Opened 8 weeks ago

Closed 3 weeks ago

#3051 closed help (completed)

Adding output vars in JULES

Reported by: NoelClancy Owned by: pmcguire
Component: JULES Keywords: output variables, NETCDF
Cc: Platform:
UM Version:

Description

Having already ran a JULES suite successfully, I have modified the suite by adding in additional output variables and ran it again successfully.

However, sometimes the suite fails to re-run successfully when many additional output variables are added in. It is not clear to me if it is because too many variables have been added in or if certain variables cannot run on a given suite configuration.

Is there a maximum number of output variables that can be included in the netCDF output files?

The two configurations that I am running are:
l_triffid, l_phenol, l_o3_damage, bgc_model = F, F, T, 1
and
l_triffid, l_phenol, l_o3_damage, bgc_model = T, T, T, 2

How do you know whether or not a given output variable is compatible with the suite configuration?

Change History (29)

comment:1 Changed 8 weeks ago by NoelClancy

Just to add more information to this ticket:

I ran the suites on MONSOON, then transferred the data to JASMIN

I did not use the cylc node in JASMIN nmc@…

I know it is recommended to use
nmc@…
rather than
nmc@…

Would it be better to run the suite in JASMIN rather than MONSOON?

comment:2 Changed 8 weeks ago by pmcguire

  • Status changed from new to accepted

comment:3 Changed 8 weeks ago by pmcguire

  • Keywords output variables, NETCDF added

comment:4 Changed 8 weeks ago by pmcguire

Hi Noel
Are you talking about variants of the suite for comparing JULES model runs to FLUXNET data? (u-al752 is the original suite ID for that). What is the suite number of your new suite? Have you checked in your new suite into MOSRS?
How many variables did you try to add? Which variables did you try to add?

Yes, it is much better to run on jasmin-cylc instead of jasmin-sci*, that's how we're supposed to do it, and you will get a cylc GUI on jasmin-cylc which you won't get on jasmin-sci*. This cylc GUI helps a lot and makes monitoring and debugging much easier.

You can run on MONSOON if you have access. I know more about JASMIN than MONSOON, so I can help more with JASMIN issues than with MONSOON issues.
Patrick

Last edited 8 weeks ago by pmcguire (previous) (diff)

comment:5 Changed 7 weeks ago by NoelClancy

Patrick,

Anytime I run a suite in JASMIN, I do so in jasmin-cylc.
However, I have been running suites in MONSOON lately because Dr. Gerd Folberth (Met Office) set up a suite that is configured for ozone damage, u-bm066

I am doing a series of experiments by copying u-bm066 (checked into MOSRS). However, in the rose-suite.conf section of the suite it's confusing what I need to change ??

Foe example, having checked out a copy of u-bm066 as suite u-bo174, I have modified the suite u-bo174 as follows in the rose-suite.conf

But there must be errors! Can you help me find them?
For simplicity I have not added any further variables I just want to try and run in JASMIN instead of MONSOON. Once the suite is running successfully in JASMIN, I will try to add in variables.

[file:bin/fluxnet_evaluation.py]
source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/fluxnet_evaluation.py@15046

[file:bin/jules.py]
source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/jules.py@15046

[file:bin/make_time_coord.py]
source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/make_time_coord.py@15046

[file:bin/parallelise.py]
source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/parallelise.py@15046

[jinja2:suite.rc]
INCLUDE_SPINUP=True
JULES_FCM='fcm:jules.x_br/dev/karinawilliams/r9227_add_gpp_unstressed_diagnostic'
JULES_REVISION='14197'
LOCATION='CEDA_JASMIN'
NYRSPIN=50
OUTPUT_FOLDER='/work/scratch/nmc/fluxnet/u-bo174/jules_output'
PLOT_FOLDER='/work/scratch/nmc/fluxnet/u-bo174/peg/plots'
PRESCRIBE_DATASETS_CONFIGURATION=0
RUNID='JULES_vn5.4'
SUBSET='peg'
SUITE_DATA='/group_workspaces/jasmin2/jules/pmcguire/fluxnet/kwilliam/suite_data'

comment:6 Changed 7 weeks ago by pmcguire

Hi Noel
What are the error messages when you try to run this suite?
Patrick

comment:7 Changed 7 weeks ago by NoelClancy

job.err

Traceback (most recent call last):

File "/apps/contrib/metomi/cylc-7.8.1/bin/cylc-cat-log", line 439, in <module>

main()

File "/apps/contrib/metomi/cylc-7.8.1/bin/cylc-cat-log", line 435, in main

tmpfile_edit(out, options.geditor)

File "/apps/contrib/metomi/cylc-7.8.1/bin/cylc-cat-log", line 268, in tmpfile_edit

proc = Popen(cmd, stderr=PIPE)

File "/usr/lib64/python2.6/subprocess.py", line 642, in init

errread, errwrite)

File "/usr/lib64/python2.6/subprocess.py", line 1238, in _execute_child

raise child_exception

OSError: [Errno 2] No such file or directory

comment:8 Changed 7 weeks ago by NoelClancy

job.out

Traceback (most recent call last):

File "/apps/contrib/metomi/cylc-7.8.1/bin/cylc-cat-log", line 439, in <module>

main()

File "/apps/contrib/metomi/cylc-7.8.1/bin/cylc-cat-log", line 435, in main

tmpfile_edit(out, options.geditor)

File "/apps/contrib/metomi/cylc-7.8.1/bin/cylc-cat-log", line 268, in tmpfile_edit

proc = Popen(cmd, stderr=PIPE)

File "/usr/lib64/python2.6/subprocess.py", line 642, in init

errread, errwrite)

File "/usr/lib64/python2.6/subprocess.py", line 1238, in _execute_child

raise child_exception

OSError: [Errno 2] No such file or directory

comment:9 Changed 7 weeks ago by NoelClancy

JULES_FCM='fcm:jules.x_br/dev/karinawilliams/r9227_add_gpp_unstressed_diagnostic'

Maybe it should be:

JULES_FCM='fcm:jules.x_br/dev/karinawilliams/r6715_add_gpp_unstressed_diagnostic'

But maybe other things are also wrong

comment:10 Changed 7 weeks ago by pmcguire

Hi Noel
Can you run the diff command to compare the rose-suite.conf file for this suite compared to the same file for the previous u-al752 FLUXNET suite that you had working on JASMIN? What differences does it report?
Patrick

comment:11 Changed 7 weeks ago by NoelClancy

[nmc@jasmin-cylc roses]$ diff /home/users/nmc/roses/u-bo174/rose-suite.conf /home/users/nmc/roses/u-ba137/rose-suite.conf
1,12d0
< [file:bin/fluxnet_evaluation.py]
< source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/fluxnet_evaluation.py@15046
<
< [file:bin/jules.py]
< source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/jules.py@15046
<
< [file:bin/make_time_coord.py]
< source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/make_time_coord.py@15046
<
< [file:bin/parallelise.py]
< source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/parallelise.py@15046
<
16,17c4
< JULES_REVISION='14197'
< LOCATION='CEDA_JASMIN'
—-

JULES_REVISION=

19,20d5
< OUTPUT_FOLDER='/work/scratch/nmc/fluxnet/u-bo174/jules_output'
< PLOT_FOLDER='/work/scratch/nmc/fluxnet/u-bo174/peg/plots'
22c7
< RUNID='JULES_vn5.4'
—-

RUNID='r9227_branch'

23a9,11

OUTPUT_FOLDER='/work/scratch/nmc/fluxnet/run_ba137/jules_output'
PLOT_FOLDER='/work/scratch/nmc/fluxnet/run_ba137/peg/plots'
LOCATION='CEDA_JASMIN'

24a13,21

[file:bin/jules.py]
source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/jules.py@12313

[file:bin/make_time_coord.py]
source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/make_time_coord.py@12313

[file:bin/fluxnet_evaluation.py]
source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/fluxnet_evaluation.py@12313

Last edited 7 weeks ago by pmcguire (previous) (diff)

comment:12 Changed 7 weeks ago by NoelClancy

I don't understand what the following line is

fcm:jules.x_br/pkg/karinawilliams/

where does it exist?

comment:13 Changed 7 weeks ago by pmcguire

Hi Noel
It looks like you successfully ran the suite u-bo137 at 11:33 GMT today. Is this true?
Is this the suite you're trying to get working?
Patrick

comment:14 Changed 7 weeks ago by NoelClancy

I'm going to have to abandon JASMIN and revert back to MONSOON.
I don't understand what is wrong

comment:15 Changed 7 weeks ago by NoelClancy

Not on JASMIN!

u-bo137

[file:bin/fluxnet_evaluation.py]
source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/fluxnet_evaluation.py@15046

[file:bin/jules.py]
source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/jules.py@15046

[file:bin/make_time_coord.py]
source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/make_time_coord.py@15046

[file:bin/parallelise.py]
source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/parallelise.py@15046

[jinja2:suite.rc]
INCLUDE_SPINUP=True
JULES_FCM='fcm:jules.xm_tr'
JULES_REVISION='14197'
LOCATION='MONSOON'
NYRSPIN=50
OUTPUT_FOLDER='/projects/ukca-read/nclancy/JULES/jules_output/u-bm225'
PLOT_FOLDER='/projects/ukca-read/nclancy/JULES/jules_plot/u-bm225'
PRESCRIBE_DATASETS_CONFIGURATION=0
RUNID='JULES_vn5.4'
SUBSET='peg'
SUITE_DATA='/projects/ukca-read/nclancy/JULES/jules_input'

comment:16 Changed 7 weeks ago by NoelClancy

I ran that suite (u-bo137) on MOONSOON but it failed after adding in variables.

hat is why I was trying JASMIN

comment:17 Changed 7 weeks ago by pmcguire

Hi Noel:
The lines:

[file:bin/fluxnet_evaluation.py]
source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/fluxnet_evaluation.py@15046

tell the suite to use FCM to copy the file fluxnet_evaluation.py at revision number 15046 from the JULES MOSRS repository at the location jules.x_br/pkg/karinawilliams/r6715_python_packages/share/. It will the save the file to ~nmc/cylc-run/bin/fluxnet_evaluation.py.

Patrick

comment:18 Changed 7 weeks ago by pmcguire

Where is the JASMIN version of u-bo174? (I originally said u-bo137, but that was a typo)
I thought it was on JASMIN at:

/home/users/nmc/roses/u-bo174
Last edited 7 weeks ago by pmcguire (previous) (diff)

comment:19 Changed 7 weeks ago by NoelClancy

Sorry, I copied u-bo137 from a previously successful run on MOONSOON but it failed on JASMIN after adding in variables.

comment:20 Changed 7 weeks ago by NoelClancy

u-bo173 (not u-bo137)

comment:21 Changed 7 weeks ago by NoelClancy

Both u-bo173 and u-bo174 are JASMIN suites (copied from MONSOON suites) but I wanted to change the output and plot folders to JASMIN but it's very complicated

comment:22 Changed 7 weeks ago by pmcguire

Hi Noel:
I see now that the JASMIN version of u-bo174 did not run successfully this morning when you ran it.
If you look at your error file on JASMIN:

/home/users/nmc/cylc-run/u-bo174/log/job/1/jules_at_neu_presc0/01/job.err

It has errors:

[FATAL ERROR] init_surface: Error reading namelist JULES_SURFACE (IOSTAT=19 IOMSG=invalid reference to variable in NAMELIST input, unit 1, file /work/scratch/nmc/cylc-run/u-bo174/work/1/jules_at_neu_presc0/./jules_surface.nml, line 18, position 9)

What do you see when you open the file it refers to, and look at line 18 position 9:

/work/scratch/nmc/cylc-run/u-bo174/work/1/jules_at_neu_presc0/./jules_surface.nml

Patrick

comment:23 Changed 7 weeks ago by pmcguire

Hi Noel:
If you look at that jules_surface.nml file referred to in the last comment, you will see that line 18 refers to l_urban2t. If you search in your browser search engine for l_urban2t, the search engine will point you to this page: http://jules-lsm.github.io/vn5.2/namelists/jules_surface.nml.html , wherein you can learn about how l_urban2t is used in JULES.

But you will note that this is a page for JULES 5.2. The revision of JULES that you are using in your suite u-bo174 has these settings in the rose-suite.conffile:

JULES_FCM='fcm:jules.x_br/dev/karinawilliams/r9227_add_gpp_unstressed_diagnostic'
JULES_REVISION='14197'

This is Karina Williams' branch of JULES. The x_br means 'branch', and dev/karinawilliams/r9227_add_gpp_unstressed_diagnostic is the branch name. From what I remember, this branch was branched from the trunk of JULES at JULES 4.9.
If you look at http://jules-lsm.github.io/vn4.9/namelists/jules_surface.nml.html , you will see that l_urban2t is not defined there in JULES 4.9.

I suspect that is the problem. You are using the wrong version of JULES.

If you look at the original suite that you are trying to get running on JASMIN u-bm066:
https://code.metoffice.gov.uk/trac/roses-u/browser/b/m/0/6/6/trunk
you will see that it is not the same as u-al752.
For example, it uses in the rose-suite.conf file:

JULES_FCM='fcm:jules.xm_tr'
JULES_REVISION='14197'

This says to use the 'trunk' version of JULES. And it says that (when you are on MONSOON) that you should use the 'mirror' repository (that's what the 'm' means in 'xm').
So maybe you can try the trunk version of JULES instead of Karina Williams' branch in u-bo174?
When you are on JASMIN, you can't use the 'mirror' repository, which means you should use the 'regular' repository (referred to by 'x' instead of 'xm'), so can you try the following instead?

JULES_FCM='fcm:jules.x_tr'
JULES_REVISION='14197'

Does that work?
Patrick

comment:24 Changed 7 weeks ago by pmcguire

Hi Noel:
To follow up on the previous comment, the following link will open a page for this JULES_REVISION number (14197) of the JULES trunk
https://code.metoffice.gov.uk/trac/jules/browser/?rev=14197
In there, you will see that the trunk version of JULES that Gerd Folberth was using for his suite u-bm066 is JULES 5.4. He is using that JULES_REVISION number of the trunk, rather than later revision numbers, presumably to make sure that later versions (5.5, 5.6) are not used.
Patrick

Last edited 7 weeks ago by pmcguire (previous) (diff)

comment:25 Changed 7 weeks ago by pmcguire

Hi Noel:
Is it working now?
Patrick

comment:26 Changed 3 weeks ago by NoelClancy

I started from scratch with a different suite run that had 67 variables.
This suite ran successfully and so I re-ran that suite with different ozone concentrations.

I read carefully through the notes in the namelists etc and the information is all there but you really have to read carefully through it all
For example, if you try to include a variables such as flux_o3_stom, you must ensure that the ozone damage switch is on, l_o3_damage = true

And another thing is that there are no other errors in the suite. You can add in more variables providing you read the notes in the namelists and ensure all the conditions are met for the new variables before re-running the suite.

comment:27 Changed 3 weeks ago by NoelClancy

yes, thanks for that Patrick, it is working now, thanks

comment:28 Changed 3 weeks ago by pmcguire

Excellent!
I will close the ticket now.
Patrick

comment:29 Changed 3 weeks ago by pmcguire

  • Resolution set to completed
  • Status changed from accepted to closed
Note: See TracTickets for help on using tickets.