[Met_help] [rt.rap.ucar.edu #67334] History for Problem with MET_TC tc_stat for models with only a track forecast

[Kathryn Newman via RT]

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

I am trying to run tc_stat on a model that only has a track forecast.
Examples include BAMS, BAMM, BAMD, or CLP5.  However, when I run tc_stat on
the attached tcst file, it can not find any cases with any of those models
and returns an empty output file.  If I include those models in a larger
group, it skips them and doesn't include them in the output.  Any thoughts
as to what the problem might be?

-Dave Zelinsky


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

Subject: Problem with MET_TC tc_stat for models with only a track forecast
From: Kathryn Newman
Time: Fri May 30 14:37:15 2014

Hi Dave,

My guess is that you are filtering on something that is requiring the
wind column to have real values. It looks like everything is correct
in your tc_pairs output file.  Could you send me either (1) your
tc_stat config file or (2) what you are typing on the command line?
That way I can try to reproduce the error and see where the track only
models are falling out.

Kathryn

------------------------------------------------
Subject: Problem with MET_TC tc_stat for models with only a track forecast
From: David Zelinsky - NOAA Affiliate
Time: Fri May 30 15:47:19 2014

Kathryn,

I was able to locate the problem.  Turns out those models have another
thing in common, no 00 hour forecast.  Typically, the NHC verification
only
verifies forecasts where the system was a tropical or subtropical
cyclone
at the initial time (Status HU, TS, TD, SD, SS), so I was filtering
with
init_str_name.  The BAM models and CLP5 were removed by that filter.
I
think one function that should be added in the future is the ability
to
filter for storm status at the initial time for a given forecast,
regardless of the availability of the model at the initial time.

-Dave


On Fri, May 30, 2014 at 4:37 PM, Kathryn Newman via RT
<met_help at ucar.edu>
wrote:

> Hi Dave,
>
> My guess is that you are filtering on something that is requiring
the wind
> column to have real values. It looks like everything is correct in
your
> tc_pairs output file.  Could you send me either (1) your tc_stat
config
> file or (2) what you are typing on the command line?  That way I can
try to
> reproduce the error and see where the track only models are falling
out.
>
> Kathryn
>
>

------------------------------------------------
Subject: Problem with MET_TC tc_stat for models with only a track forecast
From: Kathryn Newman
Time: Fri May 30 16:38:02 2014

Hi Dave,

Yep, that would do it! That is a good point... tc_stat currently
throws
out models that are missing the init time if you are filtering based
on
the init time, which isn't always optimal. John and I are looking into
how to best add the functionality you described.

Kathryn
On 5/30/14 3:47 PM, David Zelinsky - NOAA Affiliate via RT wrote:
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=67334 >
>
> Kathryn,
>
> I was able to locate the problem.  Turns out those models have
another
> thing in common, no 00 hour forecast.  Typically, the NHC
verification only
> verifies forecasts where the system was a tropical or subtropical
cyclone
> at the initial time (Status HU, TS, TD, SD, SS), so I was filtering
with
> init_str_name.  The BAM models and CLP5 were removed by that filter.
I
> think one function that should be added in the future is the ability
to
> filter for storm status at the initial time for a given forecast,
> regardless of the availability of the model at the initial time.
>
> -Dave
>
>
> On Fri, May 30, 2014 at 4:37 PM, Kathryn Newman via RT
<met_help at ucar.edu>
> wrote:
>
>> Hi Dave,
>>
>> My guess is that you are filtering on something that is requiring
the wind
>> column to have real values. It looks like everything is correct in
your
>> tc_pairs output file.  Could you send me either (1) your tc_stat
config
>> file or (2) what you are typing on the command line?  That way I
can try to
>> reproduce the error and see where the track only models are falling
out.
>>
>> Kathryn
>>
>>


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