Skip to content

2024‐03‐20

Jump to bottom
Teagan King edited this page Mar 20, 2024 · 3 revisions

March 20, 2024

Agenda:

  1. General updates
  2. Collaborative work time focused on making ADF functions available to other components

Slides

Notes

  1. General updates
    • Issue with ADF commits made after Feb 27th; PR with fix has been merged in (fix requires updating rerunning manage_externals)
    • API updates to specify component keys in config file; PR in review
    • cupid-run now has a few command line arguments (using click)
    • Single variable time series generation PR is under review. It’d be great if anyone wants to test this out and let us know what works/doesn’t work
      • ML: We may want input/output dir specified in timeseries.
    • CGD exchange on May 9th. Goal: introduce CUPiD for the rest of CGD to test
    • Comments:
      • BM: timeline for using scripts/notebooks ______
  2. Collaborative work time focused on timeseries generation
    • Examples gallery may have some additional functionality
    • Land-ocean mass gain, weird grid coordinates may be useful.
    • Vectors on polar projections (eg, windspeed) can be a particular challenge.
    • GeoCAT Viz may best play the role of supplemental support / place to host code that builds off of matplotlib
    • make_polar_plot:
    • We want to avoid circular tools (eg, don't want to have ADF have to call CUPiD to make their plots). We also want to have common functionality.
    • Who is going to support users with these functions? Issues could be posted somewhere and answered by other people (eg CUPiD folks could help with GeoCAT issues).
    • All of plotting from ADF is matplotlib and cartopy, and their functions mostly get data ready for those plots and customize annotations.
    • Abstraction could be done to ADF functions.
    • GeoCAT doesn't have capacity to host code that has to do with specific datasets or models.
    • Collaboration may include getting good latitude labels regardless of projection.
    • Code overlap with DB's plots-- differences: DB uses pcolormesh to help with arbitrary lat/lon values.
    • Intermediate places for functions to live: eg, examples gallery.
    • CESM-specific does not belong in GeoCAT; however, some of these fucntions may be relevant for model comparisons.
    • We could split pieces such as colorbar & other bits into different functions.
    • We might put vertical interpolation or climatology calculations in GeoCAT Comp.
    • mask_land_or_ocean, spatial_average, wgt_rmse, annual_mean may be useful to put in GeoCAT.
    • Timing: it may be nice to have an example of migrating a function sooner, but maybe not as much of a time crunch to move everything.
      • Currently, another notebook could call ADF functions from ADF directly.
      • Manage_externals is a bit messy; it would be nice to transition to conda install when possible.
    • Can't currently directly call make_polar_plot due to lat/lon grid, but you could probably make it work. This could be a good first step to try to have sea ice notebook call ADF functions.
    • Polar plots would be a good example to have some basic things like imposing a circular domain and boundary -- would Cartopy fix this?
    • Common plots also include simple x/y timeries plots. This should be fairly simple to do and would benefit from generic functions.
    • CUPiD plotting repo could contain some intermediate functions that might eventually be moved elsewhere. If we can standardize across components in CESM, that also helps us make more general functions.
    • A standard style across CUPiD plots would be nice (eg, style_sheets)
    • Long term concern: avoid duplicating code, maybe aim for making a base class and then adjusting code in specific components
    • PLAN FORWARD: Aim to spin up cupid plotting & Dave B. can try to use make_polar_plot() & GeoCAT first pass look through plotting tools.
Clone this wiki locally