[Met_help] [rt.rap.ucar.edu #84979] History for tc_pairs fill

John Halley Gotway via RT met_help at ucar.edu
Fri May 4 10:30:35 MDT 2018


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

Hi John,

I just noticed another issue while trying to do some verification. I'm
running a large verification experiment where I have many variants of the
same model. For the EMX experiments I've named them E00I (E002), E01I
(E012), E02I (E022), E03I (E032),... up to E16I (E162), where the I's are
for the 06 and 18 UTC cycles, and the 2's are for 00 and 12 UTC cycles. I'm
using interp12 = FILL, so that all of the cycles can be verified together.
All of the experiments seem to FILL correctly except for the experiments
with "3" in the 3rd digit of the ATCF ID. In other words, the FILL option
doesn't seem to work for experiments E03I and E13I. What's weird is that
for similar type experiments with other models such as HWRF, the FILL
option seems to work okay for all of the experiments that are named H00I
(H002), etc. I'm at a loss as to why this might be.

I've attached a very simplified ATCF file with just two dates/times and the
best track file. If you use the attached config files, hopefully you'll see
the same behavior. I'm running with version 7.0. I'm also attaching the
resulting .tcst file in case it differs from what you get.

tc_pairs -adeck ./aal112017.dat -bdeck ./bal112017.dat -config
./TCPairsConfig -out al112017
tc_stat -lookin ./al112017.tcst -config TCStatConfig

When I request "E00I" in the TCStatConfig, there are two verifiable cases,
which is correct with the FILL option. If I change this to "E03I" or
"E13I", there's only 1 verifiable case, which is incorrect.

One more thing... I also noticed that tc_pairs creates additional ATCF IDs
"E1I" and "E0I" in the .tcst file. Do you know why this happens?

Thanks for any help you can provide.

Take care,
Andy


-- 
Andrew Penny
Hurricane Model Diagnostician
SRG / National Hurricane Center
11691 SW 17th Street
Miami, FL 33165-2149
phone: 305.229.4457
email: andrew.penny at noaa.gov


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

Subject: tc_pairs fill
From: John Halley Gotway
Time: Thu May 03 14:27:22 2018

Andy,

Nice job tracking down such a pesky little issue!  And thanks for
sending
me a manageable set of data for testing.

I edited your config file to further narrow it down:
   model    = ["E02I","E022","E03I","E032" ];
   interp12 = ${INTERP12};
   lag_time = [];
   match_points = TRUE;

I ran this twice... once with the INTERP12 variable set to NONE and a
second time with it set to FILL.  Diffing the .tcst output files, I
clearly
see that the AMODEL has "E0I" in the output exactly where it should
have
"E03I" instead.

Running through the debugger, I found the problem with AFCF ID's
ending
with "32".  It's on lines 1023/1024 of tc_pairs.cc:

   1021       // Swap the '2' or '3' for an 'I' in the AMODEL
   1022       amodel = tracks[i].technique();       // amodel = E032
   1023       amodel.chomp('2');                        // amodel =
E03
   1024       amodel.chomp('3');                        // amodel = E0
   1025       amodel << 'I';                                // amodel
= E0I

After each line I've added a comment illustrating the problem... which
results in "E0I" in the output.

Here's the one line change that solves this problem:

   1021       // Swap the '2' or '3' for an 'I' in the AMODEL
   1022       amodel = tracks[i].technique();
   1023       amodel.chomp(c);
   1025       amodel << 'I';

I'll commit this change and post a new set of patches for met-7.0.

FYI, I noticed in the config file you sent that you have "model" set
to a
long list of strings.

Thanks,
John

On Wed, May 2, 2018 at 10:40 PM, Andrew Penny - NOAA Affiliate via RT
<
met_help at ucar.edu> wrote:

>
> Wed May 02 22:40:14 2018: Request 84979 was acted upon.
> Transaction: Ticket created by andrew.penny at noaa.gov
>        Queue: met_help
>      Subject: tc_pairs fill
>        Owner: Nobody
>   Requestors: andrew.penny at noaa.gov
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=84979 >
>
>
> Hi John,
>
> I just noticed another issue while trying to do some verification.
I'm
> running a large verification experiment where I have many variants
of the
> same model. For the EMX experiments I've named them E00I (E002),
E01I
> (E012), E02I (E022), E03I (E032),... up to E16I (E162), where the
I's are
> for the 06 and 18 UTC cycles, and the 2's are for 00 and 12 UTC
cycles. I'm
> using interp12 = FILL, so that all of the cycles can be verified
together.
> All of the experiments seem to FILL correctly except for the
experiments
> with "3" in the 3rd digit of the ATCF ID. In other words, the FILL
option
> doesn't seem to work for experiments E03I and E13I. What's weird is
that
> for similar type experiments with other models such as HWRF, the
FILL
> option seems to work okay for all of the experiments that are named
H00I
> (H002), etc. I'm at a loss as to why this might be.
>
> I've attached a very simplified ATCF file with just two dates/times
and the
> best track file. If you use the attached config files, hopefully
you'll see
> the same behavior. I'm running with version 7.0. I'm also attaching
the
> resulting .tcst file in case it differs from what you get.
>
> tc_pairs -adeck ./aal112017.dat -bdeck ./bal112017.dat -config
> ./TCPairsConfig -out al112017
> tc_stat -lookin ./al112017.tcst -config TCStatConfig
>
> When I request "E00I" in the TCStatConfig, there are two verifiable
cases,
> which is correct with the FILL option. If I change this to "E03I" or
> "E13I", there's only 1 verifiable case, which is incorrect.
>
> One more thing... I also noticed that tc_pairs creates additional
ATCF IDs
> "E1I" and "E0I" in the .tcst file. Do you know why this happens?
>
> Thanks for any help you can provide.
>
> Take care,
> Andy
>
>
> --
> Andrew Penny
> Hurricane Model Diagnostician
> SRG / National Hurricane Center
> 11691 SW 17th Street
> Miami, FL 33165-2149
> phone: 305.229.4457
> email: andrew.penny at noaa.gov
>
>

------------------------------------------------
Subject: tc_pairs fill
From: John Halley Gotway
Time: Thu May 03 14:46:55 2018

Sorry, I never finished my last thought there...

FYI, I noticed in the config file you sent that you have "model" set
to a
long list of strings.  By default, when the "model" list is empty,
tc_pairs
just processes all the models it finds in the input.  You only need to
specify a list if you want to exclude something.  I suspect that
that's
what you're actually doing in your real runs, but only sent me a
subset of
the data for testing.  Just wanted to make sure though.

Thanks,
John


On Thu, May 3, 2018 at 2:26 PM, John Halley Gotway <johnhg at ucar.edu>
wrote:

> Andy,
>
> Nice job tracking down such a pesky little issue!  And thanks for
sending
> me a manageable set of data for testing.
>
> I edited your config file to further narrow it down:
>    model    = ["E02I","E022","E03I","E032" ];
>    interp12 = ${INTERP12};
>    lag_time = [];
>    match_points = TRUE;
>
> I ran this twice... once with the INTERP12 variable set to NONE and
a
> second time with it set to FILL.  Diffing the .tcst output files, I
clearly
> see that the AMODEL has "E0I" in the output exactly where it should
have
> "E03I" instead.
>
> Running through the debugger, I found the problem with AFCF ID's
ending
> with "32".  It's on lines 1023/1024 of tc_pairs.cc:
>
>    1021       // Swap the '2' or '3' for an 'I' in the AMODEL
>    1022       amodel = tracks[i].technique();       // amodel = E032
>    1023       amodel.chomp('2');                        // amodel =
E03
>    1024       amodel.chomp('3');                        // amodel =
E0
>    1025       amodel << 'I';                                //
amodel =
> E0I
>
> After each line I've added a comment illustrating the problem...
which
> results in "E0I" in the output.
>
> Here's the one line change that solves this problem:
>
>    1021       // Swap the '2' or '3' for an 'I' in the AMODEL
>    1022       amodel = tracks[i].technique();
>    1023       amodel.chomp(c);
>    1025       amodel << 'I';
>
> I'll commit this change and post a new set of patches for met-7.0.
>
> FYI, I noticed in the config file you sent that you have "model" set
to a
> long list of strings.
>
> Thanks,
> John
>
> On Wed, May 2, 2018 at 10:40 PM, Andrew Penny - NOAA Affiliate via
RT <
> met_help at ucar.edu> wrote:
>
>>
>> Wed May 02 22:40:14 2018: Request 84979 was acted upon.
>> Transaction: Ticket created by andrew.penny at noaa.gov
>>        Queue: met_help
>>      Subject: tc_pairs fill
>>        Owner: Nobody
>>   Requestors: andrew.penny at noaa.gov
>>       Status: new
>>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=84979 >
>>
>>
>> Hi John,
>>
>> I just noticed another issue while trying to do some verification.
I'm
>> running a large verification experiment where I have many variants
of the
>> same model. For the EMX experiments I've named them E00I (E002),
E01I
>> (E012), E02I (E022), E03I (E032),... up to E16I (E162), where the
I's are
>> for the 06 and 18 UTC cycles, and the 2's are for 00 and 12 UTC
cycles.
>> I'm
>> using interp12 = FILL, so that all of the cycles can be verified
together.
>> All of the experiments seem to FILL correctly except for the
experiments
>> with "3" in the 3rd digit of the ATCF ID. In other words, the FILL
option
>> doesn't seem to work for experiments E03I and E13I. What's weird is
that
>> for similar type experiments with other models such as HWRF, the
FILL
>> option seems to work okay for all of the experiments that are named
H00I
>> (H002), etc. I'm at a loss as to why this might be.
>>
>> I've attached a very simplified ATCF file with just two dates/times
and
>> the
>> best track file. If you use the attached config files, hopefully
you'll
>> see
>> the same behavior. I'm running with version 7.0. I'm also attaching
the
>> resulting .tcst file in case it differs from what you get.
>>
>> tc_pairs -adeck ./aal112017.dat -bdeck ./bal112017.dat -config
>> ./TCPairsConfig -out al112017
>> tc_stat -lookin ./al112017.tcst -config TCStatConfig
>>
>> When I request "E00I" in the TCStatConfig, there are two verifiable
cases,
>> which is correct with the FILL option. If I change this to "E03I"
or
>> "E13I", there's only 1 verifiable case, which is incorrect.
>>
>> One more thing... I also noticed that tc_pairs creates additional
ATCF IDs
>> "E1I" and "E0I" in the .tcst file. Do you know why this happens?
>>
>> Thanks for any help you can provide.
>>
>> Take care,
>> Andy
>>
>>
>> --
>> Andrew Penny
>> Hurricane Model Diagnostician
>> SRG / National Hurricane Center
>> 11691 SW 17th Street
>> Miami, FL 33165-2149
>> phone: 305.229.4457
>> email: andrew.penny at noaa.gov
>>
>>
>

------------------------------------------------
Subject: tc_pairs fill
From: John Halley Gotway
Time: Thu May 03 15:02:21 2018

Andy,

OK, bugfix is posted for met-7.0 here:
    https://dtcenter.org/met/users/support/known_issues/METv7.0/index.php

Sorry about this bug, but thanks for catching it!

John

On Thu, May 3, 2018 at 2:46 PM, John Halley Gotway <johnhg at ucar.edu>
wrote:

> Sorry, I never finished my last thought there...
>
> FYI, I noticed in the config file you sent that you have "model" set
to a
> long list of strings.  By default, when the "model" list is empty,
tc_pairs
> just processes all the models it finds in the input.  You only need
to
> specify a list if you want to exclude something.  I suspect that
that's
> what you're actually doing in your real runs, but only sent me a
subset of
> the data for testing.  Just wanted to make sure though.
>
> Thanks,
> John
>
>
> On Thu, May 3, 2018 at 2:26 PM, John Halley Gotway <johnhg at ucar.edu>
> wrote:
>
>> Andy,
>>
>> Nice job tracking down such a pesky little issue!  And thanks for
sending
>> me a manageable set of data for testing.
>>
>> I edited your config file to further narrow it down:
>>    model    = ["E02I","E022","E03I","E032" ];
>>    interp12 = ${INTERP12};
>>    lag_time = [];
>>    match_points = TRUE;
>>
>> I ran this twice... once with the INTERP12 variable set to NONE and
a
>> second time with it set to FILL.  Diffing the .tcst output files, I
clearly
>> see that the AMODEL has "E0I" in the output exactly where it should
have
>> "E03I" instead.
>>
>> Running through the debugger, I found the problem with AFCF ID's
ending
>> with "32".  It's on lines 1023/1024 of tc_pairs.cc:
>>
>>    1021       // Swap the '2' or '3' for an 'I' in the AMODEL
>>    1022       amodel = tracks[i].technique();       // amodel =
E032
>>    1023       amodel.chomp('2');                        // amodel =
E03
>>    1024       amodel.chomp('3');                        // amodel =
E0
>>    1025       amodel << 'I';                                //
amodel =
>> E0I
>>
>> After each line I've added a comment illustrating the problem...
which
>> results in "E0I" in the output.
>>
>> Here's the one line change that solves this problem:
>>
>>    1021       // Swap the '2' or '3' for an 'I' in the AMODEL
>>    1022       amodel = tracks[i].technique();
>>    1023       amodel.chomp(c);
>>    1025       amodel << 'I';
>>
>> I'll commit this change and post a new set of patches for met-7.0.
>>
>> FYI, I noticed in the config file you sent that you have "model"
set to a
>> long list of strings.
>>
>> Thanks,
>> John
>>
>> On Wed, May 2, 2018 at 10:40 PM, Andrew Penny - NOAA Affiliate via
RT <
>> met_help at ucar.edu> wrote:
>>
>>>
>>> Wed May 02 22:40:14 2018: Request 84979 was acted upon.
>>> Transaction: Ticket created by andrew.penny at noaa.gov
>>>        Queue: met_help
>>>      Subject: tc_pairs fill
>>>        Owner: Nobody
>>>   Requestors: andrew.penny at noaa.gov
>>>       Status: new
>>>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=84979 >
>>>
>>>
>>> Hi John,
>>>
>>> I just noticed another issue while trying to do some verification.
I'm
>>> running a large verification experiment where I have many variants
of the
>>> same model. For the EMX experiments I've named them E00I (E002),
E01I
>>> (E012), E02I (E022), E03I (E032),... up to E16I (E162), where the
I's are
>>> for the 06 and 18 UTC cycles, and the 2's are for 00 and 12 UTC
cycles.
>>> I'm
>>> using interp12 = FILL, so that all of the cycles can be verified
>>> together.
>>> All of the experiments seem to FILL correctly except for the
experiments
>>> with "3" in the 3rd digit of the ATCF ID. In other words, the FILL
option
>>> doesn't seem to work for experiments E03I and E13I. What's weird
is that
>>> for similar type experiments with other models such as HWRF, the
FILL
>>> option seems to work okay for all of the experiments that are
named H00I
>>> (H002), etc. I'm at a loss as to why this might be.
>>>
>>> I've attached a very simplified ATCF file with just two
dates/times and
>>> the
>>> best track file. If you use the attached config files, hopefully
you'll
>>> see
>>> the same behavior. I'm running with version 7.0. I'm also
attaching the
>>> resulting .tcst file in case it differs from what you get.
>>>
>>> tc_pairs -adeck ./aal112017.dat -bdeck ./bal112017.dat -config
>>> ./TCPairsConfig -out al112017
>>> tc_stat -lookin ./al112017.tcst -config TCStatConfig
>>>
>>> When I request "E00I" in the TCStatConfig, there are two
verifiable
>>> cases,
>>> which is correct with the FILL option. If I change this to "E03I"
or
>>> "E13I", there's only 1 verifiable case, which is incorrect.
>>>
>>> One more thing... I also noticed that tc_pairs creates additional
ATCF
>>> IDs
>>> "E1I" and "E0I" in the .tcst file. Do you know why this happens?
>>>
>>> Thanks for any help you can provide.
>>>
>>> Take care,
>>> Andy
>>>
>>>
>>> --
>>> Andrew Penny
>>> Hurricane Model Diagnostician
>>> SRG / National Hurricane Center
>>> 11691 SW 17th Street
>>> Miami, FL 33165-2149
>>> phone: 305.229.4457
>>> email: andrew.penny at noaa.gov
>>>
>>>
>>
>

------------------------------------------------
Subject: tc_pairs fill
From: Andrew Penny - NOAA Affiliate
Time: Thu May 03 15:49:21 2018

Great! Thanks John. Much appreciated. I have a bunch of things running
now,
but will try to recompile the code tomorrow and see how it works.

Take care,
Andy

On Thu, May 3, 2018 at 9:02 PM, John Halley Gotway via RT
<met_help at ucar.edu
> wrote:

> Andy,
>
> OK, bugfix is posted for met-7.0 here:
>
https://dtcenter.org/met/users/support/known_issues/METv7.0/index.php
>
> Sorry about this bug, but thanks for catching it!
>
> John
>
> On Thu, May 3, 2018 at 2:46 PM, John Halley Gotway <johnhg at ucar.edu>
> wrote:
>
> > Sorry, I never finished my last thought there...
> >
> > FYI, I noticed in the config file you sent that you have "model"
set to a
> > long list of strings.  By default, when the "model" list is empty,
> tc_pairs
> > just processes all the models it finds in the input.  You only
need to
> > specify a list if you want to exclude something.  I suspect that
that's
> > what you're actually doing in your real runs, but only sent me a
subset
> of
> > the data for testing.  Just wanted to make sure though.
> >
> > Thanks,
> > John
> >
> >
> > On Thu, May 3, 2018 at 2:26 PM, John Halley Gotway
<johnhg at ucar.edu>
> > wrote:
> >
> >> Andy,
> >>
> >> Nice job tracking down such a pesky little issue!  And thanks for
> sending
> >> me a manageable set of data for testing.
> >>
> >> I edited your config file to further narrow it down:
> >>    model    = ["E02I","E022","E03I","E032" ];
> >>    interp12 = ${INTERP12};
> >>    lag_time = [];
> >>    match_points = TRUE;
> >>
> >> I ran this twice... once with the INTERP12 variable set to NONE
and a
> >> second time with it set to FILL.  Diffing the .tcst output files,
I
> clearly
> >> see that the AMODEL has "E0I" in the output exactly where it
should have
> >> "E03I" instead.
> >>
> >> Running through the debugger, I found the problem with AFCF ID's
ending
> >> with "32".  It's on lines 1023/1024 of tc_pairs.cc:
> >>
> >>    1021       // Swap the '2' or '3' for an 'I' in the AMODEL
> >>    1022       amodel = tracks[i].technique();       // amodel =
E032
> >>    1023       amodel.chomp('2');                        // amodel
= E03
> >>    1024       amodel.chomp('3');                        // amodel
= E0
> >>    1025       amodel << 'I';                                //
amodel =
> >> E0I
> >>
> >> After each line I've added a comment illustrating the problem...
which
> >> results in "E0I" in the output.
> >>
> >> Here's the one line change that solves this problem:
> >>
> >>    1021       // Swap the '2' or '3' for an 'I' in the AMODEL
> >>    1022       amodel = tracks[i].technique();
> >>    1023       amodel.chomp(c);
> >>    1025       amodel << 'I';
> >>
> >> I'll commit this change and post a new set of patches for met-
7.0.
> >>
> >> FYI, I noticed in the config file you sent that you have "model"
set to
> a
> >> long list of strings.
> >>
> >> Thanks,
> >> John
> >>
> >> On Wed, May 2, 2018 at 10:40 PM, Andrew Penny - NOAA Affiliate
via RT <
> >> met_help at ucar.edu> wrote:
> >>
> >>>
> >>> Wed May 02 22:40:14 2018: Request 84979 was acted upon.
> >>> Transaction: Ticket created by andrew.penny at noaa.gov
> >>>        Queue: met_help
> >>>      Subject: tc_pairs fill
> >>>        Owner: Nobody
> >>>   Requestors: andrew.penny at noaa.gov
> >>>       Status: new
> >>>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=84979
> >
> >>>
> >>>
> >>> Hi John,
> >>>
> >>> I just noticed another issue while trying to do some
verification. I'm
> >>> running a large verification experiment where I have many
variants of
> the
> >>> same model. For the EMX experiments I've named them E00I (E002),
E01I
> >>> (E012), E02I (E022), E03I (E032),... up to E16I (E162), where
the I's
> are
> >>> for the 06 and 18 UTC cycles, and the 2's are for 00 and 12 UTC
cycles.
> >>> I'm
> >>> using interp12 = FILL, so that all of the cycles can be verified
> >>> together.
> >>> All of the experiments seem to FILL correctly except for the
> experiments
> >>> with "3" in the 3rd digit of the ATCF ID. In other words, the
FILL
> option
> >>> doesn't seem to work for experiments E03I and E13I. What's weird
is
> that
> >>> for similar type experiments with other models such as HWRF, the
FILL
> >>> option seems to work okay for all of the experiments that are
named
> H00I
> >>> (H002), etc. I'm at a loss as to why this might be.
> >>>
> >>> I've attached a very simplified ATCF file with just two
dates/times and
> >>> the
> >>> best track file. If you use the attached config files, hopefully
you'll
> >>> see
> >>> the same behavior. I'm running with version 7.0. I'm also
attaching the
> >>> resulting .tcst file in case it differs from what you get.
> >>>
> >>> tc_pairs -adeck ./aal112017.dat -bdeck ./bal112017.dat -config
> >>> ./TCPairsConfig -out al112017
> >>> tc_stat -lookin ./al112017.tcst -config TCStatConfig
> >>>
> >>> When I request "E00I" in the TCStatConfig, there are two
verifiable
> >>> cases,
> >>> which is correct with the FILL option. If I change this to
"E03I" or
> >>> "E13I", there's only 1 verifiable case, which is incorrect.
> >>>
> >>> One more thing... I also noticed that tc_pairs creates
additional ATCF
> >>> IDs
> >>> "E1I" and "E0I" in the .tcst file. Do you know why this happens?
> >>>
> >>> Thanks for any help you can provide.
> >>>
> >>> Take care,
> >>> Andy
> >>>
> >>>
> >>> --
> >>> Andrew Penny
> >>> Hurricane Model Diagnostician
> >>> SRG / National Hurricane Center
> >>> 11691 SW 17th Street
> >>> Miami, FL 33165-2149
> >>> phone: 305.229.4457
> >>> email: andrew.penny at noaa.gov
> >>>
> >>>
> >>
> >
>
>


--
Andrew Penny
Hurricane Model Diagnostician
SRG / National Hurricane Center
11691 SW 17th Street
Miami, FL 33165-2149
phone: 305.229.4457
email: andrew.penny at noaa.gov

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


More information about the Met_help mailing list