Opened 2 years ago

Closed 2 years ago

Last modified 2 years ago

#2808 closed help (fixed)

UKESM with Offline oxidants

Reported by: lm2612 Owned by: um_support
Component: UM Model Keywords: ukca, ukesm
Cc: Platform:
UM Version:



I am running suite u-bg778, which is a copy of the UKESM suite, with land and dynamic vegetation fixed. I want to switch the UKCA chemistry to offline oxidants BE (14). I change this setting (under i_ukca_chem in Section 34 of the um app), and leave the default settings that appear with this, but the coupled task fails instantly with:

???!!!???!!!???!!!???!!!???!!! ERROR ???!!!???!!!???!!!???!!!???!!!
? Error code: 1
? Error from routine: EG_BICGSTAB
? Error message: NaNs? in error term in BiCGstab after 1 iterations
? This is a common point for the model to fail if it
? has ingested or developed NaNs? or infinities
? elsewhere in the code.
? See the following URL for more information:
? Error from processor: 0
? Error number: 446

The default input files that go into offline oxidants look the same as the files I've used in a previous suite (e.g. '$UM_NETCDF_UKCAOXID_O3_DIR/$UM_NETCDF_UKCAOXID_O3_FILE' …. ), so I didn't think these would cause the problem.

Do you have any ideas what would cause this error?



Change History (5)

comment:1 Changed 2 years ago by luke

Hi Laura,

It probably is possible to turn off StratTrop to Offline Oxidants, but it's rather fiddly as in UKESM1 you will then need to make sure that CH4 and N2O concentrations are set (this is probably causing this error), as well as other changes to cope with changing the chemistry.

Can I check why you aren't running GC3.1 instead of UKESM1 if you want to do this?

This is effectively UKESM1 without StratTrop chemistry and using Offline Oxidants (although the Ocean BGC model MEDUSA also is not used). This model will also have been tuned to cope with using this different chemistry scheme and prescribed ozone field etc.


comment:2 Changed 2 years ago by lm2612

Hi Luke,

Okay that makes sense. I wasn't sure if they would be exactly the same. (I also don't want the BGC model either)

Do you know if there is a GC3.1 standard suite for monsoon with vn11.2 (the same version as UKESM)?


comment:3 Changed 2 years ago by luke

Hi Laura,

I don't know if there is a vn11.2 GC3.1 suite. The CMIP runs will have been done at an earlier version. The page UM/Configurations/GA7.0-GC3.0 lists a vn10.7 suite on ARCHER, and this is also what is tested on the UM/Configurations/Performance page.

There is this page on the MOSRS:

that mentions a vn11.0 suite for internal use, but if you must use a vn11.2 suite then you could contact the GMED team and ask if a configuration is available on the XCS-C.


comment:4 Changed 2 years ago by lm2612

  • Resolution set to fixed
  • Status changed from new to closed

Hi Luke,

Thank you for this. I'll go with the vn10.7 GC3.1, if that is what CMIP6 runs are done with.

Thanks again for the help- I am closing the ticket now.


comment:5 Changed 2 years ago by luke

Thanks Laura,

Just to add, the GC3.1 runs are vn10.7, and the UKESM1 runs are a mixture of 10.9 for some runs, and vn11.2 for others (and possibly vn11.1), depending on the experiment. The UKESM1 release jobs are all vn11.2.


Note: See TracTickets for help on using tickets.