Enhanced initialization for LIM3

[vancop]

Current initialization for LIM3 generates negative concentrations, which are filtered out after one time step; but generate alerts in the code that can be misinterpreted.

A cleaner routine has now been implemented for v3_1.

Based on namelist values of initial ice concentration and thickness, the new initialization computes values for the thickness and area for each category. A gaussian distribution that respects the following principles is assumed:
1) volume and area are conserved
2) thickness of categories has to be within the prescribed bounds (hi_max's)
3) concentration of ice in each category has to be positive

The documentation is attached.
Any question to be addressed to Martin!

namelist_ice has to be changed because some of the initialization-related variables in the former code are useless (deformed ice thickness and area).

Best wishes to everyone,
Martin

[vancop]

routine

[vancop]

namelist_ice

[vancop]

doc

[barthele]

The new version of limistate.F90 proposed here by Martin is not perfect yet, because :

1) the initial ice thickness provided seems to be (initial thickness in namelist) * (initial concentration in namelist) ;
2) it should be cleaned to avoid disturbing warnings that appear even if the routine does its job as expected.

But, it is already way better than the current version, in which the actual initial thickness is totally different from what it should be.

Hence, in my opinion, it's a step in the right direction towards a reliable initialization of ice in LIM3.

[clem]

Fixed in branch: dev_r4028_CNRS_LIM3 to be included in the next merge party (Nov. 2013)