[MITgcm-devel] seaice_get_dynforcing.F: CDAIR SEAICE_drag and OCEAN_drag

[Martin Losch]

Hi there,
I also opt for changing the reference experiments and not the default. I agree that bad defaults are a dilemma and pkg/seaice is a collection of bad defaults, but upsetting many setups is worse.

@Ian: I am not going to change the code starting at line 247 in seaice_get_dynforcing.F because it will not affect any of my experiments, so you’ll have to do it. May I suggest that you, in order to get rid of multiple divisions by OCEAN_ocean that are my fault, introduce a temporary variable recip_ocean_drag, or maybe even a ratio and a ratio_south so that the code would read something like:
         ratio       = SEAICE_drag      /OCEAN_drag
         ratio_south = SEAICE_drag_south/OCEAN_drag
         DO j=1-OLy,sNy+OLy
          DO i=1-OLx,sNx+OLx
C now ice surface stress
           IF ( yC(i,j,bi,bj) .LT. ZERO ) THEN
            taux(i,j,bi,bj) = ratio_south*fu(i,j,bi,bj)
     &                     *_maskW(i,j,ks,bi,bj)
            tauy(i,j,bi,bj) = ratio_south*fv(i,j,bi,bj)
     &                     *_maskS(i,j,ks,bi,bj)
           ELSE
            taux(i,j,bi,bj) = ratio*fu(i,j,bi,bj)
     &                     *_maskW(i,j,ks,bi,bj)
            tauy(i,j,bi,bj) = ratio*fv(i,j,bi,bj)
     &                     *_maskS(i,j,ks,bi,bj)
           ENDIF
          ENDDO
         ENDDO

Martin


On Jan 7, 2014, at 2:59 AM, Menemenlis, Dimitris (3248) <Dimitris.Menemenlis at jpl.nasa.gov> wrote:

> I agree.  Changing a default can be a source of problems.
> Your suggestion is a more sensible way to go,
> and actually we already have started using better
> values in many of the verification experiments:
> 
> bash-3.2$ grep -i seaice_drag */i*/data.seaice
> lab_sea/input.fd/data.seaice:  SEAICE_drag        = 0.001,
> lab_sea/input.hb87/data.seaice:  SEAICE_drag        = 0.001,
> lab_sea/input/data.seaice:  SEAICE_drag        = 0.001,
> 
> and for many of the experiments checked in under
> MITgcm_contrib, for example:
> 
> bash-3.2$ grep -i seaice_drag */i*/data.seaice */*/i*/data.seaice
> arctic40km/input/data.seaice:      SEAICE_drag       = 0.0012
> arctic40km/input_ad/data.seaice:      SEAICE_drag       = 0.0012
> high_res_cube/input/data.seaice:      SEAICE_drag        = 0.00114,
> arctic/cs_18km/input/data.seaice:      SEAICE_drag        = 0.00114,
> arctic/cs_36km/input/data.seaice:      SEAICE_drag        = 0.00114,
> arctic/cs_4km/input/data.seaice:      SEAICE_drag        = 0.00114,
> high_res_cube/cs510_28lev/input/data.seaice:      SEAICE_drag        = 0.0012,
> high_res_cube/cs510_adjoint/input_ad/data.seaice:      SEAICE_drag        = 0.0012,
> high_res_cube/cube92/input/data.seaice:      SEAICE_drag        = 0.0012,
> high_res_cube/cube92up/input/data.seaice:      SEAICE_drag        = 0.00114,
> llc_hires/llc_1080/input/data.seaice:      SEAICE_drag        = 0.00114,
> llc_hires/llc_2160/input/data.seaice:      SEAICE_drag        = 0.00114,
> llc_hires/llc_4320/input/data.seaice:      SEAICE_drag        = 0.00114,
> 
> On Jan 6, 2014, at 5:42 PM, Jean-Michel Campin wrote:
> 
>> Hi Dimitris,
>> 
>> Changing default value can be a source of problems for users.
>> 
>> But if we change all the experiments that uses pkg/seaice with the 
>> default SEAICE_drag to have explicitly: SEAICE_drag = 0.0011, 
>> it will incite new users to use a better value for this parameter.
>> It could also push old users to revisit their choice (if/when
>> they update their code and check what has changed).
>> 
>> Now if you really want to change the default, changing the seaice
>> experiment will need to be done anyway, so it could be a first good step.
>> 
>> Cheers,
>> Jean-Michel
> 
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel