Loop GIS workflow planning

[lachlangrose]

Hi everyone,

Let's use this thread to plan the development of the Loop QGIS tools. Start by describing a workflow that you would like to be able to use. Include the input data, augmentation steps and user knowledge you want to include and what the end results are for the model.

@markjessell @rabii-chaarani @LaurentAilleres @ayLa-e @gpirot @vogarko @noellehmcheng @ShebMichel

Cheers,
Lachie

[LaurentAilleres]

Input: One stratigraphic column, a series of not-on-contact structural measurements assigned to a given horizon, one DTM
Knowledge: thicknesses of every horizon

Expected augmentation: extract basal contacts on DTM

[markjessell]

doesn't this require a polygon layer of stratigraphic units as input too?

[LaurentAilleres]

Yup and nup... Yes, it needs at least one point to tell us roughly where we are in the stratigraphy, then we should be able to interpolate with inequality constraints... Uncertainty is LARGE!

[markjessell]

so we have two use cases then, one with and one without polygons?

[markjessell]

Input: One geology polygon layer, one fault polyline layer
Expected Augmentation: upscaled map with topology between polygons and ebtween faults and polygons retains (Ranee's code to be rewritten)

[markjessell]

Input: One Struictural observaiton points layer
Expected Augmentation: Upscaled distribution of measurements (Ranee's code to be rewritten)

[LaurentAilleres]

yeah but the upscaling needs to be GLOBAL as per the thing I am typing in now... hahaha!!! The "same" method needs to be applied to structural data and basal contact sampling and fault trace sampling and etc...

[markjessell]

well there are topologial considerations with faults/strat polygons that don't exist in point obs so not completely global...

[LaurentAilleres]

The usual world:

• I have a geological map
• I have structural information related to strike/dip of formations
• I have a stratigraphic column
• I have faults

The ideal modeller world - am guessing this is what the current m2l process tries to achieve (tries because we are still missing bits of algorithm [upscaling] and some data/information:

• I have a geological map UPSCALED to the right scale (sampling of the contact with a GLOBAL method to decide the length of segments...)
• I have UPSCALED structural information using the same GLOBAL method (see MWJ comment re:Ranee's code)
• I have a stratigraphic column WITH Thicknesses
• I have faults, with properly UPSCALE traces, known offsets
  THe above is the holy grail...

[gpirot]

and I have drillholes
and I have a prior to put on the uncertainty of my measurements

[markjessell]

Input: One stratigraphic column, polygon geology and polyline faults
Knowledge: Validity of proposed stratigraphy

[LaurentAilleres]

yup with an extra toggle allowing to ALSO check against ASUD for Australian maps.... and any served stratigraphic database for other countries, set of countries...
Returns a second level information - not combined

[rabii-chaarani]

The expected output here is the "Validity of proposed stratigraphy"?, right?

[markjessell]

Well if we treat the stratigraphy as a graph, we can get which bits of the stratigraphy are different, not just that it is different.

[LaurentAilleres]

The Bermagui case:
I have knowledge of a deformation history and only structural measurements of foliations for up to 4 folding events...
In fact, this does not require m2l... oops... but can the LS plug in do that? I know LS can.

[lachlangrose]

not currently

[markjessell]

Input: Faults
Expected Augmentation: Fault network topology

[markjessell]

Input: One stratigraphic column, polygon geology and polyline faults
Knowledge: Relative ages of faults wrt to stratigraphy

[markjessell]

Input: Basal contacts or faults, DTM
Expected knowledge: estimate of planar dips (with large uncertainty!)

[markjessell]

Input: Structural observation points layer, geology polygon layer, fault layer
Expected knowledge: Division of stratigraphic units into groups or supergroups that may be interpolated as a single entity based on structural coherence

[markjessell]

Input: Structural observation points layer, geology polygon layer, fault layer
Expected Knowledge: How many deformation events has this rock seen and what were the structural frames associated with each one (you might as well dream big!) Probably made up of several sub-tasks

[LaurentAilleres]

I like that but yes, required proper domaining - manual or automatic? Automatic would make it more objective...

[markjessell]

Agreed, what I did in the original map2loop for domaining was only ever a very partial solution...

[markjessell]

Input: Structural observation points layer, basal contacts, fault layer
Expected Knowledge: Local and global estimates of formation thickness, local using interpolated orientatiosn withing a single structural domain

[lachlangrose]

Should we make the interpolation of orientations a separate step? And just make local/global thickness estimates from orientation data?

[markjessell]

sounds like a good idea

[rabii-chaarani]

I will start compiling these workflows in a diagram to have a clear idea of the changes that will need to be made to map2loop.
Sorry for missing the meeting...

[markjessell]

Input: spatial interpolations of orientation data
Expected Knowledge: fold axial-traces

[LaurentAilleres]

huh? why interpolate... or is that a data poor environment? I would prefer proper domaining and extraction of fold axial surface orientation and fold axis... + intersection lineations between estimated axial traces and measurements of the "previous" foliation (S2 to S1, S1 to S0 etc...).
Or is this a different functionality from what you had in mind?

[markjessell]

the % of the world with good structural data is very limited, but yes of course if we have this data then we should include that too.

[rabii-chaarani]

I tried to map the workflows proposed here, their relationships and the development tasks required for the Map2Loop plugin. This is a general map, so we can visualise the relationship between the workflows/tasks and get to the specifics.
Let me know if you have any feedback or any details you want added to the workflow

[LaurentAilleres]

Hi Rabii, I like it but we need also to think how we are going to manage the "related task" and any changes to e.g., the stratigraphy - then as we all know, thicknesses and fault offset estimation become "dirty". Similarly, for the stratigraphy, you have valid-invalid... ok but dont forget uncertain and uncertainty. At the end, i would also like to see what the user will have access to in terms of "functions" and how they are going to interfere with the "augmented data" becoming possibly "dirty". Hope that makes sense. Cheers, Laurent

…

On Mon, 16 Jun 2025 at 15:50, Rabii Chaarani ***@***.***> wrote: I tried to map the workflows proposed here, their relationships and the development tasks required. This is a general map, so we can visualise the relationship between the workflows/tasks and get to the specifics. *Let me know if you have any feedback or any details you want added to the workflow* Map2Loop.-.Initial.Diagrams.jpg (view on web) <https://github.com/user-attachments/assets/d0ed0052-4afb-4046-b748-0c450d22dea7> — Reply to this email directly, view it on GitHub <#28 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ALSB3P2IC56NCGNXQGFWWYT3DZLJLAVCNFSM6AAAAAB6UH2ZHKVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGNBXHEYDKMA> . You are receiving this because you were mentioned.Message ID: ***@***.*** .com>

[markjessell]

Yes I agree with Laurent, and to extend this further we will need to track uncertainty every time we modify the data (measurement uncertainty->subsampling data->calculation of augmented data) so we will need a logging capability I think, to track what parameters were set at each stage from data import onwards, including how we make multiple models using the uncertainty. This is probably a separate conversation but will have to work closely with this workflow. Mark From: Laurent Ailleres ***@***.***> Sent: Monday, 16 June 2025 2:55 PM To: Loop3D/plugin_loopstructural ***@***.***> Cc: Mark Jessell ***@***.***>; Mention ***@***.***> Subject: Re: [Loop3D/plugin_loopstructural] Loop GIS workflow planning (Discussion #28) Hi Rabii, I like it but we need also to think how we are going to manage the "related task" and any changes to e.g., the stratigraphy - then as we all know, thicknesses and fault offset estimation become "dirty". Similarly, for the stratigraphy, you have valid-invalid... ok but dont forget uncertain and uncertainty. At the end, i would also like to see what the user will have access to in terms of "functions" and how they are going to interfere with the "augmented data" becoming possibly "dirty". Hope that makes sense. Cheers, Laurent

On Mon, 16 Jun 2025 at 15:50, Rabii Chaarani ***@***.***<mailto:***@***.***>> wrote: I tried to map the workflows proposed here, their relationships and the development tasks required. This is a general map, so we can visualise the relationship between the workflows/tasks and get to the specifics. *Let me know if you have any feedback or any details you want added to the workflow* Map2Loop.-.Initial.Diagrams.jpg (view on web) <https://github.com/user-attachments/assets/d0ed0052-4afb-4046-b748-0c450d22dea7> - Reply to this email directly, view it on GitHub <#28 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ALSB3P2IC56NCGNXQGFWWYT3DZLJLAVCNFSM6AAAAAB6UH2ZHKVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGNBXHEYDKMA> . You are receiving this because you were mentioned.Message ID: ***@***.***<mailto:***@***.***> .com>

- Reply to this email directly, view it on GitHub<#28 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABUBSYAIG6G7L5WKEDULHE33DZS4FAVCNFSM6AAAAAB6UH2ZHKVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGNBXHE2TMNY>. You are receiving this because you were mentioned.Message ID: ***@***.******@***.***>>

[gpirot]

Here is my simplistic schematic overview, before going into more details

[rabii-chaarani]

Will you be able to write a plugin that can uses LS output to quantify uncertainty?
The idea is to have M2L plugin, LS plugin and and uncertainty plugin that can communicate together and at some point we could merge them all into one plugin

[ShebMichel]

@ rabbi-chaarani, there is already a short sneak peak of uncertainty I build but this was a prototype so far. The reason it is paused, because of a consensus of how track uncertainty, therefore, which parameters to deal with prior the ensemble of model generation. For example, if you use the left branch of the image Guillaume shared, you may have a parameters tab to change before sending the data into M2L, then maybe no tweaking before sending it through L2S. Also, there might be other parameters panels between M2L and L2S to test the uncertainty after the ensemble is generated.