Opened 2 months ago

Closed 2 months ago

#2480 closed help (fixed)

Suite failing in recon

Reported by: charlie Owned by: um_support
Priority: normal Component: UM Model
Keywords: Cc:
Platform: NEXCS UM Version: 10.7

Description

Carried over from #2464. The suite copied from is u-aw739.


On an entirely different matter, this suite is currently running fine, however a 2nd I started yesterday (u-ay314) has failed at the recon stage. I have checked all of the various log files, but can't see any obvious error. What has gone wrong this time? The suite is identical to my currently working one, with the only difference being the ancillary files going into it (which I have carefully checked, and thought they were all okay).

Charlie

Change History (9)

comment:1 Changed 2 months ago by grenville

Charlie

The problem is likely with the ancillary file(s) - the err file points to

/home/d05/cwilliams/cylc-run/u-ay314/share/fcm_make_um/preprocess-recon/src/um/src/control/dump_io/chk_look.F90 line 94, which is lookup header related

Please switch on extra diagnostic messages for the reconfiguration (search for RCF_PRINTSTATUS in the rose gui) and set prnt_force_flush to true - maybe we'll get a better idea.

Grenville

comment:2 Changed 2 months ago by charlie

Thanks Grenville. I have now done that (I don't have a prnt_force_flush, but instead have changed the RCF_PRINTSTATUS from "Normal…" to "Extra diagnostic messages" so hopefully that will do?), and have resubmitted my job.

I thought the error was probably to do with the ancillary files, because they are the only things that have changed since my last suite, which works. However, I am surprised, because the ancillary files I'm now trying to use are all the default standard versions which originally came with the suite i.e. not modified by me at all. So I can't see why the header information would be wrong.

I'll let you know when it fails…

comment:3 Changed 2 months ago by charlie

Hi again, right, it has failed. Same error as before. I have taken a look at the file it is pointing to, but can't see what the error is. Can you help?

comment:4 Changed 2 months ago by grenville

Charlie

Where did /home/d05/cwilliams/ga71/ancils/modern.d/sst/qrclim.sst come from — it looks like a file appropriate for an old UM version (with inverted latitudes)?

Grenville

comment:5 Changed 2 months ago by charlie

Hi Grenville,

Very many apologies, real schoolboy error here on my behalf. How embarrassing.

The file was originally at /projects/um1/ancil/atmos/master/sst/hadisst_6190/v1/qrclim.sst

It is a 12 monthly climatology of SST, at 1 deg resolution (i.e. 360,180). I used xconv to convert it into netcdf, then used CDO to convert it to N96 - and was very careful that the latitudes were the right way round, as I know this is a common problem (i.e. South to North, which is what the model requires). I then used xancil to convert it back to a UM file that, and here's where the schoolboy error comes in, forgot to change the version number the beginning of xancil from 4.5 to what it should be, 10.7. So in the resulting UM file, the latitudes were the wrong way round (i.e. North to South) which I failed to notice when I checked it. I made exactly the same error when making the sea ice file.

I have now re-made the file, and have checked the latitudes which are now correct. I have just resubmitted my suite, and it ran recon a bit further, but failed right at the end. What has gone wrong this time? I can see an error (in the error log), "Non-standard period for periodic data" so is this the reason?

The only ancillary files I have touched are seaice and SST (simply to use CDO to convert them to N96) so I'm wondering if this is related to one of these and something that went wrong with the xancil stage?

Very many apologies about this.

Charlie

PS. Before resubmitting, I turned the diagnostic information back to "Normal" which I assumed is what it usually should be? Unfortunately, however, this means we don't have the extra diagnostics for this new error.

comment:6 Changed 2 months ago by jeff

Hi Charlie

It looks like the time dimension in the sst and seaice ancillaries is wrong. The error occurs when using xconv to convert the UM file to netcdf, it is misinterpreting the header information. The easiest way to fix this is in xancil, in the SST section change from Use dates from NetCDF file to Specify SST ancillary file dates and fill in your own date values.

Jeff.

comment:7 Changed 2 months ago by charlie

Thanks very much Jeff, I will try doing that as soon as I can get back onto the machine (as I can't do anything right now given our home directories are all down).

comment:8 Changed 2 months ago by charlie

Okay Jeff, that was the problem - it's now got past the recon stage and is now running, many thanks.

comment:9 Changed 2 months ago by jeff

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