[Met_help] [rt.rap.ucar.edu #68435] History for point stat tool error

[John Halley Gotway via RT]

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

Hi.... 

I'm a scientist at the South African Weather Service. I'm testing MET tool the following models (WRF and Unified Model). Currently I got good results using  Gridstat tool on both models. I want to use Pointstat tool with the data I have created using the ascci2nc tool, but I get empty output text files. I loaded my data into your ftp server under the directory  /incoming/irap/met_help/phaduli_data.

I also want to test the MODE tool using TRRM data(Netcdf or bin). Please advise on this.

Thank  you.
Elelwani Phaduli
(Scientist South African Weather Service)



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

Subject: point stat tool error
From: John Halley Gotway
Time: Mon Jul 28 10:19:08 2014

Elewani,

Thanks for sending sample data.  That makes it a lot easier to debug
what's
going on.  Basically, you're able to run Point-Stat, but you're
getting 0
matched pairs.  Point-Stat does a lot of matching of observations to
forecasts... in space, in time, by message type, by vertical level or
accumulation interval, and by quality marker.  So there are many
reasons
you might end up with zero matched pairs.

One trick is to run Point-Stat at verbosity level 3 or higher (-v 3 on
the
command line).  When you do, for each verification task, you'll see a
list
of counts for why the observations were not used for that task.  The
order
of the counts is the same as the order the logic is applied in the
code.
When I run your data at verbosity level 3, I see the following:

DEBUG 2: Processing APCP/A24 versus APCP/A24, for observation type
ADPSFC,
over region FULL, for interpolation method UW_MEAN(1), using 0 pairs.
DEBUG 3: Number of matched pairs  = 0
DEBUG 3: Observations processed   = 190
DEBUG 3: Rejected: GRIB code      = 0
DEBUG 3: Rejected: valid time     = 190
DEBUG 3: Rejected: bad obs value  = 0
DEBUG 3: Rejected: off the grid   = 0
DEBUG 3: Rejected: level mismatch = 0
DEBUG 3: Rejected: quality marker = 0
DEBUG 3: Rejected: message type   = 0
DEBUG 3: Rejected: masking region = 0
DEBUG 3: Rejected: bad fcst value = 0
DEBUG 3: Rejected: duplicates     = 0

In this case, all of the observations were discarded because their
valid
time wasn't close enough to the valid time of the forecast being
evaluated.  When you run Point-Stat, the "obs_window" setting in the
configuration file defines a time window around the forecast valid
time.
I see the following in the config file you sent... beg and end are
listed
in seconds:
   obs_window = {
      beg = -5400;
      end =  5400;
   }

So we're looking for observations whose valid time is +/- 1.5 hours
around
the forecast valid time.  Typically, when you run WRF and then UPP,
you end
up with a bunch of GRIB files, each with it's own valid time.  Looking
at
the GRIB records in your file, I see that all the records have the
same
initialization time, but there are multiple lead times.  And I see
that
there are two records for 0-24hr accumulations:

[johnhg at rambler]% wgrib -V 20121020_wrf.grb | grep "24hr"
rec 17:108670:date 2012102000 APCP kpds5=61 kpds6=1 kpds7=0
levels=(0,0)
grid=255 sfc 0-24hr acc:
rec 63:399072:date 2012102000 APCP kpds5=61 kpds6=1 kpds7=0
levels=(0,0)
grid=255 sfc 0-24hr acc:

Plotting the data, it looks like those records are duplicates.  That
isn't
really a problem, but you might want to look into why you ended up
with
duplicate records in the same file.

A 0-24 accumulation initialized at 20121020_00 would have a valid time
of
20121021_00.  With a +/- 1.5 hour time window, Point-Stat will only
use
observations falling between 20121020_233000 and 20121021_013000.
However,
all of your observations are valid at 20121020_120000, which is not in
that
window.

I tried rerunning your data with the "obs_window" set to +/- 540000
seconds, and I did get plenty of matched pairs.

Hope that helps get you going.

Thanks,
John Halley Gotway
met_help at ucar.edu



On Mon, Jul 28, 2014 at 6:37 AM, Elelwani Phaduli via RT
<met_help at ucar.edu>
wrote:

>
> Mon Jul 28 06:37:50 2014: Request 68435 was acted upon.
> Transaction: Ticket created by Elelwani.Phaduli at weathersa.co.za
>        Queue: met_help
>      Subject: point stat tool error
>        Owner: Nobody
>   Requestors: Elelwani.Phaduli at weathersa.co.za
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=68435 >
>
>
> Hi....
>
> I'm a scientist at the South African Weather Service. I'm testing
MET tool
> the following models (WRF and Unified Model). Currently I got good
results
> using  Gridstat tool on both models. I want to use Pointstat tool
with the
> data I have created using the ascci2nc tool, but I get empty output
text
> files. I loaded my data into your ftp server under the directory
>  /incoming/irap/met_help/phaduli_data.
>
> I also want to test the MODE tool using TRRM data(Netcdf or bin).
Please
> advise on this.
>
> Thank  you.
> Elelwani Phaduli
> (Scientist South African Weather Service)
>
>
>

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