ENH: Add simplified export for depth prediction surfaces

[tnatt]

Resolves #1045

PR to add a new standard_result: structure_depth_surface, and a simple export function export_structure_depth_surfaces to be used to export depth horizons from within a horizon folder in RMS.

Addresses https://github.com/equinor/atlas/issues/42

[perolavsvendsen]

Do we want to state this explicitly? Or, rather, do we want to be able to change this? Currently, it sort of looks like an invitation to a discussion around which folder and filenames that should be used - but ideally nobody is going to relate to that.

[perolavsvendsen]

export_structure_depth_predictions?
export_structure_depth_prediction_surfaces?

Seen in relations to:
export_structure_thickness_predictions?

Or just drop the whole "prediction" from the name (implying that this is just "given")?
export_structure_depth_surfaces?
export_structure_thickness_surfaces?

[perolavsvendsen]

(export_)structure_depth_surfaces
(export_)structure_thickness_surfaces
(export_)structure_depth_fault_lines
(export_)structure_depth_fault_surfaces
(export_)structure_pinchout_lines
(export_)structure_xx
(export_)structure_yy

(export_)grid_depth_surfaces
(export_)grid_thickness_surfaces
(export_)grid_parameters
(export_)grid_qc

...
...
...

[perolavsvendsen]

Ref discussions offline. I think good requirements for this could include:

• Increase robustness + be implementable: Minimum footprint in FMU workflows
• Increase robustness: Changeable without disruption for end users
• Increase data quality: Provide consistency across all FMU workflows, and enable other applications to confidently consume The Prediction Of The Structural Model for any FMU run.
• Follow established conventions on naming/establishes naming conventions that scale.
• Intuitive for end users
• Documented
• ...
• ...

Probably smart to see this in relations to other similar export functions in terms of naming and the implicit conventions that are starting to form. E.g. what would one-liners look like for this list of identified items? Do they make sense when they are all combined in the same documentation? (Would it be worth trying to do that as an acid test for the naming?)

[mferrera]

Nice job 👍 have you checked a documentation build visually?

[mferrera]

This will always be relative to msl? (Just a domain curiosity from my end)

[perolavsvendsen]

Conventionally yes, but possibly no, not always 🤔 Perhaps that assumption should be noted in the docs as well?

[jcrivenaes]

Nice. I tried this in RMS now and works like charm 👍

[perolavsvendsen]

And we are (sufficiently) happy with the function name? I.e. will it be understood that you can only include this in your workflow exactly 1 time? And that there will be only 1 set of depth surfaces produced from it? That is - you must use this for the surfaces that are predictions but you cannot use it for all those surfaces that are just for QC etc.

(Since we have now taken away the term "prediction" from the method name, ref discussion yesterday)

[tnatt]

And we are (sufficiently) happy with the function name? I.e. will it be understood that you can only include this in your workflow exactly 1 time? And that there will be only 1 set of depth surfaces produced from it? That is - you must use this for the surfaces that are predictions but you cannot use it for all those surfaces that are just for QC etc.

(Since we have now taken away the term "prediction" from the method name, ref discussion yesterday)

I think so! But I can stress test it with the RM3 group today, we have a technical meeting at 14 👍

[tnatt]

And we are (sufficiently) happy with the function name? I.e. will it be understood that you can only include this in your workflow exactly 1 time? And that there will be only 1 set of depth surfaces produced from it? That is - you must use this for the surfaces that are predictions but you cannot use it for all those surfaces that are just for QC etc.
(Since we have now taken away the term "prediction" from the method name, ref discussion yesterday)

I think so! But I can stress test it with the RM3 group today, we have a technical meeting at 14 👍

Thumbs up from the group of modellers 👍 I showed them the documentation and they all understood this to be their final surfaces.
Also it seems removing the word prediction was a good choice, the consensus was that the term was much more confusing and all were in favor of the given name 🙂