Where is the info of pft dimension in clm restart file coming from

[Xueli Huo]

Hi,

I am looking at a clm restart file for global simulation. There is the dimension of pft in the restart file.
I can not find any pft dimension info in the input data, i.e., the domain and surface datasets. So I am wondering where the dimension of pft as well as the related pft location variables such as pfts1d_lon/lat, pfts1d_ixy/jxy,pfts1d_gridcell/column/landunit/_index are from. Is there any source file that defines such info ?

Same question for the dimension of column and landunit.

Sincerely,

 

[Xueli Huo]

Another question is the indexing of different plant function types in the same gridcell in the pft dimension.

Suppose that there are three kinds of pfts, broadleaf evergreen tropical tree,broadleaf deciduous tropical tree and shrub, in one grid cell. Say the index assigned is 156731, 156732, 156733. T_VEG24_VALUE(156731), T_VEG24_VALUE(156732), and T_VEG24_VALUE(156733) are the corresponding values of 24hr average of vegetation temperature for them. The question is how the index is determined in the pft dimension. I.e., why 156731 is assigned to broadleaf evergreen tropical tree other than shrub or broadleaf deciduous tropical tree ?

Sincerely,

 

[Xueli Huo]

Is the dimension of pft in the restart file related to the percentage of plant/crop function types in the input surface dataset ?

Is there any requirement for any consistency between restart and the input surface dataset or will the info read from restart file interact with the info from surface dataset and if so, how ?

Sincerely,

 

[oleson]

The pft dimension in the restart file relates to all of the pfts in the surface dataset that were used to generate the restart file, it's just in a 1d array, instead of 2d or 3d as in the surface dataset. Similarly, the column and landunit dimension refers to the columns and landunits in the surface dataset, in 1d form.
If you do an ncdump of pfts1d_itypveg in the restart file you will see something like this (for a restart file from a transient case):

pfts1d_itypveg = 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16,
17, 18, 19, 20, 21, 22, 23, 24, 41, 42, 61, 62, 67, 68, 71, 72, 73, 74,
75, 76, 77, 78, 0, 0, 1,......

So the pft index you are referring to is ordered by pft number, e.g., 0 is bare soil, 1 is Needleleaf evergreen tree – temperate, etc.
pfts1d_itypveg(1) = 0, pfts1d_itypveg(2) = 1, etc.
The ordering is also by latitude and longitude, starting with the lower left corner of the surface dataset, at least in the restart file I'm looking at:

pfts1d_lon = 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2.5, 2.5,.....

pfts1d_lat = -90, -90, -90, -90, -90, -90, -90, -90, -90, -90, -90, -90,
-90, -90, -90, -90, -90, -90, -90, -90, -90, -90, -90, -90, -90,

There is a consistency check between the restart file and the surface dataset in subgridRestMod.F90 in subroutine check_weights.

For a non-transient case, e.g., an 1850 case, you won't see an ordering of all the possible pfts. In a transient case, all pfts could potentially exist in a gridcell, and so the 1d array is much larger to account for the fact that a pft may "appear" in a gridcell at a certain time in the transient run.

For an 1850 restart file, you might see something like this:

pfts1d_itypveg =
0, 1, 5, 7, 8, 9, 10, 11, 12, 13, 15, 17, 19, 23, 41, 61, ...

Now, only the pfts that exist in the 1850 surface dataset are stored (bare soil, Broadleaf evergreen tree – temperate, Broadleaf deciduous tree – temperate, etc.)