[Met_help] [rt.rap.ucar.edu #71028] History for Issue with width function
John Halley Gotway via RT
met_help at ucar.edu
Thu Mar 26 09:35:38 MDT 2015
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hi John. I have one doubt. there are two parts in the GridStatConfig file which contain the variable width. // Interpolation methods//interp = { field = BOTH; vld_thresh = 1.0;
type = [ { method = UW_MEAN; width = 1; width = 3;
} ];};
////////////////////////////////////////////////////////////////////////////////
//// Neighborhood methods//nbrhd = { vld_thresh = 1.0; width = [ 1 ]; cov_thresh = [ >=0.5 ];}
My question is if i want to see the ETS if I UPSCALe, then which WIDTH function I have to change within the INTERP group or the Neighbourhood methods????According to me I should increase the width under the neighbourhood Group. Is that correct???
So essentially what is the difference between the two WIDTH functions under DIFFERENT GROUPS (interp and neighbourhood).
I added the Width=1 and width=3 in the first one and could not see any difference in my RESULTS.
Can U pls explain the difference.
Geeta
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Issue with width function
From: John Halley Gotway
Time: Tue Mar 10 14:26:58 2015
Geeta,
I combined your two recent questions into a single support ticket.
I see that you are asking about the "width" settings in the Grid-Stat
configuration file. "width" shows up in two places there, once in the
"interp" section and once in the "nbrhd" section. I'll do my best to
explain the difference.
The "nbrhd" section is used to define how the neighborhood
verification
methods are applied. Basically, this affects the output you see in
the
NBRCNT, NBRCTC, and NBRCTS output line types from Grid-Stat.
The "interp" section is used to do up-scaling. This is basically a
smoothing of the forecast and/or observation field. This affect the
output
you see all of the other non-neighborhood line type such as CNT, CTS,
and
so on. You're interested in the equitable threat score which is named
the
Gilbert Skill Score (GSS) in the CTS line type, and you want to see
how
upscaling impacts that score. Therefore you should set one or more
upscaling methods in the "interp" section and see how GSS varies.
You could try this:
interp = {
field = FCST;
vld_thresh = 1.0;
type = [
{ method = UW_MEAN; width = 1; },
{ method = UW_MEAN; width = 3; },
{ method = UW_MEAN; width = 5; },
{ method = UW_MEAN; width = 7; },
{ method = UW_MEAN; width = 9; }
];
};
Since I set "field = FCST;" we'll only smooth the forecast field,
leaving
the observations unchanged. We'll smooth the data using a 1x1 box (no
smoothing), 3x3 box, ..., all the way up to a 9x9 box.
I tried this on some sample data and found that the more I smoothed
the
forecast, the worst the GSS performed! However, the more I smoothed
the
forecast the better the RMSE (in the CNT line type) performed.
Generally
that is the case... the smoother the forecast, the higher the
continuous
statistics, such as RMSE.
Hope that helps.
Thanks,
John Halley Gotway
On Mon, Mar 9, 2015 at 10:54 AM, The RT System itself via RT <
met_help at ucar.edu> wrote:
>
> Mon Mar 09 10:54:50 2015: Request 71028 was acted upon.
> Transaction: Given to johnhg by RT_System
> Queue: met_help
> Subject: Issue with width function
> Owner: johnhg
> Requestors: geeta124 at hotmail.com
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71028 >
>
>
> This transaction appears to have no content
>
------------------------------------------------
Subject: Issue with width function
From: Geeta Geeta
Time: Wed Mar 11 09:52:24 2015
Hi John. I have the following entries in my config
file............nterp = { field = BOTH; vld_thresh = 1.0;
type = [ { method = UW_MEAN; width = 1;
width = 3;
} ]; };
But when I see my CNT output, I do not get different values of RMSE
and GSS corresponding to WIDTH=1 and WIDTH=3 . But I get only one
value.
Sending you my CNT output. I thought I would get one CNT output for
one width and then 2nd output CTS for the width=3. so that I can
compare them to see if RMSE improves by upscaling. Can u pls explain
the output that grid stat will write.
Geeta
> Subject: Re: [rt.rap.ucar.edu #71028] Issue with width function
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Tue, 10 Mar 2015 14:26:58 -0600
>
> Geeta,
>
> I combined your two recent questions into a single support ticket.
>
> I see that you are asking about the "width" settings in the Grid-
Stat
> configuration file. "width" shows up in two places there, once in
the
> "interp" section and once in the "nbrhd" section. I'll do my best
to
> explain the difference.
>
> The "nbrhd" section is used to define how the neighborhood
verification
> methods are applied. Basically, this affects the output you see in
the
> NBRCNT, NBRCTC, and NBRCTS output line types from Grid-Stat.
>
> The "interp" section is used to do up-scaling. This is basically a
> smoothing of the forecast and/or observation field. This affect the
output
> you see all of the other non-neighborhood line type such as CNT,
CTS, and
> so on. You're interested in the equitable threat score which is
named the
> Gilbert Skill Score (GSS) in the CTS line type, and you want to see
how
> upscaling impacts that score. Therefore you should set one or more
> upscaling methods in the "interp" section and see how GSS varies.
>
> You could try this:
> interp = {
> field = FCST;
> vld_thresh = 1.0;
>
> type = [
> { method = UW_MEAN; width = 1; },
> { method = UW_MEAN; width = 3; },
> { method = UW_MEAN; width = 5; },
> { method = UW_MEAN; width = 7; },
> { method = UW_MEAN; width = 9; }
> ];
> };
>
> Since I set "field = FCST;" we'll only smooth the forecast field,
leaving
> the observations unchanged. We'll smooth the data using a 1x1 box
(no
> smoothing), 3x3 box, ..., all the way up to a 9x9 box.
>
> I tried this on some sample data and found that the more I smoothed
the
> forecast, the worst the GSS performed! However, the more I smoothed
the
> forecast the better the RMSE (in the CNT line type) performed.
Generally
> that is the case... the smoother the forecast, the higher the
continuous
> statistics, such as RMSE.
>
> Hope that helps.
>
> Thanks,
> John Halley Gotway
>
>
> On Mon, Mar 9, 2015 at 10:54 AM, The RT System itself via RT <
> met_help at ucar.edu> wrote:
>
> >
> > Mon Mar 09 10:54:50 2015: Request 71028 was acted upon.
> > Transaction: Given to johnhg by RT_System
> > Queue: met_help
> > Subject: Issue with width function
> > Owner: johnhg
> > Requestors: geeta124 at hotmail.com
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71028 >
> >
> >
> > This transaction appears to have no content
> >
>
------------------------------------------------
Subject: Issue with width function
From: Geeta Geeta
Time: Wed Mar 11 09:52:24 2015
VERSION MODEL FCST_LEAD FCST_VALID_BEG FCST_VALID_END OBS_LEAD
OBS_VALID_BEG OBS_VALID_END FCST_VAR FCST_LEV OBS_VAR OBS_LEV
OBTYPE VX_MASK INTERP_MTHD INTERP_PNTS FCST_THRESH OBS_THRESH
COV_THRESH ALPHA LINE_TYPE TOTAL FBAR FBAR_NCL FBAR_NCU FBAR_BCL
FBAR_BCU FSTDEV FSTDEV_NCL FSTDEV_NCU FSTDEV_BCL FSTDEV_BCU OBAR
OBAR_NCL OBAR_NCU OBAR_BCL OBAR_BCU OSTDEV OSTDEV_NCL OSTDEV_NCU
OSTDEV_BCL OSTDEV_BCU PR_CORR PR_CORR_NCL PR_CORR_NCU PR_CORR_BCL
PR_CORR_BCU SP_CORR KT_CORR RANKS FRANK_TIES ORANK_TIES ME
ME_NCL ME_NCU ME_BCL ME_BCU ESTDEV ESTDEV_NCL ESTDEV_NCU
ESTDEV_BCL ESTDEV_BCU MBIAS MBIAS_BCL MBIAS_BCU MAE MAE_BCL
MAE_BCU MSE MSE_BCL MSE_BCU BCMSE BCMSE_BCL BCMSE_BCU
RMSE RMSE_BCL RMSE_BCU E10 E10_BCL E10_BCU E25 E25_BCL
E25_BCU E50 E50_BCL E50_BCU E75 E75_BCL E75_BCU E90
E90_BCL E90_BCU EIQR EIQR_BCL EIQR_BCU MAD MAD_BCL MAD_BCU
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 9 NA NA
NA 0.05000 CNT 870 9.60747 8.89059 10.32435 NA NA
10.78838 10.30418 11.32068 NA NA 11.11896 10.24189
11.99602 NA NA 13.19914 12.60674 13.85039 NA
NA 0.57308 0.52668 0.61608 NA NA NA
NA 0 0 0 -1.51148 -2.26149 -0.76148 NA
NA 11.28693 10.78036 11.84384 NA NA 0.86406
NA NA 7.46912 NA NA 129.53300 NA NA
127.24841 NA NA 11.38126 NA NA -17.01502
NA NA -5.83801 NA NA 0.00000 NA NA 3.84078
NA NA 11.18719 NA NA 9.67879 NA NA 4.60421
NA NA
------------------------------------------------
Subject: Issue with width function
From: John Halley Gotway
Time: Wed Mar 11 13:17:50 2015
Geeta,
Please refer to the example I sent previously. I'll also copy and
paste it
below. You have the config file syntax wrong...
interp = {
field = FCST;
vld_thresh = 1.0;
type = [
{ method = UW_MEAN; width = 1; },
{ method = UW_MEAN; width = 3; },
{ method = UW_MEAN; width = 5; },
{ method = UW_MEAN; width = 7; },
{ method = UW_MEAN; width = 9; }
];
};
Thanks,
John
On Wed, Mar 11, 2015 at 9:52 AM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71028 >
>
> Hi John. I have the following entries in my config
file............nterp =
> { field = BOTH; vld_thresh = 1.0;
> type = [ { method = UW_MEAN; width = 1;
> width = 3;
> } ]; };
> But when I see my CNT output, I do not get different values of RMSE
and
> GSS corresponding to WIDTH=1 and WIDTH=3 . But I get only one value.
> Sending you my CNT output. I thought I would get one CNT output for
one
> width and then 2nd output CTS for the width=3. so that I can compare
them
> to see if RMSE improves by upscaling. Can u pls explain the output
that
> grid stat will write.
> Geeta
> > Subject: Re: [rt.rap.ucar.edu #71028] Issue with width function
> > From: met_help at ucar.edu
> > To: geeta124 at hotmail.com
> > Date: Tue, 10 Mar 2015 14:26:58 -0600
> >
> > Geeta,
> >
> > I combined your two recent questions into a single support ticket.
> >
> > I see that you are asking about the "width" settings in the Grid-
Stat
> > configuration file. "width" shows up in two places there, once in
the
> > "interp" section and once in the "nbrhd" section. I'll do my best
to
> > explain the difference.
> >
> > The "nbrhd" section is used to define how the neighborhood
verification
> > methods are applied. Basically, this affects the output you see
in the
> > NBRCNT, NBRCTC, and NBRCTS output line types from Grid-Stat.
> >
> > The "interp" section is used to do up-scaling. This is basically
a
> > smoothing of the forecast and/or observation field. This affect
the
> output
> > you see all of the other non-neighborhood line type such as CNT,
CTS, and
> > so on. You're interested in the equitable threat score which is
named
> the
> > Gilbert Skill Score (GSS) in the CTS line type, and you want to
see how
> > upscaling impacts that score. Therefore you should set one or
more
> > upscaling methods in the "interp" section and see how GSS varies.
> >
> > You could try this:
> > interp = {
> > field = FCST;
> > vld_thresh = 1.0;
> >
> > type = [
> > { method = UW_MEAN; width = 1; },
> > { method = UW_MEAN; width = 3; },
> > { method = UW_MEAN; width = 5; },
> > { method = UW_MEAN; width = 7; },
> > { method = UW_MEAN; width = 9; }
> > ];
> > };
> >
> > Since I set "field = FCST;" we'll only smooth the forecast field,
leaving
> > the observations unchanged. We'll smooth the data using a 1x1 box
(no
> > smoothing), 3x3 box, ..., all the way up to a 9x9 box.
> >
> > I tried this on some sample data and found that the more I
smoothed the
> > forecast, the worst the GSS performed! However, the more I
smoothed the
> > forecast the better the RMSE (in the CNT line type) performed.
Generally
> > that is the case... the smoother the forecast, the higher the
continuous
> > statistics, such as RMSE.
> >
> > Hope that helps.
> >
> > Thanks,
> > John Halley Gotway
> >
> >
> > On Mon, Mar 9, 2015 at 10:54 AM, The RT System itself via RT <
> > met_help at ucar.edu> wrote:
> >
> > >
> > > Mon Mar 09 10:54:50 2015: Request 71028 was acted upon.
> > > Transaction: Given to johnhg by RT_System
> > > Queue: met_help
> > > Subject: Issue with width function
> > > Owner: johnhg
> > > Requestors: geeta124 at hotmail.com
> > > Status: new
> > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71028
> >
> > >
> > >
> > > This transaction appears to have no content
> > >
> >
>
>
------------------------------------------------
Subject: Issue with width function
From: Geeta Geeta
Time: Wed Mar 11 23:22:45 2015
thanks John. it worked. But further one more question, so the width of
the interp section is essentially upscaling the data (both obs and
fcst or either one on my choice) and then the neighbourhood statistics
is computed depending on the neightbourhood width and the threshold
selected. So HOW will the observations and FCST be treated, upscaled
to 3 or 5 or 1 ??? Can U pls explain. Hope I have made myself clear
!!!!!!!!!!!!!!!!Again what i mean to ask is whether the data is
considered haveing upscaling at 1, 3 or 5 level????. Following is the
config file
entries..........////////////////////////////////////////////////////////////////////////////////
//// Interpolation methods//interp = { field = BOTH;
vld_thresh = 1.0;
type = [ {method = UW_MEAN; width = 1;}, {method =
UW_MEAN; width = 3;}, {method = UW_MEAN; width = 5;} ];};
////////////////////////////////////////////////////////////////////////////////
//// Neighborhood methods//nbrhd = { vld_thresh = 1.0; width
= [ 1 ]; width = [ 3 ]; cov_thresh = [ >=0.5 ];}
I am sending you the O/P CNT output and NBRCNT output of this config
file.
If U just look at the NBRCNT o/p, then it is giving me the o/p
corresponding to only the NBR width=3. Can U pls
explain...................Geeta
> Subject: Re: [rt.rap.ucar.edu #71028] Issue with width function
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Wed, 11 Mar 2015 13:17:50 -0600
>
> Geeta,
>
> Please refer to the example I sent previously. I'll also copy and
paste it
> below. You have the config file syntax wrong...
>
> interp = {
> field = FCST;
> vld_thresh = 1.0;
>
> type = [
> { method = UW_MEAN; width = 1; },
> { method = UW_MEAN; width = 3; },
> { method = UW_MEAN; width = 5; },
> { method = UW_MEAN; width = 7; },
> { method = UW_MEAN; width = 9; }
> ];
> };
>
> Thanks,
> John
>
> On Wed, Mar 11, 2015 at 9:52 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71028 >
> >
> > Hi John. I have the following entries in my config
file............nterp =
> > { field = BOTH; vld_thresh = 1.0;
> > type = [ { method = UW_MEAN; width = 1;
> > width = 3;
> > } ]; };
> > But when I see my CNT output, I do not get different values of
RMSE and
> > GSS corresponding to WIDTH=1 and WIDTH=3 . But I get only one
value.
> > Sending you my CNT output. I thought I would get one CNT output
for one
> > width and then 2nd output CTS for the width=3. so that I can
compare them
> > to see if RMSE improves by upscaling. Can u pls explain the
output that
> > grid stat will write.
> > Geeta
> > > Subject: Re: [rt.rap.ucar.edu #71028] Issue with width function
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Tue, 10 Mar 2015 14:26:58 -0600
> > >
> > > Geeta,
> > >
> > > I combined your two recent questions into a single support
ticket.
> > >
> > > I see that you are asking about the "width" settings in the
Grid-Stat
> > > configuration file. "width" shows up in two places there, once
in the
> > > "interp" section and once in the "nbrhd" section. I'll do my
best to
> > > explain the difference.
> > >
> > > The "nbrhd" section is used to define how the neighborhood
verification
> > > methods are applied. Basically, this affects the output you see
in the
> > > NBRCNT, NBRCTC, and NBRCTS output line types from Grid-Stat.
> > >
> > > The "interp" section is used to do up-scaling. This is
basically a
> > > smoothing of the forecast and/or observation field. This affect
the
> > output
> > > you see all of the other non-neighborhood line type such as CNT,
CTS, and
> > > so on. You're interested in the equitable threat score which is
named
> > the
> > > Gilbert Skill Score (GSS) in the CTS line type, and you want to
see how
> > > upscaling impacts that score. Therefore you should set one or
more
> > > upscaling methods in the "interp" section and see how GSS
varies.
> > >
> > > You could try this:
> > > interp = {
> > > field = FCST;
> > > vld_thresh = 1.0;
> > >
> > > type = [
> > > { method = UW_MEAN; width = 1; },
> > > { method = UW_MEAN; width = 3; },
> > > { method = UW_MEAN; width = 5; },
> > > { method = UW_MEAN; width = 7; },
> > > { method = UW_MEAN; width = 9; }
> > > ];
> > > };
> > >
> > > Since I set "field = FCST;" we'll only smooth the forecast
field, leaving
> > > the observations unchanged. We'll smooth the data using a 1x1
box (no
> > > smoothing), 3x3 box, ..., all the way up to a 9x9 box.
> > >
> > > I tried this on some sample data and found that the more I
smoothed the
> > > forecast, the worst the GSS performed! However, the more I
smoothed the
> > > forecast the better the RMSE (in the CNT line type) performed.
Generally
> > > that is the case... the smoother the forecast, the higher the
continuous
> > > statistics, such as RMSE.
> > >
> > > Hope that helps.
> > >
> > > Thanks,
> > > John Halley Gotway
> > >
> > >
> > > On Mon, Mar 9, 2015 at 10:54 AM, The RT System itself via RT <
> > > met_help at ucar.edu> wrote:
> > >
> > > >
> > > > Mon Mar 09 10:54:50 2015: Request 71028 was acted upon.
> > > > Transaction: Given to johnhg by RT_System
> > > > Queue: met_help
> > > > Subject: Issue with width function
> > > > Owner: johnhg
> > > > Requestors: geeta124 at hotmail.com
> > > > Status: new
> > > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71028
> > >
> > > >
> > > >
> > > > This transaction appears to have no content
> > > >
> > >
> >
> >
>
------------------------------------------------
Subject: Issue with width function
From: Geeta Geeta
Time: Wed Mar 11 23:22:45 2015
VERSION MODEL FCST_LEAD FCST_VALID_BEG FCST_VALID_END OBS_LEAD
OBS_VALID_BEG OBS_VALID_END FCST_VAR FCST_LEV OBS_VAR OBS_LEV
OBTYPE VX_MASK INTERP_MTHD INTERP_PNTS FCST_THRESH OBS_THRESH
COV_THRESH ALPHA LINE_TYPE TOTAL FBAR FBAR_NCL FBAR_NCU FBAR_BCL
FBAR_BCU FSTDEV FSTDEV_NCL FSTDEV_NCU FSTDEV_BCL FSTDEV_BCU OBAR
OBAR_NCL OBAR_NCU OBAR_BCL OBAR_BCU OSTDEV OSTDEV_NCL OSTDEV_NCU
OSTDEV_BCL OSTDEV_BCU PR_CORR PR_CORR_NCL PR_CORR_NCU PR_CORR_BCL
PR_CORR_BCU SP_CORR KT_CORR RANKS FRANK_TIES ORANK_TIES ME
ME_NCL ME_NCU ME_BCL ME_BCU ESTDEV ESTDEV_NCL ESTDEV_NCU
ESTDEV_BCL ESTDEV_BCU MBIAS MBIAS_BCL MBIAS_BCU MAE MAE_BCL
MAE_BCU MSE MSE_BCL MSE_BCU BCMSE BCMSE_BCL BCMSE_BCU
RMSE RMSE_BCL RMSE_BCU E10 E10_BCL E10_BCU E25 E25_BCL
E25_BCU E50 E50_BCL E50_BCU E75 E75_BCL E75_BCU E90
E90_BCL E90_BCU EIQR EIQR_BCL EIQR_BCU MAD MAD_BCL MAD_BCU
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 1 NA NA
NA 0.05000 CNT 992 9.65309 8.55757 10.74860 NA NA
17.60457 16.86253 18.41542 NA NA 11.53044 10.39918
12.66170 NA NA 18.17902 17.41278 19.01634 NA
NA 0.31469 0.25749 0.36969 NA NA NA
NA 0 0 0 -1.87736 -3.18116 -0.57355 NA
NA 20.95178 20.06866 21.91681 NA NA 0.83718
NA NA 11.92132 NA NA 442.05909 NA NA
438.53463 NA NA 21.02520 NA NA -24.70651
NA NA -7.41121 NA NA 0.00000 NA NA 1.82353
NA NA 18.38433 NA NA 9.23474 NA NA 4.60727
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 9 NA NA
NA 0.05000 CNT 870 9.60747 8.89059 10.32435 NA NA
10.78838 10.30418 11.32068 NA NA 11.11896 10.24189
11.99602 NA NA 13.19914 12.60674 13.85039 NA
NA 0.57308 0.52668 0.61608 NA NA NA
NA 0 0 0 -1.51148 -2.26149 -0.76148 NA
NA 11.28693 10.78036 11.84384 NA NA 0.86406
NA NA 7.46912 NA NA 129.53300 NA NA
127.24841 NA NA 11.38126 NA NA -17.01502
NA NA -5.83801 NA NA 0.00000 NA NA 3.84078
NA NA 11.18719 NA NA 9.67879 NA NA 4.60421
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 25 NA NA
NA 0.05000 CNT 756 9.61581 8.99178 10.23984 NA NA
8.75429 8.33416 9.21937 NA NA 10.76688 10.00509
11.52868 NA NA 10.68684 10.17396 11.25459 NA
NA 0.68521 0.64544 0.72127 NA NA NA
NA 0 0 0 -1.15108 -1.71523 -0.58692 NA
NA 7.91426 7.53444 8.33471 NA NA 0.89309
NA NA 5.54217 NA NA 63.87766 NA NA
62.55269 NA NA 7.99235 NA NA -12.05399
NA NA -4.63685 NA NA 0.00527 NA NA 3.39211
NA NA 7.77246 NA NA 8.02896 NA NA 3.82150
NA NA
------------------------------------------------
Subject: Issue with width function
From: Geeta Geeta
Time: Wed Mar 11 23:22:45 2015
VERSION MODEL FCST_LEAD FCST_VALID_BEG FCST_VALID_END OBS_LEAD
OBS_VALID_BEG OBS_VALID_END FCST_VAR FCST_LEV OBS_VAR OBS_LEV
OBTYPE VX_MASK INTERP_MTHD INTERP_PNTS FCST_THRESH OBS_THRESH
COV_THRESH ALPHA LINE_TYPE TOTAL FBS FBS_BCL FBS_BCU FSS
FSS_BCL FSS_BCU AFSS AFSS_BCL AFSS_BCU UFSS UFSS_BCL UFSS_BCU
F_RATE F_RATE_BCL F_RATE_BCU O_RATE O_RATE_BCL O_RATE_BCU
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 9 >=0.000 >=0.000
NA 0.05000 NBRCNT 870 0.00000 NA NA 1.00000 NA
NA 1.00000 NA NA 1.00000 NA NA 1.00000
NA NA 1.00000 NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 9 >=5.000 >=5.000
NA 0.05000 NBRCNT 870 0.12805 NA NA 0.79377 NA
NA 0.98232 NA NA 0.72644 NA NA 0.37471
NA NA 0.45287 NA NA
------------------------------------------------
Subject: Issue with width function
From: Geeta Geeta
Time: Thu Mar 12 00:19:46 2015
Hi John. another question. So for the CONVERSION of RAW FCST and OBS
into the binary fields, we have the following section in the config
file. // Forecast and observation fields to be verified//fcst = {
wind_thresh = [ NA ];
field = [ { name = "APCP_24"; level =
[ "A03" ]; cat_thresh = [ >0.0, >5.0 ]; } ];
};obs = fcst;
which means that when FCST or OBS are >0.0 or >5.0 , it will put 1
there else 0. Now we have another section called
// Neighborhood methods//nbrhd = { vld_thresh = 1.0; width =
[ 1 ]; cov_thresh = [ >=0.5 ];}
what is this parameter cov_thresh=[>=0.5] does???? what is its role.
Can U pls explain???Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #71028] Issue with width function
Date: Thu, 12 Mar 2015 10:51:48 +0530
thanks John. it worked. But further one more question, so the width of
the interp section is essentially upscaling the data (both obs and
fcst or either one on my choice) and then the neighbourhood statistics
is computed depending on the neightbourhood width and the threshold
selected. So HOW will the observations and FCST be treated, upscaled
to 3 or 5 or 1 ??? Can U pls explain. Hope I have made myself clear
!!!!!!!!!!!!!!!!Again what i mean to ask is whether the data is
considered haveing upscaling at 1, 3 or 5 level????. Following is the
config file
entries..........////////////////////////////////////////////////////////////////////////////////
//// Interpolation methods//interp = { field = BOTH;
vld_thresh = 1.0;
type = [ {method = UW_MEAN; width = 1;}, {method =
UW_MEAN; width = 3;}, {method = UW_MEAN; width = 5;} ];};
////////////////////////////////////////////////////////////////////////////////
//// Neighborhood methods//nbrhd = { vld_thresh = 1.0; width
= [ 1 ]; width = [ 3 ]; cov_thresh = [ >=0.5 ];}
I am sending you the O/P CNT output and NBRCNT output of this config
file.
If U just look at the NBRCNT o/p, then it is giving me the o/p
corresponding to only the NBR width=3. Can U pls
explain...................Geeta
> Subject: Re: [rt.rap.ucar.edu #71028] Issue with width function
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Wed, 11 Mar 2015 13:17:50 -0600
>
> Geeta,
>
> Please refer to the example I sent previously. I'll also copy and
paste it
> below. You have the config file syntax wrong...
>
> interp = {
> field = FCST;
> vld_thresh = 1.0;
>
> type = [
> { method = UW_MEAN; width = 1; },
> { method = UW_MEAN; width = 3; },
> { method = UW_MEAN; width = 5; },
> { method = UW_MEAN; width = 7; },
> { method = UW_MEAN; width = 9; }
> ];
> };
>
> Thanks,
> John
>
> On Wed, Mar 11, 2015 at 9:52 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71028 >
> >
> > Hi John. I have the following entries in my config
file............nterp =
> > { field = BOTH; vld_thresh = 1.0;
> > type = [ { method = UW_MEAN; width = 1;
> > width = 3;
> > } ]; };
> > But when I see my CNT output, I do not get different values of
RMSE and
> > GSS corresponding to WIDTH=1 and WIDTH=3 . But I get only one
value.
> > Sending you my CNT output. I thought I would get one CNT output
for one
> > width and then 2nd output CTS for the width=3. so that I can
compare them
> > to see if RMSE improves by upscaling. Can u pls explain the
output that
> > grid stat will write.
> > Geeta
> > > Subject: Re: [rt.rap.ucar.edu #71028] Issue with width function
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Tue, 10 Mar 2015 14:26:58 -0600
> > >
> > > Geeta,
> > >
> > > I combined your two recent questions into a single support
ticket.
> > >
> > > I see that you are asking about the "width" settings in the
Grid-Stat
> > > configuration file. "width" shows up in two places there, once
in the
> > > "interp" section and once in the "nbrhd" section. I'll do my
best to
> > > explain the difference.
> > >
> > > The "nbrhd" section is used to define how the neighborhood
verification
> > > methods are applied. Basically, this affects the output you see
in the
> > > NBRCNT, NBRCTC, and NBRCTS output line types from Grid-Stat.
> > >
> > > The "interp" section is used to do up-scaling. This is
basically a
> > > smoothing of the forecast and/or observation field. This affect
the
> > output
> > > you see all of the other non-neighborhood line type such as CNT,
CTS, and
> > > so on. You're interested in the equitable threat score which is
named
> > the
> > > Gilbert Skill Score (GSS) in the CTS line type, and you want to
see how
> > > upscaling impacts that score. Therefore you should set one or
more
> > > upscaling methods in the "interp" section and see how GSS
varies.
> > >
> > > You could try this:
> > > interp = {
> > > field = FCST;
> > > vld_thresh = 1.0;
> > >
> > > type = [
> > > { method = UW_MEAN; width = 1; },
> > > { method = UW_MEAN; width = 3; },
> > > { method = UW_MEAN; width = 5; },
> > > { method = UW_MEAN; width = 7; },
> > > { method = UW_MEAN; width = 9; }
> > > ];
> > > };
> > >
> > > Since I set "field = FCST;" we'll only smooth the forecast
field, leaving
> > > the observations unchanged. We'll smooth the data using a 1x1
box (no
> > > smoothing), 3x3 box, ..., all the way up to a 9x9 box.
> > >
> > > I tried this on some sample data and found that the more I
smoothed the
> > > forecast, the worst the GSS performed! However, the more I
smoothed the
> > > forecast the better the RMSE (in the CNT line type) performed.
Generally
> > > that is the case... the smoother the forecast, the higher the
continuous
> > > statistics, such as RMSE.
> > >
> > > Hope that helps.
> > >
> > > Thanks,
> > > John Halley Gotway
> > >
> > >
> > > On Mon, Mar 9, 2015 at 10:54 AM, The RT System itself via RT <
> > > met_help at ucar.edu> wrote:
> > >
> > > >
> > > > Mon Mar 09 10:54:50 2015: Request 71028 was acted upon.
> > > > Transaction: Given to johnhg by RT_System
> > > > Queue: met_help
> > > > Subject: Issue with width function
> > > > Owner: johnhg
> > > > Requestors: geeta124 at hotmail.com
> > > > Status: new
> > > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71028
> > >
> > > >
> > > >
> > > > This transaction appears to have no content
> > > >
> > >
> >
> >
>
------------------------------------------------
Subject: Issue with width function
From: Geeta Geeta
Time: Thu Mar 12 09:49:07 2015
Hi John. This is continuation with the GSS and RMSE I am calculating
for the UPSCALE data sets. I have 37x37 grid points for FCST and
OBSNS. I upscaled BOTH FCST and OBSNS and//// Interpolation
methods//interp = { field = BOTH; vld_thresh = 1.0;
type = [ {method = UW_MEAN; width = 1;}, {method =
UW_MEAN; width = 3;}, {method = UW_MEAN; width = 5;},
{method = UW_MEAN; width = 7;}, {method = UW_MEAN; width = 9;},
{method = UW_MEAN; width = 11;}, {method = UW_MEAN; width =
13;}, {method = UW_MEAN; width = 15;}, {method = UW_MEAN;
width = 17;}, {method = UW_MEAN; width = 19;}, {method =
UW_MEAN; width = 21;}, {method = UW_MEAN; width = 23;},
{method = UW_MEAN; width = 25;}, {method = UW_MEAN; width =
29;}, {method = UW_MEAN; width = 31;}, {method = UW_MEAN;
width = 33;}, {method = UW_MEAN; width = 35;}, {method =
UW_MEAN; width = 37;} ];};
////////////////////////////////////////////////////////////////////////////////
//// Neighborhood methods//nbrhd = {
I saw that the RMSE improved. BUT did not get GSS beyond the 81 points
(width=9) Donot know whats happening. GSS=
GSS=ETS=Hits-Hitsrandom/(Hits+Misses+False
alarms-Hitsrandom)I am sending you the outputs. Can U pls explain whay
I am not able to get GSS for the 121 points and
above................................
Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #71028] Issue with width function
Date: Thu, 12 Mar 2015 11:48:20 +0530
Hi John. another question. So for the CONVERSION of RAW FCST and OBS
into the binary fields, we have the following section in the config
file. // Forecast and observation fields to be verified//fcst = {
wind_thresh = [ NA ];
field = [ { name = "APCP_24"; level =
[ "A03" ]; cat_thresh = [ >0.0, >5.0 ]; } ];
};obs = fcst;
which means that when FCST or OBS are >0.0 or >5.0 , it will put 1
there else 0. Now we have another section called
// Neighborhood methods//nbrhd = { vld_thresh = 1.0; width =
[ 1 ]; cov_thresh = [ >=0.5 ];}
what is this parameter cov_thresh=[>=0.5] does???? what is its role.
Can U pls explain???Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #71028] Issue with width function
Date: Thu, 12 Mar 2015 10:51:48 +0530
thanks John. it worked. But further one more question, so the width of
the interp section is essentially upscaling the data (both obs and
fcst or either one on my choice) and then the neighbourhood statistics
is computed depending on the neightbourhood width and the threshold
selected. So HOW will the observations and FCST be treated, upscaled
to 3 or 5 or 1 ??? Can U pls explain. Hope I have made myself clear
!!!!!!!!!!!!!!!!Again what i mean to ask is whether the data is
considered haveing upscaling at 1, 3 or 5 level????. Following is the
config file
entries..........////////////////////////////////////////////////////////////////////////////////
//// Interpolation methods//interp = { field = BOTH;
vld_thresh = 1.0;
type = [ {method = UW_MEAN; width = 1;}, {method =
UW_MEAN; width = 3;}, {method = UW_MEAN; width = 5;} ];};
////////////////////////////////////////////////////////////////////////////////
//// Neighborhood methods//nbrhd = { vld_thresh = 1.0; width
= [ 1 ]; width = [ 3 ]; cov_thresh = [ >=0.5 ];}
I am sending you the O/P CNT output and NBRCNT output of this config
file.
If U just look at the NBRCNT o/p, then it is giving me the o/p
corresponding to only the NBR width=3. Can U pls
explain...................Geeta
> Subject: Re: [rt.rap.ucar.edu #71028] Issue with width function
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Wed, 11 Mar 2015 13:17:50 -0600
>
> Geeta,
>
> Please refer to the example I sent previously. I'll also copy and
paste it
> below. You have the config file syntax wrong...
>
> interp = {
> field = FCST;
> vld_thresh = 1.0;
>
> type = [
> { method = UW_MEAN; width = 1; },
> { method = UW_MEAN; width = 3; },
> { method = UW_MEAN; width = 5; },
> { method = UW_MEAN; width = 7; },
> { method = UW_MEAN; width = 9; }
> ];
> };
>
> Thanks,
> John
>
> On Wed, Mar 11, 2015 at 9:52 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71028 >
> >
> > Hi John. I have the following entries in my config
file............nterp =
> > { field = BOTH; vld_thresh = 1.0;
> > type = [ { method = UW_MEAN; width = 1;
> > width = 3;
> > } ]; };
> > But when I see my CNT output, I do not get different values of
RMSE and
> > GSS corresponding to WIDTH=1 and WIDTH=3 . But I get only one
value.
> > Sending you my CNT output. I thought I would get one CNT output
for one
> > width and then 2nd output CTS for the width=3. so that I can
compare them
> > to see if RMSE improves by upscaling. Can u pls explain the
output that
> > grid stat will write.
> > Geeta
> > > Subject: Re: [rt.rap.ucar.edu #71028] Issue with width function
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Tue, 10 Mar 2015 14:26:58 -0600
> > >
> > > Geeta,
> > >
> > > I combined your two recent questions into a single support
ticket.
> > >
> > > I see that you are asking about the "width" settings in the
Grid-Stat
> > > configuration file. "width" shows up in two places there, once
in the
> > > "interp" section and once in the "nbrhd" section. I'll do my
best to
> > > explain the difference.
> > >
> > > The "nbrhd" section is used to define how the neighborhood
verification
> > > methods are applied. Basically, this affects the output you see
in the
> > > NBRCNT, NBRCTC, and NBRCTS output line types from Grid-Stat.
> > >
> > > The "interp" section is used to do up-scaling. This is
basically a
> > > smoothing of the forecast and/or observation field. This affect
the
> > output
> > > you see all of the other non-neighborhood line type such as CNT,
CTS, and
> > > so on. You're interested in the equitable threat score which is
named
> > the
> > > Gilbert Skill Score (GSS) in the CTS line type, and you want to
see how
> > > upscaling impacts that score. Therefore you should set one or
more
> > > upscaling methods in the "interp" section and see how GSS
varies.
> > >
> > > You could try this:
> > > interp = {
> > > field = FCST;
> > > vld_thresh = 1.0;
> > >
> > > type = [
> > > { method = UW_MEAN; width = 1; },
> > > { method = UW_MEAN; width = 3; },
> > > { method = UW_MEAN; width = 5; },
> > > { method = UW_MEAN; width = 7; },
> > > { method = UW_MEAN; width = 9; }
> > > ];
> > > };
> > >
> > > Since I set "field = FCST;" we'll only smooth the forecast
field, leaving
> > > the observations unchanged. We'll smooth the data using a 1x1
box (no
> > > smoothing), 3x3 box, ..., all the way up to a 9x9 box.
> > >
> > > I tried this on some sample data and found that the more I
smoothed the
> > > forecast, the worst the GSS performed! However, the more I
smoothed the
> > > forecast the better the RMSE (in the CNT line type) performed.
Generally
> > > that is the case... the smoother the forecast, the higher the
continuous
> > > statistics, such as RMSE.
> > >
> > > Hope that helps.
> > >
> > > Thanks,
> > > John Halley Gotway
> > >
> > >
> > > On Mon, Mar 9, 2015 at 10:54 AM, The RT System itself via RT <
> > > met_help at ucar.edu> wrote:
> > >
> > > >
> > > > Mon Mar 09 10:54:50 2015: Request 71028 was acted upon.
> > > > Transaction: Given to johnhg by RT_System
> > > > Queue: met_help
> > > > Subject: Issue with width function
> > > > Owner: johnhg
> > > > Requestors: geeta124 at hotmail.com
> > > > Status: new
> > > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71028
> > >
> > > >
> > > >
> > > > This transaction appears to have no content
> > > >
> > >
> >
> >
>
------------------------------------------------
Subject: Issue with width function
From: Geeta Geeta
Time: Thu Mar 12 09:49:07 2015
VERSION MODEL FCST_LEAD FCST_VALID_BEG FCST_VALID_END OBS_LEAD
OBS_VALID_BEG OBS_VALID_END FCST_VAR FCST_LEV OBS_VAR OBS_LEV
OBTYPE VX_MASK INTERP_MTHD INTERP_PNTS FCST_THRESH OBS_THRESH
COV_THRESH ALPHA LINE_TYPE TOTAL FBAR FBAR_NCL FBAR_NCU FBAR_BCL
FBAR_BCU FSTDEV FSTDEV_NCL FSTDEV_NCU FSTDEV_BCL FSTDEV_BCU OBAR
OBAR_NCL OBAR_NCU OBAR_BCL OBAR_BCU OSTDEV OSTDEV_NCL OSTDEV_NCU
OSTDEV_BCL OSTDEV_BCU PR_CORR PR_CORR_NCL PR_CORR_NCU PR_CORR_BCL
PR_CORR_BCU SP_CORR KT_CORR RANKS FRANK_TIES ORANK_TIES ME
ME_NCL ME_NCU ME_BCL ME_BCU ESTDEV ESTDEV_NCL ESTDEV_NCU
ESTDEV_BCL ESTDEV_BCU MBIAS MBIAS_BCL MBIAS_BCU MAE MAE_BCL
MAE_BCU MSE MSE_BCL MSE_BCU BCMSE BCMSE_BCL BCMSE_BCU
RMSE RMSE_BCL RMSE_BCU E10 E10_BCL E10_BCU E25 E25_BCL
E25_BCU E50 E50_BCL E50_BCU E75 E75_BCL E75_BCU E90
E90_BCL E90_BCU EIQR EIQR_BCL EIQR_BCU MAD MAD_BCL MAD_BCU
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 1 NA NA
NA 0.05000 CNT 992 9.65309 8.55757 10.74860 NA NA
17.60457 16.86253 18.41542 NA NA 11.53044 10.39918
12.66170 NA NA 18.17902 17.41278 19.01634 NA
NA 0.31469 0.25749 0.36969 NA NA NA
NA 0 0 0 -1.87736 -3.18116 -0.57355 NA
NA 20.95178 20.06866 21.91681 NA NA 0.83718
NA NA 11.92132 NA NA 442.05909 NA NA
438.53463 NA NA 21.02520 NA NA -24.70651
NA NA -7.41121 NA NA 0.00000 NA NA 1.82353
NA NA 18.38433 NA NA 9.23474 NA NA 4.60727
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 9 NA NA
NA 0.05000 CNT 870 9.60747 8.89059 10.32435 NA NA
10.78838 10.30418 11.32068 NA NA 11.11896 10.24189
11.99602 NA NA 13.19914 12.60674 13.85039 NA
NA 0.57308 0.52668 0.61608 NA NA NA
NA 0 0 0 -1.51148 -2.26149 -0.76148 NA
NA 11.28693 10.78036 11.84384 NA NA 0.86406
NA NA 7.46912 NA NA 129.53300 NA NA
127.24841 NA NA 11.38126 NA NA -17.01502
NA NA -5.83801 NA NA 0.00000 NA NA 3.84078
NA NA 11.18719 NA NA 9.67879 NA NA 4.60421
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 25 NA NA
NA 0.05000 CNT 756 9.61581 8.99178 10.23984 NA NA
8.75429 8.33416 9.21937 NA NA 10.76688 10.00509
11.52868 NA NA 10.68684 10.17396 11.25459 NA
NA 0.68521 0.64544 0.72127 NA NA NA
NA 0 0 0 -1.15108 -1.71523 -0.58692 NA
NA 7.91426 7.53444 8.33471 NA NA 0.89309
NA NA 5.54217 NA NA 63.87766 NA NA
62.55269 NA NA 7.99235 NA NA -12.05399
NA NA -4.63685 NA NA 0.00527 NA NA 3.39211
NA NA 7.77246 NA NA 8.02896 NA NA 3.82150
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 49 NA NA
NA 0.05000 CNT 650 9.39886 8.83082 9.96690 NA NA
7.38903 7.00801 7.81420 NA NA 10.86244 10.14447
11.58040 NA NA 9.33928 8.85770 9.87667 NA
NA 0.74089 0.70411 0.77371 NA NA NA
NA 0 0 0 -1.46357 -1.94713 -0.98002 NA
NA 6.29002 5.96567 6.65195 NA NA 0.86526
NA NA 4.64368 NA NA 41.64556 NA NA
39.50351 NA NA 6.45334 NA NA -10.22399
NA NA -4.92567 NA NA 0.00204 NA NA 2.59596
NA NA 5.14649 NA NA 7.52163 NA NA 3.29438
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 81 NA NA
NA 0.05000 CNT 552 9.29036 8.75601 9.82470 NA NA
6.40536 6.04849 6.80733 NA NA 11.19479 10.50538
11.88420 NA NA 8.26421 7.80377 8.78283 NA
NA 0.78389 0.74947 0.81409 NA NA NA
NA 0 0 0 -1.90443 -2.33253 -1.47634 NA
NA 5.13172 4.84581 5.45376 NA NA 0.82988
NA NA 4.05477 NA NA 29.91370 NA NA
26.28683 NA NA 5.46934 NA NA -8.64544
NA NA -5.09572 NA NA -0.77982 NA NA 1.69666
NA NA 3.60992 NA NA 6.79237 NA NA 3.09420
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 121 NA NA
NA 0.05000 CNT 462 9.22660 8.71945 9.73375 NA NA
5.56171 5.22472 5.94553 NA NA 11.61914 10.94767
12.29061 NA NA 7.36377 6.91759 7.87195 NA
NA 0.82959 0.79882 0.85603 NA NA NA
NA 0 0 0 -2.39254 -2.77077 -2.01430 NA
NA 4.14799 3.89666 4.43425 NA NA 0.79409
NA NA 3.70987 NA NA 22.89284 NA NA
17.16861 NA NA 4.78465 NA NA -7.98230
NA NA -5.09410 NA NA -2.01762 NA NA 1.00292
NA NA 2.43386 NA NA 6.09702 NA NA 3.04239
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 169 NA NA
NA 0.05000 CNT 380 9.23702 8.74862 9.72542 NA NA
4.85755 4.53499 5.22990 NA NA 11.95803 11.32040
12.59567 NA NA 6.34182 5.92070 6.82794 NA
NA 0.87413 0.84811 0.89594 NA NA NA
NA 0 0 0 -2.72101 -3.03830 -2.40373 NA
NA 3.15568 2.94613 3.39757 NA NA 0.77245
NA NA 3.40173 NA NA 17.33600 NA NA
9.93208 NA NA 4.16365 NA NA -7.26570
NA NA -4.63599 NA NA -2.54589 NA NA -0.51425
NA NA 1.77779 NA NA 4.12174 NA NA 2.07720
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 225 NA NA
NA 0.05000 CNT 306 9.18397 8.70456 9.66338 NA NA
4.27881 3.96452 4.64765 NA NA 12.25454 11.65585
12.85323 NA NA 5.34339 4.95090 5.80400 NA
NA 0.92956 0.91255 0.94336 NA NA NA
NA 0 0 0 -3.07057 -3.30437 -2.83677 NA
NA 2.08668 1.93340 2.26655 NA NA 0.74943
NA NA 3.21308 NA NA 13.76839 NA NA
4.33999 NA NA 3.71058 NA NA -5.81750
NA NA -4.75525 NA NA -3.05354 NA NA -1.67906
NA NA -0.26133 NA NA 3.07619 NA NA 1.52203
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 289 NA NA
NA 0.05000 CNT 240 9.17309 8.67767 9.66851 NA NA
3.91593 3.59412 4.30155 NA NA 12.31469 11.73037
12.89901 NA NA 4.61858 4.23903 5.07339 NA
NA 0.96931 0.96059 0.97613 NA NA NA
NA 0 0 0 -3.14160 -3.30181 -2.98138 NA
NA 1.26638 1.16231 1.39109 NA NA 0.74489
NA NA 3.14160 NA NA 11.46669 NA NA
1.59705 NA NA 3.38625 NA NA -4.79880
NA NA -4.21965 NA NA -3.14974 NA NA -2.18094
NA NA -1.34043 NA NA 2.03871 NA NA 1.03425
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 361 NA NA
NA 0.05000 CNT 182 9.20115 8.68352 9.71879 NA NA
3.56298 3.23068 3.97211 NA NA 12.06585 11.48672
12.64498 NA NA 3.98626 3.61448 4.44399 NA
NA 0.97275 0.96364 0.97960 NA NA NA
NA 0 0 0 -2.86470 -3.00654 -2.72286 NA
NA 0.97632 0.88526 1.08843 NA NA 0.76258
NA NA 2.86470 NA NA 9.15445 NA NA
0.94796 NA NA 3.02563 NA NA -4.10788
NA NA -3.53808 NA NA -2.87882 NA NA -2.08409
NA NA -1.61802 NA NA 1.45399 NA NA 0.76269
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 441 NA NA
NA 0.05000 CNT 132 9.21248 8.68369 9.74128 NA NA
3.09973 2.76551 3.52662 NA NA 11.71065 11.15386
12.26745 NA NA 3.26387 2.91195 3.71337 NA
NA 0.96389 0.94938 0.97429 NA NA NA
NA 0 0 0 -2.49817 -2.64666 -2.34968 NA
NA 0.87041 0.77656 0.99028 NA NA 0.78668
NA NA 2.49817 NA NA 6.99272 NA NA
0.75187 NA NA 2.64438 NA NA -3.63249
NA NA -3.00224 NA NA -2.53579 NA NA -1.86982
NA NA -1.46335 NA NA 1.13241 NA NA 0.61371
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 529 NA NA
NA 0.05000 CNT 90 9.26328 8.72448 9.80207 NA NA
2.60793 2.27464 3.05665 NA NA 11.30396 10.75510
11.85283 NA NA 2.65667 2.31715 3.11378 NA
NA 0.95965 0.93921 0.97331 NA NA NA
NA 0 0 0 -2.04069 -2.19550 -1.88588 NA
NA 0.74933 0.65357 0.87826 NA NA 0.81947
NA NA 2.04069 NA NA 4.71966 NA NA
0.55526 NA NA 2.17248 NA NA -2.96632
NA NA -2.49632 NA NA -2.14127 NA NA -1.63091
NA NA -1.02273 NA NA 0.86541 NA NA 0.44092
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 625 NA NA
NA 0.05000 CNT 56 9.31252 8.77780 9.84723 NA NA
2.04158 1.72117 2.50991 NA NA 10.89555 10.29979
11.49131 NA NA 2.27466 1.91767 2.79645 NA
NA 0.94804 0.91260 0.96934 NA NA NA
NA 0 0 0 -1.58303 -1.77494 -1.39113 NA
NA 0.73272 0.61773 0.90080 NA NA 0.85471
NA NA 1.58303 NA NA 3.03329 NA NA
0.52729 NA NA 1.74163 NA NA -2.52776
NA NA -2.09334 NA NA -1.57140 NA NA -1.05976
NA NA -0.61140 NA NA 1.03358 NA NA 0.52054
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 841 NA NA
NA 0.05000 CNT 12 9.58536 9.07032 10.10040 NA NA
0.81062 0.57424 1.37633 NA NA 10.91622 9.99332
11.83913 NA NA 1.45255 1.02898 2.46625 NA
NA 0.94003 0.76354 0.98585 NA NA NA
NA 0 0 0 -1.33086 -1.80347 -0.85825 NA
NA 0.74384 0.52693 1.26294 NA NA 0.87808
NA NA 1.33086 NA NA 2.27838 NA NA
0.50718 NA NA 1.50943 NA NA -2.16893
NA NA -1.92924 NA NA -1.41491 NA NA -0.76906
NA NA -0.37778 NA NA 1.16018 NA NA 0.60323
NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 961 NA NA
NA 0.05000 CNT 2 9.48327 3.36888 15.59765 NA NA
0.68054 0.30362 21.71604 NA NA 11.30496 3.71390
18.89602 NA NA 0.84489 0.37695 26.96066 NA
NA 1.00000 NA NA NA NA NA
NA 0 0 0 -1.82170 -3.29838 -0.34502 NA
NA 0.16436 0.07333 5.24462 NA NA 0.83886
NA NA 1.82170 NA NA 3.33209 NA NA
0.01351 NA NA 1.82540 NA NA -1.91467
NA NA -1.87981 NA NA -1.82170 NA NA -1.76359
NA NA -1.72872 NA NA 0.11622 NA NA 0.11622
NA NA
------------------------------------------------
Subject: Issue with width function
From: Geeta Geeta
Time: Thu Mar 12 09:49:07 2015
VERSION MODEL FCST_LEAD FCST_VALID_BEG FCST_VALID_END OBS_LEAD
OBS_VALID_BEG OBS_VALID_END FCST_VAR FCST_LEV OBS_VAR OBS_LEV
OBTYPE VX_MASK INTERP_MTHD INTERP_PNTS FCST_THRESH OBS_THRESH
COV_THRESH ALPHA LINE_TYPE TOTAL BASER BASER_NCL BASER_NCU
BASER_BCL BASER_BCU FMEAN FMEAN_NCL FMEAN_NCU FMEAN_BCL FMEAN_BCU
ACC ACC_NCL ACC_NCU ACC_BCL ACC_BCU FBIAS FBIAS_BCL FBIAS_BCU
PODY PODY_NCL PODY_NCU PODY_BCL PODY_BCU PODN PODN_NCL PODN_NCU
PODN_BCL PODN_BCU POFD POFD_NCL POFD_NCU POFD_BCL POFD_BCU FAR
FAR_NCL FAR_NCU FAR_BCL FAR_BCU CSI CSI_NCL CSI_NCU CSI_BCL
CSI_BCU GSS GSS_BCL GSS_BCU HK HK_NCL HK_NCU HK_BCL HK_BCU
HSS HSS_BCL HSS_BCU ODDS ODDS_NCL ODDS_NCU ODDS_BCL ODDS_BCU
LODDS LODDS_NCL LODDS_NCU LODDS_BCL LODDS_BCU ORSS ORSS_NCL
ORSS_NCU ORSS_BCL ORSS_BCU EDS EDS_NCL EDS_NCU EDS_BCL EDS_BCU
SEDS SEDS_NCL SEDS_NCU SEDS_BCL SEDS_BCU EDI EDI_NCL EDI_NCU
EDI_BCL EDI_BCU SEDI SEDI_NCL SEDI_NCU SEDI_BCL SEDI_BCU BAGSS
BAGSS_BCL BAGSS_BCU
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 1 >0.000 >0.000
NA 0.05000 CTS 992 0.57460 0.54360 0.60502 NA
NA 0.63810 0.60772 0.66742 NA NA 0.76915 0.74192
0.79431 NA NA 1.11053 NA NA 0.85439 0.83107
0.87497 NA NA 0.65403 0.62388 0.68298 NA NA
0.34597 0.31702 0.37612 NA NA 0.23065 0.20550 0.25787
NA NA 0.68017 0.65050 0.70845 NA NA 0.34995 NA
NA 0.50841 0.46402 0.55281 NA NA 0.51846 NA NA
11.09193 8.15777 15.08143 NA NA 2.40622 2.09897
2.71346 NA NA 0.83460 0.78798 0.88122 NA
NA 0.55761 0.48341 0.63181 NA NA 0.41026 0.34308 0.47744
NA NA 0.74175 0.65118 0.83232 NA NA 0.67398 0.64022
0.84328 NA NA 0.34322 NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 9 >0.000 >0.000
NA 0.05000 CTS 870 0.74253 0.71245 0.77047 NA
NA 0.84138 0.81561 0.86415 NA NA 0.84368 0.81804
0.86629 NA NA 1.13313 NA NA 0.96130 0.94632
0.97222 NA NA 0.50446 0.47129 0.53759 NA NA
0.49554 0.46241 0.52871 NA NA 0.15164 0.12934 0.17700
NA NA 0.82034 0.79344 0.84443 NA NA 0.36291 NA
NA 0.46576 0.43551 0.49601 NA NA 0.53255 NA NA
25.28757 15.67946 40.78337 NA NA 3.23031 2.75235
3.70827 NA NA 0.92392 0.88894 0.95890 NA
NA 0.76588 0.68484 0.84691 NA NA 0.39519 0.33117 0.45922
NA NA 0.89356 0.79888 0.98823 NA NA 0.69057 0.70068
1.08644 NA NA 0.38606 NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 25 >0.000 >0.000
NA 0.05000 CTS 756 0.82804 0.79950 0.85327 NA
NA 0.94444 0.92576 0.95864 NA NA 0.84656 0.81912
0.87049 NA NA 1.14058 NA NA 0.97764 0.96443
0.98601 NA NA 0.21538 0.18756 0.24609 NA NA
0.78462 0.75391 0.81244 NA NA 0.14286 0.11972 0.16961
NA NA 0.84066 0.81286 0.86502 NA NA 0.15191 NA
NA 0.19302 0.16208 0.22396 NA NA 0.26375 NA NA
12.00000 6.11034 23.56662 NA NA 2.48491 1.80998
3.15983 NA NA 0.84615 0.75031 0.94200 NA
NA 0.78592 0.68579 0.88606 NA NA 0.16346 0.09822 0.22869
NA NA 0.82941 0.41450 1.24433 NA NA 0.44367 0.23713
1.42169 NA NA 0.27028 NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 49 >0.000 >0.000
NA 0.05000 CTS 650 0.90000 0.87454 0.92076 NA
NA 0.99846 0.99134 0.99973 NA NA 0.90154 0.87622
0.92214 NA NA 1.10940 NA NA 1.00000 0.99412
1.00000 NA NA 0.01538 0.00838 0.02809 NA NA
0.98462 0.97191 0.99162 NA NA 0.09861 0.07800 0.12395
NA NA 0.90139 0.87605 0.92200 NA NA 0.01387 NA
NA 0.01538 -0.01271 0.04348 NA NA 0.02736 NA NA
NA NA NA NA NA NA NA NA
NA NA 1.00000 NA NA NA NA 1.00000
1.00000 1.00000 NA NA 0.01461 0.01461 0.01461 NA
NA 1.00000 NA NA NA NA NA NA NA
NA NA NA NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 81 >0.000 >0.000
NA 0.05000 CTS 552 0.96196 0.94254 0.97498 NA
NA 1.00000 0.99309 1.00000 NA NA 0.96196 0.94254
0.97498 NA NA 1.03955 NA NA 1.00000 0.99309
1.00000 NA NA 0.00000 0.00000 0.00691 NA NA
1.00000 0.99309 1.00000 NA NA 0.03804 0.02502 0.05746
NA NA 0.96196 0.94254 0.97498 NA NA 0.00000 NA
NA 0.00000 -0.07740 0.07740 NA NA 0.00000 NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA 1.00000
1.00000 1.00000 NA NA 0.00000 0.00000 0.00000 NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 121 >0.000 >0.000
NA 0.05000 CTS 462 1.00000 0.99175 1.00000 NA
NA 1.00000 0.99175 1.00000 NA NA 1.00000 0.99175
1.00000 NA NA 1.00000 NA NA 1.00000 0.99175
1.00000 NA NA NA NA NA NA NA
NA NA NA NA NA 0.00000 -0.00000 0.00825
NA NA 1.00000 0.99175 1.00000 NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 169 >0.000 >0.000
NA 0.05000 CTS 380 1.00000 0.98999 1.00000 NA
NA 1.00000 0.98999 1.00000 NA NA 1.00000 0.98999
1.00000 NA NA 1.00000 NA NA 1.00000 0.98999
1.00000 NA NA NA NA NA NA NA
NA NA NA NA NA 0.00000 0.00000 0.01001
NA NA 1.00000 0.98999 1.00000 NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 225 >0.000 >0.000
NA 0.05000 CTS 306 1.00000 0.98760 1.00000 NA
NA 1.00000 0.98760 1.00000 NA NA 1.00000 0.98760
1.00000 NA NA 1.00000 NA NA 1.00000 0.98760
1.00000 NA NA NA NA NA NA NA
NA NA NA NA NA 0.00000 0.00000 0.01240
NA NA 1.00000 0.98760 1.00000 NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 289 >0.000 >0.000
NA 0.05000 CTS 240 1.00000 0.98425 1.00000 NA
NA 1.00000 0.98425 1.00000 NA NA 1.00000 0.98425
1.00000 NA NA 1.00000 NA NA 1.00000 0.98425
1.00000 NA NA NA NA NA NA NA
NA NA NA NA NA 0.00000 0.00000 0.01575
NA NA 1.00000 0.98425 1.00000 NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 361 >0.000 >0.000
NA 0.05000 CTS 182 1.00000 0.97933 1.00000 NA
NA 1.00000 0.97933 1.00000 NA NA 1.00000 0.97933
1.00000 NA NA 1.00000 NA NA 1.00000 0.97933
1.00000 NA NA NA NA NA NA NA
NA NA NA NA NA 0.00000 0.00000 0.02067
NA NA 1.00000 0.97933 1.00000 NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 441 >0.000 >0.000
NA 0.05000 CTS 132 1.00000 0.97172 1.00000 NA
NA 1.00000 0.97172 1.00000 NA NA 1.00000 0.97172
1.00000 NA NA 1.00000 NA NA 1.00000 0.97172
1.00000 NA NA NA NA NA NA NA
NA NA NA NA NA 0.00000 -0.00000 0.02828
NA NA 1.00000 0.97172 1.00000 NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 529 >0.000 >0.000
NA 0.05000 CTS 90 1.00000 0.95906 1.00000 NA
NA 1.00000 0.95906 1.00000 NA NA 1.00000 0.95906
1.00000 NA NA 1.00000 NA NA 1.00000 0.95906
1.00000 NA NA NA NA NA NA NA
NA NA NA NA NA 0.00000 0.00000 0.04094
NA NA 1.00000 0.95906 1.00000 NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 625 >0.000 >0.000
NA 0.05000 CTS 56 1.00000 0.93581 1.00000 NA
NA 1.00000 0.93581 1.00000 NA NA 1.00000 0.93581
1.00000 NA NA 1.00000 NA NA 1.00000 0.93581
1.00000 NA NA NA NA NA NA NA
NA NA NA NA NA 0.00000 -0.00000 0.06419
NA NA 1.00000 0.93581 1.00000 NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 841 >0.000 >0.000
NA 0.05000 CTS 12 1.00000 0.75751 1.00000 NA
NA 1.00000 0.75751 1.00000 NA NA 1.00000 0.75751
1.00000 NA NA 1.00000 NA NA 1.00000 0.75751
1.00000 NA NA NA NA NA NA NA
NA NA NA NA NA 0.00000 -0.00000 0.24249
NA NA 1.00000 0.75751 1.00000 NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA
V5.0 WRF 270000 20060620_030000 20060620_030000 240000
20060620_000000 20060620_000000 APCP A24 APCP A24
MC_PCP FULL UW_MEAN 961 >0.000 >0.000
NA 0.05000 CTS 2 1.00000 0.34238 1.00000 NA
NA 1.00000 0.34238 1.00000 NA NA 1.00000 0.34238
1.00000 NA NA 1.00000 NA NA 1.00000 0.34238
1.00000 NA NA NA NA NA NA NA
NA NA NA NA NA 0.00000 0.00000 0.65762
NA NA 1.00000 0.34238 1.00000 NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA NA
NA NA NA NA
------------------------------------------------
Subject: Issue with width function
From: John Halley Gotway
Time: Thu Mar 12 11:03:33 2015
Geeta,
I see that you've written with 3 questions in the last 11 hours.
(1) Regarding how the logic is applied for the interp and nbrhd
sections.
Here's how it works. Grid-Stat reads the raw data values and stores
them.
Starting with the raw data values, it applies the upscaling parameters
from
the interp section and then computes output statistics in the
non-neighborhood line types, such as CNT, SL1L2, CTC, and CTS.
Then it starts over again with the raw data values and applies the
neighborhood methods defined in the nbrhd section. It computes the
neighborhoods statistics and writes them in the NBRCNT, NBRCTC, and
NBRCTS
output line type.
So there is no interaction between the logic defined in the interp and
nbrhd sections.
(2) The cov_thresh setting in the nbrhd section is used to compute the
output NBRCTC and NBRCTS line type. I suggest that you do not use
these
output line types. Instead, just focus on the FSS and FBS statistics
defined in the NBRCNT line type. Just turn off the NBRCTC and NBRCTS
lines
types and ignore cov_thresh.
(3) Look at the TOTAL column in your output CNT line type. That's the
number of matched pairs that are used to compute the statistics in
that
line. Notice that as you increase the amount of smoothing, the number
of
matches pairs decreases from around 900 down to close to 0. You're
getting
0 matched pairs for large smoothing sizes because of edge effects.
Grid-Stat only computes a smoothed value if the proportion of valid
data in
the smoothing square meets the threshold defined by the vld_thresh
setting. You have vld_thresh set to 1.0, so you only get a smoothed
data
value for an 11x11 box when all 121 grid points contains valid data.
That
will only happen at the center of your grid. Using a 13x13 box, I
suspect
that no grid point has valid data at all 169 grid points.
Please use ncview to display the NetCDF matched pairs file from
grid_stat
and look at the smoothed versions of your data. It'll become obvious
what's happening.
John
On Thu, Mar 12, 2015 at 9:49 AM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71028 >
>
> Hi John. This is continuation with the GSS and RMSE I am calculating
for
> the UPSCALE data sets. I have 37x37 grid points for FCST and OBSNS.
I
> upscaled BOTH FCST and OBSNS and//// Interpolation methods//interp =
{
> field = BOTH; vld_thresh = 1.0;
> type = [ {method = UW_MEAN; width = 1;}, {method =
UW_MEAN;
> width = 3;}, {method = UW_MEAN; width = 5;}, {method =
UW_MEAN;
> width = 7;}, {method = UW_MEAN; width = 9;}, {method =
UW_MEAN;
> width = 11;}, {method = UW_MEAN; width = 13;}, {method =
> UW_MEAN; width = 15;}, {method = UW_MEAN; width = 17;},
{method
> = UW_MEAN; width = 19;}, {method = UW_MEAN; width = 21;},
> {method = UW_MEAN; width = 23;}, {method = UW_MEAN; width =
25;},
> {method = UW_MEAN; width = 29;}, {method = UW_MEAN; width =
31;},
> {method = UW_MEAN; width = 33;}, {method = UW_MEAN; width
=
> 35;}, {method = UW_MEAN; width = 37;} ];};
>
>
////////////////////////////////////////////////////////////////////////////////
> //// Neighborhood methods//nbrhd = {
> I saw that the RMSE improved. BUT did not get GSS beyond the 81
points
> (width=9) Donot know whats happening. GSS=
>
>
>
>
>
> GSS=ETS=Hits-Hitsrandom/(Hits+Misses+False
> alarms-Hitsrandom)I am sending you the outputs. Can U pls explain
whay I
> am not able to get GSS for the 121 points and
> above................................
>
>
> Geeta
>
> From: geeta124 at hotmail.com
> To: met_help at ucar.edu
> Subject: RE: [rt.rap.ucar.edu #71028] Issue with width function
> Date: Thu, 12 Mar 2015 11:48:20 +0530
>
>
>
>
> Hi John. another question. So for the CONVERSION of RAW FCST and OBS
into
> the binary fields, we have the following section in the config
file. //
> Forecast and observation fields to be verified//fcst = {
wind_thresh = [
> NA ];
> field = [ { name = "APCP_24"; level
= [
> "A03" ]; cat_thresh = [ >0.0, >5.0 ]; } ];
> };obs = fcst;
> which means that when FCST or OBS are >0.0 or >5.0 , it will put 1
there
> else 0. Now we have another section called
> // Neighborhood methods//nbrhd = { vld_thresh = 1.0; width
= [ 1
> ]; cov_thresh = [ >=0.5 ];}
> what is this parameter cov_thresh=[>=0.5] does???? what is its role.
Can U
> pls explain???Geeta
>
>
> From: geeta124 at hotmail.com
> To: met_help at ucar.edu
> Subject: RE: [rt.rap.ucar.edu #71028] Issue with width function
> Date: Thu, 12 Mar 2015 10:51:48 +0530
>
>
>
>
> thanks John. it worked. But further one more question, so the width
of the
> interp section is essentially upscaling the data (both obs and fcst
or
> either one on my choice) and then the neighbourhood statistics is
computed
> depending on the neightbourhood width and the threshold selected. So
HOW
> will the observations and FCST be treated, upscaled to 3 or 5 or 1
??? Can
> U pls explain. Hope I have made myself clear !!!!!!!!!!!!!!!!Again
what i
> mean to ask is whether the data is considered haveing upscaling at
1, 3 or
> 5 level????. Following is the config file
>
entries..........////////////////////////////////////////////////////////////////////////////////
> //// Interpolation methods//interp = { field = BOTH;
vld_thresh =
> 1.0;
> type = [ {method = UW_MEAN; width = 1;}, {method =
UW_MEAN;
> width = 3;}, {method = UW_MEAN; width = 5;} ];};
>
>
////////////////////////////////////////////////////////////////////////////////
> //// Neighborhood methods//nbrhd = { vld_thresh = 1.0; width
= [
> 1 ]; width = [ 3 ]; cov_thresh = [ >=0.5 ];}
> I am sending you the O/P CNT output and NBRCNT output of this config
file.
> If U just look at the NBRCNT o/p, then it is giving me the o/p
> corresponding to only the NBR width=3. Can U pls
> explain...................Geeta
> > Subject: Re: [rt.rap.ucar.edu #71028] Issue with width function
> > From: met_help at ucar.edu
> > To: geeta124 at hotmail.com
> > Date: Wed, 11 Mar 2015 13:17:50 -0600
> >
> > Geeta,
> >
> > Please refer to the example I sent previously. I'll also copy and
paste
> it
> > below. You have the config file syntax wrong...
> >
> > interp = {
> > field = FCST;
> > vld_thresh = 1.0;
> >
> > type = [
> > { method = UW_MEAN; width = 1; },
> > { method = UW_MEAN; width = 3; },
> > { method = UW_MEAN; width = 5; },
> > { method = UW_MEAN; width = 7; },
> > { method = UW_MEAN; width = 9; }
> > ];
> > };
> >
> > Thanks,
> > John
> >
> > On Wed, Mar 11, 2015 at 9:52 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > wrote:
> >
> > >
> > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71028 >
> > >
> > > Hi John. I have the following entries in my config
> file............nterp =
> > > { field = BOTH; vld_thresh = 1.0;
> > > type = [ { method = UW_MEAN; width = 1;
> > > width = 3;
> > > } ]; };
> > > But when I see my CNT output, I do not get different values of
RMSE and
> > > GSS corresponding to WIDTH=1 and WIDTH=3 . But I get only one
value.
> > > Sending you my CNT output. I thought I would get one CNT output
for one
> > > width and then 2nd output CTS for the width=3. so that I can
compare
> them
> > > to see if RMSE improves by upscaling. Can u pls explain the
output
> that
> > > grid stat will write.
> > > Geeta
> > > > Subject: Re: [rt.rap.ucar.edu #71028] Issue with width
function
> > > > From: met_help at ucar.edu
> > > > To: geeta124 at hotmail.com
> > > > Date: Tue, 10 Mar 2015 14:26:58 -0600
> > > >
> > > > Geeta,
> > > >
> > > > I combined your two recent questions into a single support
ticket.
> > > >
> > > > I see that you are asking about the "width" settings in the
Grid-Stat
> > > > configuration file. "width" shows up in two places there,
once in
> the
> > > > "interp" section and once in the "nbrhd" section. I'll do my
best to
> > > > explain the difference.
> > > >
> > > > The "nbrhd" section is used to define how the neighborhood
> verification
> > > > methods are applied. Basically, this affects the output you
see in
> the
> > > > NBRCNT, NBRCTC, and NBRCTS output line types from Grid-Stat.
> > > >
> > > > The "interp" section is used to do up-scaling. This is
basically a
> > > > smoothing of the forecast and/or observation field. This
affect the
> > > output
> > > > you see all of the other non-neighborhood line type such as
CNT,
> CTS, and
> > > > so on. You're interested in the equitable threat score which
is
> named
> > > the
> > > > Gilbert Skill Score (GSS) in the CTS line type, and you want
to see
> how
> > > > upscaling impacts that score. Therefore you should set one or
more
> > > > upscaling methods in the "interp" section and see how GSS
varies.
> > > >
> > > > You could try this:
> > > > interp = {
> > > > field = FCST;
> > > > vld_thresh = 1.0;
> > > >
> > > > type = [
> > > > { method = UW_MEAN; width = 1; },
> > > > { method = UW_MEAN; width = 3; },
> > > > { method = UW_MEAN; width = 5; },
> > > > { method = UW_MEAN; width = 7; },
> > > > { method = UW_MEAN; width = 9; }
> > > > ];
> > > > };
> > > >
> > > > Since I set "field = FCST;" we'll only smooth the forecast
field,
> leaving
> > > > the observations unchanged. We'll smooth the data using a 1x1
box
> (no
> > > > smoothing), 3x3 box, ..., all the way up to a 9x9 box.
> > > >
> > > > I tried this on some sample data and found that the more I
smoothed
> the
> > > > forecast, the worst the GSS performed! However, the more I
smoothed
> the
> > > > forecast the better the RMSE (in the CNT line type) performed.
> Generally
> > > > that is the case... the smoother the forecast, the higher the
> continuous
> > > > statistics, such as RMSE.
> > > >
> > > > Hope that helps.
> > > >
> > > > Thanks,
> > > > John Halley Gotway
> > > >
> > > >
> > > > On Mon, Mar 9, 2015 at 10:54 AM, The RT System itself via RT <
> > > > met_help at ucar.edu> wrote:
> > > >
> > > > >
> > > > > Mon Mar 09 10:54:50 2015: Request 71028 was acted upon.
> > > > > Transaction: Given to johnhg by RT_System
> > > > > Queue: met_help
> > > > > Subject: Issue with width function
> > > > > Owner: johnhg
> > > > > Requestors: geeta124 at hotmail.com
> > > > > Status: new
> > > > > Ticket <URL:
> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71028
> > > >
> > > > >
> > > > >
> > > > > This transaction appears to have no content
> > > > >
> > > >
> > >
> > >
> >
>
>
------------------------------------------------
More information about the Met_help
mailing list