The index of I,J in nesting.F (version 3.6) may be wrong.

Bug reports, work arounds and fixes

Moderators: arango, robertson

Post Reply
Message
Author
liutianran
Posts: 1
Joined: Tue Dec 14, 2010 4:38 pm
Location: Institute of Oceanogr., Chinese Academy of Science

The index of I,J in nesting.F (version 3.6) may be wrong.

#1 Post by liutianran » Sun May 26, 2013 7:44 am

The codes in "Line 2417" to "Line 2420" in nesting.F (ROMS 3.6) are:

Code: Select all

          i=contact(cr)%Idg(m)
          j=contact(cr)%Idg(m)
          Ic=contact(cr)%Irg(m)
          Jc=contact(cr)%Irg(m)
These should be:

Code: Select all

          i=contact(cr)%Idg(m)
          j=contact(cr)%Jdg(m)
          Ic=contact(cr)%Irg(m)
          Jc=contact(cr)%Jrg(m)
This bug should be modified, otherwise the model can not run because the array "Fmsk" is out of bound in my experiment.

Same problems appear in Line: 2442~2445, 2481~2484, 2736~2739, 2760~2763, 2798~2801

User avatar
m.hadfield
Posts: 520
Joined: Tue Jul 01, 2003 4:12 am
Location: NIWA

Re: The index of I,J in nesting.F (version 3.6) may be wrong

#2 Post by m.hadfield » Mon May 27, 2013 12:53 am

I don't know about that, but I do know that when I run the dogbone-refined case, the water level in the refined grid trends steadily downwards in a completely unphysical way. Perhaps these two problems are related.

User avatar
arango
Site Admin
Posts: 1103
Joined: Wed Feb 26, 2003 4:41 pm
Location: IMCS, Rutgers University
Contact:

Re: The index of I,J in nesting.F (version 3.6) may be wrong

#3 Post by arango » Mon May 27, 2013 4:11 pm

Yes, thank you for bringing this to my attention. See the following svn :arrow: ticket. Please update.

User avatar
m.hadfield
Posts: 520
Joined: Tue Jul 01, 2003 4:12 am
Location: NIWA

Re: The index of I,J in nesting.F (version 3.6) may be wrong

#4 Post by m.hadfield » Tue May 28, 2013 1:46 am

I have updated my code with this fix, but the dogbone-refined case is still not working properly. I have attached a couple of videos (gzipped AVIs) showing zeta in the outer and inner domains.

In the outer domain, the positive-zeta wave moves in from the LH boundary and approaches the constriction, but then seems to disappear: e.g., see frames 5 & 6.

In the inner domain, a positive-zeta wave enters and propagates to the right, but at the left-hand end a couple of disturbances develop and grow.

The model crashes at time step 83 (outer), 250 (inner) with a CFL violation on vertical velocity at (14,10,10) on the outer domain.

Code: Select all

     82     0 00:08:12  4.351299E-02  4.939350E+01  4.943701E+01  6.325956E+07  01
            (14,09,10)  3.421707E-04  5.205159E-02  2.386312E+01  4.376280E+00
    246     0 00:08:12  4.106054E-01  5.296975E+01  5.338036E+01  8.511694E+06  02
            (01,11,10)  2.056226E-02  2.161627E-02  6.118505E-01  2.507218E+00
    247     0 00:08:14  4.339288E-01  5.291041E+01  5.334434E+01  8.489153E+06  02
            (01,08,10)  3.491709E-02  7.976886E-04  4.994146E-01  4.240518E+00
    248     0 00:08:16  4.559642E-01  5.288544E+01  5.334140E+01  8.462377E+06  02
            (02,15,10)  3.249771E-02  3.280206E-01  2.444752E+00  5.567202E+00
     83     0 00:08:18  1.173279E-01  4.978467E+01  4.990200E+01  6.326357E+07  01
            (14,10,10)  1.656612E-03  2.060149E-02  1.355544E+01  1.111991E+01
    249     0 00:08:18  5.005057E-01  5.288461E+01  5.338511E+01  8.431401E+06  02
            (02,14,10)  4.462937E-02  3.894090E-01  2.606050E+00  7.087558E+00
    250     0 00:08:20           NaN           NaN           NaN           NaN  02
            (00,00,00)  0.000000E+00  0.000000E+00  0.000000E+00  0.000000E+00

 Blowing-up: Saving latest model state into  RESTART file

      WRT_RST   - wrote re-start fields (Index=1,2) into time record = 0000001  01

 Blowing-up: Saving latest model state into  RESTART file

      WRT_RST   - wrote re-start fields (Index=1,1) into time record = 0000001  02
Attachments
dogbone_href3_zeta.avi.gz
Dogbone refined case: zeta in the inner domain
(53.84 KiB) Downloaded 119 times
dogbone_whole_zeta.avi.gz
Dogbone refined case: zeta in the outer domain
(50.21 KiB) Downloaded 116 times

User avatar
m.hadfield
Posts: 520
Joined: Tue Jul 01, 2003 4:12 am
Location: NIWA

Re: The index of I,J in nesting.F (version 3.6) may be wrong

#5 Post by m.hadfield » Tue May 28, 2013 5:39 am

The dogbone composite-grid case appears to work OK.

User avatar
arango
Site Admin
Posts: 1103
Joined: Wed Feb 26, 2003 4:41 pm
Location: IMCS, Rutgers University
Contact:

Re: The index of I,J in nesting.F (version 3.6) may be wrong

#6 Post by arango » Tue May 28, 2013 12:22 pm

Thank you for the videos Mark. I need to look in the debugger to check what it is going on. Apparently, something is wrong with the two-way nesting in fine2coarse.

Post Reply