3). By default, the output written fields are all 1D-fields (there is no longer the dimension PATCH in output files).

In agreement with the users, all written fields are now 1D, that is to say that the dimension "PATCH" doesn’t exist anymore in output files.

Consequently:

• The names of output fields defined by patches are changed:
  • with only one patch, the suffix P1 is added. Ex: TG1 => TG1P1
  • with several patches, the suffix Pk (with k = number of the current patch to be written). Ex: with 12 patches, TG1 => TG1P1, TG1P2, TG1P3, TG1P4, TG1P5, TG1P6, TG1P7, TG1P8, TG1P9, TG1P10, TG1P11, TG1P12.
• When a field name was already more than 9 characters, to stay in the limit of 12 characters for the name of a field, we are forced to change the initial name.
• To read and write the snow fields, the subroutines read_gr_snow.F90 and write_gr_snow.F90 are now called for each patch, what changes the order of writing of snow fields in the output files.

The new key LSPLIT_PATCH is added in NAM_WRITE_SURF_ATM. It is .TRUE. by default, setting .FALSE. it writes output fields 2D with the dimension PATCH, like before.