Using other land cover and soil datasets

[Melmel]

Hello there,

I have started to build WRF-Hydro.

I have already built the dependencies.

I have decided on the input datasets, which are different from your standard datasets.

“NALCMS,” 19 land cover classes for North America (30m, Landsat, 2010 from Mexico and Canada, 2011 for the U.S.) & “Global Soil Dataset for Earth System Models (GSDE)” (30” ⇡ 1km) containing eight layers of soil to a depth of 2.3m.

 

I had a couple of questions.

What is the difference between “spatially distributed parameters” for the Noah-MP and using the “lookup tables”?

Do you have any recommendations for using the above data sets instead of the usual datasets? How can I introduce these datasets to the model? Using which of the above two methods, spatially distributed parameters or lookup tables, is easier to use the new datasets?

Do I have regenerated the VEGPARAM.TBL and SOILPARAM.TBL? How can I do it? Or can I re-classify the land cover dataset?

Regards,

Melika

[Kayla Yamamoto]

Hi all,

I want to use higher resolution land cover and soil datasets with WRF-Hydro, similar to what is mentioned in Melika's post.  In my case, I'm looking to use C-CAP land cover and SoilGrids soil texture classes.  

Does anyone know the best way to do this?

I'm trying to to reclassify the CCAP land cover categories to something WPS can read (i.e. USGS, NLCD40, etc), but am not sure if I need to also modify any of the index or table files (VEGPARM.TBL, MPTABLE.TBL) that are used in WPS geogrid or WRF-Hydro.

Thanks,

Kayla

[Kevin]

This is a reply to Kayla's question regarding higher resolution land cover and soil datasets in WRF-Hydro. 

1) You will first need to reclassify your source datasets to classifications that are compatible with WRF and WRF-Hydro. I typically create a class remapping from the source data (such as C-CAP) to either USGS 21-class or MODIS IGBP. There are various tools out there that can perform the reclassification in GIS. I typically use the <LUT> (lookup table) in a GDAL VRT dataset, which preserves the source data but creates a virtual reclassification in client applications such as QGIS. 

2) You will then need to convert your reclassified raster dataset (from step 1 above) into a WPS tiled format. The best way to do this is to use the GIS4WRF plugin in QGIS. You can bring in the raster or VRT layer, and save it to WPS tile format. 

3) Your WPS tiles will need to be placed in your WPS geog_data directory and you will need to reference them in the WPS namelist and GEOGRID.TBL files. Once this is complete, you can build WRF domains using these high resolution datasets using the geogrid.exe program in WPS.

I have generated global WPS tiles for the soil top layer and bottom layers from the 2017 SoilGrids250m texture class dataset. If you have access to the NCAR Cheyenne supercomputer, I can point you to those datasets. 

Thanks,

Kevin

[Donny Kim]

Kevin's approach would be the standard way to do it.

However, if you already have well-established domain inputs, and all you want is just to use different LULC and soil datasets, there's a simpler way to workaround:
1) First, reclassify C-CAP LULC classification into something that is compatible with WRF-Hydro model, preferably noah_mp_usgs_parameters (you'll need to build your own look-up table for such reclassification).
2) Resample the reclassified LULC into GEOGRID resolution, and build GEOGRID with updated LU_Index of GEOGRID.
3) Run scripts such as "create_soilproperties.R" and "create_wrfinputs.R" to build soil_properties.nc, hydro2dtbl.nc, and wrfinput.nc that are required for WRF-Hydro simulation.

If this less conventional method works for you and comfortable with R, I can point you to my GitHub repository. My colleague initially built an R package exactly does 1) and 2) for NWM-configured WRF-Hydro, and I have been troubleshooting/improving it.

Best,
Donny

[Kayla Yamamoto]

Thank you both for your incredibly helpful responses!  

I was able to reclassify the C-CAP LULC categories to noah_mp_usgs_parameters in QGIS and then convert it to WPS binary with the GIS4WRF plugin.  My next step is to modify the namelist.wps and GEOGRID.TBL files to reference the new LULC data.  Thank you both for helping me get this far!

Kevin, unfortunately I don't have access to Cheyenne.  However I am interested in using that same soil texture class dataset and had a couple of questions about how you generated your WPS tiles:

• From what I understand, the SoilGrids250m data is provided in 6 layers (0-5, 5-15, 15-30, 30-60, 60-100, 100-200)
  • Do you have to merge these layers to get a soil top layer and bottom layer that would be compatible with WPS?

Donny, I'm still figuring out my domain setup so can't utilize your approach just yet, but I definitely will be interested in using this workaround down the line and would love to check out your GitHub repo! 

Last questions:

• If I reclassify and update my geogrid file with the new SoilGrids250m texture classes, would I still choose the option to update texture classes in the create_soilproperties.R script?
  • If so, how would my geogrid soil texture classes change?

Thanks again for all of your help and suggestions!  I truly appreciate it!

Mahalo,

Kayla

[Kayla Yamamoto]

Hi guys,

Sorry to message twice, but I just wanted to update that I came across an R soil texture wizard package that looks like it will compute the texture classes from my SoilGrids data (clay, silt, sand content).  

This may also address the question I had for Kevin about merging layers, but I'm not entirely sure yet. 

Thanks!

Kayla

[Jana Corominas]

Hi everyone, 
I face the same problem of wanting to use high resolution land cover data just in WRF-Hydro simulation without editing WPS in the WRF model. I have a 1km resolution WRF domain and 50m resolution Hydro domain. I was thinking about manually modifying Land Cover data in Fulldom_hires.nc file, but I would like to know in which computations WRF-Hydro model uses land cover data from geogrid file or from Fulldom_hires file, since maybe I have to modify this field also in geogrid file. Could this data (regridded according to the domain) work? Of course, my first step would be to resample original data to MODIS land use categories to match the tables. 

Thanks in advanced.

Jana