Error in dshr_strdata_mod: customized ERA5_Land forcing can't be read properly

[yifanc17]

Hi,

Last time I tried using customized 0.5x0.5° ERA5 bias corrected forcing with CTSM and CESM and both worked fine, so I switched to a higher resolution ERA5_Land hourly forcing at 9km resolution and tested it with CESM (cesm2_3_alpha17c). Following the similar workflow as before, I generated a masked mesh file according to the land pixels from the forcing (please see attached ERA5_Land_9km_mesh.png). However, when stream read in the file ((shr_strdata_readstrm) reading file lb: /glade/campaign/univ/uiuc0043/03_Forcing/ERA5_Land_9km/ERA5_Land_9km_TPQWL_2005_01_cdf5.nc 1), I got the error in CESM.log:

dec0003.hsn.de.hpc.ucar.edu 1: forrtl: error (65): floating invalid
dec0003.hsn.de.hpc.ucar.edu 1: Image PC Routine Line Source
dec0003.hsn.de.hpc.ucar.edu 1: libpthread-2.31.s 000014DA34B8A8C0 Unknown Unknown Unknown
dec0003.hsn.de.hpc.ucar.edu 1: libpnetcdf.so.4.0 000014DA417DB8EA ncmpix_getn_NC_FL Unknown Unknown
dec0003.hsn.de.hpc.ucar.edu 1: libpnetcdf.so.4.0 000014DA417E5B65 ncmpii_getn_NC_FL Unknown Unknown
dec0003.hsn.de.hpc.ucar.edu 1: libpnetcdf.so.4.0 000014DA417F6F09 ncmpio_unpack_xbu Unknown Unknown
dec0003.hsn.de.hpc.ucar.edu 1: libpnetcdf.so.4.0 000014DA417F07CD ncmpio_wait Unknown Unknown
dec0003.hsn.de.hpc.ucar.edu 1: libpnetcdf.so.4.0 000014DA417FAC44 ncmpio_get_varn Unknown Unknown
dec0003.hsn.de.hpc.ucar.edu 1: libpnetcdf.so.4.0 000014DA41777DDD ncmpi_get_varn_al Unknown Unknown
dec0003.hsn.de.hpc.ucar.edu 1: libpioc.so 000014DA4233097A pio_read_darray_n Unknown Unknown
dec0003.hsn.de.hpc.ucar.edu 1: libpioc.so 000014DA4232DE12 PIOc_read_darray Unknown Unknown
dec0003.hsn.de.hpc.ucar.edu 1: libpiof.so 000014DA425D2FD1 piodarray_mp_read Unknown Unknown
dec0003.hsn.de.hpc.ucar.edu 1: libpiof.so 000014DA425C8A73 piodarray_mp_read Unknown Unknown
dec0003.hsn.de.hpc.ucar.edu 1: cesm.exe 0000000003F922ED dshr_strdata_mod_ 1656 dshr_strdata_mod.F90
dec0003.hsn.de.hpc.ucar.edu 1: cesm.exe 0000000003F892F8 dshr_strdata_mod_ 1382 dshr_strdata_mod.F90
dec0003.hsn.de.hpc.ucar.edu 1: cesm.exe 0000000003F60EF5 dshr_strdata_mod_ 945 dshr_strdata_mod.F90
dec0003.hsn.de.hpc.ucar.edu 1: cesm.exe 00000000009022F3 atm_comp_nuopc_mp 684 atm_comp_nuopc.F90
dec0003.hsn.de.hpc.ucar.edu 1: cesm.exe 0000000000900209 atm_comp_nuopc_mp 482 atm_comp_nuopc.F90

I checked line 1656, 1382 and 945 in dshr_strdata_mod.F90, it seemed to encounter some issues when trying to get the lower boundary of the forcing data. But I'm not entirely sure where to look at (the lower boundary of the forcing data is zero, no negative values existed, and the range of all variables are similar with default GSWP3, the forcing data .nc file is in cdf5 format). I have attached the datm.streams.xml, atm.log, cesm.log and dshr_strdata_mod.F90. Any suggestions would be appreciated!

1655 if (per_stream%stream_pio_iodesc_set) then
1656 call pio_read_darray(pioid, varid, per_stream%stream_pio_iodesc, data_real1d, rcode)
1381 ! read lower bound of data
1382 call shr_strdata_readstrm(sdat, sdat%pstrm(ns), stream, &
1383 sdat%pstrm(ns)%fldbun_data(sdat%pstrm(ns)%stream_lb), &
1384 filename_lb, n_lb, istr=trim(istr)//'_LB', boundstr='lb', rc=rc)
1385 if (chkerr(rc,__LINE__,u_FILE_u)) return

 

[oleson]

I'd like to look at the TPQWL files but I don't seem to have permission to do so. Can you change the permissions?

 

[yifanc17]

I'd like to look at the TPQWL files but I don't seem to have permission to do so. Can you change the permissions?

Thank you so much Keith! Before I was running out of storage, I just moved the file to `/glade/derecho/scratch/yifanc17/ERA5_Land_forcing/TPHWL/ERA5_Land_9km_TPQWL_2005_01_cdf5.nc` so you should be able to access it. Thanks again!

 

[oleson]

I don't have permissions to read that either.

 

[yifanc17]

I don't have permissions to read that either.

Sorry I just added read permission for everyone, could you please try again? Thank you!

 

[oleson]

Thanks. I wonder if it is because t2m is defined as "short", while the other forcing variables are defined as "float". Maybe try making t2m a float variable.

 

[yifanc17]

Thanks. I wonder if it is because t2m is defined as "short", while the other forcing variables are defined as "float". Maybe try making t2m a float variable.

Thank you so much for your advice! I converted t2m and sp to 'float32' and the forcing file can be read in properly. I also flipped the latitude of forcing from 90~-90 to -90~90 to match with the mesh. Now the case run successfully.

 

[erik]

Moving to the land forum.