[MITgcm-support] parameter file read

[Chris Hill]

Hi Tom,

  Could you do a quick test for me?
  Can you change the line in seaice_readparms.F
       READ(UNIT=iUnit,NML=SEAICE_PARM01,IOSTAT=errIO)
       IF ( errIO .LT. 0 ) THEN
  to
       READ(UNIT=iUnit,NML=SEAICE_PARM01,IOSTAT=errIO)
       IF ( errIO .NE. 0 ) THEN

  and see if it that causes a STOP when there is an error, but works 
when there isn't one. I think the test on errIO that some packages use 
may need to be compiler dependent. It stops for "data" because the 
IOSTAT=errIO entry is commented out.

Thanks,

Chris


Thomas Haine wrote:
> Hi Folks, 
> 
> I'm experiencing some unfamiliar behaviour reading parameter files. In
> verification/lab_sea (latest release) if I corrupt one of the parameter
> names I get no error message but the subsequent parameter settings are
> ignored.
> 
> E.g.:
> 
> 
>># SEAICE parameters
>> &SEAICE_PARM01
>>      SEAICEwriteState   = .TRUE.,
>>      SEAICE_initialHEFF = 1.0,
>>      SEAICE_deltaTtherm = 3600.,
>>      XXXXXX_deltaTdyn   = 3600.,
>>#     SEAICEuseDYNAMICS  =.FALSE.,
>>      LSR_ERROR          = 1.E-12,
>>#     SEAICE_tave_mnc    = .FALSE.,
>> &
> 
> 
> gives no error and output.txt shows the initial thickness field is set
> correctly at 1.0m, but the LSR_ERROR is still at its default value. In
> other words everything after the illegal parameter is ignored.
> 
> This behaviour also seems to apply to data.kpp. But I get an runtime
> namelist error if I corrupt a parameter name in data or data.gmredi
> (which is what I normally expect to see).
> 
> This has caused me some headaches because updating my code caused my
> seaice parameter settings to change without me realising for a while
> (e.g., my seaice initial condition file was not read and I got the
> default seaice ics). I guess that my old parameters were no longer all
> valid and the settings below the defunct parameters were ignored without
> giving me a runtime error.
> 
> Can you clarify if I have this right and what is the policy on reading
> data files?
> 
> Thanks, Tom.
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support
>