Summary of key analysis tasks

[pwolfram]

Key tasks to be completed include:

• Add namelist and streams file interfaces (Adds namelist and streams file interface #27)
• Ingest namelist and streams file data, e.g., don't hard code paths to model output but instead get path from streams file (Supports ingesting arbitrary MPAS output files (in general, input info from namelist and streams files) #20 and Read paths from streams and config from namelist #38)
• Support time-series analysis over a subset of the output data (Support time-series analysis over a subset of the output data #45)
• Generalize use of variable names to support multiple versions of ACME (e.g., account for changing of variable names under different versions) (Supports ingesting arbitrary MPAS output files (in general, input info from namelist and streams files) #20)
• Update submodule in PreAndPostProcessingScripts following support for reading namelist and streams files (Update submodule in PreAndPostProcessingScripts following support for reading namelist and streams files #28)
• mpas_xarray needs to support different approaches to specification of time output for AMs (Calculation of datetime to be unified in the future  #33, Read paths from streams and config from namelist #38, others?)
• Add a check on whether the needed AM for the analysis was turned on. For example, we will always need timeSeriesStats, but of course we'll want specific am's for metrics such as the OHC, MOC, etc. (Clean return for specific analysis if a specific AM turned off in model #58)
• Deal with problem of too many open files for xarray (Xarray crashes when opening too many files are opened #49)
• See if we can generate a html interface for better sharing/visualization of MPAS standalone results
• Make sure we support changes in MPAS-SI timeSeriesStats before they are merged in ACME
• Add MHT script
• Add MOC script (for offline calculation only for the moment)
• Make sure analysis scripts do not rely on mesh info present in input files (Make sure analysis scripts do not rely on mesh info present in input files #30)

Lower priority for now:

• Generalize interpolation module (to include ncremap and possibly another method)
• Add compute-climatologies module
• Generalize modelvsobs for ocean and sea-ice (Unification of ocean / seaice scripts #31)
• Do not re-compute climatologies or do not interpolate if these actions have already been done (files exist)
• Perhaps at some point we should save time series to file (especially at high-res and when using many output files)
• Add capability to compare time series with results from a different simulation
• Improvement of documentation (Analysis documentation required (with examples) #3, Provide user-friendly templates for design documents #9, Provide user-friendly examples of analysis tools #11)
• General clean up (Robust exception handling #18, python 3 not supported #43, Installable via a single command #19, Make sure analysis scripts do not rely on mesh info present in input files #30)
• Add changes when short-term archiving will be adopted in ACME

[pwolfram]

@xylar and @milenaveneziani, we need to depreciate https://github.com/pwolfram/mpas_xarray because we really should have that in this analysis repo. Before this is done, I think we need to have the capability to install MPAS-Analysis as a package, e.g., via pip (i.e., #19). Do you all concur? If so, it can be added to the list above.

[xylar]

@pwolfram, I don't have a strong opinion either way about whether MPAS-Analysis needs to be a package. Our ACME workflow doesn't need this capability but maybe you have workflows that expect mpas_xarray to be available via pip.

For my own development, I have found it to be a bit of a pain to have POPSICLES set up as a package. Each time I modify the code, I have to re-install the package (and I sometimes have to uninstall it first with an old setup to make sure I don't have old files lying around). A setup more like we currently have in ACME or similar to CIME would better facilitate development. So I am opposed to requiring MPAS-Analysis to be installed as a package but I don't have a problem with supporting the option.

[pwolfram]

@xylar, we previously decided on requiring MPAS-Analysis to be installed as a package so at this point we are committed to doing this eventually. There are downsides as you detail. If you have any additional ideas on how to best mitigate the downsides that you list that would be really helpful. However, the upside is that we will be better able to interface with a broader community by supporting this option long term.

[milenaveneziani]

@pwolfram: is this a list of high-priority tasks for MPAS-Analysis, in order to make the package compatible with ACME alpha8 and 9?

[pwolfram]

@milenaveneziani, I believe so. We may be able to put off #33 depending upon alpha9. Does this sound reasonable to you? Please edit the list if I missed something or there is an extraneous item

[milenaveneziani]

@pwolfram: I have added all the things I can think of for now. This list does not include all the possible other scripts we can think of writing (hovmoeller diagrams, transects, other time series, etc), but just the important elements that we need to have in place to satisfy most of our requirements.
One comment about the 'Update submodule for ACME' item: this will be an on-going task, so perhaps we do not need to have it as an issue or on the list?

[xylar]

@milenaveneziani and @pwolfram, I think #28 (and check-list item "Update submodule in PreAndPostProcessingScripts") should be made renamed to be specific to the list of tasks that needs to be achieved before that update. I'm going to go ahead and rename them so they are more meaningful.

[milenaveneziani]

@pwolfram, @xylar: I have updated the list and added more items to it, also following our email exchange. I realize this has become more of a laundry list at this point, but I think it is nice to put things down in one single place.
For the next month or so, I think we should prioritize the first 6 unchecked boxes: do you agree?

[pwolfram]

@milenaveneziani, just to be clear, can we please clarify the high-priority needs right now? Can you please copy/paste them in a reply just so that this is crystal clear? The reason I ask is because we've recently checked some of the boxes above and I just want to make sure we are all on the same page and are completing the work we need for the analysis.

[pwolfram]

Note, I checked off the box mpas_xarray needs to support different approaches to specification of time output for AMs above because I would say that we have this needed support now and the cited issues / PRs are closed / merged.

[milenaveneziani]

We now have a Confluence page and various Jira tasks that include the items in this list. @pwolfram: I would probably close this issue.