SKINTEMP values becoming "missing" ($10^{-35}$) in met_em files despite correct Intermediate files - ERA5

[jcoutinho]

Hi everyone,

I am encountering an issue where SKINTEMP values are being lost during the metgrid.exe process. While the values appear correct in my source NetCDF files and the WPS Intermediate files, they show up as missing values ($1.024259e-35$) in the final met_em files.

My Workflow:

1. Data: ERA5 Reanalysis data.
2. Conversion: I used a custom era5_to_int.py script to convert the ERA5 NetCDF data into WPS Intermediate format.
3. Vtable: Used Vtable.ECMWF and also tried Vtable.ERA5-interim.pl
4. Intermediate Check: I verified the intermediate files using rd_intermediate.exe. The values look physically correct (e.g., 263.8 K) and the date/metadata headers appear correct.

Evidence from Intermediate File:

[jdirectory WPS]$ ./util/rd_intermediate.exe ERA5:1985-01-01_00 | grep -20 "FIELD = SKINTEMP"<br>================================================<br>FIELD = SKINTEMP<br>UNITS = K DESCRIPTION = Skin temperature<br>DATE = 1985-01-01_00:00:00 FCST = 0.000000<br>SOURCE = ERA5 reanalysis grid<br>LEVEL = 200100.000000<br>I,J DIMS = 1440, 721<br>IPROJ = 0 PROJECTION = LAT LON<br> REF_X, REF_Y = 1.000000, 1.000000<br> REF_LAT, REF_LON = 90.000000, 0.000000<br> DLAT, DLON = -0.250000, 0.250000<br> EARTH_RADIUS = 6371229.000000<br>DATA(1,1)=263.801544<br>

The Issue (met_em output):

Despite the intermediate files being correct, ncdump shows that the values in the met_em files are all missing values:
[directory WPS]$ ncdump -v SKINTEMP met_em.d01.1985-01-01_00:00:00.nc | tail -n 20<br> 1.024259e-35, 1.024259e-35, 1.024259e-35, 1.024259e-35, 1.024259e-35,<br> ...<br> 1.024259e-35, 1.024259e-35, 1.024259e-35, 1.024259e-35 ;<br>
Additional Context:

• The NetCDF source files for ERA5 show correct values, so no data is missing at the start.
• I am running this on a high-performance cluster (Slurm/Intel environment).
• My METGRID.TBL is the default for ARW.

Has anyone seen this behavior where metgrid successfully reads the intermediate file but fails to interpolate the values into the met_em output? Could this be related to a masking issue (LANDSEA vs LANDMASK) or a specific setting in the METGRID.TBL for ERA5 skin temperature?

Any advice on where to look in the metgrid.log or suggested changes to the namelist.wps would be greatly appreciated.

Thank you!

 

[Ming Chen]

What ERA5 data have you downloaded? Did you run the latest version of era5_to_int.py?

Please send me your namelist.wps to take a look.

 

[jcoutinho]

Hi,

I wanted to update you that I managed to resolve this issue by changing the interpolation settings to 'none' for the SKINTEMP variable in the METGRID.TBL file.

Thank you for your assistance!

 

[Ming Chen]

Thank you for the update.

Would you please confirm that you use the option below to process ERA5 SKINTEMP, and it works as you expected?

name=SKINTEMP
mpas_name=skintemp
        interp_option=none
        masked=both
        interp_land_mask  = LANDSEA(1)
        interp_water_mask = LANDSEA(0)
        fill_missing=0.

 

[Ming Chen]

Also, have you checked skintemp in met_em files ? Does it look reasonable? Thanks.

 

[jcoutinho]

Actually I had do undo the 'none' setting.
After I download the correct ERA5 files as listed in this post, the SKINTEMP and SST look reasonable.
It was not necessary to change the SST interpolation method in the METGRID.TBL.

 

[Ming Chen]

Thank you for the update. I am glad it works for you.

Just curious, where did you download the ERA5 data that always have issues in SKINTEMP?

 

[jcoutinho]

I downloaded the ERA5 data from NSF NCAR GDEX Dataset Archive using the Globus Transfer.