2024 Software Engineering meeting notes - NGEET/fates GitHub Wiki

February 26, 2024

Upcoming meetings and events

FATES software meeting check-in

  • Is there interest in moving this meeting to an every-other week cadence?
    • Consensus: those present like the weekly cadence and the reporting back of PRs and issue status.
    • Greg: What about the notes format? Would people prefer to engage with the notes in another format?
      • Consensus: those present like the transparency of the wiki format and that the notes are co-located on the repo with everything else (issues, discussion, code, etc.)

Machine outages

  • CISL: none planned
  • NERSC: 20 Mar, perlmutter
  • OLCF: none planned

PR board status update

Updates:

Latest issues and discussions of note

  • #1163 and #1165 addressed by PRs #1164 and #1163, respectively

Calibration progress update

  • Jessie: finishing up respiration work; will turn into maniscript in near future.
    • Considering adoption of Adrianna's ML method for next iteration (post SP-mode calibration updates)
    • Ryan: stem respiration for calibration?
      • Jessie: Not currently. Kolby keen to investigate this for next project phase. Discussions about field-to-model integration details on going.

Tutorial update

  • Jennifer: what's the accessibility goal? Will we explain how to setup the container, etc?
    • Jessie: Ultimate goal is to make this available all online with a very simple BCI method.
      • Greg: We also have a tutorial setup guide repo in development to host a readthedocs style walkthrough of setting up on a laptop.
    • Greg: OLMT experience? We might use or compare the single site subsetting function for pre-processing.
      • Jennifer: Yes, two years ago. Current attempts to use were hampered however.
      • Ryan: Running pre-process all at once?
        • Jessie: yes, we could use the sparse grid NCL method.

February 12, 2024

Upcoming meetings and events

Machine outages

  • CISL: none planned
  • NERSC: 21 Feb, perlmutter
  • OLCF: none planned

PR board status update

Latest issues and discussions of note

FATES speed audit update (Ryan)

  • Can we avoid leaf layer loop: Trimming routine forces us to iterate over leaf layer depth
  • Current plan: investigate Ci iteration cap and threshold. Update as necessary along with remembering the previous value.
  • Initial iteration analysis update suggests that we often hit the iteration cap.
    • Initial iteration testing with ST3 mode (audit was using full dynamics however).
  • ED2 experience: What about trying Newton method using the derivitives from reduced analytical form? Review the ED2 approach and share with team.
  • Investigate impact of making photosynthesis loosely coupled?

Calibration progress update

  • Adrianna: CLM surface data does not seem to quite match ILAMB observation. This is causing issues with gridcell to gridcell comparison.
    • Ask Nate Collier for help.
  • Calibration team to review Adrianna's brute-force method and will review selected parameter ranges.

February 5, 2024

Upcoming meetings and events

Machine outages

  • CISL: 5-7 Feb, derecho
  • NERSC: 21 Feb, perlmutter
  • OLCF: 6 Feb, frontier

PR board status update

Latest issues and discussions of note

Calibration progress update


January 29, 2024

Upcoming meetings and events

Machine outages

  • CISL: Mon 5 - Wed 7 Feb, derecho
  • NERSC: none planned
  • OLCF: none planned

PR board status update

Latest issues and discussions of note

  • LU data tool fails on derecho: probably taking too much memory for login node. This will need to be addressed eventually for CLM LU build on the fly.
    • This came up in the context of checking ctsm python environment on derecho, due to a dependency issue for xesmf.
    • Erik/Charlie: test this on the develop queue

Calibration progress update

  • Adrianna: Ran Ball-Berry without temp acclimation. Observational mean is more easily achieved, but emulator does a worse job.
    • Temperture acclimation without very low vcmax
      • Charlie: Claire Zarackas is finding something similar in the absence of hydraulics. So there is a scientific rationale for excluding temp acclimation.
  • Jessie: met with Ryan to talk about carbon budget issue; add recruitment carbon flux. Carbon budget closes now.
    • Not sure if recruitments that die in the same timestep are showing up in the diagnostics

Walk-on topics

  • Rosie: Question: Matvey fates model focus? Take over Marius work? Where does it fall in the priority?
    • Jennifer: arctic-fire stuff coming up prior to arctic-hydro. Maybe hardening?
    • Greg: This is back on API 17, so bringing it up before it gets too stale would be good
    • Charlie/Ryan/Rosie: recent hydro issues do suggest that addressing hardening is a good idea.
    • Matvey: only calibrated to 2 local sites; will it work globally?

January 22, 2024

Upcoming meetings and events

Machine outages

  • CISL: Mon 5 - Wed 7 Feb, derecho
  • NERSC: none planned
  • OLCF: none planned

PR board status update

Building parameter files on the fly

  • Ryan to look at moving this forward this week

Latest issues and discussions of note

  • hydro dry region failure: Marcos has been working with Jenny on this. Recommendation is that she not trace the issue since its out of scope for her work.
    • Ryan proposing to add a fail safe for extreme outlier cases to allow fates run to complete. Still need to make sure this is actually an outlier.

Calibration progress update

  • Calibration branch
    • Jessie still needs this branch until 2-stream gets to E3SM (probably a few weeks at this point). The calibration branch is pretty close to FATES main for her purposes, so no need to merge main into calibration branch.

Walk-on topics

  • Erik asking about status of Marius' frost branch

January 15, 2024

No meeting due to lab holiday


January 8, 2024

Upcoming meetings and events

Machine outages

  • CISL: Mon 5 - Wed 7 Feb, derecho
  • NERSC: Wed 17 Jan, perlmutter
  • OLCF: none planned

PR board status update

Latest issues and discussions of note

Calibration progress update

  • Jessie presented latest results: GPP and biomass bias improvements

January 1, 2024

No meeting due to lab holiday