Metadata error

Hi CMS,

I found errors in my suite (u-de518) Metadata when I did “check fail-in, warm-in”. They are related to (i) namelist:clmchfcg with errors on clima_fcg_levls_che4 and clim_fcg_levls_n2o, and (ii) nameless:run_radiation with errors on ch4mmr and n2ommr, see attached screenshot for details. My question is do they cause model crash? and how to correct them? Note that the job I copied is a nudged version.

Cheers,
Xin

Hi Xin,

The CH4 and N2O values will not be used if the Radiative feedback from UKCA is selected (l*_ukca_radch4=.true. ; l_ukca_radn2o=.true.*).
I believe the ch4mmr and n2ommr have been deliberately set to unrealistic values that will cause the model to crash and thus ensure that only UKCA fields are used. (The warning/ error here is because they are set to > 0.0 when the above logicals are True).
It is not clear what the warning for the clim_fcg_levels is but they also should not be read in case of UKCA feedback.

Hi Mohit,

Thanks. It seems I do not need to worry about these warnings (even they are marked error). I have just submitted a small job to see it runs smoothly.

Another thing is, I am planning to run a nudged UKCA for the period of Nov/2019-May/2020 to make a campaign comparison. The the current nudged suite version (a copy of u-de170) is only available for years before the end of 2015. I am wondering if there is a suite-id that I can take for that period?

Best,
Xin

Hi Xin,

u-de170 seems to be already set up to run for 2015+ using the CMIP6 ssp3-7.0 scenario emissions and forcings.
You will have to only replace the SST and Seaice files (currently ending in 2019-09) with those in /projects/ukca-admin/inputs/ancil/gregorian/n96e/reynolds/v2.1/reynolds_*_201401-202312.anc

(In app/um/rose-app.conf, search for ‘reynolds’ in the [namelist:items] entries and replace the reynolds_sst and reynolds_seaice files with those mentioned above).


Mohit

Hi Mohit,

Great, thanks! I have replaced both forcing files to see if they work.

Cheers,
Xin

Hi Mohit,

After I changed the for forcing file the submitted suite job (u-de523) for 2019 failed with error message shown below.
? Error code: 10
? Error from routine: INITTIME
? Error message:
? Mismatch between model_basis_time read from namelist and validity time read
? from dump fixed header.
?
? model_basis_time = 2019 9 1 0 0 0
? fixhd validity time = 2015 9 1 0 0 0

It seems the date in “fixhd validity time” is taken from the initial file (but I am not sure on this), but I have changed it to the new year 2019 in the suite. BTW, I could not find where the term “fixed validity time” was defined. Please help, thanks!

Best,
Xin

Hi Xin,

The fihxd_validity_time is read from the initialisation file (AINITIAL_FILE at the top-level).
This is currently set to use ‘de087a.da20150901_00’ from /work/n02/n02/anran.

From ‘moo ls’ it looks like neither u-de087 or u-de170 were actually run much beyond 2015, so there does not seem to be a 2019 initialisation file available from a related configuration.

If there is no suitable initialisation file for 2019, the reconfiguration step can be asked to overwrite the Year in the 2015 dump directly to 2019. Alternatively, you could start the run as ‘2018’ and run the first year as spin-up, given that you are nudging with ERA5 meteorology anyway.

Hi Mohit,

Could you please tell me a bit more in detail regarding how to overwrite the Year in reconfiguration? I could not find a proper webpage talking about this issue.
I am also trying to re-run the job from 2018 to get a 2019 re-start file, but this may take a long time.

Best,
Xin


Example from another configuration

Thanks Mohit. That is very helpful.
Xin

Hi Mohit,

The suite after changing the reconfiguration failed at "recent’ stage with error message:
? Error from routine: RCF_ANCIL_ATMOS
? Error message: replanca_rcf_replanca: Current time precedes start time of data

Not sure what caused the failure. Please help, thanks.

Xin

Hi Xin,

From the Reconfiguration log file

/home/n02/n02/emxin/cylc-run/u-dg901/work/20190901T0000Z/recon/pe_out901.fort6.pe00
=================================
IO: Open: /work/n02/n02/luke/inputs/ocean_bgc/CHL_ssp370_timeseries_bc292_be690_2014_2100_n96_greg.anc on unit  13
replanca_rcf_replanca: UPDATE REQUIRED FOR FIELD    96 : OCEAN NR. SURFACE CHLOROPHYLL(KGM-3)

????????????????????????????????????????????????????????????????????????????????
??????????????????????????????      WARNING       ??????????????????????????????
?  Warning code: -100
?  Warning from routine: REPLANCA_RCF_REPLANCA
?  Warning message: Ancillary time mean lookup items 1-12 are not compliant UMDP F03. Field no:     96
?  Warning from processor: 0
?  Warning number: 3
????????????????????????????????????????????????????????????????????????????????

Error in replanca_rcf_replanca
CMESSAGE replanca_rcf_replanca: Current time precedes start time of data
ErrorStatus 496
============================

So it looks like the recon is incorrectly interpreting dates associated with the ancillary file due to mismatch in header format.
The files seem to have been converted for Gregorian format by Luke.

@Luke have you seen this error before for other ‘future’ runs using the ancillaries?

For info, the checking and modified processing of ancillary headers was implemented at vn12.2 under um:#6438 so it would be interesting to know if the files worked fine in suites upto vn12.2, although there may have been a 15-day deviation while interpolating the data.

Hi Mohit, Xin,

It looks like I made these up using Xancil in July 2023. I was making up a nudged UM13.x job and needed Gregorian ancillaries.

My suite is u-dd451 and this should work on ARCHER2. It’s a UM13.0 UKESM1.1 AMIP suite nudged to ERA-5.

Best wishes,
Luke

Hi Luke and Mohit,

Many thanks on this. I will try suite u-dd451 and see if it works.

Cheers
Xin

@Xin,

I am not sure if this is the right suite I am looking at (u-dg901) which showed the earlier error, but it looks like while the option to ‘override the year in the dump file’ has been selected, the desired year has not been set (activated as 1979 from base UKESM1 configuration).
This is why the RCF/Ancil processing is complaining since all the ancillary data starts in 2015 while the model is trying to initialise for 1979.

Hi Mohit and Luke

I used the new sea ice and set ancillary files in a new suite (u-dh012), however the suite failed to open the seance file, see error message below. not sure what happened.

? Error from routine: io:file_open
? Error message: Failed to open file /projects/ukca-admin/inputs/ancil/gregorian/n96e/reynolds/v2.1/reynolds_seaice_201401-202312.anc

cheers,
Xin

Hi Xin,

I realised later that you are running on ARCHER2 - the files are on Monsoon !
@Luke are you able to copy the files over to ARCHER2 to the UKCA / central area?

Thanks