What are the differences between *GMET* and *GCOM* grnd_sea_temp fields?

Original question:
Regarding the *grnd_sea_temp files, I see there are *GMET*grnd_sea_temp
files and *GCOM*grnd_sea_temp files.  Is there any difference between the
two?  It is the *GMET*grnd_sea_temp files that have the gap at the 15h and
21h forecast.  The *GCOM*grnd_sea_temp files have no gaps.  The sizes look
the same so I'm guessing it's just a naming issue.

---------------------------------
Doing a 'wgrib' comparison of two of these files, same basetime, same tau,
etc. notice the following:

=================
[command_prompt]$ wgrib -V
US058GCOM-GR1mdl.0058_0240_02400F0RL2009011200_0001_000000-000000grnd_sea_te
mp

Undefined parameter table (center 58-0 table 3), using NCEP-opn rec 1:0:date
2009011200 KX kpds5=133 kpds6=1 kpds7=0 levels=(0,0) grid=240 sfc 24hr fcst:
  KX=K index [K]
  timerange 0 P1 24 P2 0 TimeU 1  nx 360 ny 181 GDS grid 0 num_in_ave 0
missing 0
  center 58 subcenter 0 process 58 Table 3
  latlon: lat  -90.000000 to 90.000000 by 1.000000  nxny 65160
          long 0.000000 to -1.000000 by 1.000000, (360 x 181) scan 64 mode
128 bdsgrid 1
  min/max data 216.63 329.17  num bits 14  BDS_Ref 21663  DecScale 2
BinScale 0 =================
[command_prompt]$ wgrib -V
US058GMET-GR1mdl.0058_0240_02400F0RL2009011200_0001_000000-000000grnd_sea_te
mp

Undefined parameter table (center 58-0 table 2), using NCEP-opn rec
1:21:date 2009011200 KX kpds5=133 kpds6=1 kpds7=0 levels=(0,0) grid=240 sfc
24hr fcst:
  KX=K index [K]
  timerange 0 P1 24 P2 0 TimeU 1  nx 360 ny 181 GDS grid 0 num_in_ave 0
missing 0
  center 58 subcenter 0 process 58 Table 2
  latlon: lat  -90.000000 to 90.000000 by 1.000000  nxny 65160
          long 0.000000 to -1.000000 by 1.000000, (360 x 181) scan 64 mode
128 bdsgrid 1
  min/max data 216.631 329.171  num bits 14  BDS_Ref 21663.1  DecScale 2
BinScale 0 =================
--------------------------------------------

Now, looks like the data is the same in terms of registration.  Looks like the
'GMET' field has more precision (look at the min/max's from above).

Also, notice 
GCOM uses Table 3 and BDS_Ref 21633
GMET uses Table 2 and BDS_Ref 21633.1

Question: What do these table differences represent?

The table 2/3 is refers to the edition of the 'parameters' table
within GRIB.   The BDS_Ref looks to be the reference value used when the
data gets packed.  Typically, this is the smallest number in the file.

So, GCOM states the the smallest value is 21633 (216.33 degrees), whereas
GMET has it a 21633.1 (216.331)  Notice the change in the precision.

------------------------------------------------------------

Now here is the real skinny:

There is a scaling factor in one of the CAGIPS tables, for grnd_sea_temp it
is 2.   This value is easily changed if necessary, but it looks like the
scaling factor is not the problem, as the scaling factor is the same in both
grids in the example (DecScale 2).

The real issue is the precision - 5 significant digits in one, 6 in the
other.   I am not sure what controls the precision, it might be something
inside the engribbing code.

The "MET" and "COM" are called parameter types or subdomains.  The correct
parameter type/subdomain for grnd_sea_temp is "COM".   We are still not sure where
the file with the name
US058GMET-GR1mdl.0058_0240_02400F0RL2009011200_0001_000000-000000grnd_sea_temp
is coming from, but do not think it is coming from CAGIPS, it has an
incorrect CAGIPS filename.

The definitions of "MET" and "COM" from the product naming standard are:

"COM" - Any product that includes two or more scientific disciples, such as
Meteorology and Oceanography.

"MET" -  Products containing just meteorological data.