[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20020411: Guidance Requested (dcmetr, NWSLI)

Daryl,

The FMH #1 states that the station identifier in a metar will have the
form "CCCC", which is 4 character:
http://205.156.54.206/oso/oso1/oso12/fmh1/fmh1ch12.htm#ch12link

The change to GEMPAK for decoding 5 character ID's is very simple:
In $GEMPAK/source/bridge/mt/mtgrpt.f, the check for 4 characters
is explicitly made- even though GEMPAK does handle up to 8 characters.

I made the following insertion into line 126:
                  ELSE IF ( lens .eq. 3 .and. saoflg ) THEN
C
C*                  Canadian SAO report.
C
                    found = .true.
                    wmo   = .false.
                  ELSE IF (lens .eq. 5) THEN    <-- begin insertion
C
C*                   Bogus metar using NWSLI
C
                     found = .true.
                     wmo = .false.              <-- end insertion
                  ELSE

Then recompiled this routine, and relinked dcmetr with:
cd $GEMPAK/source/bridge/mt/
make all

cd $GEMPAK/source/programs/dc/dcmetr
make clean
make all
make install

This allowed dcmetr to decode a bulletin of the form:
ZCZC
SAUS70 KWBC 111800 RRA
METAR
BOUC2 111753Z 30007KT 10SM FEW070 SCT150 SCT200 14/01 A3011 RMK AO2
     SLP157 T01390011 10139 20039 58002=
NNNN

But, if anyone else wanted to decode your product, and their
decoders were written strictly to the "CCCC" specification,
then you would have a problem there. 

You could use other 4 characater identifiers (the NWS does something like
this already by creating alpha numeric id's like K7V3 that don't conform to
12.6.2 in the above URL). You would want to
1) ensure they weren't already assigned elsewhere 
2) provide your own station table for decoding-
   eg, decode those 1 minute obs with dcmetr and specify the "-s" flag
   to as RWIS or school-net specific table.

The main thing is, you are taking data probably observed into 
one format, then creating a METAR like product for decoding-
and the METAR format may be ill-designed for your use. Another
thing is that since METAR doesn't provide station location,
you will also have to give the station location table along
with your data.

Other options:
You could put your data into a NetCDF file with both the station ID
and location and anyone wanting to use the data could write an
adaptor (like the dcacars, dcsuomi, dcncprof etc) decoders that
write GEMPAK files from the NetCDF data.


another question is, do you intend to mix this observations into
data files that have other hourly METAR data from the IDD? If so,
one set would have observation frequency on the order of 1 minute,
while the other is 1 hour. It gets clumsy to display data like that.

On the other hand, if you have separate files for your decoded data, then
you can use whatever station IDs you like.

Steve Chiswell
Unidata User Support



>From: Daryl Herzmann <address@hidden>
>Organization: UCAR/Unidata
>Keywords: 200204112117.g3BLHVa26469

>Hiya!
>       A mortal user seeks divine guidance.
>
>       My mesonet project is collecting 1 minute data from TV+school
>networks in the State.  Originally, I had assigned 4 character identifiers
>to them.  Recently, we got official 5 character NWSLIs for them! I am
>happy about this, since if I distribute the obs to other locations, I
>don't need to worry about ID conflicts, since mine will be official :)
>
>       Anyway, dcmetr does not like 5 character IDs in the METARs that I 
>generate from the raw data feed. Looking at the METAR format, 5 characters 
>is probably not legal.  I poked around at some of the GEMPAK code and it 
>seemed that the ID length limit was 8, but I am not sure.
>
>       Basically, I would like to make an announcement, saying "Point 
>your LDM to mesonet.agron.iastate.edu and feed 1 minute SchoolNet data in 
>METAR format", but that won't work with dcmetr or perhaps any other 
>METAR decoder.  I am currently disseminating schoolnet data in CSV files 
>via LDM to the NWS and not using the METAR format.
>
>       What should I do?  If I invent 4 character IDs, I am worried about 
>conflicts.  If I started the ID with something other than K or C, would 
>that keep me out of trouble?
>
>       Ohh, I have RWIS data as well, which I would like to distribute 
>in the same manner.  Currently, I have assigned 4 char IDs beginning with 
>'R' .  Should I change them as well to something better?  I won't be 
>getting NWSLI for the RWIS stations, since their precip reporting is 
>basically meaningless.  I may be getting RWIS data from other states as 
>well, so I would like to hammer down some standard, if possible.
>
>Thanks!
>   Daryl
>
>
>-- 
>/**
> * Daryl Herzmann (address@hidden)
> * Program Assistant -- Iowa Environmental Mesonet
> * http://mesonet.agron.iastate.edu
> */
>
>
>