[Met_help] [rt.rap.ucar.edu #89605] History for questions on PB2NCConfig and PointStatConfig

[John Halley Gotway via RT]

----------------------------------------------------------------
  Initial Request
----------------------------------------------------------------

I have a question on the "obs_window" for both PB2NCConfig & PointStatConfig.
I made a mistake on the "obs_window" in PB2NCConfig by putting -5400 & 5400.
It ended up with only 3 hours obs data being processed. As a result, all obs values in the MPR files are the same.
I found this bug and re-generated all obs files with the "obs_window" = -24*60*60 & 24*60*60.
My question is when I run point_stat, do I have to set the "obs_window" = -24*60*60 & 24*60*60?
Is it matter if I set the "obs_window" = -12*60*60 & 12*60*60? Will it change the MPR result?
If so, why they are different? Which one the correct one?
Thank you for help.

----------------------------------------------------------------
  Complete Ticket History
----------------------------------------------------------------

Subject: questions on PB2NCConfig and PointStatConfig
From: John Halley Gotway
Time: Thu Apr 04 13:05:20 2019

Ka Yee,

I see you have a question about the obs_window config file option for
both PB2NC and Point-Stat.  This option applies the same basic logic
in both tools.  It specifies a time window in seconds around some
reference time that controls which observations are included in the
processing logic.

For PB2NC, the "reference time" is the PrepBUFR center time... which
is printed out in one of the log messages when you run pb2nc.  Any
observations falling inside the time window are written to the output
pb2nc file.  The only reason to restrict observations here is if
you're only verifying a specific valid time.  This just makes the
output data smaller and enables downstream tools like Point-Stat to
run faster.

For Point-Stat, the "reference time" is the forecast valid time.  Any
observations falling inside the time window are included in the
computation of statistics, and would therefore be included in the MPR
output from Point-Stat.

However, you should also think about the obs_summary config file
option, which tells Point-Stat what to do if multiple observations
occur at a single location in the time window you defined.  We often
set "obs_summary = NEAREST;"  Below is a selection from the user's
guide to explain that.

Does this answer your question?

Thanks,
John

//
// The "obs_summary" entry specifies how to compute statistics on
// observations that appear at a single location (lat,lon,level,elev)
// in Point-Stat and Ensemble-Stat.  Eight techniques are
// currently supported:
//
//    - "NONE" to use all point observations (legacy behavior)
//    - "NEAREST" use only the observation that has the valid
//      time closest to the forecast valid time
//    - "MIN" use only the observation that has the lowest value
//    - "MAX" use only the observation that has the highest value
//    - "UW_MEAN" compute an unweighted mean of the observations
//    - "DW_MEAN" compute a weighted mean of the observations based
//      on the time of the observation
//    - "MEDIAN" use the median observation
//    - "PERC" use the Nth percentile observation where N =
obs_perc_value
//
// The reporting mechanism for this feature can be activated by
specifying
// a verbosity level of three or higher.  The report will show
information
// about where duplicates were detected and which observations were
used
// in those cases.
//
obs_summary = NONE;



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