Sources of GFS data matters the ungrib FILE

[jianglizhi]

Hello,
I used download the GFS 0.25 degree data through the AWS cloud website. https://noaa-gfs-bdp-pds.s3.amazonaws.com/index.html

It is strange that ungribbed FILE data only contains 21 levels data.

the output of vars attribute
gfs.t00z.pgrb2b.0p25.f000

https://noaa-gfs-bdp-pds.s3.amazonaws.com/gfs.20230727/18/atmos/gfs.t18z.pgrb2.0p25.f000

Variable: f
Type: file
filename: gfs.t00z.pgrb2b.0p25.f000
path: gfs.t00z.pgrb2b.0p25.f000
file global attributes:
dimensions:
lv_ISBL0 = 21
lat_0 = 721
lon_0 = 1440
lv_AMSL1 = 5
lv_SPDL2 = 5
lv_PVL3 = 6
lv_ISBL4 = 17
lv_ISBL5 = 16
lv_ISBL6 = 5
lv_DBLL7 = 4
variables:
float TMP_P0_L100_GLL0 ( lv_ISBL0, lat_0, lon_0 )
center : US National Weather Service - NCEP (WMC)
production_status : Operational products

and from the ungrib.log, it not contains the soil level. see ungrib.log_AWS

Then I redownload the data from the RDA website,
It contains 41 levels data
gfs.0p25.2023072718.f000.grib2
https://data.rda.ucar.edu/d084001/2023/20230727/gfs.0p25.2023072718.f000.grib2

Variable: f1
Type: file
filename: gfs.0p25.2023072718.f000
path: gfs.0p25.2023072718.f000.grib2
file global attributes:
dimensions:
lat_0 = 721
lon_0 = 1440
lv_ISBL0 = 41
lv_AMSL1 = 3
lv_HTGL2 = 3
lv_PVL3 = 2
lv_HTGL4 = 2
lv_SIGL5 = 4
lv_ISBL6 = 22
lv_HTGL7 = 7
lv_SPDL8 = 3
lv_HTGL9 = 2
lv_HYBL10 = 2
lv_DBLL11 = 4
variables:
float TMP_P0_L1_GLL0 ( lat_0, lon_0 )
center : US National Weather Service - NCEP (WMC)
production_status : Operational products

I don't know the difference between the two src data. It should have the same dimensions. However, the results of ungrib are different.

So the result is that It is better download the data from the RDA website, instead of AWS/ NCEI

 

[jianglizhi]

Update:
Today I redownloded the GFS AWS data and run the ungrib, metgrid, real program again. No error occurs today. It seems that the GFS AWS data may also suitable for the use of WRF. NO PROBLEMS FOR BOTH DATA SOURCES, IT'S PROBABLY MY FAULT.
Visually speaking, it was found that the two attritubes num_metgrid_levels = 22 and BOTTOM-TOP_GRID_DIMENSION = 32 (or the reverse value?) of the met_ file were inconsistent, which led to errors in real.exe later on.
This is similar to the previous post.(RESOLVED) ERROR: Mismatch between namelist and input file dimensions
Because I have deleted the erroneous source data and rsl files, I am unable to determine the cause of the error. I think it may cause by linked to dismatched met_ files.