MSNoise 2.0 - Checklist

[ThomasLecocq]

Getting ready for MSNoise 2.0

For those who have not followed the last changes: for MSNoise 2.0 I plan on:

• Changing the default outputs of all steps after CC (stack, mwcs, etc) to HDF5, aka "a single dataframe h5 file per filter/movstack/comp/pair/" and no longer "single MSEED per .....", reducing dramatically the number of outputs
• Adding PSD computations, PPSD/spectrograms and PSD-SR capabilities
• Adding simple Ballmer-style locations
• Adding support for location-codes

With some still "under heavy discussion with myself":

• (maybe) PSD to channel-ratio (see Pedersen et al 2020))
• (maybe) Adding wavelet-based dv/v (Mordret)
• (maybe) Beamforming from cross-correlations (Ruigrok)

Remaining Work:

General

• Agree on the naming of the "workflows" and "command groups": CCF, DVV, PSD, QC, etc.
• [ ]

Documentation

Installation

• Rewrite Installation instructions to make them simpler (conda install)
• Rewrite Databases part, to stress all is doable on sqlite, but MySQL/MariaDB or PostgreSQL are recommended
• Write PostgresSQL documentation
• Test !new! MySQL installation with the latest version, I think the changes imposed to make PostgreSQL work will solve the legendary GROUP-BY issue

Graphics / visuals

• Create new visuals for the workflow(s) - including the worfklow branching model etc
• Update all plots in documentation with enough data (Workshop example from Undervolc? or full undervolc)
• Create new screenshots from the web Admin

API-Recipes

• Add interaction examples
• Add API calls examples
• Check all methods in API for signature & docstrings

Misc

• Fix/add citations in all pages requiring it (add references to key papers if needed)
• Edit readme.md
• Important documentation needed for the location-code support!
• Decide where to host msnoise-tomo documentation (but would be good to have it here, no? at least linked)

Review

• will be useful for the MSNoise2.0 paper anyway: would be cool to have 1-pager intro on "what can be done with MSNoise", citing the right refs from the >100 articles using msnoise

Processing

HDF5-related

• Check all steps are producing identical results as single-MSEED files
• Make worfklow work with the "I process the whole archive at once" scheme
• Make workflow work with the "I'm re-processing just a few jobs" scheme (requires read-update/append-write HDF5)
• API: add all the read/save HDF dataframes from all steps to the API

Methods

• Compare my PCC implementation with @RDePlaen & Schimmel
• Add basic Ballmer-style location method (+config bits)
• Add the "msnoise compute_dvv" step (HDF5)
• Replace dtt_maxdt by dtt_maxdtt =dt/t to allow small changes at large time lag

Command Line

• Add the possibility to add a Filter or a Station from the command line (not using db execute)

Plotting

• Single pair plotting issue plot dvv fails when only one pair is being used #118
• rewrite plot_dvv to make use of the results of "compute_dvv" output
• More verbosity when plot fail Adding some additional information why some plots were not produced #111
• Add stretching plot from Jannik Kühn
• Add "plot all" mode (looping over all pairs/filters/stacks/components) - + define data structure
• Add "apparent velocity" dashed lines on the ccftime/interferogram plots
• Add new "ccf_spectrogram" plot (image version of the spectime plot)
• Clean station map plot (cartopy + leaflet only)
• Add "group by stations" plot (aka aurelogram) Add "group by stations" plot (aka aurelogram) #214
• Add station groups for dvv plots (cf PdF "volcan", "gpentes", "cratère")

Testing

• write a second set of tests to check contents
• (maybe) re-re-retry database tests on TravisCI
• (possible?) to switch to another CI infrastructure

Rules of engagement:

• Read carefully
• If an item has no reference next to it, open a new issue with the appropriate name, and mention it here, I'll update the link next to the item, it will make things easier to follow: Discussion (and agreement if needed), Pull-request, merge
• I'll mark items done when they are solved/merged.

[ThomasLecocq]

reminder: current master documentation is built manually and hosted at http://msnoise.org/doc/master/