Error ‘unbound variable’, related to ancillaries / using ERA5

Unified Model
1

Hi Helpdesk.

I am running the nested suite on ARCHER for the South Orkney Islands region, but having difficulties using ERA5 data to force the model. The model id is u-ct746.

I used the RAS suite to produce ancils for a ‘cutout’ region which is larger than my first nested domain. However, I keep getting the error ‘unbound variable’, which is related to ancillaries used by the glm_um. See attached screenshot. One error was for STASH 357, 358 (aerosols) which I addressed by switching off aerosols in the science options. But another error referred to orography, while the attached screenshot refers to vegfrac … Can you please advise how to deal with these errors?

Many thanks for any help.

Best wishes,

Andrew

2

for_helpdesk
for_helpdesk1825×696 223 KB

3

Hi NCAS,

Just to update that I got passed this error with the help of Claudio Sanchez from the MO. He advised that you need to complete the RAS suite run (so set any ancils that fail to ‘succeeded’) so it can complete its workflow and build the ancil_versions & ancil_filenames files. This worked.

However, I’m now getting a failure in creating the LBCs for the 4 km domain. I have never come across this error.

I would be grateful for any suggestions.

Thanks

Andrew

4

for_ncas
for_ncas1815×756 126 KB

5

Andrew

Looks like you have fixed it?

Grenville

6

for_ncas
for_ncas1815×756 126 KB

7

Hi Grenville,
Thanks for replying.
I now have a further issue related to producing LBCs for the 4km model.
See previous screenshot
Andrew

8

for_claudio
for_claudio1574×733 73.9 KB

9

Hi Grenville

But I just looked at the 12km output and the surface temperatures are -150C or so. This is the rhs panel of the attached screenshot, using ERA5 + cutout to force a 12 km domain. The output on the LHS is from the same 12 km domain but using ERA-Interim to force the N320 GLM to produce start data and LBCs. So there seems to something wrong with using the combination of ERA5/cutouts.

Have you came across this type of error before? I’m new to using ERA5/cutouts.

Also, can I just ask whether I can use ERA5 to drive a GLM model? So can I just replace ERA-Interim fields with ERA5 fields?

Thanks

Andrew

10

Andrew

I’m a bit confused

/work/n02/n02/anmcr/cylc-run/u-ct746/log/job/19910115T1200Z/south_orkney_islands_4km_ra2m_um_createbc_000/03 indicates that createbc succeeded at 14:46

CYLC_JOB_INIT_TIME=2023-01-25T14:46:43Z
CYLC_JOB_EXIT=SUCCEEDED
CYLC_JOB_EXIT_TIME=2023-01-25T14:46:52Z

Is there still a problem with the 4km lbcs?

Grenville

11

Hi Grenville,

I’m confused as well, as yes the 4km model has completed successfully. I did trigger it again, so maybe it finally ran.

But the issue with very cold temperatures persists. This is also in the 4 km model (I just checked).

Any advise would be great.

Thanks

Andrew

12

Hi Grenville,

Is the ERA5 cutout method appropriate for a domain with a rotated pole?

That might explain why my results are so strange.

Thanks

Andrew

13

Hi Grenville,

As a fallback, I’m trying to get a suite running using the global model to drive the LAM, with ERA5 driving the global model. This is u-cs823. The global model is failing at the reconfiguration stage as its asking for STASH 0, 243 (OBS/CLIM SNOW-FREE SURF SW ALBEDO) which is not in the input dump. Do you know where I can find this ancillary file on ARCHER?

Many thanks,

Andrew

???
???!!!???!!!???!!!???!!!???!!! ERROR ???!!!???!!!???!!!???!!!???!!!
? Error code: 30
? Error from routine: RCF_RESET_DATA_SOURCE
? Error message: Section 0 Item 243 : Required field is not in input dump!

???
???!!!???!!!???!!!???!!!???!!! ERROR ???!!!???!!!???!!!???!!!???!!!

14

Hi Grenville,

I’ve been making progress getting the LAM to run on ARCHER, using ERA5 to force a global model. This is u-ct823. I have an ‘unbound variable’ error related to the ‘topographic index’ ancillary, which the global model requires. I have added a path to this file (see attached screenshot), but I think I need to also define in this in another place, which is why I am getting the error.

FAIL] namelist:items(7ea74479)=ancilfilename: UM_ANCIL_LSH_TIND_FILE: unbound variable
[FAIL] source: namelist:items(:slight_smile:
2023-01-26T22:31:32Z CRITICAL - failed/EXIT

Can you please advise?

Thanks

Andrew

15

for_ncas
for_ncas1773×910 136 KB

16

Andrew

Can you not simply specify the full path to the topographic index file rather than using environment variables?

Grenville

17

Hi Grenville,

I don’t know what the full path is. I don’t know where the data for this ancillary is stored on Archer. I also couldn’t find it on Monsoon.

Thanks

Andrew

18

Hi Andrew

The topographic index file has cropped up of late. I confess to knowing very little about it - Ella might be a better source of information than us (until we’ve had time to understand its use).

Grenville

19

Hi Grenville,

I got past this stage by setting the topographic index to fixed values. Now the reconfiguration of the global start dump works now. However, the global model forecast (glm_um_fcst_00) fails with the EG_BICGSTAB error. Note that I am trying to get the LAM running with the global model being forced by ERA5, rather than the cutouts, as I am more familiar with this. This is still u-cs823.

Are you able to please help. I am spending so long on this and getting nowhere.

Has anyone got a LAM running on ARCHER2 using ERA5 that I can email? Or take a copy of their model?

Finally, I know that version 11.1 of the UM on Monsoon runs using ERA5, so maybe there is a version 11.1 of the model on ARCHER that I can use.

Thanks

Andrew

???
???!!!???!!!???!!!???!!!???!!! 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:
? https://code.metoffice.gov.uk/trac/um/wiki/KnownUMFailurePoints
? Error from processor: 653
? Error number: 49
???

20

Hi Grenville,

Claudio Sanchez at MO suggested that I update to latest version of the RNS at archer (branch archer2_um12p0 [1]). Can you please tell me how I check out a copy of the RNS branch? Or can you do this for me and give me a suite ID that I can copy?

Thanks,

Andrew