Issue Implementing More/Deeper Soil Layers in LIS

[mcthreems]

We are trying to run LIS with the Noah-MP 4.0.1 surface scheme using more/deeper soil layers than the standard 4 with a 2m total depth. Going to 5 layers with 4m total depth seems to work fine, but adding one more layer causes the model to crash due to water loss. Is there an established way to add more soil layers to the model that can avoid this issue? I have attached the lis configuration files for both the 5-layer and 6-layer cases for reference.
lis.config_noahmp401_snake_sl5.txt
lis.config_noahmp401_snake_sl6.txt

[dmocko]

Hi @mcthreems

Thanks for sharing your lis.config files.

I notice that you have set the initial soil moisture of the lower layers to 0.00. Why haven't you set these to 0.20?

I suspect that setting these values to 0.00 is what's causing your model crash.

Furthermore, the Noah-MP-4.0.1 runoff and groundwater option "1" for SIMGM groundwater assumes that the total soil depth
is 2.0-meters, with the groundwater below that. It's hard-coded this way in the original model physics. Thus, it is recommended
that if you change the total soil depth that you change this option to "3" and not run the model with groundwater. If you do this,
the initial water table depth value won't matter, as it's not calculated with this setting.

[dmocko]

Thanks for the update. I'm not sure what's going on. Can you please post the file with the water loss error? And your lislog.0000 file? Please look in all of your lislog files for any warning/error messages, especially at the end of the file when the crash occurs. Finally, can you please share your MPTABLE_NROOT6.TBL file?

[mcthreems]

Sure, I've attached the log file and MPTABLE files, as well as the terminal output from the attempted run. The only change we made to the MPTABLE file was making the rooting depth equal to the number of soil layers for all plant types that had a non-zero rooting depth in the default file.
gw3_out.txt
lislog_noahmp401_snake_sl6_gw3.0000.txt
MPTABLE_NROOT6.TBL.txt

[dmocko]

Thanks. These lines in your gw3_out.txt file are concerning:
problem with initial snow fields: snow/snowh>0 while snowh/snow=0 at i,j 1 1 288.000000 0.00000000

Your lis.config file has initial snow as zero, so this shouldn't be popping up.

What I think is happening is due to the way Noah-MP calculates snow/soil layers. The LSM puts values (for example, of temperature) into a single array with dimensions from -2 to 4 (for a standard 4-layer soil scheme). Thus layer "-2" is the top soil layer, "-1" is the middle soil layer, "0" is the bottom soil layer, and then "1" through "4" are the 4 soil layers from top to bottom.

The code SHOULD automatically change this when "NSOIL" is changed that way that you have in your lis.config file. But looking through the code, I think I have found an issue:
https://github.com/NASA-LIS/LISF/blob/master/lis/surfacemodels/land/noahmp.4.0.1/phys/module_sf_noahmplsm_401.F90#L179

This line hard-codes NSOIL to 4, and only dimensions the soil parameter values to 4. Can you please try changing this hard-coded value to 6, re-compile, and then re-run?

Obviously, if you change the number of layers, you'll have to change the hard-coded value. We'll look for a way for the code to only use the NSOIL value everywhere in the code to stamp out this bug.

[mcthreems]

I tried out replacing the hard coded value with 6 as suggested and rerunning, but it seems to have produced the same error. I've attached the terminal output and log file again for the new attempt.
gw3_out-2.txt
lislog_noahmp401_snake_sl6_gw3.0000.txt

[dmocko]

Hi @mcthreems

Apologies for the slow response. I have been hoping to look further into this, but haven't had the time. It will take me looking
deeper into the code to see what might be occurring. Honestly, I haven't tried running Noah-MP with more soil layers before,
and it's possible that this capability was never fully functional. I hope to get to it soon.