Negative soil moisture values

[Vaishnavi_198]

Hello everyone,


I am generating WRF intermediate files using pywinter and running WPS (metgrid.exe directly, no ungrib involved) using CESM2 CMIP6 bias-corrected data as input. I am encountering small negative soil moisture values in the met_em files, even though I have masked water points and set appropriate missing value handling in METGRID.TBL.

Input data​

• Model: CESM2 (CMIP6)
• Variables: mrsol (soil moisture, 4 layers)
• Data are bias-corrected
• Due to bias correction, the original NetCDF files contain small negative values at a few grid points
• Fraction of affected grid points is < 2%
• I explicitly set all negative soil moisture values to 0.0 before creating WPS intermediate files
• Missing / masked points are set to -1e30
  

and the intermediate files HAVE "landsea" mask as mention in a previous forum post

METGRID.TBL configuration used
name=SM000007
interp_option=sixteen_pt+four_pt+wt_average_4pt+wt_average_16pt+search
masked=water
interp_mask=LANDSEA(0)
missing_value=-1.E30
fill_missing=1.
flag_in_output=FLAG_SM000007
========================================
name=SM007028
interp_option=sixteen_pt+four_pt+wt_average_4pt+wt_average_16pt+search
masked=water
interp_mask=LANDSEA(0)
missing_value=-1.E30
fill_missing=1.
flag_in_output=FLAG_SM007028
========================================
name=SM028100
interp_option=sixteen_pt+four_pt+wt_average_4pt+wt_average_16pt+search
masked=water
interp_mask=LANDSEA(0)
missing_value=-1.E30
fill_missing=1.
flag_in_output=FLAG_SM028100
========================================
name=SM100289
interp_option=sixteen_pt+four_pt+wt_average_4pt+wt_average_16pt+search
masked=water
interp_mask=LANDSEA(0)
missing_value=-1.E30
fill_missing=1.
flag_in_output=FLAG_SM100289
========================================







How do I fix this? Can real.exe and wrf.exe run regardless?

 

[Ming Chen]

WRF doesn't allow negative soil moisture in its input data.

I am not sure whether there is any bugs in pywinter. We need to figure out whether the negative SM exists in your intermediate file. If so, we need to fix those unreasonble values. Can you check your files first?

 

[Vaishnavi_198]

Thank you for your reply @Ming Chen

I double checked for negative values in the intermediate files using NCL. There were no negative values. I found out that when I set interp_opt= wt_average_16pt in the METGRID.TBL I had no negative values in met_em files.I get NO negative values. But for interp_opt = sixteen_pt, I get negative values. But the interpolation is not as smooth.

SM000007 using interp_opt= wt_average_16pt


SM100289 interp_opt= wt_average_16pt

 

[Ming Chen]

This is a weird model behavior. The option average_16pt calculates the average of data at the surrounding 16 points. The option wt_average_16pt performs an area-weighted average of data. There is no reason the option average_16pt yields negative value.

Are there any water points close to the point where negative SM occurs? If so, what is the value of SM at water points in your intermediate file?

 

[Vaishnavi_198]

@Ming Chen I checked for points if there were water bodies around the point where there are negative values. There weren't any. Those negative points happen at the follwoing places and at only 39 grid points

• Western part, Saudi Arabia (land)
• Eastern part, Arabian Sea / Gulf of Oman​
• Western China (Tibetan Plateau region)​

I tried again setting missing values as 99999 instead of a negative value (-1E30). I still get the same result. I use WPS version 4.6.0 which is compiled for no leap year option.

I have attached here a zip file which contains the namelist.wps, ncl code (to read intermediate files to print the min and max values of the soil moisture fields),few intermediate files and geo_em files, METGRID.TBL (set for interp_opt=sixteen_pt+four_pt+wt_average_4pt+wt_average_16pt+search), and the screenshot of ncview window for different interp_opt.

intermediate and geo_em files: Drive link