[Met_help] [rt.rap.ucar.edu #70433] History for output of grid_stat
John Halley Gotway via RT
met_help at ucar.edu
Fri Feb 20 09:52:33 MST 2015
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hi John.
I had run grid stat on my data which has 25km resolution and got the output.
As per my header file, I have 37x37 points in the X n Y directions. Output of ncdump -h for both datasets is given below.
aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h Obs.024_19june2006.nc
netcdf Obs.024_19june2006 {
dimensions:
lon = 37 ;
lat = 37 ;
variables:
double lon(lon) ;
lon:units = "degrees_east" ;
double lat(lat) ;
lat:units = "degrees_north" ;
float APCP_24(lat, lon) ;
APCP_24:units = "kg/m^2" ;
APCP_24:missing_value = -9999.f ;
APCP_24:long_name = "Total precipitation" ;
APCP_24:name = "APCP" ;
APCP_24:level = "A24" ;
APCP_24:grib_code = 61.f ;
APCP_24:_FillValue = -9999.f ;
APCP_24:init_time = "20060619_000000" ;
APCP_24:init_time_ut = 1150675200 ;
APCP_24:valid_time = "20060620_000000" ;
APCP_24:valid_time_ut = 1150761600 ;
APCP_24:accum_time = "240000" ;
APCP_24:accum_time_sec = 86400 ;
// global attributes:
:FileOrigins = "File Obs.024_19june2006.nc generated 20150113_221613 on host agniilap by the Rscript trmmbin2nc.R" ;
:MET_version = "V4.1" ;
:Projection = "LatLon" ;
:lat_ll = "9 degrees_north" ;
:lon_ll = "74 degrees_east" ;
:delta_lat = "0.25 degrees" ;
:delta_lon = "0.25 degrees" ;
:Nlat = "37 grid_points" ;
:Nlon = "37 grid_points" ;
}
aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$
FCST FILE:;;;;;;;;;;;;;;;;;;;;;;;;;;
aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h WRFPRS_d03.024.pcp.nc
netcdf WRFPRS_d03.024.pcp {
dimensions:
lat = 37 ;
lon = 37 ;
variables:
float lat(lat, lon) ;
lat:long_name = "latitude" ;
lat:units = "degrees_north" ;
lat:standard_name = "latitude" ;
float lon(lat, lon) ;
lon:long_name = "longitude" ;
lon:units = "degrees_east" ;
lon:standard_name = "longitude" ;
float APCP_24(lat, lon) ;
APCP_24:name = "APCP" ;
APCP_24:long_name = "Total precipitation" ;
APCP_24:level = "A24" ;
APCP_24:units = "kg/m^2" ;
APCP_24:grib_code = 61 ;
APCP_24:_FillValue = -9999.f ;
APCP_24:init_time = "20060619_000000" ;
APCP_24:init_time_ut = 1150675200 ;
APCP_24:valid_time = "20060620_030000" ;
APCP_24:valid_time_ut = 1150772400 ;
APCP_24:accum_time = "240000" ;
APCP_24:accum_time_sec = 86400 ;
// global attributes:
:FileOrigins = "File WRFPRS_d03.024.pcp.nc generated 20150113_113544 UTC on host rmcdlh by the MET pcp_combine tool" ;
:MET_version = "V3.0" ;
:MET_tool = "pcp_combine" ;
:RunCommand = "Subtraction: WRFPRS_d03.027.grb25 with accumulation of 270000 minus WRFPRS_d03.003.grb25 with accumulation of 030000." ;
:Projection = "LatLon" ;
:lat_ll = "9.000000 degrees_north" ;
:lon_ll = "74.000000 degrees_east" ;
:delta_lat = "0.250000 degrees" ;
:delta_lon = "0.250000 degrees" ;
:Nlat = "37 grid_points" ;
:Nlon = "37 grid_points" ;
Now when I had a look at my output,
(say
Why do I get 992 MPR's (output attatched).
There should be 37*37 points=1369.
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 N_CAT F1_O1 F1_O2 F1_O3 F2_O1 F2_O2 F2_O3 F3_O1 F3_O2 F3_O3
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,>=5.000 >=0.000,>=5.000 NA NA MCTC 992 3 0 0 0 0 433 204 0 111 244
2. where in the Config file we specify the MCTC threholds exclusively?? Attatching the config file I used for your reference.
I saw following parts in config file specifying that
cst = {
wind_thresh = [ NA ];
field = [
{
name = "APCP_24";
level = [ "(*,*)" ];
cat_thresh = [ >=0.0, >=5.0 ];
}
];
};
obs = {
wind_thresh = [ NA ];
field = [
{
name = "APCP_24";
level = [ "(*,*)" ];
cat_thresh = [ >=0.0, >=5.0 ];
}
Now since 2 thresholds are specified, I should have 3x3 contingency tables.
>=0.0
0.0-4.9
>=5.0
For finite values of RF, how is 3x3 Table defined in this case????
3. Query reg FHO output.
in case of threhold >=0.0 , How the F_RATE, OBS_RATE will all be =1.
Kindly explain.
Geeta
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: output of grid_stat
From: Geeta Geeta
Time: Tue Jan 20 10:29:13 2015
Hi John.
This is in continuation with the earlier mail.
1. Pls clarify the range of MCTC categories what they will be in this
case???
2. Secondly, for the neighbourhood methods, U had told me to look at
the FSS.
Now for the POINTS AT the boundaries, how does MET define the 3x3 or
say 5x5 NEIGHBOURHOOD???
Pls explain.
Geeta
> Subject: [rt.rap.ucar.edu #70433] AutoReply: output of grid_stat
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Tue, 20 Jan 2015 10:23:28 -0700
>
> Greetings,
>
> This message has been automatically generated in response to the
> creation of a trouble ticket regarding:
> "output of grid_stat",
> a summary of which appears below.
>
> There is no need to reply to this message right now. Your ticket
has been
> assigned an ID of [rt.rap.ucar.edu #70433].
>
> Please include the string:
>
> [rt.rap.ucar.edu #70433]
>
> in the subject line of all future correspondence about this issue.
To do so,
> you may reply to this message.
>
> Thank you,
> met_help at ucar.edu
>
>
-------------------------------------------------------------------------
> Hi John.
> I had run grid stat on my data which has 25km resolution and got the
output.
> As per my header file, I have 37x37 points in the X n Y directions.
Output of ncdump -h for both datasets is given below.
> aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h
Obs.024_19june2006.nc
> netcdf Obs.024_19june2006 {
> dimensions:
> lon = 37 ;
> lat = 37 ;
> variables:
> double lon(lon) ;
> lon:units = "degrees_east" ;
> double lat(lat) ;
> lat:units = "degrees_north" ;
> float APCP_24(lat, lon) ;
> APCP_24:units = "kg/m^2" ;
> APCP_24:missing_value = -9999.f ;
> APCP_24:long_name = "Total precipitation" ;
> APCP_24:name = "APCP" ;
> APCP_24:level = "A24" ;
> APCP_24:grib_code = 61.f ;
> APCP_24:_FillValue = -9999.f ;
> APCP_24:init_time = "20060619_000000" ;
> APCP_24:init_time_ut = 1150675200 ;
> APCP_24:valid_time = "20060620_000000" ;
> APCP_24:valid_time_ut = 1150761600 ;
> APCP_24:accum_time = "240000" ;
> APCP_24:accum_time_sec = 86400 ;
>
> // global attributes:
> :FileOrigins = "File Obs.024_19june2006.nc generated
20150113_221613 on host agniilap by the Rscript trmmbin2nc.R" ;
> :MET_version = "V4.1" ;
> :Projection = "LatLon" ;
> :lat_ll = "9 degrees_north" ;
> :lon_ll = "74 degrees_east" ;
> :delta_lat = "0.25 degrees" ;
> :delta_lon = "0.25 degrees" ;
> :Nlat = "37 grid_points" ;
> :Nlon = "37 grid_points" ;
> }
> aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$
>
> FCST FILE:;;;;;;;;;;;;;;;;;;;;;;;;;;
>
> aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h
WRFPRS_d03.024.pcp.nc
> netcdf WRFPRS_d03.024.pcp {
> dimensions:
> lat = 37 ;
> lon = 37 ;
> variables:
> float lat(lat, lon) ;
> lat:long_name = "latitude" ;
> lat:units = "degrees_north" ;
> lat:standard_name = "latitude" ;
> float lon(lat, lon) ;
> lon:long_name = "longitude" ;
> lon:units = "degrees_east" ;
> lon:standard_name = "longitude" ;
> float APCP_24(lat, lon) ;
> APCP_24:name = "APCP" ;
> APCP_24:long_name = "Total precipitation" ;
> APCP_24:level = "A24" ;
> APCP_24:units = "kg/m^2" ;
> APCP_24:grib_code = 61 ;
> APCP_24:_FillValue = -9999.f ;
> APCP_24:init_time = "20060619_000000" ;
> APCP_24:init_time_ut = 1150675200 ;
> APCP_24:valid_time = "20060620_030000" ;
> APCP_24:valid_time_ut = 1150772400 ;
> APCP_24:accum_time = "240000" ;
> APCP_24:accum_time_sec = 86400 ;
>
> // global attributes:
> :FileOrigins = "File WRFPRS_d03.024.pcp.nc generated
20150113_113544 UTC on host rmcdlh by the MET pcp_combine tool" ;
> :MET_version = "V3.0" ;
> :MET_tool = "pcp_combine" ;
> :RunCommand = "Subtraction: WRFPRS_d03.027.grb25 with
accumulation of 270000 minus WRFPRS_d03.003.grb25 with accumulation of
030000." ;
> :Projection = "LatLon" ;
> :lat_ll = "9.000000 degrees_north" ;
> :lon_ll = "74.000000 degrees_east" ;
> :delta_lat = "0.250000 degrees" ;
> :delta_lon = "0.250000 degrees" ;
> :Nlat = "37 grid_points" ;
> :Nlon = "37 grid_points" ;
>
> Now when I had a look at my output,
> (say
> Why do I get 992 MPR's (output attatched).
> There should be 37*37 points=1369.
> 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 N_CAT F1_O1 F1_O2 F1_O3 F2_O1 F2_O2
F2_O3 F3_O1 F3_O2 F3_O3
> 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,>=5.000 >=0.000,>=5.000
NA NA MCTC 992 3 0 0 0 0 433 204
0 111 244
>
> 2. where in the Config file we specify the MCTC threholds
exclusively?? Attatching the config file I used for your reference.
> I saw following parts in config file specifying that
> cst = {
> wind_thresh = [ NA ];
>
> field = [
> {
> name = "APCP_24";
> level = [ "(*,*)" ];
> cat_thresh = [ >=0.0, >=5.0 ];
> }
> ];
>
> };
> obs = {
> wind_thresh = [ NA ];
>
> field = [
> {
> name = "APCP_24";
> level = [ "(*,*)" ];
> cat_thresh = [ >=0.0, >=5.0 ];
> }
> Now since 2 thresholds are specified, I should have 3x3 contingency
tables.
> >=0.0
> 0.0-4.9
> >=5.0
> For finite values of RF, how is 3x3 Table defined in this case????
>
> 3. Query reg FHO output.
> in case of threhold >=0.0 , How the F_RATE, OBS_RATE will all be =1.
>
> Kindly explain.
> Geeta
>
------------------------------------------------
Subject: output of grid_stat
From: John Halley Gotway
Time: Tue Jan 20 14:21:02 2015
Geeta,
1. Since the dimension of your grid is 37x37, you could get up to 1369
matched pairs. However, if the forecast or observation contains bad
data
at a point, that matched pair is not included in the calculations.
Using
the ncview tool to look at the
grid_stat_270000L_20060620_030000V_pairs.nc
NetCDF file you sent, I see that the forecast field contains missing
data
around the edge of the domain. That's the reason you're getting 992
matched pairs instead of 1369.
2. Yes, specifying 2 thresholds (>=0.0 and >=5.0) will result in a 3x3
contingency table. The MCTC counts for that 3x3 table are listed on
line 8
of the grid_stat_270000L_20060620_030000V.stat file you sent. The
MCTS
statistics corresponding to that 3x3 table are listed on line 9. The
3
categories are:
1: less than 0
2: 0 to 5
3: greater than 5
3. You're using grid_stat to evaluate precipitation, whose minimum
value
should be 0. If you're thresholding the data greater-than-or-equal-to
0
(>= 0), that will always evaluate to true for precipitation. I'd
suggest
using strictly greater-than 0 (>0) instead.
John
On Tue, Jan 20, 2015 at 10:23 AM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:
>
> Tue Jan 20 10:23:28 2015: Request 70433 was acted upon.
> Transaction: Ticket created by geeta124 at hotmail.com
> Queue: met_help
> Subject: output of grid_stat
> Owner: Nobody
> Requestors: geeta124 at hotmail.com
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433 >
>
>
> Hi John.
> I had run grid stat on my data which has 25km resolution and got the
> output.
> As per my header file, I have 37x37 points in the X n Y directions.
Output
> of ncdump -h for both datasets is given below.
> aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h
> Obs.024_19june2006.nc
> netcdf Obs.024_19june2006 {
> dimensions:
> lon = 37 ;
> lat = 37 ;
> variables:
> double lon(lon) ;
> lon:units = "degrees_east" ;
> double lat(lat) ;
> lat:units = "degrees_north" ;
> float APCP_24(lat, lon) ;
> APCP_24:units = "kg/m^2" ;
> APCP_24:missing_value = -9999.f ;
> APCP_24:long_name = "Total precipitation" ;
> APCP_24:name = "APCP" ;
> APCP_24:level = "A24" ;
> APCP_24:grib_code = 61.f ;
> APCP_24:_FillValue = -9999.f ;
> APCP_24:init_time = "20060619_000000" ;
> APCP_24:init_time_ut = 1150675200 ;
> APCP_24:valid_time = "20060620_000000" ;
> APCP_24:valid_time_ut = 1150761600 ;
> APCP_24:accum_time = "240000" ;
> APCP_24:accum_time_sec = 86400 ;
>
> // global attributes:
> :FileOrigins = "File Obs.024_19june2006.nc generated
> 20150113_221613 on host agniilap by the Rscript trmmbin2nc.R" ;
> :MET_version = "V4.1" ;
> :Projection = "LatLon" ;
> :lat_ll = "9 degrees_north" ;
> :lon_ll = "74 degrees_east" ;
> :delta_lat = "0.25 degrees" ;
> :delta_lon = "0.25 degrees" ;
> :Nlat = "37 grid_points" ;
> :Nlon = "37 grid_points" ;
> }
> aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$
>
> FCST FILE:;;;;;;;;;;;;;;;;;;;;;;;;;;
>
> aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h
> WRFPRS_d03.024.pcp.nc
> netcdf WRFPRS_d03.024.pcp {
> dimensions:
> lat = 37 ;
> lon = 37 ;
> variables:
> float lat(lat, lon) ;
> lat:long_name = "latitude" ;
> lat:units = "degrees_north" ;
> lat:standard_name = "latitude" ;
> float lon(lat, lon) ;
> lon:long_name = "longitude" ;
> lon:units = "degrees_east" ;
> lon:standard_name = "longitude" ;
> float APCP_24(lat, lon) ;
> APCP_24:name = "APCP" ;
> APCP_24:long_name = "Total precipitation" ;
> APCP_24:level = "A24" ;
> APCP_24:units = "kg/m^2" ;
> APCP_24:grib_code = 61 ;
> APCP_24:_FillValue = -9999.f ;
> APCP_24:init_time = "20060619_000000" ;
> APCP_24:init_time_ut = 1150675200 ;
> APCP_24:valid_time = "20060620_030000" ;
> APCP_24:valid_time_ut = 1150772400 ;
> APCP_24:accum_time = "240000" ;
> APCP_24:accum_time_sec = 86400 ;
>
> // global attributes:
> :FileOrigins = "File WRFPRS_d03.024.pcp.nc generated
> 20150113_113544 UTC on host rmcdlh by the MET pcp_combine tool" ;
> :MET_version = "V3.0" ;
> :MET_tool = "pcp_combine" ;
> :RunCommand = "Subtraction: WRFPRS_d03.027.grb25 with
accumulation
> of 270000 minus WRFPRS_d03.003.grb25 with accumulation of 030000." ;
> :Projection = "LatLon" ;
> :lat_ll = "9.000000 degrees_north" ;
> :lon_ll = "74.000000 degrees_east" ;
> :delta_lat = "0.250000 degrees" ;
> :delta_lon = "0.250000 degrees" ;
> :Nlat = "37 grid_points" ;
> :Nlon = "37 grid_points" ;
>
> Now when I had a look at my output,
> (say
> Why do I get 992 MPR's (output attatched).
> There should be 37*37 points=1369.
> 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 N_CAT F1_O1 F1_O2 F1_O3 F2_O1 F2_O2 F2_O3
F3_O1 F3_O2
> F3_O3
> 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,>=5.000 >=0.000,>=5.000
NA
> NA MCTC 992 3 0 0 0 0 433 204 0
111
> 244
>
> 2. where in the Config file we specify the MCTC threholds
exclusively??
> Attatching the config file I used for your reference.
> I saw following parts in config file specifying that
> cst = {
> wind_thresh = [ NA ];
>
> field = [
> {
> name = "APCP_24";
> level = [ "(*,*)" ];
> cat_thresh = [ >=0.0, >=5.0 ];
> }
> ];
>
> };
> obs = {
> wind_thresh = [ NA ];
>
> field = [
> {
> name = "APCP_24";
> level = [ "(*,*)" ];
> cat_thresh = [ >=0.0, >=5.0 ];
> }
> Now since 2 thresholds are specified, I should have 3x3 contingency
tables.
> >=0.0
> 0.0-4.9
> >=5.0
> For finite values of RF, how is 3x3 Table defined in this case????
>
> 3. Query reg FHO output.
> in case of threhold >=0.0 , How the F_RATE, OBS_RATE will all be =1.
>
> Kindly explain.
> Geeta
>
>
------------------------------------------------
Subject: output of grid_stat
From: John Halley Gotway
Time: Tue Jan 20 14:27:42 2015
Geeta,
1. I already covered the multi-category definitions in the previous
email.
2. When computing fractions skill score, MET uses the "vld_thresh"
setting
in the configuration file to decide how to handle data along the edge
of
the domain. Let's say it's computing a fractional coverage field
using a
5x5 neighborhood and it's at the edge of the domain. 15 points
contain
valid data and 10 points are outside the domain. Grid-Stat computes
the
valid data ratio as 15/25 = 0.6. Then it applies the valid data
threshold. Suppose vld_thresh = 0.5 ... since 0.6 > 0.5 it'll compute
a
fractional coverage value for that point using the 15 valid data
points.
Next suppose vld_thresh = 1.0 ... since 0.6 is less than 1.0, it'll
just
skip that point by setting it to bad data.
Setting vld_thresh = 1.0 ensure that FSS will only be computed at
points
where all NxN values contain valid data. Setting it to 0.5 only
requires
half of them.
So it is up to you to decide how you'd like to handle it.
Thanks,
John
On Tue, Jan 20, 2015 at 10:29 AM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433 >
>
> Hi John.
> This is in continuation with the earlier mail.
> 1. Pls clarify the range of MCTC categories what they will be in
this
> case???
> 2. Secondly, for the neighbourhood methods, U had told me to look at
the
> FSS.
> Now for the POINTS AT the boundaries, how does MET define the 3x3 or
say
> 5x5 NEIGHBOURHOOD???
> Pls explain.
> Geeta
>
> > Subject: [rt.rap.ucar.edu #70433] AutoReply: output of grid_stat
> > From: met_help at ucar.edu
> > To: geeta124 at hotmail.com
> > Date: Tue, 20 Jan 2015 10:23:28 -0700
> >
> > Greetings,
> >
> > This message has been automatically generated in response to the
> > creation of a trouble ticket regarding:
> > "output of grid_stat",
> > a summary of which appears below.
> >
> > There is no need to reply to this message right now. Your ticket
has
> been
> > assigned an ID of [rt.rap.ucar.edu #70433].
> >
> > Please include the string:
> >
> > [rt.rap.ucar.edu #70433]
> >
> > in the subject line of all future correspondence about this issue.
To do
> so,
> > you may reply to this message.
> >
> > Thank you,
> > met_help at ucar.edu
> >
> >
-------------------------------------------------------------------------
> > Hi John.
> > I had run grid stat on my data which has 25km resolution and got
the
> output.
> > As per my header file, I have 37x37 points in the X n Y
directions.
> Output of ncdump -h for both datasets is given below.
> > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h
> Obs.024_19june2006.nc
> > netcdf Obs.024_19june2006 {
> > dimensions:
> > lon = 37 ;
> > lat = 37 ;
> > variables:
> > double lon(lon) ;
> > lon:units = "degrees_east" ;
> > double lat(lat) ;
> > lat:units = "degrees_north" ;
> > float APCP_24(lat, lon) ;
> > APCP_24:units = "kg/m^2" ;
> > APCP_24:missing_value = -9999.f ;
> > APCP_24:long_name = "Total precipitation" ;
> > APCP_24:name = "APCP" ;
> > APCP_24:level = "A24" ;
> > APCP_24:grib_code = 61.f ;
> > APCP_24:_FillValue = -9999.f ;
> > APCP_24:init_time = "20060619_000000" ;
> > APCP_24:init_time_ut = 1150675200 ;
> > APCP_24:valid_time = "20060620_000000" ;
> > APCP_24:valid_time_ut = 1150761600 ;
> > APCP_24:accum_time = "240000" ;
> > APCP_24:accum_time_sec = 86400 ;
> >
> > // global attributes:
> > :FileOrigins = "File Obs.024_19june2006.nc generated
> 20150113_221613 on host agniilap by the Rscript trmmbin2nc.R" ;
> > :MET_version = "V4.1" ;
> > :Projection = "LatLon" ;
> > :lat_ll = "9 degrees_north" ;
> > :lon_ll = "74 degrees_east" ;
> > :delta_lat = "0.25 degrees" ;
> > :delta_lon = "0.25 degrees" ;
> > :Nlat = "37 grid_points" ;
> > :Nlon = "37 grid_points" ;
> > }
> > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$
> >
> > FCST FILE:;;;;;;;;;;;;;;;;;;;;;;;;;;
> >
> > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h
> WRFPRS_d03.024.pcp.nc
> > netcdf WRFPRS_d03.024.pcp {
> > dimensions:
> > lat = 37 ;
> > lon = 37 ;
> > variables:
> > float lat(lat, lon) ;
> > lat:long_name = "latitude" ;
> > lat:units = "degrees_north" ;
> > lat:standard_name = "latitude" ;
> > float lon(lat, lon) ;
> > lon:long_name = "longitude" ;
> > lon:units = "degrees_east" ;
> > lon:standard_name = "longitude" ;
> > float APCP_24(lat, lon) ;
> > APCP_24:name = "APCP" ;
> > APCP_24:long_name = "Total precipitation" ;
> > APCP_24:level = "A24" ;
> > APCP_24:units = "kg/m^2" ;
> > APCP_24:grib_code = 61 ;
> > APCP_24:_FillValue = -9999.f ;
> > APCP_24:init_time = "20060619_000000" ;
> > APCP_24:init_time_ut = 1150675200 ;
> > APCP_24:valid_time = "20060620_030000" ;
> > APCP_24:valid_time_ut = 1150772400 ;
> > APCP_24:accum_time = "240000" ;
> > APCP_24:accum_time_sec = 86400 ;
> >
> > // global attributes:
> > :FileOrigins = "File WRFPRS_d03.024.pcp.nc generated
> 20150113_113544 UTC on host rmcdlh by the MET pcp_combine tool" ;
> > :MET_version = "V3.0" ;
> > :MET_tool = "pcp_combine" ;
> > :RunCommand = "Subtraction: WRFPRS_d03.027.grb25 with
> accumulation of 270000 minus WRFPRS_d03.003.grb25 with accumulation
of
> 030000." ;
> > :Projection = "LatLon" ;
> > :lat_ll = "9.000000 degrees_north" ;
> > :lon_ll = "74.000000 degrees_east" ;
> > :delta_lat = "0.250000 degrees" ;
> > :delta_lon = "0.250000 degrees" ;
> > :Nlat = "37 grid_points" ;
> > :Nlon = "37 grid_points" ;
> >
> > Now when I had a look at my output,
> > (say
> > Why do I get 992 MPR's (output attatched).
> > There should be 37*37 points=1369.
> > 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 N_CAT F1_O1 F1_O2 F1_O3 F2_O1 F2_O2 F2_O3
F3_O1 F3_O2
> F3_O3
> > 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,>=5.000 >=0.000,>=5.000
NA
> NA MCTC 992 3 0 0 0 0 433 204 0
111
> 244
> >
> > 2. where in the Config file we specify the MCTC threholds
exclusively??
> Attatching the config file I used for your reference.
> > I saw following parts in config file specifying that
> > cst = {
> > wind_thresh = [ NA ];
> >
> > field = [
> > {
> > name = "APCP_24";
> > level = [ "(*,*)" ];
> > cat_thresh = [ >=0.0, >=5.0 ];
> > }
> > ];
> >
> > };
> > obs = {
> > wind_thresh = [ NA ];
> >
> > field = [
> > {
> > name = "APCP_24";
> > level = [ "(*,*)" ];
> > cat_thresh = [ >=0.0, >=5.0 ];
> > }
> > Now since 2 thresholds are specified, I should have 3x3
contingency
> tables.
> > >=0.0
> > 0.0-4.9
> > >=5.0
> > For finite values of RF, how is 3x3 Table defined in this case????
> >
> > 3. Query reg FHO output.
> > in case of threhold >=0.0 , How the F_RATE, OBS_RATE will all be
=1.
> >
> > Kindly explain.
> > Geeta
> >
>
>
------------------------------------------------
Subject: output of grid_stat
From: Geeta Geeta
Time: Wed Jan 28 03:02:54 2015
Hi John.
I viewed my output using the ncview and saw 3 variables.
FCST_APCP_A24_FULL, OBS_APCP_A24_FULL and DIFF_APCP_A24_FULL.
It shows a spatial plot with a scale.
I could not seee the MPRS. Want to know if I can count the bad ones
where there was no data.
How can I recognise those points where there was no data.
Geeta
> Subject: Re: [rt.rap.ucar.edu #70433] output of grid_stat
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Tue, 20 Jan 2015 14:21:03 -0700
>
> Geeta,
>
> 1. Since the dimension of your grid is 37x37, you could get up to
1369
> matched pairs. However, if the forecast or observation contains bad
data
> at a point, that matched pair is not included in the calculations.
Using
> the ncview tool to look at the
grid_stat_270000L_20060620_030000V_pairs.nc
> NetCDF file you sent, I see that the forecast field contains missing
data
> around the edge of the domain. That's the reason you're getting 992
> matched pairs instead of 1369.
>
> 2. Yes, specifying 2 thresholds (>=0.0 and >=5.0) will result in a
3x3
> contingency table. The MCTC counts for that 3x3 table are listed on
line 8
> of the grid_stat_270000L_20060620_030000V.stat file you sent. The
MCTS
> statistics corresponding to that 3x3 table are listed on line 9.
The 3
> categories are:
> 1: less than 0
> 2: 0 to 5
> 3: greater than 5
>
> 3. You're using grid_stat to evaluate precipitation, whose minimum
value
> should be 0. If you're thresholding the data greater-than-or-equal-
to 0
> (>= 0), that will always evaluate to true for precipitation. I'd
suggest
> using strictly greater-than 0 (>0) instead.
>
> John
>
>
>
>
> On Tue, Jan 20, 2015 at 10:23 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > Tue Jan 20 10:23:28 2015: Request 70433 was acted upon.
> > Transaction: Ticket created by geeta124 at hotmail.com
> > Queue: met_help
> > Subject: output of grid_stat
> > Owner: Nobody
> > Requestors: geeta124 at hotmail.com
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433 >
> >
> >
> > Hi John.
> > I had run grid stat on my data which has 25km resolution and got
the
> > output.
> > As per my header file, I have 37x37 points in the X n Y
directions. Output
> > of ncdump -h for both datasets is given below.
> > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h
> > Obs.024_19june2006.nc
> > netcdf Obs.024_19june2006 {
> > dimensions:
> > lon = 37 ;
> > lat = 37 ;
> > variables:
> > double lon(lon) ;
> > lon:units = "degrees_east" ;
> > double lat(lat) ;
> > lat:units = "degrees_north" ;
> > float APCP_24(lat, lon) ;
> > APCP_24:units = "kg/m^2" ;
> > APCP_24:missing_value = -9999.f ;
> > APCP_24:long_name = "Total precipitation" ;
> > APCP_24:name = "APCP" ;
> > APCP_24:level = "A24" ;
> > APCP_24:grib_code = 61.f ;
> > APCP_24:_FillValue = -9999.f ;
> > APCP_24:init_time = "20060619_000000" ;
> > APCP_24:init_time_ut = 1150675200 ;
> > APCP_24:valid_time = "20060620_000000" ;
> > APCP_24:valid_time_ut = 1150761600 ;
> > APCP_24:accum_time = "240000" ;
> > APCP_24:accum_time_sec = 86400 ;
> >
> > // global attributes:
> > :FileOrigins = "File Obs.024_19june2006.nc generated
> > 20150113_221613 on host agniilap by the Rscript trmmbin2nc.R" ;
> > :MET_version = "V4.1" ;
> > :Projection = "LatLon" ;
> > :lat_ll = "9 degrees_north" ;
> > :lon_ll = "74 degrees_east" ;
> > :delta_lat = "0.25 degrees" ;
> > :delta_lon = "0.25 degrees" ;
> > :Nlat = "37 grid_points" ;
> > :Nlon = "37 grid_points" ;
> > }
> > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$
> >
> > FCST FILE:;;;;;;;;;;;;;;;;;;;;;;;;;;
> >
> > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h
> > WRFPRS_d03.024.pcp.nc
> > netcdf WRFPRS_d03.024.pcp {
> > dimensions:
> > lat = 37 ;
> > lon = 37 ;
> > variables:
> > float lat(lat, lon) ;
> > lat:long_name = "latitude" ;
> > lat:units = "degrees_north" ;
> > lat:standard_name = "latitude" ;
> > float lon(lat, lon) ;
> > lon:long_name = "longitude" ;
> > lon:units = "degrees_east" ;
> > lon:standard_name = "longitude" ;
> > float APCP_24(lat, lon) ;
> > APCP_24:name = "APCP" ;
> > APCP_24:long_name = "Total precipitation" ;
> > APCP_24:level = "A24" ;
> > APCP_24:units = "kg/m^2" ;
> > APCP_24:grib_code = 61 ;
> > APCP_24:_FillValue = -9999.f ;
> > APCP_24:init_time = "20060619_000000" ;
> > APCP_24:init_time_ut = 1150675200 ;
> > APCP_24:valid_time = "20060620_030000" ;
> > APCP_24:valid_time_ut = 1150772400 ;
> > APCP_24:accum_time = "240000" ;
> > APCP_24:accum_time_sec = 86400 ;
> >
> > // global attributes:
> > :FileOrigins = "File WRFPRS_d03.024.pcp.nc generated
> > 20150113_113544 UTC on host rmcdlh by the MET pcp_combine tool" ;
> > :MET_version = "V3.0" ;
> > :MET_tool = "pcp_combine" ;
> > :RunCommand = "Subtraction: WRFPRS_d03.027.grb25 with
accumulation
> > of 270000 minus WRFPRS_d03.003.grb25 with accumulation of 030000."
;
> > :Projection = "LatLon" ;
> > :lat_ll = "9.000000 degrees_north" ;
> > :lon_ll = "74.000000 degrees_east" ;
> > :delta_lat = "0.250000 degrees" ;
> > :delta_lon = "0.250000 degrees" ;
> > :Nlat = "37 grid_points" ;
> > :Nlon = "37 grid_points" ;
> >
> > Now when I had a look at my output,
> > (say
> > Why do I get 992 MPR's (output attatched).
> > There should be 37*37 points=1369.
> > 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 N_CAT F1_O1 F1_O2 F1_O3 F2_O1 F2_O2 F2_O3
F3_O1 F3_O2
> > F3_O3
> > 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,>=5.000 >=0.000,>=5.000
NA
> > NA MCTC 992 3 0 0 0 0 433 204
0 111
> > 244
> >
> > 2. where in the Config file we specify the MCTC threholds
exclusively??
> > Attatching the config file I used for your reference.
> > I saw following parts in config file specifying that
> > cst = {
> > wind_thresh = [ NA ];
> >
> > field = [
> > {
> > name = "APCP_24";
> > level = [ "(*,*)" ];
> > cat_thresh = [ >=0.0, >=5.0 ];
> > }
> > ];
> >
> > };
> > obs = {
> > wind_thresh = [ NA ];
> >
> > field = [
> > {
> > name = "APCP_24";
> > level = [ "(*,*)" ];
> > cat_thresh = [ >=0.0, >=5.0 ];
> > }
> > Now since 2 thresholds are specified, I should have 3x3
contingency tables.
> > >=0.0
> > 0.0-4.9
> > >=5.0
> > For finite values of RF, how is 3x3 Table defined in this case????
> >
> > 3. Query reg FHO output.
> > in case of threhold >=0.0 , How the F_RATE, OBS_RATE will all be
=1.
> >
> > Kindly explain.
> > Geeta
> >
> >
>
------------------------------------------------
Subject: output of grid_stat
From: Geeta Geeta
Time: Wed Jan 28 03:06:51 2015
Hi John.
The scale in case of FCST was 0-50 while that of 0-100. the scale of
DIFF was -50 to 50.
Could not interpret the spatial plots. and I could not save this
output into any of the image formats.
Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
Date: Wed, 28 Jan 2015 15:31:37 +0530
Hi John.
I viewed my output using the ncview and saw 3 variables.
FCST_APCP_A24_FULL, OBS_APCP_A24_FULL and DIFF_APCP_A24_FULL.
It shows a spatial plot with a scale.
I could not seee the MPRS. Want to know if I can count the bad ones
where there was no data.
How can I recognise those points where there was no data.
Geeta
> Subject: Re: [rt.rap.ucar.edu #70433] output of grid_stat
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Tue, 20 Jan 2015 14:21:03 -0700
>
> Geeta,
>
> 1. Since the dimension of your grid is 37x37, you could get up to
1369
> matched pairs. However, if the forecast or observation contains bad
data
> at a point, that matched pair is not included in the calculations.
Using
> the ncview tool to look at the
grid_stat_270000L_20060620_030000V_pairs.nc
> NetCDF file you sent, I see that the forecast field contains missing
data
> around the edge of the domain. That's the reason you're getting 992
> matched pairs instead of 1369.
>
> 2. Yes, specifying 2 thresholds (>=0.0 and >=5.0) will result in a
3x3
> contingency table. The MCTC counts for that 3x3 table are listed on
line 8
> of the grid_stat_270000L_20060620_030000V.stat file you sent. The
MCTS
> statistics corresponding to that 3x3 table are listed on line 9.
The 3
> categories are:
> 1: less than 0
> 2: 0 to 5
> 3: greater than 5
>
> 3. You're using grid_stat to evaluate precipitation, whose minimum
value
> should be 0. If you're thresholding the data greater-than-or-equal-
to 0
> (>= 0), that will always evaluate to true for precipitation. I'd
suggest
> using strictly greater-than 0 (>0) instead.
>
> John
>
>
>
>
> On Tue, Jan 20, 2015 at 10:23 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > Tue Jan 20 10:23:28 2015: Request 70433 was acted upon.
> > Transaction: Ticket created by geeta124 at hotmail.com
> > Queue: met_help
> > Subject: output of grid_stat
> > Owner: Nobody
> > Requestors: geeta124 at hotmail.com
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433 >
> >
> >
> > Hi John.
> > I had run grid stat on my data which has 25km resolution and got
the
> > output.
> > As per my header file, I have 37x37 points in the X n Y
directions. Output
> > of ncdump -h for both datasets is given below.
> > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h
> > Obs.024_19june2006.nc
> > netcdf Obs.024_19june2006 {
> > dimensions:
> > lon = 37 ;
> > lat = 37 ;
> > variables:
> > double lon(lon) ;
> > lon:units = "degrees_east" ;
> > double lat(lat) ;
> > lat:units = "degrees_north" ;
> > float APCP_24(lat, lon) ;
> > APCP_24:units = "kg/m^2" ;
> > APCP_24:missing_value = -9999.f ;
> > APCP_24:long_name = "Total precipitation" ;
> > APCP_24:name = "APCP" ;
> > APCP_24:level = "A24" ;
> > APCP_24:grib_code = 61.f ;
> > APCP_24:_FillValue = -9999.f ;
> > APCP_24:init_time = "20060619_000000" ;
> > APCP_24:init_time_ut = 1150675200 ;
> > APCP_24:valid_time = "20060620_000000" ;
> > APCP_24:valid_time_ut = 1150761600 ;
> > APCP_24:accum_time = "240000" ;
> > APCP_24:accum_time_sec = 86400 ;
> >
> > // global attributes:
> > :FileOrigins = "File Obs.024_19june2006.nc generated
> > 20150113_221613 on host agniilap by the Rscript trmmbin2nc.R" ;
> > :MET_version = "V4.1" ;
> > :Projection = "LatLon" ;
> > :lat_ll = "9 degrees_north" ;
> > :lon_ll = "74 degrees_east" ;
> > :delta_lat = "0.25 degrees" ;
> > :delta_lon = "0.25 degrees" ;
> > :Nlat = "37 grid_points" ;
> > :Nlon = "37 grid_points" ;
> > }
> > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$
> >
> > FCST FILE:;;;;;;;;;;;;;;;;;;;;;;;;;;
> >
> > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h
> > WRFPRS_d03.024.pcp.nc
> > netcdf WRFPRS_d03.024.pcp {
> > dimensions:
> > lat = 37 ;
> > lon = 37 ;
> > variables:
> > float lat(lat, lon) ;
> > lat:long_name = "latitude" ;
> > lat:units = "degrees_north" ;
> > lat:standard_name = "latitude" ;
> > float lon(lat, lon) ;
> > lon:long_name = "longitude" ;
> > lon:units = "degrees_east" ;
> > lon:standard_name = "longitude" ;
> > float APCP_24(lat, lon) ;
> > APCP_24:name = "APCP" ;
> > APCP_24:long_name = "Total precipitation" ;
> > APCP_24:level = "A24" ;
> > APCP_24:units = "kg/m^2" ;
> > APCP_24:grib_code = 61 ;
> > APCP_24:_FillValue = -9999.f ;
> > APCP_24:init_time = "20060619_000000" ;
> > APCP_24:init_time_ut = 1150675200 ;
> > APCP_24:valid_time = "20060620_030000" ;
> > APCP_24:valid_time_ut = 1150772400 ;
> > APCP_24:accum_time = "240000" ;
> > APCP_24:accum_time_sec = 86400 ;
> >
> > // global attributes:
> > :FileOrigins = "File WRFPRS_d03.024.pcp.nc generated
> > 20150113_113544 UTC on host rmcdlh by the MET pcp_combine tool" ;
> > :MET_version = "V3.0" ;
> > :MET_tool = "pcp_combine" ;
> > :RunCommand = "Subtraction: WRFPRS_d03.027.grb25 with
accumulation
> > of 270000 minus WRFPRS_d03.003.grb25 with accumulation of 030000."
;
> > :Projection = "LatLon" ;
> > :lat_ll = "9.000000 degrees_north" ;
> > :lon_ll = "74.000000 degrees_east" ;
> > :delta_lat = "0.250000 degrees" ;
> > :delta_lon = "0.250000 degrees" ;
> > :Nlat = "37 grid_points" ;
> > :Nlon = "37 grid_points" ;
> >
> > Now when I had a look at my output,
> > (say
> > Why do I get 992 MPR's (output attatched).
> > There should be 37*37 points=1369.
> > 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 N_CAT F1_O1 F1_O2 F1_O3 F2_O1 F2_O2 F2_O3
F3_O1 F3_O2
> > F3_O3
> > 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,>=5.000 >=0.000,>=5.000
NA
> > NA MCTC 992 3 0 0 0 0 433 204
0 111
> > 244
> >
> > 2. where in the Config file we specify the MCTC threholds
exclusively??
> > Attatching the config file I used for your reference.
> > I saw following parts in config file specifying that
> > cst = {
> > wind_thresh = [ NA ];
> >
> > field = [
> > {
> > name = "APCP_24";
> > level = [ "(*,*)" ];
> > cat_thresh = [ >=0.0, >=5.0 ];
> > }
> > ];
> >
> > };
> > obs = {
> > wind_thresh = [ NA ];
> >
> > field = [
> > {
> > name = "APCP_24";
> > level = [ "(*,*)" ];
> > cat_thresh = [ >=0.0, >=5.0 ];
> > }
> > Now since 2 thresholds are specified, I should have 3x3
contingency tables.
> > >=0.0
> > 0.0-4.9
> > >=5.0
> > For finite values of RF, how is 3x3 Table defined in this case????
> >
> > 3. Query reg FHO output.
> > in case of threhold >=0.0 , How the F_RATE, OBS_RATE will all be
=1.
> >
> > Kindly explain.
> > Geeta
> >
> >
>
------------------------------------------------
Subject: output of grid_stat
From: Geeta Geeta
Time: Wed Jan 28 04:19:21 2015
Hi John.
I want to know about the interp group in the GRID stat config file.
It says, the FCST and OBS are interpolated if BOTH is specified.
Now after both datasets on the same grid and same resolution, what is
interpolated???
Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
Date: Wed, 28 Jan 2015 15:45:36 +0530
Hi John.
So I could save the data in the ps files. (attached for Ur reference).
But need your help in interpretation.
Also the LAT/LON are not as per Indian Domain.
it shows -8................
Kindly clarify.
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
Date: Wed, 28 Jan 2015 15:36:48 +0530
Hi John.
The scale in case of FCST was 0-50 while that of 0-100. the scale of
DIFF was -50 to 50.
Could not interpret the spatial plots. and I could not save this
output into any of the image formats.
Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
Date: Wed, 28 Jan 2015 15:31:37 +0530
Hi John.
I viewed my output using the ncview and saw 3 variables.
FCST_APCP_A24_FULL, OBS_APCP_A24_FULL and DIFF_APCP_A24_FULL.
It shows a spatial plot with a scale.
I could not seee the MPRS. Want to know if I can count the bad ones
where there was no data.
How can I recognise those points where there was no data.
Geeta
> Subject: Re: [rt.rap.ucar.edu #70433] output of grid_stat
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Tue, 20 Jan 2015 14:21:03 -0700
>
> Geeta,
>
> 1. Since the dimension of your grid is 37x37, you could get up to
1369
> matched pairs. However, if the forecast or observation contains bad
data
> at a point, that matched pair is not included in the calculations.
Using
> the ncview tool to look at the
grid_stat_270000L_20060620_030000V_pairs.nc
> NetCDF file you sent, I see that the forecast field contains missing
data
> around the edge of the domain. That's the reason you're getting 992
> matched pairs instead of 1369.
>
> 2. Yes, specifying 2 thresholds (>=0.0 and >=5.0) will result in a
3x3
> contingency table. The MCTC counts for that 3x3 table are listed on
line 8
> of the grid_stat_270000L_20060620_030000V.stat file you sent. The
MCTS
> statistics corresponding to that 3x3 table are listed on line 9.
The 3
> categories are:
> 1: less than 0
> 2: 0 to 5
> 3: greater than 5
>
> 3. You're using grid_stat to evaluate precipitation, whose minimum
value
> should be 0. If you're thresholding the data greater-than-or-equal-
to 0
> (>= 0), that will always evaluate to true for precipitation. I'd
suggest
> using strictly greater-than 0 (>0) instead.
>
> John
>
>
>
>
> On Tue, Jan 20, 2015 at 10:23 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > Tue Jan 20 10:23:28 2015: Request 70433 was acted upon.
> > Transaction: Ticket created by geeta124 at hotmail.com
> > Queue: met_help
> > Subject: output of grid_stat
> > Owner: Nobody
> > Requestors: geeta124 at hotmail.com
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433 >
> >
> >
> > Hi John.
> > I had run grid stat on my data which has 25km resolution and got
the
> > output.
> > As per my header file, I have 37x37 points in the X n Y
directions. Output
> > of ncdump -h for both datasets is given below.
> > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h
> > Obs.024_19june2006.nc
> > netcdf Obs.024_19june2006 {
> > dimensions:
> > lon = 37 ;
> > lat = 37 ;
> > variables:
> > double lon(lon) ;
> > lon:units = "degrees_east" ;
> > double lat(lat) ;
> > lat:units = "degrees_north" ;
> > float APCP_24(lat, lon) ;
> > APCP_24:units = "kg/m^2" ;
> > APCP_24:missing_value = -9999.f ;
> > APCP_24:long_name = "Total precipitation" ;
> > APCP_24:name = "APCP" ;
> > APCP_24:level = "A24" ;
> > APCP_24:grib_code = 61.f ;
> > APCP_24:_FillValue = -9999.f ;
> > APCP_24:init_time = "20060619_000000" ;
> > APCP_24:init_time_ut = 1150675200 ;
> > APCP_24:valid_time = "20060620_000000" ;
> > APCP_24:valid_time_ut = 1150761600 ;
> > APCP_24:accum_time = "240000" ;
> > APCP_24:accum_time_sec = 86400 ;
> >
> > // global attributes:
> > :FileOrigins = "File Obs.024_19june2006.nc generated
> > 20150113_221613 on host agniilap by the Rscript trmmbin2nc.R" ;
> > :MET_version = "V4.1" ;
> > :Projection = "LatLon" ;
> > :lat_ll = "9 degrees_north" ;
> > :lon_ll = "74 degrees_east" ;
> > :delta_lat = "0.25 degrees" ;
> > :delta_lon = "0.25 degrees" ;
> > :Nlat = "37 grid_points" ;
> > :Nlon = "37 grid_points" ;
> > }
> > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$
> >
> > FCST FILE:;;;;;;;;;;;;;;;;;;;;;;;;;;
> >
> > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump -h
> > WRFPRS_d03.024.pcp.nc
> > netcdf WRFPRS_d03.024.pcp {
> > dimensions:
> > lat = 37 ;
> > lon = 37 ;
> > variables:
> > float lat(lat, lon) ;
> > lat:long_name = "latitude" ;
> > lat:units = "degrees_north" ;
> > lat:standard_name = "latitude" ;
> > float lon(lat, lon) ;
> > lon:long_name = "longitude" ;
> > lon:units = "degrees_east" ;
> > lon:standard_name = "longitude" ;
> > float APCP_24(lat, lon) ;
> > APCP_24:name = "APCP" ;
> > APCP_24:long_name = "Total precipitation" ;
> > APCP_24:level = "A24" ;
> > APCP_24:units = "kg/m^2" ;
> > APCP_24:grib_code = 61 ;
> > APCP_24:_FillValue = -9999.f ;
> > APCP_24:init_time = "20060619_000000" ;
> > APCP_24:init_time_ut = 1150675200 ;
> > APCP_24:valid_time = "20060620_030000" ;
> > APCP_24:valid_time_ut = 1150772400 ;
> > APCP_24:accum_time = "240000" ;
> > APCP_24:accum_time_sec = 86400 ;
> >
> > // global attributes:
> > :FileOrigins = "File WRFPRS_d03.024.pcp.nc generated
> > 20150113_113544 UTC on host rmcdlh by the MET pcp_combine tool" ;
> > :MET_version = "V3.0" ;
> > :MET_tool = "pcp_combine" ;
> > :RunCommand = "Subtraction: WRFPRS_d03.027.grb25 with
accumulation
> > of 270000 minus WRFPRS_d03.003.grb25 with accumulation of 030000."
;
> > :Projection = "LatLon" ;
> > :lat_ll = "9.000000 degrees_north" ;
> > :lon_ll = "74.000000 degrees_east" ;
> > :delta_lat = "0.250000 degrees" ;
> > :delta_lon = "0.250000 degrees" ;
> > :Nlat = "37 grid_points" ;
> > :Nlon = "37 grid_points" ;
> >
> > Now when I had a look at my output,
> > (say
> > Why do I get 992 MPR's (output attatched).
> > There should be 37*37 points=1369.
> > 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 N_CAT F1_O1 F1_O2 F1_O3 F2_O1 F2_O2 F2_O3
F3_O1 F3_O2
> > F3_O3
> > 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,>=5.000 >=0.000,>=5.000
NA
> > NA MCTC 992 3 0 0 0 0 433 204
0 111
> > 244
> >
> > 2. where in the Config file we specify the MCTC threholds
exclusively??
> > Attatching the config file I used for your reference.
> > I saw following parts in config file specifying that
> > cst = {
> > wind_thresh = [ NA ];
> >
> > field = [
> > {
> > name = "APCP_24";
> > level = [ "(*,*)" ];
> > cat_thresh = [ >=0.0, >=5.0 ];
> > }
> > ];
> >
> > };
> > obs = {
> > wind_thresh = [ NA ];
> >
> > field = [
> > {
> > name = "APCP_24";
> > level = [ "(*,*)" ];
> > cat_thresh = [ >=0.0, >=5.0 ];
> > }
> > Now since 2 thresholds are specified, I should have 3x3
contingency tables.
> > >=0.0
> > 0.0-4.9
> > >=5.0
> > For finite values of RF, how is 3x3 Table defined in this case????
> >
> > 3. Query reg FHO output.
> > in case of threhold >=0.0 , How the F_RATE, OBS_RATE will all be
=1.
> >
> > Kindly explain.
> > Geeta
> >
> >
>
------------------------------------------------
Subject: output of grid_stat
From: John Halley Gotway
Time: Wed Jan 28 14:14:55 2015
Geeta,
You're correct. In Grid-Stat, the forecast and observation data must
already be on the same grid. In this context, the "interpolation"
option
is really more of a "smoothing" option. We only call it "interp" in
the
configuration file to be consistent with the same option in the Point-
Stat
configuration file. And logically speaking, the same steps are
performed
in Grid-Stat and Point-Stat for this option.
Here's how it works...
- The default interpolation option in Grid-Stat is unweighted mean of
the
nearest point (UW_MEAN of 1). That basically means do no
interpolation.
- Suppose you change the interpolation width to 5. For each grid
point,
it'll look at the 5x5 box around it and take the average of those 25
points. If FCST or BOTH is selected, it'll do the smoothing in the
forecast field. If OBS or BOTH is selected, it'll do the smoothing in
the
observation field.
- Choose one or more widths to vary the amount of smoothing.
- Choose one or more methods to vary the type summary (min, max,
unweighted
mean, and median).
Then output statistics are computed using the smoothed data rather
than the
raw data.
Why would you want to do this? It is well-known that when using
traditional statistics (like RMSE), smoother forecasts score better
the
higher resolution ones. Giving users the option to smooth their data
enables them to quantify how much the traditional scores would be with
a
smoother version of their forecast.
This answers similar questions that fractions skill score does, but
without
defining a categorical threshold first.
If you want to try this, I'd suggest applying it only to the forecast
field. Keep the observations fixed, and smooth the forecast data
using
widths of 1, 3, 5, 7, and 9. Then look to see how your scores change
as
your increase the amount of smoothing.
Hope that helps clarify.
John
On Wed, Jan 28, 2015 at 4:19 AM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433 >
>
> Hi John.
> I want to know about the interp group in the GRID stat config file.
> It says, the FCST and OBS are interpolated if BOTH is specified.
> Now after both datasets on the same grid and same resolution, what
is
> interpolated???
> Geeta
>
> From: geeta124 at hotmail.com
> To: met_help at ucar.edu
> Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
> Date: Wed, 28 Jan 2015 15:45:36 +0530
>
>
>
>
> Hi John.
> So I could save the data in the ps files. (attached for Ur
reference).
> But need your help in interpretation.
> Also the LAT/LON are not as per Indian Domain.
> it shows -8................
> Kindly clarify.
>
> From: geeta124 at hotmail.com
> To: met_help at ucar.edu
> Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
> Date: Wed, 28 Jan 2015 15:36:48 +0530
>
>
>
>
> Hi John.
> The scale in case of FCST was 0-50 while that of 0-100. the scale of
DIFF
> was -50 to 50.
> Could not interpret the spatial plots. and I could not save this
output
> into any of the image formats.
> Geeta
>
> From: geeta124 at hotmail.com
> To: met_help at ucar.edu
> Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
> Date: Wed, 28 Jan 2015 15:31:37 +0530
>
>
>
>
> Hi John.
> I viewed my output using the ncview and saw 3 variables.
> FCST_APCP_A24_FULL, OBS_APCP_A24_FULL and DIFF_APCP_A24_FULL.
> It shows a spatial plot with a scale.
>
> I could not seee the MPRS. Want to know if I can count the bad ones
where
> there was no data.
> How can I recognise those points where there was no data.
> Geeta
>
> > Subject: Re: [rt.rap.ucar.edu #70433] output of grid_stat
> > From: met_help at ucar.edu
> > To: geeta124 at hotmail.com
> > Date: Tue, 20 Jan 2015 14:21:03 -0700
> >
> > Geeta,
> >
> > 1. Since the dimension of your grid is 37x37, you could get up to
1369
> > matched pairs. However, if the forecast or observation contains
bad data
> > at a point, that matched pair is not included in the calculations.
Using
> > the ncview tool to look at the
> grid_stat_270000L_20060620_030000V_pairs.nc
> > NetCDF file you sent, I see that the forecast field contains
missing data
> > around the edge of the domain. That's the reason you're getting
992
> > matched pairs instead of 1369.
> >
> > 2. Yes, specifying 2 thresholds (>=0.0 and >=5.0) will result in a
3x3
> > contingency table. The MCTC counts for that 3x3 table are listed
on
> line 8
> > of the grid_stat_270000L_20060620_030000V.stat file you sent. The
MCTS
> > statistics corresponding to that 3x3 table are listed on line 9.
The 3
> > categories are:
> > 1: less than 0
> > 2: 0 to 5
> > 3: greater than 5
> >
> > 3. You're using grid_stat to evaluate precipitation, whose minimum
value
> > should be 0. If you're thresholding the data greater-than-or-
equal-to 0
> > (>= 0), that will always evaluate to true for precipitation. I'd
suggest
> > using strictly greater-than 0 (>0) instead.
> >
> > John
> >
> >
> >
> >
> > On Tue, Jan 20, 2015 at 10:23 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > wrote:
> >
> > >
> > > Tue Jan 20 10:23:28 2015: Request 70433 was acted upon.
> > > Transaction: Ticket created by geeta124 at hotmail.com
> > > Queue: met_help
> > > Subject: output of grid_stat
> > > Owner: Nobody
> > > Requestors: geeta124 at hotmail.com
> > > Status: new
> > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433
> >
> > >
> > >
> > > Hi John.
> > > I had run grid stat on my data which has 25km resolution and got
the
> > > output.
> > > As per my header file, I have 37x37 points in the X n Y
directions.
> Output
> > > of ncdump -h for both datasets is given below.
> > > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump
-h
> > > Obs.024_19june2006.nc
> > > netcdf Obs.024_19june2006 {
> > > dimensions:
> > > lon = 37 ;
> > > lat = 37 ;
> > > variables:
> > > double lon(lon) ;
> > > lon:units = "degrees_east" ;
> > > double lat(lat) ;
> > > lat:units = "degrees_north" ;
> > > float APCP_24(lat, lon) ;
> > > APCP_24:units = "kg/m^2" ;
> > > APCP_24:missing_value = -9999.f ;
> > > APCP_24:long_name = "Total precipitation" ;
> > > APCP_24:name = "APCP" ;
> > > APCP_24:level = "A24" ;
> > > APCP_24:grib_code = 61.f ;
> > > APCP_24:_FillValue = -9999.f ;
> > > APCP_24:init_time = "20060619_000000" ;
> > > APCP_24:init_time_ut = 1150675200 ;
> > > APCP_24:valid_time = "20060620_000000" ;
> > > APCP_24:valid_time_ut = 1150761600 ;
> > > APCP_24:accum_time = "240000" ;
> > > APCP_24:accum_time_sec = 86400 ;
> > >
> > > // global attributes:
> > > :FileOrigins = "File Obs.024_19june2006.nc generated
> > > 20150113_221613 on host agniilap by the Rscript trmmbin2nc.R" ;
> > > :MET_version = "V4.1" ;
> > > :Projection = "LatLon" ;
> > > :lat_ll = "9 degrees_north" ;
> > > :lon_ll = "74 degrees_east" ;
> > > :delta_lat = "0.25 degrees" ;
> > > :delta_lon = "0.25 degrees" ;
> > > :Nlat = "37 grid_points" ;
> > > :Nlon = "37 grid_points" ;
> > > }
> > > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$
> > >
> > > FCST FILE:;;;;;;;;;;;;;;;;;;;;;;;;;;
> > >
> > > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump
-h
> > > WRFPRS_d03.024.pcp.nc
> > > netcdf WRFPRS_d03.024.pcp {
> > > dimensions:
> > > lat = 37 ;
> > > lon = 37 ;
> > > variables:
> > > float lat(lat, lon) ;
> > > lat:long_name = "latitude" ;
> > > lat:units = "degrees_north" ;
> > > lat:standard_name = "latitude" ;
> > > float lon(lat, lon) ;
> > > lon:long_name = "longitude" ;
> > > lon:units = "degrees_east" ;
> > > lon:standard_name = "longitude" ;
> > > float APCP_24(lat, lon) ;
> > > APCP_24:name = "APCP" ;
> > > APCP_24:long_name = "Total precipitation" ;
> > > APCP_24:level = "A24" ;
> > > APCP_24:units = "kg/m^2" ;
> > > APCP_24:grib_code = 61 ;
> > > APCP_24:_FillValue = -9999.f ;
> > > APCP_24:init_time = "20060619_000000" ;
> > > APCP_24:init_time_ut = 1150675200 ;
> > > APCP_24:valid_time = "20060620_030000" ;
> > > APCP_24:valid_time_ut = 1150772400 ;
> > > APCP_24:accum_time = "240000" ;
> > > APCP_24:accum_time_sec = 86400 ;
> > >
> > > // global attributes:
> > > :FileOrigins = "File WRFPRS_d03.024.pcp.nc generated
> > > 20150113_113544 UTC on host rmcdlh by the MET pcp_combine tool"
;
> > > :MET_version = "V3.0" ;
> > > :MET_tool = "pcp_combine" ;
> > > :RunCommand = "Subtraction: WRFPRS_d03.027.grb25 with
> accumulation
> > > of 270000 minus WRFPRS_d03.003.grb25 with accumulation of
030000." ;
> > > :Projection = "LatLon" ;
> > > :lat_ll = "9.000000 degrees_north" ;
> > > :lon_ll = "74.000000 degrees_east" ;
> > > :delta_lat = "0.250000 degrees" ;
> > > :delta_lon = "0.250000 degrees" ;
> > > :Nlat = "37 grid_points" ;
> > > :Nlon = "37 grid_points" ;
> > >
> > > Now when I had a look at my output,
> > > (say
> > > Why do I get 992 MPR's (output attatched).
> > > There should be 37*37 points=1369.
> > > 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 N_CAT F1_O1 F1_O2 F1_O3 F2_O1 F2_O2 F2_O3
F3_O1
> F3_O2
> > > F3_O3
> > > 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,>=5.000 >=0.000,>=5.000
> NA
> > > NA MCTC 992 3 0 0 0 0 433 204
0
> 111
> > > 244
> > >
> > > 2. where in the Config file we specify the MCTC threholds
exclusively??
> > > Attatching the config file I used for your reference.
> > > I saw following parts in config file specifying that
> > > cst = {
> > > wind_thresh = [ NA ];
> > >
> > > field = [
> > > {
> > > name = "APCP_24";
> > > level = [ "(*,*)" ];
> > > cat_thresh = [ >=0.0, >=5.0 ];
> > > }
> > > ];
> > >
> > > };
> > > obs = {
> > > wind_thresh = [ NA ];
> > >
> > > field = [
> > > {
> > > name = "APCP_24";
> > > level = [ "(*,*)" ];
> > > cat_thresh = [ >=0.0, >=5.0 ];
> > > }
> > > Now since 2 thresholds are specified, I should have 3x3
contingency
> tables.
> > > >=0.0
> > > 0.0-4.9
> > > >=5.0
> > > For finite values of RF, how is 3x3 Table defined in this
case????
> > >
> > > 3. Query reg FHO output.
> > > in case of threhold >=0.0 , How the F_RATE, OBS_RATE will all be
=1.
> > >
> > > Kindly explain.
> > > Geeta
> > >
> > >
> >
>
>
------------------------------------------------
Subject: output of grid_stat
From: Geeta Geeta
Time: Thu Jan 29 22:48:25 2015
Hi John.
One further query to this aspect. When You say interp option in Grid
stat, Do U mean the interpolation at the copygb Level???.
When we run copyGb on the model data (BEFORE GRID STAT), it transforms
it from 9/5km grid to 25 km grid for it to be consistent with the
observed data.
After this step, do u mean there is another round of interpolation
which you have explained below???.
If that is the case, what is the difference between running GRID_STAT
and upscaling??? Kindly clarify this.
Also pls clarify my doubt reg the ncview(MPR plots) and the error that
I got for the TRMM acumulated data.
I shall be waiting for your response.
Geeta
> Subject: Re: [rt.rap.ucar.edu #70433] output of grid_stat
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Wed, 28 Jan 2015 14:14:56 -0700
>
> Geeta,
>
> You're correct. In Grid-Stat, the forecast and observation data
must
> already be on the same grid. In this context, the "interpolation"
option
> is really more of a "smoothing" option. We only call it "interp" in
the
> configuration file to be consistent with the same option in the
Point-Stat
> configuration file. And logically speaking, the same steps are
performed
> in Grid-Stat and Point-Stat for this option.
>
> Here's how it works...
> - The default interpolation option in Grid-Stat is unweighted mean
of the
> nearest point (UW_MEAN of 1). That basically means do no
interpolation.
> - Suppose you change the interpolation width to 5. For each grid
point,
> it'll look at the 5x5 box around it and take the average of those 25
> points. If FCST or BOTH is selected, it'll do the smoothing in the
> forecast field. If OBS or BOTH is selected, it'll do the smoothing
in the
> observation field.
> - Choose one or more widths to vary the amount of smoothing.
> - Choose one or more methods to vary the type summary (min, max,
unweighted
> mean, and median).
>
> Then output statistics are computed using the smoothed data rather
than the
> raw data.
>
> Why would you want to do this? It is well-known that when using
> traditional statistics (like RMSE), smoother forecasts score better
the
> higher resolution ones. Giving users the option to smooth their
data
> enables them to quantify how much the traditional scores would be
with a
> smoother version of their forecast.
>
> This answers similar questions that fractions skill score does, but
without
> defining a categorical threshold first.
>
> If you want to try this, I'd suggest applying it only to the
forecast
> field. Keep the observations fixed, and smooth the forecast data
using
> widths of 1, 3, 5, 7, and 9. Then look to see how your scores
change as
> your increase the amount of smoothing.
>
> Hope that helps clarify.
>
> John
>
>
> On Wed, Jan 28, 2015 at 4:19 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433 >
> >
> > Hi John.
> > I want to know about the interp group in the GRID stat config
file.
> > It says, the FCST and OBS are interpolated if BOTH is specified.
> > Now after both datasets on the same grid and same resolution, what
is
> > interpolated???
> > Geeta
> >
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
> > Date: Wed, 28 Jan 2015 15:45:36 +0530
> >
> >
> >
> >
> > Hi John.
> > So I could save the data in the ps files. (attached for Ur
reference).
> > But need your help in interpretation.
> > Also the LAT/LON are not as per Indian Domain.
> > it shows -8................
> > Kindly clarify.
> >
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
> > Date: Wed, 28 Jan 2015 15:36:48 +0530
> >
> >
> >
> >
> > Hi John.
> > The scale in case of FCST was 0-50 while that of 0-100. the scale
of DIFF
> > was -50 to 50.
> > Could not interpret the spatial plots. and I could not save this
output
> > into any of the image formats.
> > Geeta
> >
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
> > Date: Wed, 28 Jan 2015 15:31:37 +0530
> >
> >
> >
> >
> > Hi John.
> > I viewed my output using the ncview and saw 3 variables.
> > FCST_APCP_A24_FULL, OBS_APCP_A24_FULL and DIFF_APCP_A24_FULL.
> > It shows a spatial plot with a scale.
> >
> > I could not seee the MPRS. Want to know if I can count the bad
ones where
> > there was no data.
> > How can I recognise those points where there was no data.
> > Geeta
> >
> > > Subject: Re: [rt.rap.ucar.edu #70433] output of grid_stat
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Tue, 20 Jan 2015 14:21:03 -0700
> > >
> > > Geeta,
> > >
> > > 1. Since the dimension of your grid is 37x37, you could get up
to 1369
> > > matched pairs. However, if the forecast or observation contains
bad data
> > > at a point, that matched pair is not included in the
calculations. Using
> > > the ncview tool to look at the
> > grid_stat_270000L_20060620_030000V_pairs.nc
> > > NetCDF file you sent, I see that the forecast field contains
missing data
> > > around the edge of the domain. That's the reason you're getting
992
> > > matched pairs instead of 1369.
> > >
> > > 2. Yes, specifying 2 thresholds (>=0.0 and >=5.0) will result in
a 3x3
> > > contingency table. The MCTC counts for that 3x3 table are
listed on
> > line 8
> > > of the grid_stat_270000L_20060620_030000V.stat file you sent.
The MCTS
> > > statistics corresponding to that 3x3 table are listed on line 9.
The 3
> > > categories are:
> > > 1: less than 0
> > > 2: 0 to 5
> > > 3: greater than 5
> > >
> > > 3. You're using grid_stat to evaluate precipitation, whose
minimum value
> > > should be 0. If you're thresholding the data greater-than-or-
equal-to 0
> > > (>= 0), that will always evaluate to true for precipitation.
I'd suggest
> > > using strictly greater-than 0 (>0) instead.
> > >
> > > John
> > >
> > >
> > >
> > >
> > > On Tue, Jan 20, 2015 at 10:23 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > Tue Jan 20 10:23:28 2015: Request 70433 was acted upon.
> > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > Queue: met_help
> > > > Subject: output of grid_stat
> > > > Owner: Nobody
> > > > Requestors: geeta124 at hotmail.com
> > > > Status: new
> > > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433
> > >
> > > >
> > > >
> > > > Hi John.
> > > > I had run grid stat on my data which has 25km resolution and
got the
> > > > output.
> > > > As per my header file, I have 37x37 points in the X n Y
directions.
> > Output
> > > > of ncdump -h for both datasets is given below.
> > > > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump
-h
> > > > Obs.024_19june2006.nc
> > > > netcdf Obs.024_19june2006 {
> > > > dimensions:
> > > > lon = 37 ;
> > > > lat = 37 ;
> > > > variables:
> > > > double lon(lon) ;
> > > > lon:units = "degrees_east" ;
> > > > double lat(lat) ;
> > > > lat:units = "degrees_north" ;
> > > > float APCP_24(lat, lon) ;
> > > > APCP_24:units = "kg/m^2" ;
> > > > APCP_24:missing_value = -9999.f ;
> > > > APCP_24:long_name = "Total precipitation" ;
> > > > APCP_24:name = "APCP" ;
> > > > APCP_24:level = "A24" ;
> > > > APCP_24:grib_code = 61.f ;
> > > > APCP_24:_FillValue = -9999.f ;
> > > > APCP_24:init_time = "20060619_000000" ;
> > > > APCP_24:init_time_ut = 1150675200 ;
> > > > APCP_24:valid_time = "20060620_000000" ;
> > > > APCP_24:valid_time_ut = 1150761600 ;
> > > > APCP_24:accum_time = "240000" ;
> > > > APCP_24:accum_time_sec = 86400 ;
> > > >
> > > > // global attributes:
> > > > :FileOrigins = "File Obs.024_19june2006.nc generated
> > > > 20150113_221613 on host agniilap by the Rscript trmmbin2nc.R"
;
> > > > :MET_version = "V4.1" ;
> > > > :Projection = "LatLon" ;
> > > > :lat_ll = "9 degrees_north" ;
> > > > :lon_ll = "74 degrees_east" ;
> > > > :delta_lat = "0.25 degrees" ;
> > > > :delta_lon = "0.25 degrees" ;
> > > > :Nlat = "37 grid_points" ;
> > > > :Nlon = "37 grid_points" ;
> > > > }
> > > > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$
> > > >
> > > > FCST FILE:;;;;;;;;;;;;;;;;;;;;;;;;;;
> > > >
> > > > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump
-h
> > > > WRFPRS_d03.024.pcp.nc
> > > > netcdf WRFPRS_d03.024.pcp {
> > > > dimensions:
> > > > lat = 37 ;
> > > > lon = 37 ;
> > > > variables:
> > > > float lat(lat, lon) ;
> > > > lat:long_name = "latitude" ;
> > > > lat:units = "degrees_north" ;
> > > > lat:standard_name = "latitude" ;
> > > > float lon(lat, lon) ;
> > > > lon:long_name = "longitude" ;
> > > > lon:units = "degrees_east" ;
> > > > lon:standard_name = "longitude" ;
> > > > float APCP_24(lat, lon) ;
> > > > APCP_24:name = "APCP" ;
> > > > APCP_24:long_name = "Total precipitation" ;
> > > > APCP_24:level = "A24" ;
> > > > APCP_24:units = "kg/m^2" ;
> > > > APCP_24:grib_code = 61 ;
> > > > APCP_24:_FillValue = -9999.f ;
> > > > APCP_24:init_time = "20060619_000000" ;
> > > > APCP_24:init_time_ut = 1150675200 ;
> > > > APCP_24:valid_time = "20060620_030000" ;
> > > > APCP_24:valid_time_ut = 1150772400 ;
> > > > APCP_24:accum_time = "240000" ;
> > > > APCP_24:accum_time_sec = 86400 ;
> > > >
> > > > // global attributes:
> > > > :FileOrigins = "File WRFPRS_d03.024.pcp.nc generated
> > > > 20150113_113544 UTC on host rmcdlh by the MET pcp_combine
tool" ;
> > > > :MET_version = "V3.0" ;
> > > > :MET_tool = "pcp_combine" ;
> > > > :RunCommand = "Subtraction: WRFPRS_d03.027.grb25 with
> > accumulation
> > > > of 270000 minus WRFPRS_d03.003.grb25 with accumulation of
030000." ;
> > > > :Projection = "LatLon" ;
> > > > :lat_ll = "9.000000 degrees_north" ;
> > > > :lon_ll = "74.000000 degrees_east" ;
> > > > :delta_lat = "0.250000 degrees" ;
> > > > :delta_lon = "0.250000 degrees" ;
> > > > :Nlat = "37 grid_points" ;
> > > > :Nlon = "37 grid_points" ;
> > > >
> > > > Now when I had a look at my output,
> > > > (say
> > > > Why do I get 992 MPR's (output attatched).
> > > > There should be 37*37 points=1369.
> > > > 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 N_CAT F1_O1 F1_O2 F1_O3 F2_O1 F2_O2
F2_O3 F3_O1
> > F3_O2
> > > > F3_O3
> > > > 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,>=5.000
>=0.000,>=5.000
> > NA
> > > > NA MCTC 992 3 0 0 0 0 433 204
0
> > 111
> > > > 244
> > > >
> > > > 2. where in the Config file we specify the MCTC threholds
exclusively??
> > > > Attatching the config file I used for your reference.
> > > > I saw following parts in config file specifying that
> > > > cst = {
> > > > wind_thresh = [ NA ];
> > > >
> > > > field = [
> > > > {
> > > > name = "APCP_24";
> > > > level = [ "(*,*)" ];
> > > > cat_thresh = [ >=0.0, >=5.0 ];
> > > > }
> > > > ];
> > > >
> > > > };
> > > > obs = {
> > > > wind_thresh = [ NA ];
> > > >
> > > > field = [
> > > > {
> > > > name = "APCP_24";
> > > > level = [ "(*,*)" ];
> > > > cat_thresh = [ >=0.0, >=5.0 ];
> > > > }
> > > > Now since 2 thresholds are specified, I should have 3x3
contingency
> > tables.
> > > > >=0.0
> > > > 0.0-4.9
> > > > >=5.0
> > > > For finite values of RF, how is 3x3 Table defined in this
case????
> > > >
> > > > 3. Query reg FHO output.
> > > > in case of threhold >=0.0 , How the F_RATE, OBS_RATE will all
be =1.
> > > >
> > > > Kindly explain.
> > > > Geeta
> > > >
> > > >
> > >
> >
> >
>
------------------------------------------------
Subject: output of grid_stat
From: Geeta Geeta
Time: Thu Jan 29 23:04:47 2015
John.
I was going through the GRID_STAT config file.
Since I m discussing the smoothing/upscaling aspect of grid_stat. I
came across two things in this file---
1. // Interpolation methods
//
interp = {
field = BOTH;
vld_thresh = 1.0;
type = [
{
method = UW_MEAN;
width = 1;
}
];
};
What you have suggested in your earlier email is that I should smooth
the FCST field by specifying as FCST (ie. field = FCST;) under
this HEADER and putting width =3,5 or 7..... etc and then check for
the variation in the traditional Scores, how they vary with smoothed
forecast.
2. NOW, We also have the following HEADER called as the NEIGHBOURHOOD
methods. In this also R nt we are doing the same thing????
//
// Neighborhood methods
//
nbrhd = {
vld_thresh = 1.0;
width = [ 3 ];
cov_thresh = [ >=0.5 ];
}
In this part also, we are varying the width as 1, 3, 5 n so on.
Can U pls explain the difference between the two things specified in
the CONFIG file???.
If I change any of the above parameters, THE FSS will change!!!!.
Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
Date: Fri, 30 Jan 2015 11:17:13 +0530
Hi John.
One further query to this aspect. When You say interp option in Grid
stat, Do U mean the interpolation at the copygb Level???.
When we run copyGb on the model data (BEFORE GRID STAT), it transforms
it from 9/5km grid to 25 km grid for it to be consistent with the
observed data.
After this step, do u mean there is another round of interpolation
which you have explained below???.
If that is the case, what is the difference between running GRID_STAT
and upscaling??? Kindly clarify this.
Also pls clarify my doubt reg the ncview(MPR plots) and the error that
I got for the TRMM acumulated data.
I shall be waiting for your response.
Geeta
> Subject: Re: [rt.rap.ucar.edu #70433] output of grid_stat
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Wed, 28 Jan 2015 14:14:56 -0700
>
> Geeta,
>
> You're correct. In Grid-Stat, the forecast and observation data
must
> already be on the same grid. In this context, the "interpolation"
option
> is really more of a "smoothing" option. We only call it "interp" in
the
> configuration file to be consistent with the same option in the
Point-Stat
> configuration file. And logically speaking, the same steps are
performed
> in Grid-Stat and Point-Stat for this option.
>
> Here's how it works...
> - The default interpolation option in Grid-Stat is unweighted mean
of the
> nearest point (UW_MEAN of 1). That basically means do no
interpolation.
> - Suppose you change the interpolation width to 5. For each grid
point,
> it'll look at the 5x5 box around it and take the average of those 25
> points. If FCST or BOTH is selected, it'll do the smoothing in the
> forecast field. If OBS or BOTH is selected, it'll do the smoothing
in the
> observation field.
> - Choose one or more widths to vary the amount of smoothing.
> - Choose one or more methods to vary the type summary (min, max,
unweighted
> mean, and median).
>
> Then output statistics are computed using the smoothed data rather
than the
> raw data.
>
> Why would you want to do this? It is well-known that when using
> traditional statistics (like RMSE), smoother forecasts score better
the
> higher resolution ones. Giving users the option to smooth their
data
> enables them to quantify how much the traditional scores would be
with a
> smoother version of their forecast.
>
> This answers similar questions that fractions skill score does, but
without
> defining a categorical threshold first.
>
> If you want to try this, I'd suggest applying it only to the
forecast
> field. Keep the observations fixed, and smooth the forecast data
using
> widths of 1, 3, 5, 7, and 9. Then look to see how your scores
change as
> your increase the amount of smoothing.
>
> Hope that helps clarify.
>
> John
>
>
> On Wed, Jan 28, 2015 at 4:19 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433 >
> >
> > Hi John.
> > I want to know about the interp group in the GRID stat config
file.
> > It says, the FCST and OBS are interpolated if BOTH is specified.
> > Now after both datasets on the same grid and same resolution, what
is
> > interpolated???
> > Geeta
> >
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
> > Date: Wed, 28 Jan 2015 15:45:36 +0530
> >
> >
> >
> >
> > Hi John.
> > So I could save the data in the ps files. (attached for Ur
reference).
> > But need your help in interpretation.
> > Also the LAT/LON are not as per Indian Domain.
> > it shows -8................
> > Kindly clarify.
> >
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
> > Date: Wed, 28 Jan 2015 15:36:48 +0530
> >
> >
> >
> >
> > Hi John.
> > The scale in case of FCST was 0-50 while that of 0-100. the scale
of DIFF
> > was -50 to 50.
> > Could not interpret the spatial plots. and I could not save this
output
> > into any of the image formats.
> > Geeta
> >
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
> > Date: Wed, 28 Jan 2015 15:31:37 +0530
> >
> >
> >
> >
> > Hi John.
> > I viewed my output using the ncview and saw 3 variables.
> > FCST_APCP_A24_FULL, OBS_APCP_A24_FULL and DIFF_APCP_A24_FULL.
> > It shows a spatial plot with a scale.
> >
> > I could not seee the MPRS. Want to know if I can count the bad
ones where
> > there was no data.
> > How can I recognise those points where there was no data.
> > Geeta
> >
> > > Subject: Re: [rt.rap.ucar.edu #70433] output of grid_stat
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Tue, 20 Jan 2015 14:21:03 -0700
> > >
> > > Geeta,
> > >
> > > 1. Since the dimension of your grid is 37x37, you could get up
to 1369
> > > matched pairs. However, if the forecast or observation contains
bad data
> > > at a point, that matched pair is not included in the
calculations. Using
> > > the ncview tool to look at the
> > grid_stat_270000L_20060620_030000V_pairs.nc
> > > NetCDF file you sent, I see that the forecast field contains
missing data
> > > around the edge of the domain. That's the reason you're getting
992
> > > matched pairs instead of 1369.
> > >
> > > 2. Yes, specifying 2 thresholds (>=0.0 and >=5.0) will result in
a 3x3
> > > contingency table. The MCTC counts for that 3x3 table are
listed on
> > line 8
> > > of the grid_stat_270000L_20060620_030000V.stat file you sent.
The MCTS
> > > statistics corresponding to that 3x3 table are listed on line 9.
The 3
> > > categories are:
> > > 1: less than 0
> > > 2: 0 to 5
> > > 3: greater than 5
> > >
> > > 3. You're using grid_stat to evaluate precipitation, whose
minimum value
> > > should be 0. If you're thresholding the data greater-than-or-
equal-to 0
> > > (>= 0), that will always evaluate to true for precipitation.
I'd suggest
> > > using strictly greater-than 0 (>0) instead.
> > >
> > > John
> > >
> > >
> > >
> > >
> > > On Tue, Jan 20, 2015 at 10:23 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > Tue Jan 20 10:23:28 2015: Request 70433 was acted upon.
> > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > Queue: met_help
> > > > Subject: output of grid_stat
> > > > Owner: Nobody
> > > > Requestors: geeta124 at hotmail.com
> > > > Status: new
> > > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433
> > >
> > > >
> > > >
> > > > Hi John.
> > > > I had run grid stat on my data which has 25km resolution and
got the
> > > > output.
> > > > As per my header file, I have 37x37 points in the X n Y
directions.
> > Output
> > > > of ncdump -h for both datasets is given below.
> > > > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump
-h
> > > > Obs.024_19june2006.nc
> > > > netcdf Obs.024_19june2006 {
> > > > dimensions:
> > > > lon = 37 ;
> > > > lat = 37 ;
> > > > variables:
> > > > double lon(lon) ;
> > > > lon:units = "degrees_east" ;
> > > > double lat(lat) ;
> > > > lat:units = "degrees_north" ;
> > > > float APCP_24(lat, lon) ;
> > > > APCP_24:units = "kg/m^2" ;
> > > > APCP_24:missing_value = -9999.f ;
> > > > APCP_24:long_name = "Total precipitation" ;
> > > > APCP_24:name = "APCP" ;
> > > > APCP_24:level = "A24" ;
> > > > APCP_24:grib_code = 61.f ;
> > > > APCP_24:_FillValue = -9999.f ;
> > > > APCP_24:init_time = "20060619_000000" ;
> > > > APCP_24:init_time_ut = 1150675200 ;
> > > > APCP_24:valid_time = "20060620_000000" ;
> > > > APCP_24:valid_time_ut = 1150761600 ;
> > > > APCP_24:accum_time = "240000" ;
> > > > APCP_24:accum_time_sec = 86400 ;
> > > >
> > > > // global attributes:
> > > > :FileOrigins = "File Obs.024_19june2006.nc generated
> > > > 20150113_221613 on host agniilap by the Rscript trmmbin2nc.R"
;
> > > > :MET_version = "V4.1" ;
> > > > :Projection = "LatLon" ;
> > > > :lat_ll = "9 degrees_north" ;
> > > > :lon_ll = "74 degrees_east" ;
> > > > :delta_lat = "0.25 degrees" ;
> > > > :delta_lon = "0.25 degrees" ;
> > > > :Nlat = "37 grid_points" ;
> > > > :Nlon = "37 grid_points" ;
> > > > }
> > > > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$
> > > >
> > > > FCST FILE:;;;;;;;;;;;;;;;;;;;;;;;;;;
> > > >
> > > > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$ ncdump
-h
> > > > WRFPRS_d03.024.pcp.nc
> > > > netcdf WRFPRS_d03.024.pcp {
> > > > dimensions:
> > > > lat = 37 ;
> > > > lon = 37 ;
> > > > variables:
> > > > float lat(lat, lon) ;
> > > > lat:long_name = "latitude" ;
> > > > lat:units = "degrees_north" ;
> > > > lat:standard_name = "latitude" ;
> > > > float lon(lat, lon) ;
> > > > lon:long_name = "longitude" ;
> > > > lon:units = "degrees_east" ;
> > > > lon:standard_name = "longitude" ;
> > > > float APCP_24(lat, lon) ;
> > > > APCP_24:name = "APCP" ;
> > > > APCP_24:long_name = "Total precipitation" ;
> > > > APCP_24:level = "A24" ;
> > > > APCP_24:units = "kg/m^2" ;
> > > > APCP_24:grib_code = 61 ;
> > > > APCP_24:_FillValue = -9999.f ;
> > > > APCP_24:init_time = "20060619_000000" ;
> > > > APCP_24:init_time_ut = 1150675200 ;
> > > > APCP_24:valid_time = "20060620_030000" ;
> > > > APCP_24:valid_time_ut = 1150772400 ;
> > > > APCP_24:accum_time = "240000" ;
> > > > APCP_24:accum_time_sec = 86400 ;
> > > >
> > > > // global attributes:
> > > > :FileOrigins = "File WRFPRS_d03.024.pcp.nc generated
> > > > 20150113_113544 UTC on host rmcdlh by the MET pcp_combine
tool" ;
> > > > :MET_version = "V3.0" ;
> > > > :MET_tool = "pcp_combine" ;
> > > > :RunCommand = "Subtraction: WRFPRS_d03.027.grb25 with
> > accumulation
> > > > of 270000 minus WRFPRS_d03.003.grb25 with accumulation of
030000." ;
> > > > :Projection = "LatLon" ;
> > > > :lat_ll = "9.000000 degrees_north" ;
> > > > :lon_ll = "74.000000 degrees_east" ;
> > > > :delta_lat = "0.250000 degrees" ;
> > > > :delta_lon = "0.250000 degrees" ;
> > > > :Nlat = "37 grid_points" ;
> > > > :Nlon = "37 grid_points" ;
> > > >
> > > > Now when I had a look at my output,
> > > > (say
> > > > Why do I get 992 MPR's (output attatched).
> > > > There should be 37*37 points=1369.
> > > > 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 N_CAT F1_O1 F1_O2 F1_O3 F2_O1 F2_O2
F2_O3 F3_O1
> > F3_O2
> > > > F3_O3
> > > > 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,>=5.000
>=0.000,>=5.000
> > NA
> > > > NA MCTC 992 3 0 0 0 0 433 204
0
> > 111
> > > > 244
> > > >
> > > > 2. where in the Config file we specify the MCTC threholds
exclusively??
> > > > Attatching the config file I used for your reference.
> > > > I saw following parts in config file specifying that
> > > > cst = {
> > > > wind_thresh = [ NA ];
> > > >
> > > > field = [
> > > > {
> > > > name = "APCP_24";
> > > > level = [ "(*,*)" ];
> > > > cat_thresh = [ >=0.0, >=5.0 ];
> > > > }
> > > > ];
> > > >
> > > > };
> > > > obs = {
> > > > wind_thresh = [ NA ];
> > > >
> > > > field = [
> > > > {
> > > > name = "APCP_24";
> > > > level = [ "(*,*)" ];
> > > > cat_thresh = [ >=0.0, >=5.0 ];
> > > > }
> > > > Now since 2 thresholds are specified, I should have 3x3
contingency
> > tables.
> > > > >=0.0
> > > > 0.0-4.9
> > > > >=5.0
> > > > For finite values of RF, how is 3x3 Table defined in this
case????
> > > >
> > > > 3. Query reg FHO output.
> > > > in case of threhold >=0.0 , How the F_RATE, OBS_RATE will all
be =1.
> > > >
> > > > Kindly explain.
> > > > Geeta
> > > >
> > > >
> > >
> >
> >
>
------------------------------------------------
Subject: output of grid_stat
From: John Halley Gotway
Time: Fri Jan 30 10:37:16 2015
Geeta,
The "interp" section and the "nbrd" section of the configuration file
do
similar things, but they are slightly different.
Changing the "nbrhd" section will affect the output values in the
NBRCTC,
NBRCTS, and NBRCNT lines.
Changing the "interp" section will affect the output values in all the
other line types (such as CTC, CTS, SL1L2, CNT, and MPR).
If you remember my previous description, the "nbrhd" section is
applies as
follows...
For each field, apply each of the categorical thresholds (cat_thresh)
to
convert it to a 0/1 binary field. Then apply each neighborhood width
(nbrhd.width) and compute statistics like the fractions skill score
(FSS).
And the "interp" section is applied as follows...
For each field, apply the smoothing options (interp.type) to the
fields
specified (interp.field). Then compute all the traditional statistics
output type (CNT, CTS, and so on).
Thanks,
John
On Thu, Jan 29, 2015 at 11:04 PM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433 >
>
> John.
> I was going through the GRID_STAT config file.
> Since I m discussing the smoothing/upscaling aspect of grid_stat. I
came
> across two things in this file---
> 1. // Interpolation methods
> //
> interp = {
> field = BOTH;
> vld_thresh = 1.0;
>
> type = [
> {
> method = UW_MEAN;
> width = 1;
> }
> ];
> };
> What you have suggested in your earlier email is that I should
smooth the
> FCST field by specifying as FCST (ie. field = FCST;) under this
HEADER
> and putting width =3,5 or 7..... etc and then check for the
variation in
> the traditional Scores, how they vary with smoothed forecast.
> 2. NOW, We also have the following HEADER called as the
NEIGHBOURHOOD
> methods. In this also R nt we are doing the same thing????
> //
> // Neighborhood methods
> //
> nbrhd = {
> vld_thresh = 1.0;
> width = [ 3 ];
> cov_thresh = [ >=0.5 ];
> }
> In this part also, we are varying the width as 1, 3, 5 n so on.
> Can U pls explain the difference between the two things specified in
the
> CONFIG file???.
> If I change any of the above parameters, THE FSS will change!!!!.
>
> Geeta
> From: geeta124 at hotmail.com
> To: met_help at ucar.edu
> Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
> Date: Fri, 30 Jan 2015 11:17:13 +0530
>
>
>
>
> Hi John.
> One further query to this aspect. When You say interp option in Grid
stat,
> Do U mean the interpolation at the copygb Level???.
> When we run copyGb on the model data (BEFORE GRID STAT), it
transforms it
> from 9/5km grid to 25 km grid for it to be consistent with the
observed
> data.
> After this step, do u mean there is another round of interpolation
which
> you have explained below???.
> If that is the case, what is the difference between running
GRID_STAT and
> upscaling??? Kindly clarify this.
>
> Also pls clarify my doubt reg the ncview(MPR plots) and the error
that I
> got for the TRMM acumulated data.
> I shall be waiting for your response.
> Geeta
>
> > Subject: Re: [rt.rap.ucar.edu #70433] output of grid_stat
> > From: met_help at ucar.edu
> > To: geeta124 at hotmail.com
> > Date: Wed, 28 Jan 2015 14:14:56 -0700
> >
> > Geeta,
> >
> > You're correct. In Grid-Stat, the forecast and observation data
must
> > already be on the same grid. In this context, the "interpolation"
option
> > is really more of a "smoothing" option. We only call it "interp"
in the
> > configuration file to be consistent with the same option in the
> Point-Stat
> > configuration file. And logically speaking, the same steps are
performed
> > in Grid-Stat and Point-Stat for this option.
> >
> > Here's how it works...
> > - The default interpolation option in Grid-Stat is unweighted mean
of the
> > nearest point (UW_MEAN of 1). That basically means do no
interpolation.
> > - Suppose you change the interpolation width to 5. For each grid
point,
> > it'll look at the 5x5 box around it and take the average of those
25
> > points. If FCST or BOTH is selected, it'll do the smoothing in
the
> > forecast field. If OBS or BOTH is selected, it'll do the
smoothing in
> the
> > observation field.
> > - Choose one or more widths to vary the amount of smoothing.
> > - Choose one or more methods to vary the type summary (min, max,
> unweighted
> > mean, and median).
> >
> > Then output statistics are computed using the smoothed data rather
than
> the
> > raw data.
> >
> > Why would you want to do this? It is well-known that when using
> > traditional statistics (like RMSE), smoother forecasts score
better the
> > higher resolution ones. Giving users the option to smooth their
data
> > enables them to quantify how much the traditional scores would be
with a
> > smoother version of their forecast.
> >
> > This answers similar questions that fractions skill score does,
but
> without
> > defining a categorical threshold first.
> >
> > If you want to try this, I'd suggest applying it only to the
forecast
> > field. Keep the observations fixed, and smooth the forecast data
using
> > widths of 1, 3, 5, 7, and 9. Then look to see how your scores
change as
> > your increase the amount of smoothing.
> >
> > Hope that helps clarify.
> >
> > John
> >
> >
> > On Wed, Jan 28, 2015 at 4:19 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > wrote:
> >
> > >
> > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433 >
> > >
> > > Hi John.
> > > I want to know about the interp group in the GRID stat config
file.
> > > It says, the FCST and OBS are interpolated if BOTH is specified.
> > > Now after both datasets on the same grid and same resolution,
what is
> > > interpolated???
> > > Geeta
> > >
> > > From: geeta124 at hotmail.com
> > > To: met_help at ucar.edu
> > > Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
> > > Date: Wed, 28 Jan 2015 15:45:36 +0530
> > >
> > >
> > >
> > >
> > > Hi John.
> > > So I could save the data in the ps files. (attached for Ur
reference).
> > > But need your help in interpretation.
> > > Also the LAT/LON are not as per Indian Domain.
> > > it shows -8................
> > > Kindly clarify.
> > >
> > > From: geeta124 at hotmail.com
> > > To: met_help at ucar.edu
> > > Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
> > > Date: Wed, 28 Jan 2015 15:36:48 +0530
> > >
> > >
> > >
> > >
> > > Hi John.
> > > The scale in case of FCST was 0-50 while that of 0-100. the
scale of
> DIFF
> > > was -50 to 50.
> > > Could not interpret the spatial plots. and I could not save this
output
> > > into any of the image formats.
> > > Geeta
> > >
> > > From: geeta124 at hotmail.com
> > > To: met_help at ucar.edu
> > > Subject: RE: [rt.rap.ucar.edu #70433] output of grid_stat
> > > Date: Wed, 28 Jan 2015 15:31:37 +0530
> > >
> > >
> > >
> > >
> > > Hi John.
> > > I viewed my output using the ncview and saw 3 variables.
> > > FCST_APCP_A24_FULL, OBS_APCP_A24_FULL and DIFF_APCP_A24_FULL.
> > > It shows a spatial plot with a scale.
> > >
> > > I could not seee the MPRS. Want to know if I can count the bad
ones
> where
> > > there was no data.
> > > How can I recognise those points where there was no data.
> > > Geeta
> > >
> > > > Subject: Re: [rt.rap.ucar.edu #70433] output of grid_stat
> > > > From: met_help at ucar.edu
> > > > To: geeta124 at hotmail.com
> > > > Date: Tue, 20 Jan 2015 14:21:03 -0700
> > > >
> > > > Geeta,
> > > >
> > > > 1. Since the dimension of your grid is 37x37, you could get up
to
> 1369
> > > > matched pairs. However, if the forecast or observation
contains bad
> data
> > > > at a point, that matched pair is not included in the
calculations.
> Using
> > > > the ncview tool to look at the
> > > grid_stat_270000L_20060620_030000V_pairs.nc
> > > > NetCDF file you sent, I see that the forecast field contains
missing
> data
> > > > around the edge of the domain. That's the reason you're
getting 992
> > > > matched pairs instead of 1369.
> > > >
> > > > 2. Yes, specifying 2 thresholds (>=0.0 and >=5.0) will result
in a
> 3x3
> > > > contingency table. The MCTC counts for that 3x3 table are
listed on
> > > line 8
> > > > of the grid_stat_270000L_20060620_030000V.stat file you sent.
The
> MCTS
> > > > statistics corresponding to that 3x3 table are listed on line
9.
> The 3
> > > > categories are:
> > > > 1: less than 0
> > > > 2: 0 to 5
> > > > 3: greater than 5
> > > >
> > > > 3. You're using grid_stat to evaluate precipitation, whose
minimum
> value
> > > > should be 0. If you're thresholding the data
> greater-than-or-equal-to 0
> > > > (>= 0), that will always evaluate to true for precipitation.
I'd
> suggest
> > > > using strictly greater-than 0 (>0) instead.
> > > >
> > > > John
> > > >
> > > >
> > > >
> > > >
> > > > On Tue, Jan 20, 2015 at 10:23 AM, Geeta Geeta via RT <
> met_help at ucar.edu>
> > > > wrote:
> > > >
> > > > >
> > > > > Tue Jan 20 10:23:28 2015: Request 70433 was acted upon.
> > > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > > Queue: met_help
> > > > > Subject: output of grid_stat
> > > > > Owner: Nobody
> > > > > Requestors: geeta124 at hotmail.com
> > > > > Status: new
> > > > > Ticket <URL:
> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70433
> > > >
> > > > >
> > > > >
> > > > > Hi John.
> > > > > I had run grid stat on my data which has 25km resolution and
got
> the
> > > > > output.
> > > > > As per my header file, I have 37x37 points in the X n Y
directions.
> > > Output
> > > > > of ncdump -h for both datasets is given below.
> > > > > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$
ncdump -h
> > > > > Obs.024_19june2006.nc
> > > > > netcdf Obs.024_19june2006 {
> > > > > dimensions:
> > > > > lon = 37 ;
> > > > > lat = 37 ;
> > > > > variables:
> > > > > double lon(lon) ;
> > > > > lon:units = "degrees_east" ;
> > > > > double lat(lat) ;
> > > > > lat:units = "degrees_north" ;
> > > > > float APCP_24(lat, lon) ;
> > > > > APCP_24:units = "kg/m^2" ;
> > > > > APCP_24:missing_value = -9999.f ;
> > > > > APCP_24:long_name = "Total precipitation" ;
> > > > > APCP_24:name = "APCP" ;
> > > > > APCP_24:level = "A24" ;
> > > > > APCP_24:grib_code = 61.f ;
> > > > > APCP_24:_FillValue = -9999.f ;
> > > > > APCP_24:init_time = "20060619_000000" ;
> > > > > APCP_24:init_time_ut = 1150675200 ;
> > > > > APCP_24:valid_time = "20060620_000000" ;
> > > > > APCP_24:valid_time_ut = 1150761600 ;
> > > > > APCP_24:accum_time = "240000" ;
> > > > > APCP_24:accum_time_sec = 86400 ;
> > > > >
> > > > > // global attributes:
> > > > > :FileOrigins = "File Obs.024_19june2006.nc generated
> > > > > 20150113_221613 on host agniilap by the Rscript
trmmbin2nc.R" ;
> > > > > :MET_version = "V4.1" ;
> > > > > :Projection = "LatLon" ;
> > > > > :lat_ll = "9 degrees_north" ;
> > > > > :lon_ll = "74 degrees_east" ;
> > > > > :delta_lat = "0.25 degrees" ;
> > > > > :delta_lon = "0.25 degrees" ;
> > > > > :Nlat = "37 grid_points" ;
> > > > > :Nlon = "37 grid_points" ;
> > > > > }
> > > > > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$
> > > > >
> > > > > FCST FILE:;;;;;;;;;;;;;;;;;;;;;;;;;;
> > > > >
> > > > > aditya at agniilap:~/geeta/installs/met/RUN_MET-19jun2006$
ncdump -h
> > > > > WRFPRS_d03.024.pcp.nc
> > > > > netcdf WRFPRS_d03.024.pcp {
> > > > > dimensions:
> > > > > lat = 37 ;
> > > > > lon = 37 ;
> > > > > variables:
> > > > > float lat(lat, lon) ;
> > > > > lat:long_name = "latitude" ;
> > > > > lat:units = "degrees_north" ;
> > > > > lat:standard_name = "latitude" ;
> > > > > float lon(lat, lon) ;
> > > > > lon:long_name = "longitude" ;
> > > > > lon:units = "degrees_east" ;
> > > > > lon:standard_name = "longitude" ;
> > > > > float APCP_24(lat, lon) ;
> > > > > APCP_24:name = "APCP" ;
> > > > > APCP_24:long_name = "Total precipitation" ;
> > > > > APCP_24:level = "A24" ;
> > > > > APCP_24:units = "kg/m^2" ;
> > > > > APCP_24:grib_code = 61 ;
> > > > > APCP_24:_FillValue = -9999.f ;
> > > > > APCP_24:init_time = "20060619_000000" ;
> > > > > APCP_24:init_time_ut = 1150675200 ;
> > > > > APCP_24:valid_time = "20060620_030000" ;
> > > > > APCP_24:valid_time_ut = 1150772400 ;
> > > > > APCP_24:accum_time = "240000" ;
> > > > > APCP_24:accum_time_sec = 86400 ;
> > > > >
> > > > > // global attributes:
> > > > > :FileOrigins = "File WRFPRS_d03.024.pcp.nc generated
> > > > > 20150113_113544 UTC on host rmcdlh by the MET pcp_combine
tool" ;
> > > > > :MET_version = "V3.0" ;
> > > > > :MET_tool = "pcp_combine" ;
> > > > > :RunCommand = "Subtraction: WRFPRS_d03.027.grb25
with
> > > accumulation
> > > > > of 270000 minus WRFPRS_d03.003.grb25 with accumulation of
030000."
> ;
> > > > > :Projection = "LatLon" ;
> > > > > :lat_ll = "9.000000 degrees_north" ;
> > > > > :lon_ll = "74.000000 degrees_east" ;
> > > > > :delta_lat = "0.250000 degrees" ;
> > > > > :delta_lon = "0.250000 degrees" ;
> > > > > :Nlat = "37 grid_points" ;
> > > > > :Nlon = "37 grid_points" ;
> > > > >
> > > > > Now when I had a look at my output,
> > > > > (say
> > > > > Why do I get 992 MPR's (output attatched).
> > > > > There should be 37*37 points=1369.
> > > > > 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 N_CAT F1_O1 F1_O2 F1_O3 F2_O1 F2_O2
F2_O3
> F3_O1
> > > F3_O2
> > > > > F3_O3
> > > > > 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,>=5.000
>=0.000,>=5.000
> > > NA
> > > > > NA MCTC 992 3 0 0 0 0 433
204
> 0
> > > 111
> > > > > 244
> > > > >
> > > > > 2. where in the Config file we specify the MCTC threholds
> exclusively??
> > > > > Attatching the config file I used for your reference.
> > > > > I saw following parts in config file specifying that
> > > > > cst = {
> > > > > wind_thresh = [ NA ];
> > > > >
> > > > > field = [
> > > > > {
> > > > > name = "APCP_24";
> > > > > level = [ "(*,*)" ];
> > > > > cat_thresh = [ >=0.0, >=5.0 ];
> > > > > }
> > > > > ];
> > > > >
> > > > > };
> > > > > obs = {
> > > > > wind_thresh = [ NA ];
> > > > >
> > > > > field = [
> > > > > {
> > > > > name = "APCP_24";
> > > > > level = [ "(*,*)" ];
> > > > > cat_thresh = [ >=0.0, >=5.0 ];
> > > > > }
> > > > > Now since 2 thresholds are specified, I should have 3x3
contingency
> > > tables.
> > > > > >=0.0
> > > > > 0.0-4.9
> > > > > >=5.0
> > > > > For finite values of RF, how is 3x3 Table defined in this
case????
> > > > >
> > > > > 3. Query reg FHO output.
> > > > > in case of threhold >=0.0 , How the F_RATE, OBS_RATE will
all be
> =1.
> > > > >
> > > > > Kindly explain.
> > > > > Geeta
> > > > >
> > > > >
> > > >
> > >
> > >
> >
>
>
------------------------------------------------
More information about the Met_help
mailing list