Milestone Reference configurations


GYRE and ORCA2_LIM are the current reference configurations available and on which regular tests are performed.


  • GYRE_LOBSTER integration

Steps to be done

- modify assessment.ksh script to perform an 'automatic" check for the Biogeo. component 

Steps done

- modules list which are different from the trunk:
# initrc.F90
# par_trc.F90
# trcdit.F90
# trcdtr.F90
# trcrst.F90
modifs related to trcbio.F  &trcsed.F have been already committed

- integration under NVTK OK (2do: make links of *[Ffh] files in the HEAD/Makefile)
- bugs in trcbio.F and trcsed.F90 missing "#" before endif
- BUG in trcrst.F90 which doesn't allow the restartability: just after read the trb(:,:,:,:) field a swap is done trb() = trn() ..... 

For the Bio this can be checked only for the 6 tracers MEAN and the 
"Integral of all tracers over the full domain" not for the "Drift of the sum of all tracers"
since it is computed since the beginning of the run not the simulation ..... 
The awk script must compare the following lines extracetd from ocean.output
----TRACER STAT----                                                             
   MEAN NO   1  DET          
   MEAN NO   2  ZOO          
   MEAN NO   3  PHY           
   MEAN NO   4  NO3          
   MEAN NO   5  NH4            
   MEAN NO   6  DOM          
   Integral of all tracers over the full domain

- Mon:  OK over 30 days for both physics (solver.stat) and bio (TRACERS STAT in ocean.output)
- Mpi:  OK over 30 days for both physics (solver.stat) and bio (TRACERS STAT in ocean.output_xxxx)

To check this, we just compare the MAX & MIN value of each tracer at the end of each mon & mpi run. 
We can't make differences on vairable computed with an mpp_sum CALL
- Reproductibility (mon .vs. mpi ) has been checked for both physics and bio 

Last modified 4 years ago Last modified on 2016-07-06T12:22:32+02:00