[MITgcm-support] OBCS errors in OB location vs Mask

michael schaferkotter schaferk at bellsouth.net
Sun Sep 14 07:26:11 EDT 2014


Liu.

you/ll also need to set useOBCS=.FALSE. in data.pkg and set nTimesSteps=1 in data.

this will permit the executable to run for one timestep without halting because of the obcs check.

the hFac[WS].??ta (meta and data) files will be created (among others)

in matlab you can read the hFac data using readbin.m, (after specifying nx,ny,nr, prec )

HFACS   = readbin('hFacS.data' ,[nx ny nr],1,prec);
HFACW  = readbin('hFacW.data',[nx ny nr],1,prec);

readbin.m is located:

http://mitgcm.org/viewvc/MITgcm/MITgcm/utils/matlab/cs_grid/read_cs/readbin.m?view=log


in fortran, the array sections are accessed (after reading in), as
>       OBE = int(hFacW(NNX  ,1:NNY,1))
>       OBW = int(hFacW(2    ,1:NNY,1))
>       OBN = int(hFacS(1:NNX,NNY  ,1)) 
>       OBS = int(hFacS(1:NNX,2    ,1)) 


On Sep 13, 2014, at 6:46 AM, michael schaferkotter wrote:

> Liu;
> 
> looks like you/ll need a more 'irregular' specification of the open boundary, as in OB_Jsouth example (from an experiment) below.
> 
> 
> (PID.TID 0000.0001) // =======================================================
> (PID.TID 0000.0001) // Parameter file "data.obcs"
> (PID.TID 0000.0001) // =======================================================
> (PID.TID 0000.0001) ># Open-boundaries
> (PID.TID 0000.0001) > &OBCS_PARM01
> (PID.TID 0000.0001) > OB_Ieast=720*-1,
> (PID.TID 0000.0001) > OB_Iwest=720*1,
> (PID.TID 0000.0001) > OB_Jnorth=1036*-1,
> (PID.TID 0000.0001) > OB_Jsouth=332*1,155*0,166*1,24*0,359*1,
> 
> the model is calculating the mask and comparing hFac[WS] array sections to the corresponding namelist entries, and warning that there is
> some disagreement.
> 
> you can begin to study the proper set of namelist entries by looking at (after running the model one timestep to get the files hFac[WS].
> 
>       OBE = int(hFacW(NNX  ,1:NNY,1))
>       OBW = int(hFacW(2    ,1:NNY,1))
>       OBN = int(hFacS(1:NNX,NNY  ,1)) 
>       OBS = int(hFacS(1:NNX,2    ,1)) 
> 
> the namelist entries are simply a run length encoding of the array sections, with the added wrinkle that the "1s" are '-'  on east and north.
> 
> iirc, the mask arrays are printed in the STDOUT files.
> If you are using mpi, then there will be one STDOUT file per tile. to use that information is cumbersome.
> 
> michael
> 
> 
> On Sep 13, 2014, at 1:20 AM, 刘成彦 wrote:
> 
>> 
>> Dear all
>> 
>>  I am trying to set up a regional experiment with OBCS in the Prydz Bay. We have three open boundaries in the eastern, western and northern edges, respectively. However, I have encountered a problem. The report from STDERR.0000 is in the follows:
>> *** ERROR *** OBCS_CHECK:   742 errors in OB location vs Mask
>> *** ERROR *** S/R ALL_PROC_DIE: ending the run
>> First of all, I think there should be something wrong in my namelistfile of data.obcs as follows:
>> &OBCS_PARM01
>>  OB_Jnorth =   640*1,
>>  OB_Jsouth =   640*0,
>>  OB_Iwest  =   640*1,
>>  OB_Ieast  =   640*1,
>> After reading discussions in former emails, I realize that the setting of the OB_Jnorth and OB_Ieast should be changed with:
>> &OBCS_PARM01
>>  OB_Jnorth =   640*-1,
>>  OB_Jsouth =   640*0,
>>  OB_Iwest  =   640*1,
>>  OB_Ieast  =   640*-1,
>> Again, I submit this mission. At this time, I have encountered the same problem, but with a small difference!
>> The message in STDERR.0000 is
>> *** ERROR *** OBCS_CHECK:   222 errors in OB location vs Mask
>> *** ERROR *** S/R ALL_PROC_DIE: ending the run
>> 742 errors becomes 222 errors may imply that there is something related in my modification in OB_Jnorth and OB_Ieast.
>> By reading the information of STDOUT.0000, the model has been manipulate at the following step:
>> OBCS_CHECK: end summary.
>> OBCS_CHECK: set-up OK
>>  
>> So, does anyone have encounter the similar problems as I did, any response will be great help to me!
>>  
>> Thank you for your considerations!
>> 
>> 
>> 
>> ----------------
>> *********************************************
>> Dr. Chengyan Liu
>> Nanjing University of Information
>> Science & Technology
>> No.219, Ningliu Road, Nanjing, Jiangsu,
>> China(Postcode: 210044)
>> Phone: 18251854735
>> E-Mail: killy at nuist.edu.cn
>> 
>> *********************************************
>>  
>> 
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-support
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20140914/d39a02d0/attachment-0001.htm>


More information about the MITgcm-support mailing list