Issue when running single-point Noah-MP with 7 soil layers

[mcthreems]

I attempted to run a 7-layer simulation of the single-point Noah-MP by adding 3 additional 1m layers below the default 4 layers (for a total soil column depth of 5m). I could get the input forcing data created fine, but when running the hrldas.exe it encountered an error:
At line 192 of file EnergyVarInTransferMod.f90
Fortran runtime error: Index '-9999' of dimension 1 of array 'noahmpio%quartz_table' below lower bound of 1

I have not encountered this error previously and it does not seem clearly connected to the added soil layers, so I am wondering if anyone else has seen something similar. I have attached the input and setup files I used to attempt this simulation.

model_files.zip

[cenlinhe]

You also need to modify this line of code since you are using soil_data_option = 2 (soil types vary with depth): https://github.com/NCAR/noahmp/blob/9fbe672420f4468e7d78aecfd10626927ecc8156/drivers/hrldas/ConfigVarInTransferMod.F90#L142-L146
Please include the following two lines:
noahmp%config%domain%SoilType(5) = nint(NoahmpIO%SOILCL5(I,J)) ! soil type in layer5
noahmp%config%domain%SoilType(6) = nint(NoahmpIO%SOILCL6(I,J)) ! soil type in layer6
noahmp%config%domain%SoilType(7) = nint(NoahmpIO%SOILCL7(I,J)) ! soil type in layer7

[mcthreems]

So to make sure I understand, if I have specified in the source code 7 soil layers, but then only define 6 layers in my model setup files when running a simulation, will the model crash or have various issues?

[cenlinhe]

the model will crash, because the missing one layer will have a value of -9999.

[mcthreems]

OK, good to know.

[mcthreems]

You also need to modify this line of code since you are using soil_data_option = 2 (soil types vary with depth): https://github.com/NCAR/noahmp/blob/9fbe672420f4468e7d78aecfd10626927ecc8156/drivers/hrldas/ConfigVarInTransferMod.F90#L142-L146 Please include the following two lines: noahmp%config%domain%SoilType(5) = nint(NoahmpIO%SOILCL5(I,J)) ! soil type in layer5 noahmp%config%domain%SoilType(6) = nint(NoahmpIO%SOILCL6(I,J)) ! soil type in layer6 noahmp%config%domain%SoilType(7) = nint(NoahmpIO%SOILCL7(I,J)) ! soil type in layer7

In addition to this edit it seems like I will need to also edit the initialization of these SOILCL variables, which are in a different file. I believe the file in question is "/hrldas/noahmp/drivers/hrldas/NoahmpIOVarInitMod.F90", with edits needed at both lines 81 and 895 to add the additional soil layer variables. I can update this if I run into any additional errors when trying to recompile.

good catch! Yes, you will also need to change the SOILCL initialization code.

I made this additional change but the model still seems to need some other initialization to compile. Do you know where else in the code I would need to add these additional soil layers?

[cenlinhe]

what is your error message?

[cenlinhe]

You also need to add SOILCL5, SOILCL6, SOILCL7 to this subroutine following SOILCL1~4:

hrldas/hrldas/IO_code/module_hrldas_netcdf_io.F

Line 1460 in 0dc57ea

subroutine read_soil_texture(spatial_filename,xstart, xend,ystart, yend, &

and modify this subroutine call interface accordingly:

hrldas/hrldas/IO_code/module_NoahMP_hrldas_driver.F

Lines 233 to 235 in 0dc57ea

	call READ_SOIL_TEXTURE(NoahmpIO%HRLDAS_SETUP_FILE, NoahmpIO%XSTART, NoahmpIO%XEND, NoahmpIO%YSTART, NoahmpIO%YEND, &
	NoahmpIO%NSOIL, NoahmpIO%IVGTYP, NoahmpIO%SOILCL1, NoahmpIO%SOILCL2, NoahmpIO%SOILCL3, &
	NoahmpIO%SOILCL4, NoahmpIO%ISICE, NoahmpIO%ISWATER)

[mcthreems]

This is the header from the .dat file

!----------------------------------------------------------------------------------------------------------------------
!Location Information
!----------------------------------------------------------------------------------------------------------------------
&location
latitude = 40.0 ! latitude [degrees]
longitude = -117.0 ! longitude [degrees]
vegetation_category = 10 ! category based on land_cover_source
soil_category = 6 ! soil texture class
deep_soil_temperature = 279.67 ! deep soil temperature [K]
elevation = 1791.7598 ! elevation [m]
sea_ice = 0.0 ! sea ice fraction [-] (should be 0.0 for a land point)
maximum_vegetation_pct = 20.43 ! maximum annual vegetation percentage [0-100]
minimum_vegetation_pct = 2.37 ! minimum annual vegetation percentage [0-100]
land_mask = 1 ! land mask [-] (should be 1 for a land point)
nsoil = 7 ! number of soil layers
/
!----------------------------------------------------------------------------------------------------------------------
!Initial States
!----------------------------------------------------------------------------------------------------------------------
&initial
snow_depth = 0.01 ! snow depth [m]
snow_water_equivalent = 1.0 ! snow water equivalent [mm]
canopy_water = 0.01 ! canopy water [mm]
skin_temperature = 263.7 ! skin temperature [K]
soil_layer_thickness = 0.1, 0.3, 0.6, 1.0, 1.0, 1.0, 1.0 ! layer thicknesses [m]
soil_layer_nodes = 0.05, 0.25, 0.7, 1.5, 2.5, 3.5, 4.5 ! layer centroids from surface [m]
soil_temperature = 266.1, 274.0, 276.9, 279.9, 279.9, 279.9, 279.9 ! layer soil temperature [K]
soil_moisture = 0.298, 0.294, 0.271, 0.307, 0.307, 0.307, 0.307 ! layer volumetric total water content [m3/m3]
leaf_area_index = 2.0 ! leaf area index [m2/m2] used to initialize carbon budgets
soilcl1 = 6
soilcl2 = 6
soilcl3 = 6
soilcl4 = 6
soilcl5 = 6
soilcl6 = 6
soilcl7 = 6
/
!----------------------------------------------------------------------------------------------------------------------
!Metadata
!----------------------------------------------------------------------------------------------------------------------
&metadata
grid_id = 1 ! used for grid labeling
water_classification = 17 ! water type in land classification (17 for MODIS; 16 for USGS)
urban_classification = 13 ! urban type in land classification (13 for MODIS; 1 for USGS)
ice_classification = 15 ! snow/ice type in land classification (15 for MODIS; 24 for USGS)
land_cover_source = "MODIFIED_IGBP_MODIS_NOAH" ! land classification (USGS or MODIFIED_IGBP_MODIS_NOAH)
/
!----------------------------------------------------------------------------------------------------------------------
!Conversion information
!----------------------------------------------------------------------------------------------------------------------
&conversion
have_relative_humidity = .false. ! set to true if you need to convert relative to specific
temperature_offset = 0.0 ! 273.15 to convert C to K
temperature_scale = 1.0 ! can be used to convert F to C
pressure_scale = 100.0 ! 100.0 to convert mb to Pa
precipitation_scale = 0.000278 ! 0.014111 to convert inches/30min to mm/s
/

[cenlinhe]

This header looks fine to me. Which version of HRLDAS are you using? I did not find "soilcl*" in our latest HRLDAS version .dat sample file in the GitHub. Looks like this variable is not used at all in the single point processing script. What you need to do is to either modify the script to include the read-in capability of SOILCL variable: https://github.com/NCAR/hrldas/blob/master/hrldas/HRLDAS_forcing/run/examples/single_point/create_point_data.f90
Or you can directly use nco command to add SOILCL1~7 variables to your hrldas_setup.nc file.

[mcthreems]

I should be using the most recent version since I just reinstalled it from a git command for this new compilation. I did edit the create_point_data.f90 file for this run, and have attached that edited version here. I believe this should allow for those variables to be read in, and have previously used this approach to read in variable soil texture runs for 4-layer simulations without issue.
loop.f90.txt

[cenlinhe]

ok, I think the issue is that you cannot comment out the "count = (1,1,4,1) ..." in these two lines:
https://github.com/NCAR/hrldas/blob/master/hrldas/HRLDAS_forcing/run/examples/single_point/create_point_data.f90#L241-L242
you need to keep it and modify the "4" to "7".

[mcthreems]

Ah ok I think this was it. I probably commented that bit out previously thinking I'd get back to it then forgot, thanks for checking these. Currently seems like it runs and produces real outputs for the soil variables.