[MITgcm-devel] lab_sea.hb87 restart problem

[Martin Losch]

OK, there error was somewhere between keyboard and chair: your script  
is case sensitive and I was sloppy in editing data.tst.

Now I have repeated the tests with your script and this is what I get:
original lab_sea.hb87
> csysm3::tr_run.hb87> ./tst1+1 2
> 10 12 14
> 0000000010 0000000012 0000000014
> gcmExc=mitgcmuv
> -- compare cg2d_init_res :
> run 1iA:
>     6.08551932893086E-02
>     6.55586521348751E-02
> run 1iB:
>     7.02066252452855E-02
>     7.29377130933022E-02
> run 2it:
>     6.08551932893086E-02
>     6.55586521348751E-02
>     7.01960854205512E-02
>     7.29481919712837E-02
lab_sea.hb87 with useHB87stressCoupling = .false.,
> csysm3::tr_run.hb87> ./tst1+1 2
> 10 12 14
> 0000000010 0000000012 0000000014
> gcmExc=mitgcmuv
> -- compare cg2d_init_res :
> run 1iA:
>     6.08586878524103E-02
>     6.55715983076715E-02
> run 1iB:
>     7.02079681169673E-02
>     7.29487338830253E-02
> run 2it:
>     6.08586878524103E-02
>     6.55715983076715E-02
>     7.02078133431693E-02
>     7.29489594851236E-02
with useHB87stressCoupling = .false., and evp turned off (#  
SEAICE_deltaTevp   = 60.,)
> 10 12 14
> 0000000010 0000000012 0000000014
> gcmExc=mitgcmuv
> -- compare cg2d_init_res :
> run 1iA:
>     6.08355827055190E-02
>     1.16426574689837E-01
> run 1iB:
>     7.80912267704927E-02
>     7.36159469063700E-02
> run 2it:
>     6.08355827055190E-02
>     1.16426574689837E-01
>     7.80912267704927E-02
>     7.36159469063700E-02
with useHB87stressCoupling = .true., and evp turned off (#  
SEAICE_deltaTevp   = 60.,)
and with useSeaice=.false.,
> 10 12 14
> 0000000010 0000000012 0000000014
> gcmExc=mitgcmuv
> -- compare cg2d_init_res :
> run 1iA:
>     6.13113158130375E-02
>     6.57463167891239E-02
> run 1iB:
>     7.04816593980862E-02
>     7.33615824658603E-02
> run 2it:
>     6.13113158130375E-02
>     6.57463167891239E-02
>     7.04816593980862E-02
>     7.33615824658603E-02

My conclusions from these numbers:
1. Your script is great!!!
2. The restart problem is related to EVP.
However, I can reproduce your result in lab_sea/run, that the restart  
does work with EVP. Puzzling. Therefore I reran the test in lab_sea/ 
run with SEAICE_no_slip = .true. (which is also used in tr_run.hb87),  
and then:
> -- compare cg2d_init_res :
> run 1iA:
>     6.08981197561294E-02
>     6.55972956406261E-02
> run 1iB:
>     7.02007081962388E-02
>     7.29011820001111E-02
> run 2it:
>     6.08981197561294E-02
>     6.55972956406261E-02
>     7.02007451116285E-02
>     7.29010160470697E-02
For tr_run.lsr everything is OK even with SEAICE_no_slip = .true.

So the problem is related to a combination of EVP and no-slip  
boundary conditions (and not the hb87 stress formulation). Now, the  
flag SEAICE_no_slip is not used anywhere in the EVP code, except in  
seaice_calc_strainrates.F, which is also used by the LSR code. I  
don't have no clue where this problem comes from. Why should break  
this flag the pickup for the evp code only, if it is not used in the  
code that is exclusive to EVP?

Any idea?

Martin