[Met_help] [rt.rap.ucar.edu #79618] History for SID Filters

[John Halley Gotway via RT]

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

John or Julie, I have a question the manual is unclear on.  I know I can filter the observations used in point stat with the SID list of observation ids to use.  Can I do the same thing in Statanalysis?   I have several sub regions to verify in a global model.  I would rather run point stat and get the MPR file of model ob pairs for the entire global domain and then use stat analysis to only calculate statistics over a set of observations (based on IDS) in the domain of interest.

Thanks
Bob Craig



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

Subject: SID Filters
From: John Halley Gotway
Time: Thu Feb 23 10:04:41 2017

Bob,

The answer is yes and no.

First, let me clarify something.  In STAT-Analysis, there is a "-
vx_mask"
job filtering option.  That option reads the VX_MASK column from the
input
STAT lines and applies string matching with the values in that column.
Presumably, all of your MPR lines will have the value of "FULL" in the
VX_MASK column... so that filtering option is *NOT* what you want.

STAT-Analysis has the ability to read MPR lines and recompute
statistics
from them using the same library code that the other MET tools use.
The
job command options which begin with "-out..." are used to specify
settings
to be applied to the output of that process.  For example, the
"-fcst_thresh" option filters strings from the input "FCST_THRESH"
header
column.  The "-out_fcst_thresh" option defines the threshold to be
applied
to the output of STAT-Analysis.  So reading MPR lines and applying a
threshold to define contingency table statistics (CTS) would be done
using
the "-out_fcst_thresh" option.

STAT-Analysis does have the ability to filter MPR lat/lon locations
using...
  - the "-mask_poly" option for a lat/lon polyline
  - the "-mask_grid" option to define a retention grid.

However, there is currently no "-mask_sid" option.  But that would be
a
logical enhancement for a future release.

With met-5.2, one option is to apply column string matching using the
"-column_str" option to define the list of station ID's you'd like to
aggregate.
That job would look something like this:

   stat_analysis -lookin path/to/mpr/directory -job aggregate_stat
-line_type MPR -out_line_type CNT \
      -column_str OBS_SID SID1,SID2,SID3,...,SIDN \
      -set_hdr VX_MASK SID_GROUP_NAME \
      -out_stat mpr_to_cnt.stat

Where SID1...SIDN is a comma-separated list of the station id's in
your
group.  Notice that I've specified a value for the output VX_MASK
column
using the "-set_hdr" option.  Otherwise, you'd get a list of the
unique
values found in that column.  Presumably, all the input VX_MASK
columns say
"FULL" so that's what the output would say.  Use "-set_hdr" to
explicitly
set the output value.

I'll add a development task to add a "-mask_sid" option to a future
release.

Hope that helps clarify.

Thanks,
John


On Thu, Feb 23, 2017 at 7:26 AM, robert.craig.2 at us.af.mil via RT <
met_help at ucar.edu> wrote:

>
> Thu Feb 23 07:26:58 2017: Request 79618 was acted upon.
> Transaction: Ticket created by robert.craig.2 at us.af.mil
>        Queue: met_help
>      Subject: SID Filters
>        Owner: Nobody
>   Requestors: robert.craig.2 at us.af.mil
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=79618 >
>
>
> John or Julie, I have a question the manual is unclear on.  I know I
can
> filter the observations used in point stat with the SID list of
observation
> ids to use.  Can I do the same thing in Statanalysis?   I have
several sub
> regions to verify in a global model.  I would rather run point stat
and get
> the MPR file of model ob pairs for the entire global domain and then
use
> stat analysis to only calculate statistics over a set of
observations
> (based on IDS) in the domain of interest.
>
> Thanks
> Bob Craig
>
>
>

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