Skip to content

Z. General Tips

rtodling edited this page Sep 25, 2021 · 24 revisions

General Tips

  1. Failed compilation in CRTM: temporary remedy:

./parallel_build.csh -tmpdir /tmp

  1. Getting GMAOpy capability:

Setting up env:
setenv MODULEPATH /home/dao_ops/gmao_packages/modules:$MODULEPATH
module load gpy/sles12-v1.0.0
setenv SEMPERPY_CONFIG /home/dao_ops/gmao_packages/o2h/o2h/config

  1. Viewing scorecard: replace control and experiment names as well as applicable dates in line below, and paste it in web browser command line

https://dphttpdev01.nccs.nasa.gov/data-services/scorecard/v1/?exp=x0044.21z&cntrl=x0043.21z&bdate=20191122&edate=20191219

  1. GAAS analysis efficient run. When a particular QOS is used to run the jobs in the DAS this should serve as a warning that the job associated with GAAS must adjusted by hand to accommodate the particular QOS used by the user. Edit file:

EXPID/run/gaas/ana_aod.j.tmpl

and adjust the SLURM/PBS options as needed.

  1. Getting monthly plots to run on priority queue: if you have access to a priority queue and want the monthly plots to run off that queue, you need to edit the file

EXPID/run/monthly_plots/plot/gcm_plot.tmpl

and add the SLURM QOS and other parameters related to your priority queue access.

  1. Producing energy fields from error fields

dyndiff.x -g5 -txt mynorm.txt -ene_scale -rc initadj.rc fcst.nc4 verif.nc4

initadj.rc is typically found in the fcst directory of x-like experiments.

The above will produce a text file in the format of Jnormf.txt (produced in the DAS when running fsens); for example:

Date: 20210105 Time: 000000 Norm type: twe
sum, u, v, t, ps, q
4.987192259405 1.832882427525 1.829440192886 0.635617107003 0.015451936209 0.673800595781

The execution of dyndiff (w/ the opts above) will also produce an nc4 file named enepert.nc4 (this name is wired – sorry) which holds the energy partition in different terms: kinetic, moist-static, etc …

If, for example, you open the file in grads, choose (say) txe and do

define aa = sum(twe,z=1,z=72)
d asum(aa,g)

you will get the first number in the text file (in the case above, 4.9871…); in my case the energy norm is twe; you can get txe (w/o moist component) from the text file by subtracting total from last term in the numbers line; in grads you can do the calculation above using the txe field instead … an so, you can reproduce in grads all the terms in the text file.

Notice also that in the typical way of running this – meaning something equivalent to what we do in fsens – the input files fcst.nc4 and verif.nc4 in the command line above, correspond to prog.eta and asm.eta files.

Futhermore, notice also that dyndiff expects the two input files to be at the same resolution; prog.eta and asm.eta are at the same resolution for the x-exps.

Also know that, if you are going to compare what you get from the command above with what you see come out from the x-exps (Jnorm) you need to remember that because the analysis runs at a reduced resolution, the files prog.eta and asm.eta are converted to a consistent resolution w/ that of GSI (in x-exps: 576x361). To do the proper comparison you’d need to run dyn2dyn in both prog.eta and asm.eta (as done online), e.g.,

dyn2dyn.x -g5 -res d -o dresfile.nc4 input.nc4

The change in resolution also implies a change in the value of total energy.

  1. Overriding default Background Error Covariance

See: https://github.com/GEOS-ESM/GEOSadas/wiki/G.-Updating-Climatological-Background-Error-Statistics

  1. Posting experiments
  • Log into train (or polar)
  • Create a tunnel(*) to CVS
  • Go to the model web server location:
  • If train, web server location is: /gmao/intranet/research/modeling/agcm/geos5
  • If polar, web server location is: /www/html/intranet/research/modeling/agcm/geos5
  • Enter the command: ./makexp
  • and answer the questions related to your experiment

Back on discover, go to the directory containing your plots.
Issue the command from your build:
…src/GMAO_Shared/GEOS_Util/plots/moveplot -expid EXPID -year Yyyyy
where EXPID and Yyyyy are the ID and year of your experiment

(*) For how to open a tunnel look here: https://geos5.org/wiki/index.php?title=Setting_up_to_create_an_experiment_page_on_GMAO_Intranet

  1. GMAO Python Tutorial

git clone https://git.mysmce.com/astg/training/py_materials.git