search

equinor / fmu-dataio

FMU data standard and data export with rich metadata in the FMU context
https://fmu-dataio.readthedocs.io/en/latest/
Apache License 2.0
10 stars 15 forks source link

Abstract away common `fmu-dataio` use cases #668

Closed anders-kiaer closed 1 month ago

anders-kiaer commented 4 months ago

Based on discussion between fmu-dataio + webviz May 29th:

Today: Currently the examples for different data types in fmu-dataio consist of different semi-long scripts (e.g. https://fmu-dataio.readthedocs.io/en/latest/examples.html#exporting-fault-polygons). Maintaining these centrally will not be possible if modelling projects copy them from the docs.

First improvement step: To both make it easier for users, and to improve/maintain common data exports centrally, these should be abstracted away to ~one-liner which can be called in RMS, and expose as few arguments as possible to the user (we can later add new arguments with default values if needed).

Second improvement step: Experiment with simple RMS plugins to give a simple GUI to user in order to fill in values for the high-level functions from step one. This would require distribution system internally for RMS plugins.

perolavsvendsen commented 4 months ago

Yes, this is good. It increases robustness and sustainability (centrally vs locally maintained), it reduces the amount of locally implemented code and it will dramatically improve/ease updates and changes, I assume.

I think this could/should be seen in relations to the discussions around more "product" thinking around FMU results. If/when possible, wrappers should be defined around which data is exported, not which data type is exported. I.e. a wrapper for "inplace volumes", not for "tables" and then provide:

  1. Definitions in the data model
  2. Export function/wrapper
  3. Documentation
  4. Endpoint for consumption (from Sumo)

It's fine to focus on the fmu-dataio stuff first, but important to keep in mind that "getting the data into Sumo" is not an end state.

mferrera commented 1 month ago

I am closing this issue because we have taken it as our current dataio milestone and adopted a description here https://github.com/equinor/atlas/issues/42

Input there is very welcome, but this will be a difficult issue to know when to close here otherwise