CLM5: Are the 10 layers in the surface dataset the same as the top 10 layers of 20SL_8.5m?

[kylin]

Hi all,

I have a question about how the soil layer depths defined in the CLM5 technical note relate to the soil layers in the surface dataset.

In the CLM5 Technical Note (Section 2.2.2.1), there is a table that defines the vertical soil structure (node depths, layer thicknesses, and interface depths), which I understand corresponds to the default soil_layerstruct (e.g., 20SL_8.5m).

At the same time, the surface dataset (e.g., PCT_SAND, PCT_CLAY) typically provides soil properties with nlevsoi = 10 layers.

From previous discussions, I understand that during initialization, these 10 layers are mapped onto the upper 10 layers of the model soil structure, and deeper layers (11–20) reuse the 10th layer properties.

My confusion is about the layer-depth relationship:

Is the table in Section 2.2.2.1 (soil layer depths)

• directly defining the physical depth structure used internally by the model (soil_layerstruct),
• or is it also intended to define the depth ranges corresponding to the 10 layers in the surface dataset?

In other words:

• Do the 10 layers in the surface dataset correspond directly to the top 10 layers in the 20SL_8.5m structure (as defined in the table)?
• Or are the 10 surface dataset layers simply index-based, without a strictly defined physical depth, and only mapped to the model layers during initialization?

Finally, when preparing a custom surface dataset:
Should soil properties be interpolated according to the depth boundaries given in the Technical Note (Section 2.2.2.1) to match the model soil layers?

Thanks very much for your help!

 

[oleson]

The 10 surface dataset layers are index-based (there are no soil layer structure variables on the surface dataset) and they correspond to the top 10 layers in the 20SL_8.5m structure and mapped during initialization.
If you have soil properties, they should be mapped to the 20SL_8.5m layer structure if feasible.