[Met_help] [rt.rap.ucar.edu #83674] History for Question in Stat_Analysis tool

[John Halley Gotway via RT]

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

I have question for Stat_Analysis tool.

What is difference between "fcst_valid_hour" and "obsv_valid_hour".
In Metviewer, there is no such option in "Fixed Values". Only "VALID HOUR"
option.  This "VALID HOUR" is correspondent to which one, "fcst_valid_hour",
or "obsv_valid_hour". In my think, these two is same thing?

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

Subject: Question in Stat_Analysis tool
From: John Halley Gotway
Time: Fri Jan 19 16:00:36 2018

Hi Binbin,

I moved this question over to met_help since the answer may be helpful
for other users.

You're right, the .stat output files generated by MET contain columns
to specify the timing info separately for the forecast and observation
data:
- FCST_LEAD, FCST_VALID_BEG, FCST_VALID_END
- OBS_LEAD, OBS_VALID_BEG, OBS_VALID_END

The vast majority of the time, we run MET to compare forecast data to
observations that match up exactly in time.  In those cases, the valid
times are the same.  However, here are some use cases when the timing
info may look different:

(1) Compare a forecast to an observation valid 1 hour later to study
timing errors.  The forecast and observation valid times won't match.
(2) Compare the output of one model to a different model (i.e. not an
analysis).  The FCST_LEAD would indicate lead time of the first model
and OBS_LEAD would indicate the lead time of the second model.
(3) In Point-Stat, OBS_VALID_BEG and OBS_VALID_END define the
beginning/ending times of the matching time window.  So those times
would be set to the forecast valid time +/- 30 minutes (or however you
configured the matching time window).
(4) When running STAT-Analysis to summarize statistics through time,
the output FCST_VALID_BEG and FCST_VALID_END columns contain the
min/max of the timestamps from the input.

But let me answer your specific question about METViewer.
FCST_VALID_BEG and FCST_LEAD are the only two columns from the MET
output that show up in METViewer's drop-down lists.  However,
METViewer derives some more timing info on the fly:
- FCST_INIT_BEG = FCST_VALID_BEG - FCST_LEAD
- INIT_HOUR = only the HH portion of FCST_INIT_BEG
- VALID_HOUR = only the HH portion of the FCST_VALID_BEG

In the METViewer GUI, there currently is no way of accessing the
OBS_VALID_BEG, OBS_VALID_END, and OBS_LEAD timing info from the MET
output.  Since we (almost always) compare forecast and observations
with the same timing info, we haven't had a need to specify them
separately.

Does this logic suffice?  Or do you need to be able to specify
observation timing information in METViewer?

Thanks,
John

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