in the repo. Binary files cannot be tracked with git and they take a lot of space. It would probably make more sense to put them in Zenodo (separate repo) and download using datahugger? But you can ask @NiklasHohmann for a second opinion

In rastering.jl I get:
ERROR: KeyError: key "trackID" not found
Consider not tracking files that are generated in the code, just to keep the repo lean

Generally agree, keeping binary data on Zenodo and then autodownload it would be the smoothest option.
Another option would be to automate the whole stacker workflow and programmatically generate the .mat file instead of archiving it
No strong feeling either way

xyl96 · 2026-03-03T13:18:26Z

Not related to this PR specifically but more generally: please add README (don't hesitate to ask for help)
I think the main problem is storing all the mat files in the repo. Binary files cannot be tracked with git and they take a lot of space. It would probably make more sense to put them in Zenodo (separate repo) and download using datahugger? But you can ask @NiklasHohmann for a second opinion
In rastering.jl I get:
ERROR: KeyError: key "trackID" not found
Consider not tracking files that are generated in the code, just to keep the repo lean
Generally agree, keeping binary data on Zenodo and then autodownload it would be the smoothest option. Another option would be to automate the whole stacker workflow and programmatically generate the .mat file instead of archiving it No strong feeling either way

Yes, the results mat file is provided in the zenodo.

EmiliaJarochowska · 2026-03-03T13:20:25Z

Yes, the results mat file is provided in the zenodo.

I think that wasn't the point of this comment ;-) The point was not to put them on GH but download from Zenodo

xyl96 · 2026-03-03T15:55:24Z

Yes, the results mat file is provided in the zenodo.

I think that wasn't the point of this comment ;-) The point was not to put them on GH but download from Zenodo

ok, will add 'download' function in the plottingcores.jl

delete useless files

EmiliaJarochowska · 2026-03-07T16:42:46Z

I need the current version of the data for #5 so I'll merge off this branch

EmiliaJarochowska · 2026-03-07T17:09:13Z

In order to do this I need some info from @xyl96
My understanding is that FaciesMosaic.mat contains also the Z (time) dimension, so all model output in terms of facies.
So

TrackingHabitat/src/PlottingCores/plottingcores.jl

Lines 14 to 15 in e73fac8

    
           facies = stratadata["facies"]  
        
           layers = stratadata["layers"]

creates arrays for facies codes and for thicknesses?
However, layers contains negative values:

julia> layers
100×100×1000 Array{Float64, 3}:
[:, :, 1] =
 -2.1681   -2.07705  -2.12417  -2.24804  …  -6.96664  -6.74394  -7.00432  -6.97156
 -1.82665  -2.04633  -2.02343  -2.1511      -6.7736   -6.80748  -6.90505  -6.96027
 -1.99421  -2.0791   -2.06234  -2.04844     -6.87824  -6.80586  -6.85596  -6.78426
 -2.25707  -1.90442  -2.14208  -2.29268     -6.84399  -6.86519  -6.7282   -7.11105
  ⋮                                      ⋱                                
 -2.1038   -1.98354  -2.10764  -2.32388     -6.81002  -6.76294  -7.03121  -6.84482
 -2.01357  -2.11726  -2.15233  -2.17706     -6.71198  -6.76788  -7.05157  -6.96443
 -2.01388  -2.10531  -2.10149  -2.16529     -6.68735  -6.95486  -6.91334  -6.95715

etc
so my questions:

how are these values measured? With reference to what?
why are there only 1000 values if there were 5000 time steps?

What I need for the power spectral densities are:

x(t) : a vector showing facies class at the sampling point as a function of time step
x(z) : a vector of the stratigraphic column, i.e. x resampled at uniform depth intervals
Can you add extracting them from Stacker, @xyl96 ? I thought I could do this but I stumbled on the above questions.

Actually what I am trying to do would probably benefit from some depth-to-time transfers of admtools by @NiklasHohmann but I guess it's easier to stick to Julia at this point

EmiliaJarochowska · 2026-03-07T17:10:34Z

PS what I am trying to do is to compute $|H(k)|^2$ — the empirical power transfer function from the time series
to the depth series.

Because the depth series has a different length than the time series, their PSDs are computed on their own grids and then interpolated to a common spatial frequency axis before taking the ratio.

xyl96 · 2026-05-19T19:55:02Z

In order to do this I need some info from @xyl96 My understanding is that FaciesMosaic.mat contains also the Z (time) dimension, so all model output in terms of facies. So

TrackingHabitat/src/PlottingCores/plottingcores.jl

Lines 14 to 15 in e73fac8

facies = stratadata["facies"]

layers = stratadata["layers"]

creates arrays for facies codes and for thicknesses?
However, layers contains negative values:
julia> layers
100×100×1000 Array{Float64, 3}:
[:, :, 1] =
 -2.1681   -2.07705  -2.12417  -2.24804  …  -6.96664  -6.74394  -7.00432  -6.97156
 -1.82665  -2.04633  -2.02343  -2.1511      -6.7736   -6.80748  -6.90505  -6.96027
 -1.99421  -2.0791   -2.06234  -2.04844     -6.87824  -6.80586  -6.85596  -6.78426
 -2.25707  -1.90442  -2.14208  -2.29268     -6.84399  -6.86519  -6.7282   -7.11105
  ⋮                                      ⋱                                
 -2.1038   -1.98354  -2.10764  -2.32388     -6.81002  -6.76294  -7.03121  -6.84482
 -2.01357  -2.11726  -2.15233  -2.17706     -6.71198  -6.76788  -7.05157  -6.96443
 -2.01388  -2.10531  -2.10149  -2.16529     -6.68735  -6.95486  -6.91334  -6.95715
etc so my questions:

how are these values measured? With reference to what?

why are there only 1000 values if there were 5000 time steps?

What I need for the power spectral densities are:

x(t) : a vector showing facies class at the sampling point as a function of time step x(z) : a vector of the stratigraphic column, i.e. x resampled at uniform depth intervals Can you add extracting them from Stacker, @xyl96 ? I thought I could do this but I stumbled on the above questions.

Actually what I am trying to do would probably benefit from some depth-to-time transfers of admtools by @NiklasHohmann but I guess it's easier to stick to Julia at this point

The layers means the 'water depth' so that's why they are negative. I think now it's 100 layers.
Now the code is able to download mat files from zenodo.

extract data from mat file

16bb29d

xyl96 requested a review from EmiliaJarochowska December 16, 2025 13:04

fix the plotting issue

70f8120

EmiliaJarochowska requested changes Dec 17, 2025

View reviewed changes

xyl96 changed the title ~~extract data from mat file~~ extract data from mat file, reproject the facies and plot the virtual cores Dec 18, 2025

Liu and others added 8 commits December 19, 2025 11:59

draw real cores

b42ded1

add proportion comparison

2a50393

change core ploting

3557b3d

change tracking param

b4b8e66

add spatial entropy

55b3faf

works, but not elegant

updated working code add plotting of Abacos cores

70d27a5

statistic plot

6398d63

correct calculate_all_cores_proportions() so it writes a file

51cfd5c

EmiliaJarochowska reviewed Feb 26, 2026

View reviewed changes

Comment thread data/Abacos.xlsx

EmiliaJarochowska added 3 commits February 26, 2026 10:46

fix calculate_distances - it wouldn't work with these column names

ec061c4

remove redundant file writing

dd072a4

remove another unused object

d4a34d4

EmiliaJarochowska requested changes Feb 26, 2026

View reviewed changes

organise files

e73fac8

delete useless files

Liu added 2 commits March 24, 2026 17:16

change according to comments add automatic downloading

709a3b6

updated version for paper

8cbe1cb

fix typo, update zenodo download link

2369761

xyl96 requested a review from EmiliaJarochowska May 20, 2026 11:27

Conversation

xyl96 commented Dec 16, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

xyl96 commented Dec 16, 2025

Uh oh!

xyl96 commented Dec 17, 2025

Uh oh!

xyl96 commented Dec 17, 2025

Uh oh!

EmiliaJarochowska commented Dec 17, 2025

Uh oh!

Uh oh!

EmiliaJarochowska Dec 17, 2025

Choose a reason for hiding this comment

Uh oh!

xyl96 Dec 18, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

xyl96 commented Dec 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

EmiliaJarochowska left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

NiklasHohmann commented Feb 26, 2026

Uh oh!

xyl96 commented Mar 3, 2026

Uh oh!

EmiliaJarochowska commented Mar 3, 2026

Uh oh!

xyl96 commented Mar 3, 2026

Uh oh!

EmiliaJarochowska commented Mar 7, 2026

Uh oh!

EmiliaJarochowska commented Mar 7, 2026

Uh oh!

EmiliaJarochowska commented Mar 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

xyl96 commented May 19, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

xyl96 commented Dec 16, 2025 •

edited

Loading

xyl96 commented Dec 18, 2025 •

edited

Loading

EmiliaJarochowska commented Mar 7, 2026 •

edited

Loading