How does CLM handle grids in the ocean

[KeerZ]

Hi, I downloaded surface data at /glade/p/cesm/cseg/inputdata/lnd/clm2/surfdata_map/surfdata_0.9x1.25_16pfts_simyr2000_c170428.nc and plot contour maps of tree fraction. Oddly, there are some grids with trees in the ocean (see the left figure).

Later I found that those grids in the ocean are actually not used (or not output) in my simulation. Say if I run an I2000 simulation with this surface data, the output at tree PFTs actually does not include those weird grids over the ocean (see the right figure). But the surface data do indicate that there are trees in those grids...

I wonder if there is any reason for this? My guess is that the output from CLM automatically masks out some grids beyond the major continents? I also notice this phenomenon for the lake land units. i.e. some grids in the surface data with positive lake fractions cannot be found in the lake output.

Any help is appreciated. Thank you!

 

[erik]

Likely these tree fraction points are due to grid cells that include islands that have non-zero vegetation fractions. But, in terms of what the model actually runs as land that's determined by the domain.lnd file that gives both a land mask which determines if the land is run at all (and a land fraction that shows what fraction of that grid cell is covered by land (as opposed to ocean).

I think this might be the same for lakes. The difference between what is considered "ocean" and "lake" is somewhat arbitrary. So the domain.lnd file will determine if a gridcell is considered "land" which means lake area will be run in CTSM as lake. However, if it's marked as ocean, CTSM won't run that gridcell at all.

 

[KeerZ]

Likely these tree fraction points are due to grid cells that include islands that have non-zero vegetation fractions. But, in terms of what the model actually runs as land that's determined by the domain.lnd file that gives both a land mask which determines if the land is run at all (and a land fraction that shows what fraction of that grid cell is covered by land (as opposed to ocean).

I think this might be the same for lakes. The difference between what is considered "ocean" and "lake" is somewhat arbitrary. So the domain.lnd file will determine if a gridcell is considered "land" which means lake area will be run in CTSM as lake. However, if it's marked as ocean, CTSM won't run that gridcell at all.

This makes sense. Thank you, Erik!