MERRA-2 BLM2 soil parameters

[egibney]

Hello, I'm a researcher working on calculating H2 deposition velocity and my model requires certain soil parameters used internally within MERRA-2, but seemingly difficult to find. Specifically I'm looking for the "BLM2" soil parameters (Brooks-Corey b and soil saturated matric potential ψₛ) as described in De Lannoy et al. 2014. Are these parameters available anywhere to download? Alternatively, is there a way to reconstruct them? I can access several MERRA-2 soil variables (e.g., porosity, wetness, etc.) through the public data pages but cannot find these specific parameters I'm looking for, even though I know they must be internally used within the Catchment model.

[mbosilov]

First, to be sure, have you gone through the MERRA-2 Documentation www?
https://gmao.gsfc.nasa.gov/gmao-products/merra-2/documentation_merra-2/

Then, Have you looked at the MERRA-2 Land Constants data file?
https://search.earthdata.nasa.gov/search/granules?p=C1278787757-GES_DISC&pg[0][v]=f&pg[0][gsk]=-start_date&q=MERRA-2%20Land%20constants

[egibney]

Thanks for the reply. Yes, I've read through the documentation in detail and have also checked the Land Constants file. Unfortunately these specific parameters are not available there.

[reichle]

For the record, the following email response was sent to the user in response to their direct email inquiry.

Hi Evan,

Here are the values for the soil saturated matric potential (lpsis), the b exponent (lbee), and the porosity (lporo) associated with the each of the 12 USDA texture classes of Fig 1a of De Lannoy et al (2014):

data lbee /3.30, 3.80, 4.34, 5.25, 3.63, 5.96, 7.32, &
8.41, 8.34, 9.70, 10.78, 12.93/
data lpsis /-0.05, -0.07, -0.16, -0.65, -0.84, -0.24, &
-0.12, -0.63, -0.28, -0.12, -0.58, -0.27/
data lporo /0.373, 0.386, 0.419, 0.476, 0.471, 0.437, &
0.412, 0.478, 0.447, 0.415, 0.478, 0.450/

This is a straight cut-and-paste from the GEOS Fortran code here:

https://github.com/GEOS-ESM/GEOSgcm_GridComp/blob/065ae1e611a4d309e9774a42944952e207773e3c/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/Utils/Raster/makebcs/rmTinyCatchParaMod.F90#L563

The linked version of the code is from the most recent release of the component repository in GitHub. The actual code used in MERRA-2 production is much older than our transition to GitHub and not easily accessible, but the subroutine with the relevant data statements is exactly what was used for MERRA-2. (For background, this subroutine is part of a workflow that creates spatially distributed soil parameters in model tile space during pre-processing.)

I should note that there are additional data statements for the saturated hydraulic conductivity (lcond) and the wilting point (lwpwet). Note that these two variables are used in MERRA-2 only after the following transformations:
cond=lcond(soil_gswp)/exp(-1.*zks*gnu) ! zks=2, gnu=1. In MERRA-2
wpwet=lwpwet(soil_gswp)/lporo(soil_gswp)

With the above information and information about the soil class, you could create global maps of the parameter values. But soil class is defined on the model's (sub-grid) "tile" space. For an illustration, see Appendix E of the File Specification Document from the original ("version 1") MERRA product
https://gmao.gsfc.nasa.gov/publications/office_notes/ (Lucchesi, R., 2012: File Specification for MERRA Products. GMAO Office Note No. 1 (Version 2.3), 82 pp)

The tile vs. grid space aspect is also why we only provide porosity, but not bee or psis, in the MERRA-2 land constants Collection (https://cmr.earthdata.nasa.gov/search/concepts/C1278787757-GES_DISC.html). Porosity can be averaged from tiles to grid cells, but such averaging does not make sense for bee and psis. (A further complication is that the MERRA-2 lat/lon output grid is different from the MERRA-2 computational "cube-sphere" grid. The illustration of the model tiles in App E of the MERRA file specs only illustrates the tiles within the computational grid cells, which were lat/lon in the original ("version 1") MERRA system.)

I suspect you used the gridded porosity to infer the (lat/lon) gridded bee and psis values in the REV version of the soil parameters. This makes sense only where the soil class is homogeneous within a (lat/lon) output grid cell. Where there are two or more soil classes within a grid cell, this inference is not technically possible and introduces errors in the bee and psis maps.

I could share the location and soil class information about the model tiles if that's of interest.

I hope this helps, Rolf