ED2 documentation / scripts

[ashiklom]

These will need some combination of documentation and scripts/functions/R package:

• Setup
  • Compilation
  • On MacOS (for local runs)
  • On PIC (for large simulations)
  • Shared input datasets (land-soil mask, etc.)
  • General structure of the code base
  • How to do a test run (to make sure everything is linked, etc.)
• Simple run configuration
  • ED2IN file flags:
  • Model run dates
  • Input and output file locations
  • Output file format and frequency
  • Initialization mode
  • Some relevant model configurations (RTM, finite canopy, etc.)
  • Parameters -- what they are (names, units, etc.) and how to set them (config.xml)
  • Meteorology driver files
• Processing outputs
  • Output variable names
  • File structure
  • Scripts/functions for efficiently reading these files into tidy data structures
• Advanced run configuration
  • Vegetation initial conditions
  • Prescribing an exogenous disturbance
  • Performance considerations
  • PEcAn trait meta-analysis
  • Ensemble simulations
  • Running many runs in parallel on PIC
  • PEcAn-like variance decomposition analysis
• Additional documentation
  • ED2 wiki
  • Relevant literature
  • Layperson's overview of key ED2 features

See #63.

[bpbond]

Is it worth coding these (in some way) by difficulty? I don't know, just as a guide? Maybe an oddball thought.

[ashiklom]

So the major bullets are already basically in order of increasing difficulty. I could think about re-arranging some of the sub-bullets to be in a more logical order.

A related idea that might be even more useful is to sketch out a plan for the modeling part of FoRTE paper 2 (disturbance treatment?) in order of increasing priority (e.g. bare minimum -> recommended -> wishlist/ideal scenario). For instance, if the disturbance treatment is what we're after then something like this:

• Bare minimum
  • One run per treatment with default parameters; no uncertainty analysis
  • Initialize from bare ground starting ~1900 (don't worry about initial condition)
  • Hybrid meteorology (historical reanalysis + more current reanalysis + random meteorology for several years into future)
  • Prescribe disturbance treatments for year 2019
• Additional "features" that can be added, in order of increasing difficulty:
  • Start from observed initial conditions (inventory data) in a more recent year
  • Use "best-guess" parameters from GCB paper (possibly with some additional calibration)
  • Parameter uncertainty analysis
  • Separate initial conditions for each FoRTE plot (i.e. initialize from plot-specific inventories)

[bpbond]

@ashiklom Sketching out a plan for paper 2 in the manner you propose sounds hugely valuable to me.

Adding @kdorheim to the thread and repository.

[ashiklom]

ICYMI, I created a repository that will contain a combination of documentation and helper scripts for FoRTE-related ED2 activities: https://github.com/FoRTExperiment/ed4forte/

So far, I only have documentation on compiling ED2 on MacOS and doing a simple test run. More to come, but I wanted to get this piece out there early because getting ED2 to compile and run at all is a prerequisite to everything else.

Feedback (and PRs) welcome!

[kdorheim]

@ashiklom Thanks!!