[Met_help] [rt.rap.ucar.edu #78783] History for pcp_combine data format
John Halley Gotway via RT
met_help at ucar.edu
Wed Dec 7 09:52:51 MST 2016
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hi all,
I try to start MET with pcp_combine, I found wrfoutput can't be directly used. Form the former questions and answers , it is said :
Here's what's required:
- The dimensions should be named lat and lon.
- The lat and lon variables are actually not used by MET, but it's
fine if they're present.
- The variable attributes should contain the "name" and "level"
attributes. In this example "A1" means that this is precip
accumulated over 1 hour.
- The variable attributes should contain timing information:
init_time_ut, valid_time_ut, and accum_time_sec for the initialization
and valid times in unixtime (date +%s) and the accumulation interval
as a number of seconds.
- The global attributes should contain "MET_version" so that MET knows
how to parse the projection information.
- The global attributes for a lat/lon projection are: Projection,
lat_ll, lon_ll, delat_lat, delta_lon, Nlat, and Nlon (pretty self-
explanatory).
Give that a shot and just let me know if you get stuck.
Thanks,
John Halley Gotway
// global attributes:
:FileOrigins = "File ex_latlon.nc generated
20120726_134701 UTC on host rambler by the MET pcp_combine tool" ;
:MET_version = "V4.0" ;
:MET_tool = "pcp_combine" ;
:RunCommand = "Addition: 1 files." ;
:Projection = "LatLon" ;
:lat_ll = "25.000000 degrees_north" ;
:lon_ll = "-130.000000 degrees_east" ;
:delta_lat = "0.500000 degrees" ;
:delta_lon = "0.500000 degrees" ;
:Nlat = "60 grid_points" ;
:Nlon = "120 grid_points" ; }
************************************
So when I use UPP convert wrfoutput to grib2, whether the projection has to be "lat-lon" ?
I found lots of input data in MET is grib format, does that mean my input data has to be grib format ?
If I use the observations which is netcdf and generated by ourselfs , how can I convert it to the grib format?
Thank you very much!
Mandy
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: pcp_combine data format
From: John Halley Gotway
Time: Tue Nov 29 12:40:49 2016
Mandy,
MET can read gridded data in GRIB1, GRIB2, or 3 different flavors of
NetCDF:
- The "internal" NetCDF format that looks like the output of the
pcp_combine tool. That's what you've referenced above.
- CF-compliant NetCDF3 files.
- The output of the wrf_interp utility.
If you have the NetCDF output from WRF, yes, I would suggest using UPP
to
post-process it. It does not need to be "lat-lon"... it can be
post-processed to whatever projection you'd like.
There is in general no easy way to convert NetCDF to GRIB. If you
have
NetCDF data you've generated yourself, I'd suggest either making it
look
like the NetCDF output from pcp_combine, or preferably, make it
CF-compliant.
Hope that helps.
Thanks,
John Halley Gotway
On Tue, Nov 29, 2016 at 10:35 AM, Zhang, Mandy via RT
<met_help at ucar.edu>
wrote:
>
> Tue Nov 29 10:35:54 2016: Request 78783 was acted upon.
> Transaction: Ticket created by Man.Zhang-1 at ou.edu
> Queue: met_help
> Subject: pcp_combine data format
> Owner: Nobody
> Requestors: Man.Zhang-1 at ou.edu
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78783 >
>
>
> Hi all,
>
> I try to start MET with pcp_combine, I found wrfoutput can't be
> directly used. Form the former questions and answers , it is said :
>
> Here's what's required:
>
> - The dimensions should be named lat and lon.
> - The lat and lon variables are actually not used by MET, but it's
> fine if they're present.
> - The variable attributes should contain the "name" and "level"
> attributes. In this example "A1" means that this is precip
> accumulated over 1 hour.
> - The variable attributes should contain timing information:
> init_time_ut, valid_time_ut, and accum_time_sec for the
initialization
> and valid times in unixtime (date +%s) and the accumulation interval
> as a number of seconds.
> - The global attributes should contain "MET_version" so that MET
knows
> how to parse the projection information.
> - The global attributes for a lat/lon projection are: Projection,
> lat_ll, lon_ll, delat_lat, delta_lon, Nlat, and Nlon (pretty self-
> explanatory).
>
> Give that a shot and just let me know if you get stuck.
>
> Thanks,
> John Halley Gotway
>
> // global attributes:
> :FileOrigins = "File ex_latlon.nc generated
> 20120726_134701 UTC on host rambler by the MET pcp_combine tool" ;
> :MET_version = "V4.0" ;
> :MET_tool = "pcp_combine" ;
> :RunCommand = "Addition: 1 files." ;
> :Projection = "LatLon" ;
> :lat_ll = "25.000000 degrees_north" ;
> :lon_ll = "-130.000000 degrees_east" ;
> :delta_lat = "0.500000 degrees" ;
> :delta_lon = "0.500000 degrees" ;
> :Nlat = "60 grid_points" ;
> :Nlon = "120 grid_points" ; }
>
>
>
> ************************************
>
> So when I use UPP convert wrfoutput to grib2, whether the projection
has
> to be "lat-lon" ?
>
> I found lots of input data in MET is grib format, does that mean my
input
> data has to be grib format ?
>
> If I use the observations which is netcdf and generated by ourselfs
, how
> can I convert it to the grib format?
>
> Thank you very much!
>
> Mandy
>
>
------------------------------------------------
Subject: pcp_combine data format
From: Zhang, Mandy
Time: Wed Nov 30 23:25:35 2016
Hi John,
I'm still working with pcp_combine test..
I have WRF output and postprocess with UPP. the data is grib format,
wrfprs_d01.00, wrfprs_d01.01,wrfprs_d01.02...then I do pcp_combine
like :
#~bin/bash
echo
/home/mzhang/tool/met/bin/pcp_combine \
00000000_000000 1 20160427_010000 1 \
pcp_20160427_24hr.nc \
-pcpdir /home/mzhang/output/model/wrfdata/grib/postprd/pcp
******************
it works well!
but if I do
#~bin/bash
echo
/home/mzhang/tool/met/bin/pcp_combine \
00000000_000000 1 20160427_020000 2 \
pcp_20160427_24hr.nc \
-pcpdir /home/mzhang/output/model/wrfdata/grib/postprd/pcp
******
the error follows blow:
ERROR : sum_data_files() -> Cannot find a file with a valid time of
20160427_020000 and accumulation time of 010000 matching the regular
expression ".*"
ERROR :
test_pcp.sh: line 7: /home/mzhang: Is a directory
test_pcp.sh: line 8: /home/mzhang: Is a directory
test_pcp.sh: line 9: /home/mzhang: Is a directory
test_pcp.sh: line 10: /home/mzhang: Is a directory
test_pcp.sh: line 11: /home/mzhang: Is a directory
*****************
the grib data wrfprs_d01.02 header can be show by Cdo -sinfo:
....
31 : generic (ltype=108) : : levels=1
lev : 12750
bounds : 25500-0
32 : top_of_atmosphere : levels=1
toa : 0
Time coordinate : 1 step
RefTime = 2016-04-27 00:00:00 Units = hours Calendar =
proleptic_gregorian
YYYY-MM-DD hh:mm:ss YYYY-MM-DD hh:mm:ss YYYY-MM-DD hh:mm:ss YYYY-
MM-DD hh:mm:ss
2016-04-27 02:00:00
cdo sinfo: Processed 85 variables over 1 timestep ( 1.71s )
**********
And I repeat it many times, the same case occurs each time.
Do you have clues what's wrong with it?
Thank you very much!
Mandy
________________________________
From: John Halley Gotway via RT <met_help at ucar.edu>
Sent: Tuesday, November 29, 2016 1:40:49 PM
To: Zhang, Mandy
Subject: Re: [rt.rap.ucar.edu #78783] pcp_combine data format
Mandy,
MET can read gridded data in GRIB1, GRIB2, or 3 different flavors of
NetCDF:
- The "internal" NetCDF format that looks like the output of the
pcp_combine tool. That's what you've referenced above.
- CF-compliant NetCDF3 files.
- The output of the wrf_interp utility.
If you have the NetCDF output from WRF, yes, I would suggest using UPP
to
post-process it. It does not need to be "lat-lon"... it can be
post-processed to whatever projection you'd like.
There is in general no easy way to convert NetCDF to GRIB. If you
have
NetCDF data you've generated yourself, I'd suggest either making it
look
like the NetCDF output from pcp_combine, or preferably, make it
CF-compliant.
Hope that helps.
Thanks,
John Halley Gotway
On Tue, Nov 29, 2016 at 10:35 AM, Zhang, Mandy via RT
<met_help at ucar.edu>
wrote:
>
> Tue Nov 29 10:35:54 2016: Request 78783 was acted upon.
> Transaction: Ticket created by Man.Zhang-1 at ou.edu
> Queue: met_help
> Subject: pcp_combine data format
> Owner: Nobody
> Requestors: Man.Zhang-1 at ou.edu
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78783 >
>
>
> Hi all,
>
> I try to start MET with pcp_combine, I found wrfoutput can't be
> directly used. Form the former questions and answers , it is said :
>
> Here's what's required:
>
> - The dimensions should be named lat and lon.
> - The lat and lon variables are actually not used by MET, but it's
> fine if they're present.
> - The variable attributes should contain the "name" and "level"
> attributes. In this example "A1" means that this is precip
> accumulated over 1 hour.
> - The variable attributes should contain timing information:
> init_time_ut, valid_time_ut, and accum_time_sec for the
initialization
> and valid times in unixtime (date +%s) and the accumulation interval
> as a number of seconds.
> - The global attributes should contain "MET_version" so that MET
knows
> how to parse the projection information.
> - The global attributes for a lat/lon projection are: Projection,
> lat_ll, lon_ll, delat_lat, delta_lon, Nlat, and Nlon (pretty self-
> explanatory).
>
> Give that a shot and just let me know if you get stuck.
>
> Thanks,
> John Halley Gotway
>
> // global attributes:
> :FileOrigins = "File ex_latlon.nc generated
> 20120726_134701 UTC on host rambler by the MET pcp_combine tool" ;
> :MET_version = "V4.0" ;
> :MET_tool = "pcp_combine" ;
> :RunCommand = "Addition: 1 files." ;
> :Projection = "LatLon" ;
> :lat_ll = "25.000000 degrees_north" ;
> :lon_ll = "-130.000000 degrees_east" ;
> :delta_lat = "0.500000 degrees" ;
> :delta_lon = "0.500000 degrees" ;
> :Nlat = "60 grid_points" ;
> :Nlon = "120 grid_points" ; }
>
>
>
> ************************************
>
> So when I use UPP convert wrfoutput to grib2, whether the projection
has
> to be "lat-lon" ?
>
> I found lots of input data in MET is grib format, does that mean my
input
> data has to be grib format ?
>
> If I use the observations which is netcdf and generated by ourselfs
, how
> can I convert it to the grib format?
>
> Thank you very much!
>
> Mandy
>
>
------------------------------------------------
Subject: pcp_combine data format
From: John Halley Gotway
Time: Thu Dec 01 15:35:15 2016
Mandy,
I'd suggest that you start by looking at the accumulation interval of
the
precipitation data in your WRF output files from UPP. I always use
the
"wgrib" utility to dump out that sort of information:
wgrib wrfprs_d01.02
wgrib wrfprs_d01.03
... and so on
The question is whether your output actually contains 1-hourly
accumulated
precip, or does it contain runtime accumulation. Runtime accumulation
means that the 6-hour wrf file contains 0 to 6 hours of precip; the 7-
hour
file contains 0 to 7 hours of precip; and so on. The precip values
just
accumulate over the course of the entire model integration.
The default for WRF-ARW is runtime accumulation. So if you're running
WRF-ARW and you didn't specifically change the output bucket interval,
then
that's very likely what you have.
If that's the case, it's totally fine. You should just use the
pcp_combine
"-subtract" option instead of "-sum". Suppose you want the 6 hours of
precip between the 6hr and 12hr forecasts, you'd run the following:
pcp_combine -subtract wrfprs_d01.12 12 wrfprs_d01.06 06
apcp_06_to_12.nc
That says... get 12 hours of precip from the first file and 6 hours
from
the second file and subtract them.
Hope that helps.
Thanks,
John
On Wed, Nov 30, 2016 at 11:25 PM, Zhang, Mandy via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78783 >
>
> Hi John,
>
> I'm still working with pcp_combine test..
>
> I have WRF output and postprocess with UPP. the data is grib format,
> wrfprs_d01.00, wrfprs_d01.01,wrfprs_d01.02...then I do pcp_combine
like :
>
> #~bin/bash
> echo
> /home/mzhang/tool/met/bin/pcp_combine \
> 00000000_000000 1 20160427_010000 1 \
> pcp_20160427_24hr.nc \
> -pcpdir /home/mzhang/output/model/wrfdata/grib/postprd/pcp
> ******************
>
> it works well!
>
> but if I do
>
> #~bin/bash
> echo
> /home/mzhang/tool/met/bin/pcp_combine \
> 00000000_000000 1 20160427_020000 2 \
> pcp_20160427_24hr.nc \
> -pcpdir /home/mzhang/output/model/wrfdata/grib/postprd/pcp
> ******
>
> the error follows blow:
>
> ERROR : sum_data_files() -> Cannot find a file with a valid time of
> 20160427_020000 and accumulation time of 010000 matching the regular
> expression ".*"
> ERROR :
> test_pcp.sh: line 7: /home/mzhang: Is a directory
> test_pcp.sh: line 8: /home/mzhang: Is a directory
> test_pcp.sh: line 9: /home/mzhang: Is a directory
> test_pcp.sh: line 10: /home/mzhang: Is a directory
> test_pcp.sh: line 11: /home/mzhang: Is a directory
> *****************
>
> the grib data wrfprs_d01.02 header can be show by Cdo -sinfo:
>
> ....
>
> 31 : generic (ltype=108) : : levels=1
> lev : 12750
> bounds : 25500-0
> 32 : top_of_atmosphere : levels=1
> toa : 0
> Time coordinate : 1 step
> RefTime = 2016-04-27 00:00:00 Units = hours Calendar =
> proleptic_gregorian
> YYYY-MM-DD hh:mm:ss YYYY-MM-DD hh:mm:ss YYYY-MM-DD hh:mm:ss
> YYYY-MM-DD hh:mm:ss
> 2016-04-27 02:00:00
> cdo sinfo: Processed 85 variables over 1 timestep ( 1.71s )
> **********
>
> And I repeat it many times, the same case occurs each time.
>
> Do you have clues what's wrong with it?
>
> Thank you very much!
>
> Mandy
>
>
> ________________________________
> From: John Halley Gotway via RT <met_help at ucar.edu>
> Sent: Tuesday, November 29, 2016 1:40:49 PM
> To: Zhang, Mandy
> Subject: Re: [rt.rap.ucar.edu #78783] pcp_combine data format
>
> Mandy,
>
> MET can read gridded data in GRIB1, GRIB2, or 3 different flavors of
> NetCDF:
> - The "internal" NetCDF format that looks like the output of the
> pcp_combine tool. That's what you've referenced above.
> - CF-compliant NetCDF3 files.
> - The output of the wrf_interp utility.
>
> If you have the NetCDF output from WRF, yes, I would suggest using
UPP to
> post-process it. It does not need to be "lat-lon"... it can be
> post-processed to whatever projection you'd like.
>
> There is in general no easy way to convert NetCDF to GRIB. If you
have
> NetCDF data you've generated yourself, I'd suggest either making it
look
> like the NetCDF output from pcp_combine, or preferably, make it
> CF-compliant.
>
> Hope that helps.
>
> Thanks,
> John Halley Gotway
>
>
>
> On Tue, Nov 29, 2016 at 10:35 AM, Zhang, Mandy via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > Tue Nov 29 10:35:54 2016: Request 78783 was acted upon.
> > Transaction: Ticket created by Man.Zhang-1 at ou.edu
> > Queue: met_help
> > Subject: pcp_combine data format
> > Owner: Nobody
> > Requestors: Man.Zhang-1 at ou.edu
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78783 >
> >
> >
> > Hi all,
> >
> > I try to start MET with pcp_combine, I found wrfoutput can't
be
> > directly used. Form the former questions and answers , it is said
:
> >
> > Here's what's required:
> >
> > - The dimensions should be named lat and lon.
> > - The lat and lon variables are actually not used by MET, but it's
> > fine if they're present.
> > - The variable attributes should contain the "name" and "level"
> > attributes. In this example "A1" means that this is precip
> > accumulated over 1 hour.
> > - The variable attributes should contain timing information:
> > init_time_ut, valid_time_ut, and accum_time_sec for the
initialization
> > and valid times in unixtime (date +%s) and the accumulation
interval
> > as a number of seconds.
> > - The global attributes should contain "MET_version" so that MET
knows
> > how to parse the projection information.
> > - The global attributes for a lat/lon projection are: Projection,
> > lat_ll, lon_ll, delat_lat, delta_lon, Nlat, and Nlon (pretty self-
> > explanatory).
> >
> > Give that a shot and just let me know if you get stuck.
> >
> > Thanks,
> > John Halley Gotway
> >
> > // global attributes:
> > :FileOrigins = "File ex_latlon.nc generated
> > 20120726_134701 UTC on host rambler by the MET pcp_combine tool" ;
> > :MET_version = "V4.0" ;
> > :MET_tool = "pcp_combine" ;
> > :RunCommand = "Addition: 1 files." ;
> > :Projection = "LatLon" ;
> > :lat_ll = "25.000000 degrees_north" ;
> > :lon_ll = "-130.000000 degrees_east" ;
> > :delta_lat = "0.500000 degrees" ;
> > :delta_lon = "0.500000 degrees" ;
> > :Nlat = "60 grid_points" ;
> > :Nlon = "120 grid_points" ; }
> >
> >
> >
> > ************************************
> >
> > So when I use UPP convert wrfoutput to grib2, whether the
projection has
> > to be "lat-lon" ?
> >
> > I found lots of input data in MET is grib format, does that mean
my input
> > data has to be grib format ?
> >
> > If I use the observations which is netcdf and generated by
ourselfs ,
> how
> > can I convert it to the grib format?
> >
> > Thank you very much!
> >
> > Mandy
> >
> >
>
>
>
------------------------------------------------
Subject: pcp_combine data format
From: Zhang, Mandy
Time: Sun Dec 04 20:46:48 2016
Hi John,
Thank you for your attentions. I'm still working with the
PCP_combine to start MET.
Now the input data from WRF forecast is resolved, but I stucked
with observations. From our former emails, the input data with Netcdf
format should follow with the examples, I created the netcdf data, but
the global attribution is not clear. For instant, the input data is
hourly radar oberservation named 2016042701, how to define the
valid_time_ut, init_time_ut, valid time, init_time? Could you please
help me to define it? Also I'm not understand the meaning with X_pin,
Y_pin,lat-pin and lon-pin.
the following is my current process with Netcdf data:
dimensions:
lat = 1152 ;
lon = 1680 ;
variables:
float lat(lat, lon) ;
lat:standard_name = "latitude" ;
lat:units = "degrees_north" ;
lat:long_name = "latitude" ;
float lon(lat, lon) ;
lon:standard_name = "longitude" ;
lon:units = "degrees_east" ;
lon:long_name = "longitude" ;
float pres(lat, lon) ;
pres:accum_time_sec = "3600" ;
pres:accum_time = "100000" ;
pres:valid_time_ut = "1123459200" ;
pres:valid_time = "20160427_000000" ;
pres:init_time_ut = "1123459200" ;
pres:init_time = "20160427_000000" ;
pres:_FillValue = -9999.f ;
pres:units = "kg/m" ;
pres:level = "A12" ;
pres:long_name = "MRMS QPE" ;
// global attributes:
:ny = "1153 grid_points" ;
:nx = "1681" ;
:r_km = "6371.200000" ;
:d_km = "3.0000" ;
:lon_orient = "-97.000000" ;
:y_pin = "0.000000" ;
:x_pin = "0.000000" ;
:lon_pin = "-97.0" ;
:lat_pin = "38.50001" ;
:scale_lat_2 = "30" ;
:scale_lat_1 = "60" ;
:project = "Lambert Conformal" ;
:tile = "NCL generate QPE data" ;
Thank you very much!
Mandy
________________________________
From: John Halley Gotway via RT <met_help at ucar.edu>
Sent: Friday, December 2, 2016 10:17:33 AM
To: Zhang, Mandy
Subject: [rt.rap.ucar.edu #78783] Resolved: pcp_combine data format
According to our records, your request has been resolved. If you have
any
further questions or concerns, please respond to this message.
------------------------------------------------
Subject: pcp_combine data format
From: Zhang, Mandy
Time: Mon Dec 05 14:50:19 2016
Hi John,
I'm confused with the set of Wavelet_stat_tool, it is said the
Haar wavelet 's member value can only be set to 2 in waveletconfig, so
how to get different wavelet components like figure10.2 shown?
Thank you very much!
Mandy
________________________________
From: Zhang, Mandy
Sent: Sunday, December 4, 2016 9:46:44 PM
To: met_help at ucar.edu
Subject: Re: [rt.rap.ucar.edu #78783] Resolved: pcp_combine data
format
Hi John,
Thank you for your attentions. I'm still working with the
PCP_combine to start MET.
Now the input data from WRF forecast is resolved, but I stucked
with observations. From our former emails, the input data with Netcdf
format should follow with the examples, I created the netcdf data, but
the global attribution is not clear. For instant, the input data is
hourly radar oberservation named 2016042701, how to define the
valid_time_ut, init_time_ut, valid time, init_time? Could you please
help me to define it? Also I'm not understand the meaning with X_pin,
Y_pin,lat-pin and lon-pin.
the following is my current process with Netcdf data:
dimensions:
lat = 1152 ;
lon = 1680 ;
variables:
float lat(lat, lon) ;
lat:standard_name = "latitude" ;
lat:units = "degrees_north" ;
lat:long_name = "latitude" ;
float lon(lat, lon) ;
lon:standard_name = "longitude" ;
lon:units = "degrees_east" ;
lon:long_name = "longitude" ;
float pres(lat, lon) ;
pres:accum_time_sec = "3600" ;
pres:accum_time = "100000" ;
pres:valid_time_ut = "1123459200" ;
pres:valid_time = "20160427_000000" ;
pres:init_time_ut = "1123459200" ;
pres:init_time = "20160427_000000" ;
pres:_FillValue = -9999.f ;
pres:units = "kg/m" ;
pres:level = "A12" ;
pres:long_name = "MRMS QPE" ;
// global attributes:
:ny = "1153 grid_points" ;
:nx = "1681" ;
:r_km = "6371.200000" ;
:d_km = "3.0000" ;
:lon_orient = "-97.000000" ;
:y_pin = "0.000000" ;
:x_pin = "0.000000" ;
:lon_pin = "-97.0" ;
:lat_pin = "38.50001" ;
:scale_lat_2 = "30" ;
:scale_lat_1 = "60" ;
:project = "Lambert Conformal" ;
:tile = "NCL generate QPE data" ;
Thank you very much!
Mandy
________________________________
From: John Halley Gotway via RT <met_help at ucar.edu>
Sent: Friday, December 2, 2016 10:17:33 AM
To: Zhang, Mandy
Subject: [rt.rap.ucar.edu #78783] Resolved: pcp_combine data format
According to our records, your request has been resolved. If you have
any
further questions or concerns, please respond to this message.
------------------------------------------------
Subject: pcp_combine data format
From: John Halley Gotway
Time: Tue Dec 06 11:55:19 2016
Mandy,
I think you just need a little clarification on how the wavelets are
used
to decompose the error field in wavelet_stat. You pick a single
wavelet
shape to do the entire decomposition, and the Haar wavelet is the best
default. We included support for other wavelet types and shapes only
because they are options available in the GNU scientific library, to
which
MET links.
The wavelet_stat tool is based on work by a scientist named Barbara
Casati
and, to my knowledge, all of her wavelet decomposition results are
based on
the Haar wavelet.
For more info about the Haar wavelet, I'd refer you to this wikipedia
page:
https://en.wikipedia.org/wiki/Haar_wavelet
But to answer your question more directly about figure 10.2 from the
MET
User's Guide, let's just take a look at some output from the MET test
scripts. After you installed the MET tools, hopefully you ran "make
test". If so, that should have produced some output in the
"met-5.2/out/wavelet_stat" directory. I converted one of the
PostScript
output files from that directory to PDF and have attached it to this
message.
This output was generated using the default, Haar wavelet. Looking at
the
attached file, one page 1 we see the location of the dyadic tile (128
x
128) over which wavelet_stat was run. Page 2 shows the binary
difference
(fcst - obs) field after applying the threshold of >0 to the raw data.
The
following pages show the decomposition of that binary difference field
from
scales 1 to 7. There are 7 scales because 2^7 = 128, which is the
dimension of the tile.
The goal of the wavelet_stat tool is to apply a wavelet decomposition
to
determine the scales at which the forecast error occurs. One really
nice
feature is that the overall mean-squared error (MSE) can be decomposed
across the scales. So the sum of the MSE at each scale = the total
MSE of
the binary difference field.
Looking at our example, 22.67% of the error occurs at the finest grid
scale
(scale 1) and 20.83% occurs at the largest scale (scale 7). The rest
of
the error falls somewhere in between.
So you do not need to change the shape of the wavelet to generate
figure
10.2. All of the output necessary for fig 10.2 is created with a
single
choice of wavelet.
Hope that helps clarify.
Thanks,
John
On Mon, Dec 5, 2016 at 2:50 PM, Zhang, Mandy via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78783 >
>
> Hi John,
>
> I'm confused with the set of Wavelet_stat_tool, it is said the
Haar
> wavelet 's member value can only be set to 2 in waveletconfig, so
how to
> get different wavelet components like figure10.2 shown?
>
> Thank you very much!
>
> Mandy
>
> ________________________________
> From: Zhang, Mandy
> Sent: Sunday, December 4, 2016 9:46:44 PM
> To: met_help at ucar.edu
> Subject: Re: [rt.rap.ucar.edu #78783] Resolved: pcp_combine data
format
>
>
> Hi John,
>
> Thank you for your attentions. I'm still working with the
PCP_combine
> to start MET.
>
> Now the input data from WRF forecast is resolved, but I
stucked with
> observations. From our former emails, the input data with Netcdf
format
> should follow with the examples, I created the netcdf data, but the
global
> attribution is not clear. For instant, the input data is hourly
radar
> oberservation named 2016042701, how to define the valid_time_ut,
> init_time_ut, valid time, init_time? Could you please help me to
define it?
> Also I'm not understand the meaning with X_pin, Y_pin,lat-pin and
lon-pin.
>
> the following is my current process with Netcdf data:
>
>
>
> dimensions:
> lat = 1152 ;
> lon = 1680 ;
> variables:
> float lat(lat, lon) ;
> lat:standard_name = "latitude" ;
> lat:units = "degrees_north" ;
> lat:long_name = "latitude" ;
> float lon(lat, lon) ;
> lon:standard_name = "longitude" ;
> lon:units = "degrees_east" ;
> lon:long_name = "longitude" ;
> float pres(lat, lon) ;
> pres:accum_time_sec = "3600" ;
> pres:accum_time = "100000" ;
> pres:valid_time_ut = "1123459200" ;
> pres:valid_time = "20160427_000000" ;
> pres:init_time_ut = "1123459200" ;
> pres:init_time = "20160427_000000" ;
> pres:_FillValue = -9999.f ;
> pres:units = "kg/m" ;
> pres:level = "A12" ;
> pres:long_name = "MRMS QPE" ;
>
> // global attributes:
> :ny = "1153 grid_points" ;
> :nx = "1681" ;
> :r_km = "6371.200000" ;
> :d_km = "3.0000" ;
> :lon_orient = "-97.000000" ;
> :y_pin = "0.000000" ;
> :x_pin = "0.000000" ;
> :lon_pin = "-97.0" ;
> :lat_pin = "38.50001" ;
> :scale_lat_2 = "30" ;
> :scale_lat_1 = "60" ;
> :project = "Lambert Conformal" ;
> :tile = "NCL generate QPE data" ;
> Thank you very much!
> Mandy
>
>
>
> ________________________________
> From: John Halley Gotway via RT <met_help at ucar.edu>
> Sent: Friday, December 2, 2016 10:17:33 AM
> To: Zhang, Mandy
> Subject: [rt.rap.ucar.edu #78783] Resolved: pcp_combine data format
>
> According to our records, your request has been resolved. If you
have any
> further questions or concerns, please respond to this message.
>
>
------------------------------------------------
Subject: pcp_combine data format
From: Zhang, Mandy
Time: Wed Dec 07 08:27:14 2016
Hi John,
Thank You very much!
Mandy
________________________________
From: John Halley Gotway via RT <met_help at ucar.edu>
Sent: Tuesday, December 6, 2016 12:55:20 PM
To: Zhang, Mandy
Subject: Re: [rt.rap.ucar.edu #78783] Resolved: pcp_combine data
format
Mandy,
I think you just need a little clarification on how the wavelets are
used
to decompose the error field in wavelet_stat. You pick a single
wavelet
shape to do the entire decomposition, and the Haar wavelet is the best
default. We included support for other wavelet types and shapes only
because they are options available in the GNU scientific library, to
which
MET links.
The wavelet_stat tool is based on work by a scientist named Barbara
Casati
and, to my knowledge, all of her wavelet decomposition results are
based on
the Haar wavelet.
For more info about the Haar wavelet, I'd refer you to this wikipedia
page:
https://en.wikipedia.org/wiki/Haar_wavelet
But to answer your question more directly about figure 10.2 from the
MET
User's Guide, let's just take a look at some output from the MET test
scripts. After you installed the MET tools, hopefully you ran "make
test". If so, that should have produced some output in the
"met-5.2/out/wavelet_stat" directory. I converted one of the
PostScript
output files from that directory to PDF and have attached it to this
message.
This output was generated using the default, Haar wavelet. Looking at
the
attached file, one page 1 we see the location of the dyadic tile (128
x
128) over which wavelet_stat was run. Page 2 shows the binary
difference
(fcst - obs) field after applying the threshold of >0 to the raw data.
The
following pages show the decomposition of that binary difference field
from
scales 1 to 7. There are 7 scales because 2^7 = 128, which is the
dimension of the tile.
The goal of the wavelet_stat tool is to apply a wavelet decomposition
to
determine the scales at which the forecast error occurs. One really
nice
feature is that the overall mean-squared error (MSE) can be decomposed
across the scales. So the sum of the MSE at each scale = the total
MSE of
the binary difference field.
Looking at our example, 22.67% of the error occurs at the finest grid
scale
(scale 1) and 20.83% occurs at the largest scale (scale 7). The rest
of
the error falls somewhere in between.
So you do not need to change the shape of the wavelet to generate
figure
10.2. All of the output necessary for fig 10.2 is created with a
single
choice of wavelet.
Hope that helps clarify.
Thanks,
John
On Mon, Dec 5, 2016 at 2:50 PM, Zhang, Mandy via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78783 >
>
> Hi John,
>
> I'm confused with the set of Wavelet_stat_tool, it is said the
Haar
> wavelet 's member value can only be set to 2 in waveletconfig, so
how to
> get different wavelet components like figure10.2 shown?
>
> Thank you very much!
>
> Mandy
>
> ________________________________
> From: Zhang, Mandy
> Sent: Sunday, December 4, 2016 9:46:44 PM
> To: met_help at ucar.edu
> Subject: Re: [rt.rap.ucar.edu #78783] Resolved: pcp_combine data
format
>
>
> Hi John,
>
> Thank you for your attentions. I'm still working with the
PCP_combine
> to start MET.
>
> Now the input data from WRF forecast is resolved, but I
stucked with
> observations. From our former emails, the input data with Netcdf
format
> should follow with the examples, I created the netcdf data, but the
global
> attribution is not clear. For instant, the input data is hourly
radar
> oberservation named 2016042701, how to define the valid_time_ut,
> init_time_ut, valid time, init_time? Could you please help me to
define it?
> Also I'm not understand the meaning with X_pin, Y_pin,lat-pin and
lon-pin.
>
> the following is my current process with Netcdf data:
>
>
>
> dimensions:
> lat = 1152 ;
> lon = 1680 ;
> variables:
> float lat(lat, lon) ;
> lat:standard_name = "latitude" ;
> lat:units = "degrees_north" ;
> lat:long_name = "latitude" ;
> float lon(lat, lon) ;
> lon:standard_name = "longitude" ;
> lon:units = "degrees_east" ;
> lon:long_name = "longitude" ;
> float pres(lat, lon) ;
> pres:accum_time_sec = "3600" ;
> pres:accum_time = "100000" ;
> pres:valid_time_ut = "1123459200" ;
> pres:valid_time = "20160427_000000" ;
> pres:init_time_ut = "1123459200" ;
> pres:init_time = "20160427_000000" ;
> pres:_FillValue = -9999.f ;
> pres:units = "kg/m" ;
> pres:level = "A12" ;
> pres:long_name = "MRMS QPE" ;
>
> // global attributes:
> :ny = "1153 grid_points" ;
> :nx = "1681" ;
> :r_km = "6371.200000" ;
> :d_km = "3.0000" ;
> :lon_orient = "-97.000000" ;
> :y_pin = "0.000000" ;
> :x_pin = "0.000000" ;
> :lon_pin = "-97.0" ;
> :lat_pin = "38.50001" ;
> :scale_lat_2 = "30" ;
> :scale_lat_1 = "60" ;
> :project = "Lambert Conformal" ;
> :tile = "NCL generate QPE data" ;
> Thank you very much!
> Mandy
>
>
>
> ________________________________
> From: John Halley Gotway via RT <met_help at ucar.edu>
> Sent: Friday, December 2, 2016 10:17:33 AM
> To: Zhang, Mandy
> Subject: [rt.rap.ucar.edu #78783] Resolved: pcp_combine data format
>
> According to our records, your request has been resolved. If you
have any
> further questions or concerns, please respond to this message.
>
>
------------------------------------------------
More information about the Met_help
mailing list