[Met_help] Question about MET Point-Stat (UNCLASSIFIED)

[John Halley Gotway]

John,

If the same observation is listed twice in Point-Stat, it will use that observation twice.  It doesn't filter the observations to skip duplicates.

Adding the logic do throw out duplicate observations would be doable but would require code changes to the vx_met_util library.  Right now, for each observation we're remembering the lat/lon,
elevation, vertical level, and forecast value.  We'd probably need to add code to remember the observation valid time (and maybe station id).  And then, before adding a new observation, we'd need to
sift through the old ones to see if that combination of lat/lon/level/time/forecast value already exists.  That extra filtering would likely cause Point-Stat to run slower.

Thanks,
John

Raby, John (Civ, ARL/CISD) wrote:
> Classification: UNCLASSIFIED
> Caveats: NONE
> 
> John or Tressa - 
> 
> Right now we provide the observations to Point-Stat using the mandatory
> obs_file as specified in the usage statement. The file we provide is the
> PrepBUFR data obtained from NCEP. We have acquired access to the MADIS data
> and are downloading it in anticipation of using it alongside the PrepBUFR
> data (using the optional argument -ncfile netcdf_file) to maximize the
> number of observations used in the Point-Stat calculations. How does
> Point-Stat deal with any duplicate obs between the two different data
> sources?
> 
> Thanks.
> 
> R/
> John
> 
> Classification: UNCLASSIFIED
> Caveats: NONE
> 
>