[netCDFJava #TOX-206476]: Problems converting GEMPAK point (METAR) data to a usable NetCDF file

I've reviewed our IOSP and point writer code. A couple points to note:

1. The example dataset contains 25 stations that "do not appear in the station 
table" (GEMPAK's wording). Such stations lack latitude/longitude information, 
which is causing the point writer to reject them (we need that info for 
subsetting operations, among other things). That explains the drop from 216 
stations to 191. This seems like a reasonable thing to do.

2. In GEMPAK, doing a SFCHCK with:
    SFFILE = 141002_mtr_us.gem
    AREA   = DSET
    DATTIM = ALL
    OUTPUT = f/stations.txt
    IDNTYP = STID
    STNTYP = L
gives observation counts that are different than what we're seeing in the point 
writer output. For example, SFCHCK claims that "ORD" has 22 observations, but 
there are only 17 in the point writer result. I've determined that the missing 
observations are "special reports" (GEMPAK's wording). I'm not exactly sure 
what that means, but if we SFLIST ORD's observations with:
    SFFILE   = 141002_mtr_us.gem
    AREA     = @ORD
    DATTIM   = ALL
    SFPARM   = STID
    OUTPUT   = f/stations.txt
    IDNTYP   = STID
It only lists 17 of the 22. So, the IOSP's decision to ignore those special 
report observations also seems reasonable.

So, the only remaining thing to do is add command-line access to the point 
writer. I'll try to have that finished by tomorrow.

-Christian

Ticket Details
===================
Ticket ID: TOX-206476
Department: Support netCDF Java
Priority: Urgent
Status: Open