Determining expected output filenames from configuration

[maddenp-noaa]

Hi. I'm wondering how to determine the names of the files UPP will create when run, given its itag namelist file and its postxconfig-NT.txt control file. The docs suggest an answer, but it doesn't appear to reflect the behavior of UPP when run by SRW (I'm seeing filenames with extensions like .GrbF06), and looking at the source code, it seems like the behavior might be more complicated. Thanks in advance for ideas.

[gspetro-NOAA]

Hi @maddenp-noaa ,

I think @WenMeng-NOAA and/or @MichaelLueken are most likely to know the answer offhand if there is an easy answer. Otherwise, I'll probably have to work with them to see what we can figure out, which may take some time.

Best,
Gillian

[MichaelLueken]

Hi @maddenp-noaa,

I can only speak about the SRW, but it looks like the DateStr entry in the itag namelist file dictates the forecast hour in the GrbFHHH filename. Looking in scripts/exregional_run_post.sh, the post filename is constructed as GrbF${post_fhr}. The post_fhr variable is set up as follows:

len_fhr=${#fhr}
if [ ${len_fhr} -eq 2 ]; then
  post_fhr=${fhr}
elif [ ${len_fhr} -eq 3 ]; then
  if [ "${fhr:0:1}" = "0" ]; then
    post_fhr="${fhr:1}"
  else
    post_fhr="${fhr}"
  fi
fi

and fhr is the forecast hour. The DateStr itag namelist option is set as '${post_yyyy}-${post_mm}-${post_dd}_${post_hh}:${post_mn}:00'

where post_yyyy, post_mm, post_dd, post_hh, and post_mn are set as:

post_time=$( $DATE_UTIL --utc --date "${yyyymmdd} ${hh} UTC + ${fhr} hours + ${fmn} minutes" "+%Y%m%d%H%M" )
post_yyyy=${post_time:0:4}
post_mm=${post_time:4:2}
post_dd=${post_time:6:2}
post_hh=${post_time:8:2}
post_mn=${post_time:10:2}

So, for the SRW App, post_hh = post_fhr = fhr.

As an example, take a 6 hour forecast, starting at 0z, on July 22, 2019. In the subdirectory that runs the post for the 00z cycle, the DateStr option in itag is set as DateStr='2019-07-22_00:00:00'. The post output filename would be *.GrbF00 (PRSLEV.GrbF00 and NATLEV.GrbF00). Then, in the subdirectory that runs the post for the 01z cycle, the DateStr option in itag would be set as DateStr='2019-07-22_01:00:00', with the post output filename would be *.GrbF01 (PRSLEV.Grb01 and NATLEV.GrbF01).

As for the names before the .GrbFHHH entries, I believe the defaults are PRSLEV and NATLEV. Sometimes I see other entries, like CMAQ, AVIATION, and IFIFIP, but I don't know where they come from and when they are used.

[MichaelLueken]

After some additional looking, the PRSLEV, NATLEV, CMAQ, AVIATION, and IFIFIP entries are from postxconfig-NT.txt control file.

[maddenp-noaa]

Thank you @MichaelLueken.

It seems like the behavior of UPP is somewhat constrained by SRW, but I was hoping for something more definitive, as I have a need to be able to determine the UPP output filenames in the general case. I find the code in get_postfilename.f difficult to interpret, but it appears that the overall behavior conditionally depends on a number of environment variables in addition to the namelist and control files, so a straightforward answer is looking unlikely.

As for PRSLEV et al., those are "from postxconfig-NT.txt", but how would one locate them / know what to look for? Are there a fixed number of strings that could ever appear? I'm guessing that the answer might be "yes" for SRW, at least in the state it's in today, but "no" in general.

It's unfortunate if the only way to know what the output of UPP will be is to replicate, in another language (bash, Python, etc.) something as complex as a big chunk of UPP code, and to have to keep that in sync with any UPP changes. It seems that SRW is dealing with that difficulty already, and is somehow arranging -- by accident or by design -- to only get the .GrbF* suffixes, which seem (maybe?) to be just one possibility. It's hard to tell.

In any case, I think the documentation is currently too specific, mentioning only the NATLEV and PRSLEV prefixes, which are not even sufficient to describe SRW's behavior; and incorrect in indicating a .HHH suffix (for SRW I think it would be .GrbFHH for leadtimes < 100, .GrbFHHH for leadtimes between 100 and 999, etc.).