How to update the depth of the active soil layer in ELM?

[mpaiao]

I am running ELM-FATES for multiple sites across the Amazon and in California, using single-site configuration because I want to drive the model with in situ meteorological observations.

I recently noticed that ELM assumes a fixed soil profile with the number of layers defined by nlevgrnd=15 and going down to ~ 35 m, and a fixed number of active soil layers (nlevsoi=10). If I understood correctly, this means that roots can extract water from the top 10 layers, or the top 2.8 m (or 3.8 m based on the bottom of soil layer 10). Is this correct?

Assuming it is, then this depth is somewhat shallower the typical rooting zone in many parts of the Amazon, and also deeper than the sites in the Sierra Nevada. I tried to reset nlevsoi=15 by editing the surface file for a site in the Amazon with very deep soils, however the model did not run:

 check_dim ERROR: mismatch of input dimension           15  with expected value 
          10  for variable nlevsoi
 ERROR: 
 ERROR in /path/to/E3SM/components/elm/src/main/ncdio_pio.F90.
 in at line 363

I checked the sub-routine surfrd_special, and it seems it expects the surface file to have 10 soil layers (as defined by nlevsoifl). I see how I could manually change these defaults and the soil layers, though this is not ideal because it would require hardcoding the changes for each site. Is there a way to reset these soil depth layers settings when I create and set the case? If not, I can try to make a branch that reads in optional fields from the surface file and resets the soil layer settings.

If I did not correctly understand the way soils and the rooting zone are defined, any input or guidance would be much appreciated too, thanks!

[LianGong98]

Hi @mpaiao I’ve come across the same issue. I attempted to define the soil depth manually but wasn’t successful. Have you been able to resolve it?

[mpaiao]

Hi @LianGong98, the depth to bedrock can be defined through the surface file, you can add a variable aveDTB to your surface containing the depth to bedrock, then the code will use that to define the active layers.

In addition, my pull request #7319 should allow changing the soil layer depths, also through the surface file. For this you would need to specify the maximum number of active soil layers by updating nlevsoi, setting the layer mid points with variable ZSOI, and making sure all variables that depend on nlevsoi have the right dimensions (namely PCT_SAND, PCT_CLAY and ORGANIC, maybe a few others that are optional).

[LianGong98]

Oh, that's awesome! thank you so much for your help!

[LianGong98]

Hi @mpaiao
I'm trying to add the variable aveDTB to the surface dataset to represent spatial variability in depth to bedrock. However, I found that simply modifying aveDTB does not affect the number of active soil layers in the model.

To enable this functionality, I checked the code and set use_var_soil_thick = .true. in the configuration. However, when I run the model with FATES enabled, I encounter an error stating that use_fates and use_var_soil_thick cannot be used simultaneously.

Therefore, besides applying pull request #7319, are there any other recommended ways to address this problem?

[mpaiao]

Thanks for the feedback @LianGong98. I mostly tested it with single-point simulations, so I wonder if this is what is causing the differences. I was not aware of the lack of compatibility between variable soil thickness and FATES, I am tagging @ckoven @rgknox and @glemieux, maybe they know a bit more about this.