Version 6 (modified by ctlod, 12 years ago) (diff)

Last edited Timestamp?

Sensitivity experiments performed with LIM 3.0


 background-white, blue 


Exp: LIM3_30

# Based on LIM3_29 one BUT:

nn_ico_cpl = 0 & nn_fsbc = 5 & increase iteration number for the rheology nevp=600 instead of 300

Results: stops when reaching 10th year in March in the North Canadian archipelago in the sea-ice accumlulation
area ixj=66,146 i.e. always in the same area for LIM 3.0

relaunch from scratch with nevp=900

Results: idem as above resulting with no changes in results

new attempt with coupling frequency ocean/sea-ice nn_fsbc set to 3 instead 5 and with nn_ico_cpl = 1

i.e. computation of the wind stress each time step but with a wind module computed each nn_fsbc only
Results: stops after just 9 years run with a negative salinity in the Canadian North


Exp: LIM3_29

# Updated with the revision number r1170 as for the CLIO_LIM2 configuration:[BR]]

  • last modification concerning sbcfwb.F90 with the sign of the term empold which must be +
  • implementation of modifications related to exp. PHC_29 for comparison: daily SSS damping
  • implementation of the Kz tides which was missing, i.e. zdftmx.F90 zdf_oce.F90, step.F90
  • implementation of dommsk.F90, diafwb.F90 et diawri.F90 modifications

This run will be directly comparable with the one based on LIM 2.0 PHC_29
Results: stops after 20 years run with salinity of -4 PSU in the Canadian North

23/12/2008 :

relaunch with this time nn_ico_cpl = 1

Results: not sufficient, stops after 11 years

29/12/2008 :

Results: stops again after relaunching after 19 years with negative salinity O(-104) PSU in i,j=65,146
whereas the initial state looks like normal, it seems to be related with an instability which grows in winter

relaunch with nn_fsbc = 1 instead 5 i.e. coupling every time step with nn_ico_cpl = 0

Results: it seems to work fine but the cost is huge, ~250 years performed



# Based on exp. USE_PHC_10:

  • take into account bug corrections in:
    • sbcice_lim.F90
    • sbcmod.F90
    • limsbc.F90
    • sbcssm.F90
  • coupling method is different and better handled. Keep case (0) correcting 4 bugs:
    • module mean is now under root square for the 3 options
    • add quotes in the leads fraction mean in option (1)
    • use mean values ss[uv]_m based on [uv]b instead [uv]n