Partial sum file inconsistent - nudged suite error

Unified Model
1

Hi,

I converted u-be303/archer2 to run as a nudged run (u-cy789). Everything was fine for the first ~20 years of model run time, but it fails in the atmos_main step with the following error:

???!!!???!!!???!!!???!!!???!!! ERROR ???!!!???!!!???!!!???!!!???!!!
? Error code: 2
? Error from routine: U_MODEL_4A
? Error message: ACUMPS1: Partial sum file inconsistent. See Output
? Error from processor: 149
? Error number: 71
???

When looking in the job.out file I see the following issue:

FINAL TOTAL ENERGY = 0.13066E+25 J/
INITIAL TOTAL ENERGY = 0.13067E+25 J/
CHG IN TOTAL ENERGY O. P. = -0.87759E+20 J/
FLUXES INTO ATM OVER PERIOD = -0.19855E+22 J/
ERROR IN ENERGY BUDGET = -0.18977E+22 J/
TEMP CORRECTION OVER A DAY = -0.51534E+00 K
TEMPERATURE CORRECTION RATE = -0.59646E-05 K/S
FLUX CORRECTION (ATM) = -0.43058E+02 W/M2

Where the line ‘FLUXES INTO ATM OVER PERIOD = -0.19855E+22 J/’ is highlighted red, implying this is the error.

I was just wondering if this was an obvious error as I can’t seem to find any previous issues of it occurring, and if not, if you had any idea about how to solve it.

All the best,
Liam

2

Liam

How did the model recover from the first error from Sept 10?

????????????????????????????????????????????????????????????????????????????????
???!!!???!!!???!!!???!!!???!!!       ERROR        ???!!!???!!!???!!!???!!!???!!!
?  Error code: 1
?  Error from routine: NDG_GET_DATA
?  Error message: Dimension mismatch for: T      3    60    91
?  Error from processor: 0
?  Error number: 71
????????????????????????????????????????????????????????????????????????????????

Grenville

3

Hi Grenville,

Ok, that interesting, so I would have retriggered the atmos_main routine, probably by literally setting the status to ‘waiting’ in the gui, I guess this is not best practise, but in the past I have found it will often reload modules correctly if I do this. However, the model seems to have failed again, but this time with this different error. Perhaps I am best to just restart the model at an earlier date using an older restart file (say, 20050101) and see if the error pops up again and try to solve from there?

Liam

4

Hi Grenville,

I have tried rerunning the model triggering from 20050101 and again it works correctly until 20060101 where I get the error you sent over with the dimension mismatch. Any suggestions?

All the best,
Liam

5

what are your ARCHER2 & puma user names

6

lpb20 on both, although I am now running this suite from puma2 with the same username.

7

Hi Liam

I’ll need run the model to get it into the state where the error occurs - I’ll assume it needs to be fixed unless you tell me otherwise.

Grenville

8

Hi Grenville,

Thanks, that’s great. Have just committed changes to latest point where issue was occurring.

All the best,
Liam

9

Hi Liam

The nudging problem comes about because of a problem with /work/y07/shared/umshared/ukca/nudging/era-40/ecm-e40_1deg-model-levs_2006020100_all.nc

We are looking into it.

Grenville

10

Liam

This message from the MO:

For information, the official ERA-40 dataset only runs from 1957-2002. From /projects/ukca-admin/analyses/ERA_Nudging_readme.txt , it looks like the data needed for Nudging the UM was temporarily extended to 2009 using Operational ECMWF data which was being output on 91 levels. So, unless the user wants to specifically nudge with ERA-40 and/or for periods before 1979 it would be better for them to switch to ERA-Interim (or ERA5 using a branch at vn11.1).

Can you not use a different nudging dataset?

Grenville

11

Hi Grenville,

I see, that makes sense, the files for era-40 go up to 2010 on ARCHER2, so assumed I would be able to run up to then with the suite. I have now changed over to era5 nudging and all seems to be working correctly, so this issue is solved.

Thanks for your help,
Liam

12

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.