"Tile" structure in the analyzed soil moisture field

[tais958]

Hi, I have done a few experiments with Noah-MP model which assimilates SMAP soil moisture data. I found that there is usually
"tiles" shown in the 2D map of analyzed soil moisture and suspect it may be due to the coarser grids of SMAP data compared to my model grids. Wondering if anyone has noticed this issue and has came up with a good way to mitigate it? For example, maybe there's a better treatment for soil texture input or option for a soother transition between analyzing grids. Thanks in advance!

[dmocko]

HI @tais958

What version of Noah-MP are you using here?

We have seen this behavior in the regular Noah LSM, where the "slope type" parameter can give a blocky pattern to the soil moisture, particularly for the lower soil moisture levels. This behavior is from the 1-degree "slope type" source dataset. To prevent this, we recommend that users set the "fixed slope type:" value in the lis.config for Noah - typically a value of "1" is used.

In Noah-MP, this "slope type" is hard-coded to a value of "1" for all grid points, as written by the Noah-MP LSM developers.

So, that's likely not the source of what you are seeing. Do you see this more in the upper soil layers, or in lower layers as well?

If you want to use a more modern soil texture input, you can try the ISRIC soil texture data in LDT.

[dmocko]

I have not personally seen this. @sujayvkumar and/or @LIS-navari - do you have any idea what may be causing this pattern, and how to mitigate it?

@tais958 - can you share a sample image, and also share your SMAP and LSM attribs/pertattribs files?

[LIS-navari]

I have not run a 1km test case.
Could you please share your LIS parameter file and a sample of the tile pattern in the output?

[tais958]

Thanks, @dmocko and @LIS-navari ! Please see the example figures and LIS config file as attached. Please let me know if more info is needed for you to diagnose.

lis.config_noahmp401_smapda_9km.large.txt

[dmocko]

Thanks! Please also share these files:

Forcing attributes file:                  ./DA_INPUT/forcing_attribs.txt
Forcing perturbation attributes file:     ./DA_INPUT/forcing_pert_attribs.txt
State attributes file:                    ./DA_INPUT/noahmp_sm_attribs.txt
State perturbation attributes file:       ./DA_INPUT/noahmp_sm_pertattribs.txt
Observation attributes file:              ./DA_INPUT/smap_attribs.txt
Observation perturbation attributes file: ./DA_INPUT/smap_pertattribs.txt

[tais958]

Thank you. Please find the files as in attached zip file.
attribs.zip

[LIS-navari]

Hi @tais958

I do see a similar pattern in the OL simulation.
Could you please share your parameter file (the netCDF file from the LDT run)?

Thanks

[tais958]

Thanks @LIS-navari! Please check the attached parameter file.

ldt.config.noahmp401_params.1km.txt

[LIS-navari]

Hi @tais958

Thanks. I meant the output of LDT after you run this config file. It should be a NetCDF file.

Thanks

[tais958]

Hi @LIS-navari. Sorry for the confusion. I tried to upload but it did not go through. Maybe due to the excessive size. Here's the NetCDF file that was the output from LDT run from my google drive. Please let me know if you can access. Thanks!

Link:
https://drive.google.com/file/d/1Mx_CpoFlIoQABDNDjza2nBBI7pDACkmg/view?usp=sharing

[LIS-navari]

Hi @tais958, @dmocko

The parameter file looks fine.
How did you downscale the forcing data?
Maybe this feature is coming from the forcing data.

[tais958]

Hi @LIS-navari, below are the relevant options I used when running LIS:

Number of met forcing sources: 2
Blending method for forcings: "overlay"
Met forcing sources: "GDAS" "NLDAS2"
Topographic correction method (met forcing): "none" "none"
Enable spatial downscaling of precipitation: 0
Spatial interpolation method (met forcing): "budget-bilinear" "budget-bilinear"
Spatial upscaling method (met forcing): "average" "average"
Temporal interpolation method (met forcing): "linear" "linear"

Not sure if this answers your question? Please let me know what I may provide for your assessment. Thanks!

[tais958]

Hi @LIS-navari @dmocko. Wondering if you could provide any guidance on this? Thanks for looking into this.