Implement NOAA/ARL FENGSHA dust scheme

[rmontuoro]

This PR implements the FENGSHA dust scheme as an option for the GOCART dust component.

The FENGSHA scheme, developed at NOAA/ARL, is added to the GOCART process library along with few auxiliary procedures. It requires the additional input fields below, provided as surface map files via ExtData:

# dust component
FRCLAY                '1'  Y E - none none clayfrac     ExtData/Dust/FENGSHA_DUST_INPUTS_v0.6.nc
FRSAND                '1'  Y E - none none sandfrac     ExtData/Dust/FENGSHA_DUST_INPUTS_v0.6.nc
FRSILT                '1'  Y E - none none siltfrac     ExtData/Dust/FENGSHA_DUST_INPUTS_v0.6.nc
SSM                   '1'  Y E - none none ssm          ExtData/Dust/FENGSHA_DUST_INPUTS_v0.6.nc
RDRAG                 '1'  Y E - none none drag_part    ExtData/Dust/FENGSHA_DUST_INPUTS_v0.6.nc
UTHRES                '1'  Y E - none none uthres       ExtData/Dust/FENGSHA_DUST_INPUTS_v0.6.nc

Note the E option for nearest-neighbor interpolation enabled by MAPL PR #788. Until this PR has been merged, the scheme can work (not ideally) with other interpolation methods, too.

FENGSHA also requires the liquid water content (SLC) in the soil top layer, as well as the surface snow fraction (FRSNOW). All required fields fields have been added to DU2G_StateSpecs.rc:

FRSNOW     | 1      | xy   | N   |      | surface_snow_area_fraction
SLC        | 1      | xy   | N   |      | liquid_water_content_of_soil_layer
FRCLAY     | 1      | xy   | N   |      | volume_fraction_of_clay_in_soil
FRSAND     | 1      | xy   | N   |      | volume_fraction_of_sand_in_soil
FRSILT     | 1      | xy   | N   |      | volume_fraction_of_silt_in_soil
RDRAG      | m-1    | xy   | N   |      | drag_partition
SSM        | 1      | xy   | N   |      | sediment_supply_map
UTHRES     | m s-1  | xy   | N   |      | surface_dry_threshold_velocity

Additional input parameters have also been added to the dust component's resource file (DU2G_GridComp_DU.rc):

• emission_scheme - an integer selector for the original GOCART (1) or the FENGSHA (2) dust scheme
• alpha, gamma - FENGSHA scaling parameters (size-independent)
• vertical_to_horizontal_flux_ratio_limit - Maximum vertical-to-horizontal dust flux ratio, according to: B. Marticorena, G. Bergametti, J. Geophys. Res., 100(D8), 16415–16430, 1995.

# Emissions methods
emission_scheme: 2       # 1 for GOCART, 2 for FENGSHA

# FENGSHA settings
alpha: 0.3
gamma: 1.3
vertical_to_horizontal_flux_ratio_limit: 2.e-04

All FENGSHA-related parameters are read only if emission_scheme is 2.

Note also that no attempt is currently made to verify all pointers used by FENGSHA are actually allocated. I would like to have a general discussion to identify a common strategy to implement such sanity checks in GOCART, since this may affect other methods, too.

This PR was tested within UFS-GOCART on the NOAA RDHPCS Hera platform, and the generated dust distributions verified by NOAA/ARL.

[tclune]

Doesn't this impose a requirement on users of the default DU scheme to provide these variables? I think an additional parameter is needed in the table to make these conditional.

[rmontuoro]

Conditional import variables have been implemented with commit 261a4ac

[tclune]

This is probably a good opportunity to use newer syntax that avoids counting by the end user:

character(len=*), dimension(*), parameter :: importFieldNames =
   ["inst_pres_interface         ", &
    "inst_sensi_heat_flx          ", 
...

[rmontuoro]

Commit e8425cd incorporates the suggested syntax. Thank you!

[adarmenov]

@rmontuoro Volume/mass fractions (the 'distribution' variable) can be computed once. Can't we afford more precise integration?

[rmontuoro]

@adarmenov - Not sure I understand your question. Could you please clarify?

[adarmenov]

@rmontuoro Let me elaborate... lines 156-163 are a fast and likely imprecise way to numerically integrate over the size distribution. I was pointing out that the integration needs to be done once, hence one can try to be numerically more precise, e.g. use more than 5 sub-bins to approximate the integral.

[rmontuoro]

@adarmenov - Thank you for the clarification, I see your point. I will bring this up with NOAA/ARL, where the scheme was developed. For the time being, however, I don't expect any changes since we would like to compare the performance of the FENGSHA scheme in GOCART with the operational version, which implements this integration technique.