Opened 2 years ago

Closed 2 years ago

#2126 closed help (answered)

Error in plume deposition diagnostics

Reported by: ggxmy Owned by: um_support
Component: UM Model Keywords: log plume scavenging
Cc: Platform: Monsoon2
UM Version: 10.6

Description

Dear CMS,

My suite u-al036, which is based on Luke's vn10.6.1 nudged job failed. It gives me error messages like this;

???!!!???!!!???!!!???!!!???!!!       ERROR        ???!!!???!!!???!!!???!!!???!!!
?  Error code: 1
?  Error from routine: UKCA_MAIN1
?  Error message:  No valid item number identified for this mode and component
?  Error from processor: 103
?  Error number: 21
????????????????????????????????????????????????????????????????????????????????

I found this error message in ukca_main1-ukca_main1.F90 where it is dealing with plume scavenging diagnostics. But that part of the code goes like this;

          item = item_no_plume(imode,icp)   ! find correct stash item no.
          IF (item < 0) THEN
            icode = 1
            cmessage = ' No valid item number identified for this mode'// &
                       ' and component'
            WRITE(umMessage,'(A70,2I6)') cmessage,imode,icp
            CALL umPrint(umMessage,src='ukca_main1')
            CALL ereport('UKCA_MAIN1',icode,cmessage)
          END IF

so it should have output not only the text message but also imode and icp so I can see for which diagnostic the model crashes, but these are actually not seen in the error messages. Moreover, as you can see in the attached image, I cannot find job.err file in the place shown at the top of the window ( $HOME/cylc-run/u-al036/log/job/19880901T0000Z/atmos_main/01/ ).

So I have a couple of questions and requests:

1) Where are the log files?

2) Am I right in that the run crashed where one of plume scavenging diagnostics is being dealt with in ukca_main1-ukca_main1.F90?

3) If so how can I get imode and icp for which the run crashed?

4) If you already know which plume scavenging diagnostic is likely to cause a problem, could you please tell me that? The diagnostics I requested are shown in the attached image.

5) job.err also contains lines like below;

[103] exceptions: Program is "/home/myosh/cylc-run/u-al036/share/fcm_make_um/build-atmos/bin/um-atmos.exe"

but my home directory is actually /home/d03/myosh/. Did this cause a problem? If so how can I fix this?

Thanks for your help in advance.
Masaru

Attachments (2)

job.err.jpg (140.9 KB) - added by ggxmy 2 years ago.
plume scavenging STASH requests.jpg (102.7 KB) - added by ggxmy 2 years ago.

Download all attachments as: .zip

Change History (18)

Changed 2 years ago by ggxmy

Changed 2 years ago by ggxmy

comment:1 Changed 2 years ago by ggxmy

Ah, I just noticed that I was requesting SO (secondary organic?) which is not included in my model setting. I will try without this.

comment:2 Changed 2 years ago by ros

Hi Masaru,

I can't help with the ukca error, hopefully Luke will be able to help out here, although it sounds like you have thought of something else to try, however, I can help with the directory naming issue.

You are still submitting this suite to the XCM where your home directory is /home/myosh. To submit to the new XCS-C you will need to change the host to xcsc. In the file roses/u-al036/site/MONSooN.rc file change xcm to xcsc.

Cheers,
Ros.

comment:3 Changed 2 years ago by ggxmy

I tried to run the suite with plume scavenging of SO removed, but it failed exactly the same way…

comment:4 Changed 2 years ago by ggxmy

Hi Ros.,

Making that change caused a new problem. This should be a problem associated with the change from xcm to xcsc so I thought there should be an instruction to follow? If not, I will open a new ticket to ask about the error I got.

Thanks,
Masaru

comment:5 Changed 2 years ago by ros

Hi Masaru,

For Rose suites all you should need to do is change the host and paths to any input files, so yes do open a separate ticket for that error and I'll take a look.

Cheers,
Ros.

comment:6 Changed 2 years ago by luke

Hi Masaru,

Can you comment here whether or not you still have this problem once you've made the changes I've suggested in #2127?

Thanks,
Luke

comment:7 Changed 2 years ago by ggxmy

#2127 has resolved by doing the option 1. Now I'm back with this issue.

The relatively minor problems like 1) and 5) above have been solved now by porting the job to xsc-s but the main problem persists.

I'll try this again including a print statement to write out imode and icp for which model crashed.

Masaru

comment:8 Changed 2 years ago by luke

Hi Masaru,

Can I just double-check: are you running a straight copy of my u-ak353 suite, or are you copying it, making some changes, and then trying to run it?

If the latter, could you just try running a straight copy of my suite with no changes, other than those required from #2127?

Thanks,
Luke

comment:9 Changed 2 years ago by ggxmy

Hi Luke,

Now I'm trying to run a suite u-al036 including modifications on top of the copy of your u-ak353. But u-ak996, my copy of your u-ak353 without any modification, ran OK for me. That time I was running on old monsoon but that situation hadn't changed when I created this ticket yesterday.

Masaru

comment:10 Changed 2 years ago by luke

Hi Masaru,

Can you please fcm commit all your changes in u-al036 so that I can see them in the repository.

Thanks,
Luke

comment:11 Changed 2 years ago by ggxmy

Hi Luke,

I have done that less than an hour ago. I only added a print statement after that….OK, I did that again now.

Thanks,
Masaru

comment:12 Changed 2 years ago by ggxmy

OK, I found this in the log file;

imode,icp = 6 5

I can't remember what component 5 is (OC?), but it is mode 6, which is accumulation insoluble mode. I'm not requesting plume scavenging diagnostics for anything in insoluble mode…something is wrong.

Masaru

comment:13 Changed 2 years ago by luke

Hi Masaru,

On the face of things, you aren't trying to do anything the code shouldn't do. However, this configuration isn't widely tested, and may not even have been used with Rose. The item_no_plume array is set in ukca_scavenging_diags_mod.F90:

! Define section 38 stash item numbers for 2D plume scavenging diagnostics
INTEGER, PUBLIC :: item_no_plume(nmodes,ncp) =     &
  RESHAPE( (/ 900, 905, 911, 920,  -1, -1, -1,     &      ! SU
               -1, 906, 912, 921, 929, -1, -1,     &      ! BC
              901, 907, 913, 922, 930, -1, -1,     &      ! OC
               -1,  -1, 914, 923,  -1, -1, -1,     &      ! Cl
               -1,  -1, 915, 924,  -1, -1, -1,     &      ! Na
               -1,  -1, 916, 925,  -1,931,932,     &      ! DU
              902, 908, 917, 926,  -1, -1, -1,     &      ! SO
              903, 909, 918, 927,  -1, -1, -1,     &      ! NT
              904, 910, 919, 928,  -1, -1, -1  /), &      ! NH
              (/nmodes, ncp/))

I suspect that you'll need to populate this correctly for the configuration you are running.

Thanks,
Luke

comment:14 Changed 2 years ago by ggxmy

After consulting other people off-line, Colin Johnson suggested a change to this array in the email on 28/03/2017. Below is slightly modified by me.

INTEGER, PUBLIC :: item_no_plume =                   &
    RESHAPE( (/ 900, 905, 911, 920,  -1, -1, -1,     &      ! SU
                 -1, 906, 912, 921, 929, -1, -1,     &      ! BC
                901, 907, 913, 922, 930, -1, -1,     &      ! OC
                 -1,  -1, 915, 924,  -1, -1, -1,     &      ! Na
                 -1,  -1, 916, 925,  -1,931,932,     &      ! DU
                902, 908, 917, 926,  -1, -1, -1/),   &      ! SO
                (/nmodes, ncp/) )

I included this change in my local working copy;

/home/d03/myosh/Branches/vn10.6.1_debugging

and the suite ran without stopping. Now I created a new branch to make this change;

branches/dev/masaruyoshioka/vn10.6.1_plume_scav_diag_fixed@36238

although I haven't run the suite with this branch.

Also from Mohit, I learned that plume scavenging should be requested for DIAG, not D1TH like dry depositions.

Thank you, Ros. and Luke for your help so far.

However, the job did not create pm files in which plume scavenging were requested. So I could not confirm if plume scavenging diagnostics come out correctly. I will open another ticket for this issue.

Masaru

comment:15 Changed 2 years ago by ggxmy

Mohit is helping me about this issue. Also I found a similar ticket #1902. So now I have a couple of things to try. I will come back if these do not work.

Masaru

comment:16 Changed 2 years ago by ggxmy

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

I tried couple of things but all failed in the same way. Anyway this is off the main topic of this thread so I'm closing this one.

Thank you for your help.
Masaru

Note: See TracTickets for help on using tickets.