Playing with the u-al752 Fluxnet suite

Hi Patrick (and others),
I’ve been playing around with the Fluxnet suite and have two of my own suites building up to full 4DEnVar:

  1. u-dg263 is a copy of the Fluxnet suite that I have adapted to show outputs of GPP, LE and SH for two values of g1_stomata_io (2 and 7 – see u-dg263.png attached).
  2. u-dg123 is also a copy of the Fluxnet suite and I have adapted this one to run for a parameter ensemble. This one will ultimately run with the full 4DEnVar method which I will finish once I’m back from leave next week.

I noticed some things/issues that as I played around. I thought you might like to know them as you have been involved with the updating of the u-al752 suite.

  1. If you try to run the suite with a large ‘selection’, you get a ‘submit-failed’ error for some of the runs when you submit to the test partition.
  2. I can’t seem to include the ‘LBA_RJA’ site I any runs – ‘make_plots’ fails complaining about array size (from memory).
  3. fluxnet_evaluation.get_site_list_from_info_inc requires the only_neo argument and the following snippet of code:
    if only_neo:
    list_sites = [s for s in list_sites if ‘neo’ in s.lower()]
  4. when I used the selection [‘AT_Neu’,‘CH_Cha’, ‘CH_Oe2’,‘CN_HaM’, ‘FR_Gri’, ‘RU_Che’] (the C3 grass sites), I had to comment out lines 205-222 in fluxnet_evaluation for the plots to work – something about a scalar a day attribute.

I hope that’s helpful – possibly not, lol.Thanks!

Hi Natalie:
Thanks for the feedback. Your plots look rather interesting. I will try to fix up the trunk version of the suite with your feedback suggestions. Have you tried to run the suite using the short-serial or short-serial-4hr partition instead of the test partition? Does that fix problem #1?

Hi Patrick,
Yes I did and it did fix that problem, thanks!