run ./real.exe get "grid%tsk unreasonable"

I use GFS 0.25 data and use “sf_surface_physics = 2”，when run ./real.exe ,its get errors.
but use FNL data,./real.exe excute success.
what case this error??

Domain 1: Current date being processed: 2020-05-09_00:00:00.0000, which is loop # 1 out of 26
configflags%julyr, %julday, %gmt: 2020 130 0.00000000
d01 2020-05-09_00:00:00 Yes, this special data is acceptable to use: OUTPUT FROM METGRID V4.1
d01 2020-05-09_00:00:00 Input data is acceptable to use: met_em.d01.2020-05-09_00:00:00.nc
metgrid input_wrf.F first_date_input = 2020-05-09_00:00:00
metgrid input_wrf.F first_date_nml = 2020-05-09_00:00:00
d01 2020-05-09_00:00:00 Timing for input 0 s.
d01 2020-05-09_00:00:00 flag_soil_layers read from met_em file is 1
d01 2020-05-09_00:00:00 Turning off use of MAX WIND level data in vertical interpolation
d01 2020-05-09_00:00:00 Turning off use of TROPOPAUSE level data in vertical interpolation
Max map factor in domain 1 = 1.00. Scale the dt in the model accordingly.
Using sfcprs3 to compute psfc
d01 2020-05-09_00:00:00 No average surface temperature for use with inland lakes
Assume Noah LSM input
d01 2020-05-09_00:00:00 forcing artificial silty clay loam at 1 points, out of 10197
error in the grid%tsk
i,j= 80 55
grid%landmask= 0.00000000
grid%tsk, grid%sst, grid%tmn= 0.00000000 0.00000000 0.00000000
-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 2977
grid%tsk unreasonable
-------------------------------------------

Where did you download the GFS data? Please also send me your namelist.wps and namelist.input to take a look.
Also, which version of WRF/WPS are you using?