Alexey-Ben meeting 2019-01-29

[ashiklom]

Updates from last week

• FoRTE
  • UMBS PFTs in PEcAn -- done
  • TRY data for UMBS in PEcAn -- done(-ish).
    • I added data for the traits that PEcAn already knows about. That doesn't cover all of ED's parameters, but it does cover many (should be all of the parameters from Mike's 2014 paper). I also only added traits that could be added directly (i.e. measurements correspond directly to ED parameters, like SLA, leaf C:N). There are other measurements that could be used to inform traits (e.g. A-Ci curves for photosynthetic parameters; paired DBH and height/biomass measurements for allometries) that I haven't considered yet.
  • Start ED ensembles -- done.
    • Started a 50-member ED ensemble on Thursday. Still running as of AM Monday. A bit slower than I expected.
    • Runs are parallelized over 8 cores, which is the max for a single OpenStack Cloud Node. However, it should possible to run connected Docker instances across multiple machines via Docker Swarm. I would have to learn how to do this, though. For now, I'll refine other aspects of the workflow with the setup we already have, but this is something to keep in mind for the "production" runs, especially if performance becomes a bottleneck.
  • Initial conditions -- Didn't get to.
    • Non-trivial, so I decided I would start some runs first to make sure everything else in the workflow works before diving into this.
  • Paper reading and outline -- Didn't spend enough time on this. Higher priority this week.
• Hector
  • Biome debugging -- done. PR was merged.
  • Reading and paper outline -- Didn't spend enough time on this. Higher priority this week.
  • Parameter/initial conditions database -- Didn't get to.
  • Played around with documentation website on Friday. Not strictly required for work, but hopefully useful (productive procrastination?).

Goals for this week

• FoRTE
  • Reading and paper outline
  • Process ED2 ensemble output
  • Look into soil initial conditions
• Hector
  • Reading and paper outline
  • Exploration of simple permafrost model

Other topics

• Teaching opportunity
• Rebecca teaching materials

[bpbond]

productive procrastination

👍

Let's also discuss Mike's EFI meeting in May.

[ashiklom]

Another thing to discuss: I dumped a bunch of information into my FoRTE paper outline (link).

I started very-rough-drafting an introduction, but there's a good chance much (or all) of that will be scrapped as this paper evolves.

More importantly, I went through a recent ED2 configuration file, pulled out a brief description of all the key "switches", and took a first pass at grouping them according to priority. Even after removing ones I was fairly certain weren't relevant, there are still ~20 switches, many of them with more than 2 options, which is probably past the limits of what we can do in a factorial experiment (2^20 = 1,048,576!).

From an informatics standpoint, it should be pretty straightforward to figure out which of these even work, and if some of them don't, that's an easy way to narrow the scope. But, I think it's worth making a decision based on biological/ecological hypotheses, and it's worth taking a bit of time to discuss these switches in that context. This would also be good to discuss with Chris and Jeff on the next FoRTE call.