Computation of the mean z0 over a grid cell

[nvuilsce]

When working with different value of z0 for momentum and heat (z0m and z0h), there is an error in the computation of the mean z0h. The method used is to average the drag coefficients (and not the z0 values or their log) over the grid cell to deduce the mean z0m and mean z0h.
The drag coefficient for momentum equals :
dragm = ln(zref/z0m)*ln(zref/z0m)
The drag coefficient for heat equals :
dragh = ln(zref/z0m)*ln(zref/z0h)

The error was that dragh was computed as ln(zref/z0h)*ln(zref/z0h).

The correction of this error leads to the same computation of z0m but lower values for z0h.
Test simulations have been performed in force (fG3) and coupled (CL4) mode.
FG3 simulation:
outputs are stored here: /ccc/store/cont003/dsm/p529vui/IGCM_OUT/OL2/PROD/ref3977/FG3WFDEIv1.comb.rsolcor.ws.testz0h.bis
To be compared to /ccc/store/cont003/dsm/p529vui/IGCM_OUT/OL2/PROD/ref3977/FG3WFDEIv1.comb.rsolcor.ws
Diagnostics of the differences between these 2 versions are accessible here: ​https://vesg.ipsl.upmc.fr/thredds/fileServer/work/p529vui/OL2/DEVT/Eval-ORCHIDEE/r3977/FG3/FG3WFDEIv1.comb.rsolcor.ws.testz0h.bis_VS_FG3WFDEIv1.comb.rsolcor.ws/index.html

CL4 simulations:
outputs are stored here: /ccc/store/cont003/gencmip6/p86ghatt/IGCM_OUT/LMDZOR/DEVT/CM609climpdcontrl/CM609.m12b/
To be compared to /ccc/store/cont003/gencmip6/p86ghatt/IGCM_OUT/LMDZOR/DEVT/CM609climpdcontrl/CM609.refb
Diagnostics of the differences between these 2 versions are accessible here: ​https://vesg.ipsl.upmc.fr/thredds/fileServer/work/p529vui/OL2/DEVT/Eval-ORCHIDEE/CM609climpdcontrol/CM609.m12b_VS_CM609.refb/index.html

Change in annual mean of z0h with and without the correction is attached to the ticket (z0h_wmodif/z0h_womodif, ratio_z0h.gif).
Difference in the annual mean kB-1 value is also attached.

[nvuilsce]

ratio z0h_modif/z0h

[nvuilsce]

diff kB-1

[nvuilsce]

ORCHIDEE trunk has been modified in revision r4432.