[MITgcm-support] confusion about parameters diagstats

[Jean-Michel Campin]

Hi Martin,

The only thing I could think of to explain this:
> the first 15 regions have reasonable values and then everything in wrong: 
would be to check that you really changed :
>>  stat_region(1,1)= 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,10,11,12,13,14,
If you did not change the list of selected regions you want,
then it goes back to our discussion (on the developpment list,
from July 30):
> (if I remember well, the problem I had was that all the regions 
>  and not only the selected ones were written in the mnc file).
and this would explain what you are getting.

Now, this is just a guess, but before I try to reproduce the problem, 
generating a region mask with 4 sets of regions and a total of 
382 regions, I would propose that you try 1rst to run a short test, 
with at least 1 reccord in the diagstat output file (you can do 2 
iterations and set stat_freq(1) equal to the time-step to get 2 reccords)
but without using mnc (just setting diagSt_mnc = .FALSE. should
do it). If it's still so "confusing" after this, then I will 
try to reproduce the problem.

Cheers,
Jean-Michel

On Fri, Aug 29, 2008 at 04:40:54PM +0200, Martin Losch wrote:
> Hi Jean-Michel,
> I am making strange mistakes with the diagnostics package, in particular 
> with the diagstats part:
>
> I am abusing the "regMsk" capability to save time averages at a lot of 
> individual points (and small areas) for comparison with CTD  
> sections/moorings etc. First I tried this (and it worked):
>>  &DIAG_STATIS_PARMS
>>  diagSt_mnc   = .true.,
>>  diagSt_regMaskFile='arcticNetMask.bin',
>>  nSetRegMskFile= 4,
>>  set_regMask   = 1 , 1 , 2 , 2 , 3 , 1 , 2 , 3 , 4 , 1 , 1 , 1 , 1 , 2 
>> ,
>>  val_regMask   = 1., 2., 3., 4., 5., 6., 7., 8.,  
>> 9.,10.,11.,12.,13.,14.,
>> #---
>>  stat_fields(1,1)= 'THETA   ','SALT    ','DRHODR  ',
>>                    'TRAC04  ','TRAC07  ','TRAC09  ',
>>                    'TRAC05  ','TRAC08  ','TRAC10  ',
>>                    'TRAC14  ','TRAC15  ',
>>                    'TRAC01  ','TRAC06  ','TRAC13  ','TRAC16  ',
>>  stat_fname(1)   = 'ArcticNetMoorings',
>>  stat_freq(1)    = 86400.,
>>  stat_region(1,1)= 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,10,11,12,13,14,
>>  &
> So obviously, I have a mask with 4 different levels (because there is  
> overlap between the different regions). For each region I asign the  
> level in set_regMask, the mask value in val_regMask and the actual  
> regions to be used in output stream 1 in stat_region. The maskfile  
> contains the appropriate masks with values 1:14. and I get a netcdf file 
> with 15 regions.
>
>  Now I tried to extend the number of regions, but still not using them 
> with new mask file the following modifications:
>>  nSetRegMskFile= 4,
>>  set_regMask   = 1 , 1 , 2 , 2 , 3 , 1 , 2 , 3 , 4 , 1 , 1 , 1 , 1 , 2 
>> , 368*1,
>>  val_regMask   = 1., 2., 3., 4., 5., 6., 7., 8.,  
>> 9.,10.,11.,12.,13.,14.,
>>                   15., 16., 17., 18., 19., 20., [...], 382.,
>
> and now I get a netcdf file with 386 regions (which I don't understand), 
> the first 15 regions have reasonable values and then everything in wrong: 
> a lot of regions of the 386 that are completely zeros and some values 
> that are obviously not from the regions I wanted, eg. too high 
> temperatures.
>
> What am I doing wrong? Or is this a bug in the interplay of mnc and  
> diagstats?
>
> Martin
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support