[Met_help] [rt.rap.ucar.edu #63192] History for ascii2nc using 24 hour apcp
John Halley Gotway via RT
met_help at ucar.edu
Fri Sep 27 09:53:09 MDT 2013
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Good Morning MET helpdesk
I am trying to use point-stat to compare 24hr apcp from WRF to my observed data.
I used ascii2nc to create my observed netcdf file. The observed data I entered had already been converted to 24hr totals.
When I ncdump -v obs_arr observed.nc, I think it is still showing the data as 1hr data (3600s).
obs_arr =
0, 61, 3600, -9999, 0,
1, 61, 3600, -9999, 0,
2, 61, 3600, -9999, 0,
3, 61, 3600, -9999, 0,
4, 61, 3600, -9999, 0,
Sooooo, when I run point-stat. I get this error message.
NetCDF: Attribute not found
My questions are:
1. How do I create 24hr observed netcdf file? (if this is in fact the reason for my error)
2. In creating my observed file, I entered multiple stations. Does MET handle this?
I am asking because the manual says it only handles 1 point station, yet ascii2nc ran with no problems and all the data is written to the netcdf.
FYI, this is the process following on from a previous inquiry.
[rt.rap.ucar.edu #62959] POINT_STAT Error APCP
I have placed my data on ftp.rap.ucar.edu fyr
/incoming/irap/met_help/weston_data/
Thanks guys!
Michael Weston
EScience Associates (Pty) Ltd
E-mail: michael at escience.co.za
Web: www.escience.co.za
PO Box 2950, Saxonwold, 2132
9 Victoria Street, Oaklands, Johannesburg, 2192
Tel: +27 (0)11 718 6380
Fax: 086 527 5241
VAT No: 473 025 4416
Reg No: 2009/014472/07
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #63192] ascii2nc using 24 hour apcp
From: John Halley Gotway
Time: Thu Sep 26 10:15:17 2013
Michael,
Thanks for sending your data. I grabbed it from the ftp site and ran
it. There's several things going on here...
(1) I tried to figure out what version of MET you're using...
ncdump -h pcp_subtract.nc | grep MET_version
:MET_version = "V4.0" ;
ncdump -h lieb_Oct2000_Apr2000_24hour.nc | grep MET_version
:MET_version = "V3.1" ;
grep version PointStatConfig_24hour
version = "V4.0";
So it looks like you're using a mixture of METv3.1 and METv4.0. When
I run METv3.1/bin/point_stat, I get the error you're seeing (NetCDF:
Attribute not found). That's because METv3.1 doesn't know
about the format of the data generated by later versions of MET. When
I run METv4.0/bin/point_stat, I do not get that error. While we
attempt to make things backwards compatible when possible, it's
obviously preferable to use data that's all been generated by the same
version of the code.
(2) When running METv4.0/bin/point_stat, I do get other errors. In
the Point-Stat config file, you need to adjust the "fcst" setting as
follows....
field = [
{
name = "APCP_24";
level = "(*,*)";
cat_thresh = [ >0, >1, >5, >10, >25 ];
}
The reason for this is that you're passing a NetCDF file to Point-
Stat. When using gridded NetCDF files, you set "name" to the NetCDF
variable name to be used. And you set "level" to indicate which
dimensions contain the gridded data. For that output of pcp_combine,
there are only 2 dimensions, so "(*,*)" is what you use.
(3) Even after I got Point-Stat to run, I got 0 matched pairs.
Running at debug level 3 (-v 3), I see the following information:
DEBUG 2: Processing APCP_24(*,*) versus APCP/A24, for observation type
ADPSFC, over region FULL, for interpolation method UW_MEAN(1), using 0
pairs.
DEBUG 3: Number of matched pairs = 0
DEBUG 3: Observations processed = 7955
DEBUG 3: Rejected: GRIB code = 0
DEBUG 3: Rejected: valid time = 7918
DEBUG 3: Rejected: bad obs value = 0
DEBUG 3: Rejected: off the grid = 0
DEBUG 3: Rejected: level mismatch = 37
DEBUG 3: Rejected: message type = 0
DEBUG 3: Rejected: masking region = 0
DEBUG 3: Rejected: bad fcst value = 0
So most of the observations are rejected due to the valid time and
some are a "level mismatch". Since you're verifying accumulated
precip, that means that the observation accumulation interval
doesn't match what you've requested in the config file. You requested
24-hour accumulation intervals, but your observation data contains
hourly intervals (3600 seconds). So I modified the "obs"
section of the config file like this:
field = [
{
name = "APCP";
level = "A01";
cat_thresh = [ >0, >1, >5, >10, >25 ];
}
];
Then when I reran, I got 37 matched pairs. You should figure out if
your point observations really do contain 24-hour accumulations, and
if so, modify the accumulation interval in the input to
ASCII2NC to indicate that: 24-hours = 86400 seconds.
Hope this helps get you going. Just let me know if more issues or
questions arise.
Thanks,
John
On 09/26/2013 04:14 AM, Michael Weston via RT wrote:
>
> Thu Sep 26 04:14:24 2013: Request 63192 was acted upon.
> Transaction: Ticket created by michael at escience.co.za
> Queue: met_help
> Subject: ascii2nc using 24 hour apcp
> Owner: Nobody
> Requestors: michael at escience.co.za
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63192 >
>
>
> Good Morning MET helpdesk
>
> I am trying to use point-stat to compare 24hr apcp from WRF to my
observed data.
>
> I used ascii2nc to create my observed netcdf file. The observed data
I entered had already been converted to 24hr totals.
> When I ncdump -v obs_arr observed.nc, I think it is still showing
the data as 1hr data (3600s).
>
> obs_arr =
> 0, 61, 3600, -9999, 0,
> 1, 61, 3600, -9999, 0,
> 2, 61, 3600, -9999, 0,
> 3, 61, 3600, -9999, 0,
> 4, 61, 3600, -9999, 0,
>
> Sooooo, when I run point-stat. I get this error message.
> NetCDF: Attribute not found
>
> My questions are:
> 1. How do I create 24hr observed netcdf file? (if this is in fact
the reason for my error)
> 2. In creating my observed file, I entered multiple stations. Does
MET handle this?
> I am asking because the manual says it only handles 1 point station,
yet ascii2nc ran with no problems and all the data is written to the
netcdf.
>
> FYI, this is the process following on from a previous inquiry.
> [rt.rap.ucar.edu #62959] POINT_STAT Error APCP
>
> I have placed my data on ftp.rap.ucar.edu fyr
> /incoming/irap/met_help/weston_data/
>
> Thanks guys!
> Michael Weston
>
> EScience Associates (Pty) Ltd
> E-mail: michael at escience.co.za
> Web: www.escience.co.za
> PO Box 2950, Saxonwold, 2132
> 9 Victoria Street, Oaklands, Johannesburg, 2192
> Tel: +27 (0)11 718 6380
> Fax: 086 527 5241
> VAT No: 473 025 4416
> Reg No: 2009/014472/07
>
------------------------------------------------
Subject: RE: [rt.rap.ucar.edu #63192] ascii2nc using 24 hour apcp
From: Michael Weston
Time: Thu Sep 26 10:42:30 2013
Hi John
Thanks for the feedback.
1.) Re: different versions.
I am working on several machines. Some have 4.0, some don't.
I will try use 3.1.
2.) I will adjust my Config file accordingly
3.) I have uploaded my text file used as input for ascii2nc. Could you
please check it for any obvious errors?
If I get the conversion to netcdf right with ascii2nc, it should solve
the matched pairs issue.
Thanks
Mike
EScience Associates (Pty) Ltd
E-mail: michael at escience.co.za
Web: www.escience.co.za
PO Box 2950, Saxonwold, 2132
9 Victoria Street, Oaklands, Johannesburg, 2192
Tel: +27 (0)11 718 6380
Fax: 086 527 5241
VAT No: 473 025 4416
Reg No: 2009/014472/07
________________________________________
From: John Halley Gotway via RT [met_help at ucar.edu]
Sent: Thursday, September 26, 2013 6:15 PM
To: Michael Weston
Subject: Re: [rt.rap.ucar.edu #63192] ascii2nc using 24 hour apcp
Michael,
Thanks for sending your data. I grabbed it from the ftp site and ran
it. There's several things going on here...
(1) I tried to figure out what version of MET you're using...
ncdump -h pcp_subtract.nc | grep MET_version
:MET_version = "V4.0" ;
ncdump -h lieb_Oct2000_Apr2000_24hour.nc | grep MET_version
:MET_version = "V3.1" ;
grep version PointStatConfig_24hour
version = "V4.0";
So it looks like you're using a mixture of METv3.1 and METv4.0. When
I run METv3.1/bin/point_stat, I get the error you're seeing (NetCDF:
Attribute not found). That's because METv3.1 doesn't know
about the format of the data generated by later versions of MET. When
I run METv4.0/bin/point_stat, I do not get that error. While we
attempt to make things backwards compatible when possible, it's
obviously preferable to use data that's all been generated by the same
version of the code.
(2) When running METv4.0/bin/point_stat, I do get other errors. In
the Point-Stat config file, you need to adjust the "fcst" setting as
follows....
field = [
{
name = "APCP_24";
level = "(*,*)";
cat_thresh = [ >0, >1, >5, >10, >25 ];
}
The reason for this is that you're passing a NetCDF file to Point-
Stat. When using gridded NetCDF files, you set "name" to the NetCDF
variable name to be used. And you set "level" to indicate which
dimensions contain the gridded data. For that output of pcp_combine,
there are only 2 dimensions, so "(*,*)" is what you use.
(3) Even after I got Point-Stat to run, I got 0 matched pairs.
Running at debug level 3 (-v 3), I see the following information:
DEBUG 2: Processing APCP_24(*,*) versus APCP/A24, for observation type
ADPSFC, over region FULL, for interpolation method UW_MEAN(1), using 0
pairs.
DEBUG 3: Number of matched pairs = 0
DEBUG 3: Observations processed = 7955
DEBUG 3: Rejected: GRIB code = 0
DEBUG 3: Rejected: valid time = 7918
DEBUG 3: Rejected: bad obs value = 0
DEBUG 3: Rejected: off the grid = 0
DEBUG 3: Rejected: level mismatch = 37
DEBUG 3: Rejected: message type = 0
DEBUG 3: Rejected: masking region = 0
DEBUG 3: Rejected: bad fcst value = 0
So most of the observations are rejected due to the valid time and
some are a "level mismatch". Since you're verifying accumulated
precip, that means that the observation accumulation interval
doesn't match what you've requested in the config file. You requested
24-hour accumulation intervals, but your observation data contains
hourly intervals (3600 seconds). So I modified the "obs"
section of the config file like this:
field = [
{
name = "APCP";
level = "A01";
cat_thresh = [ >0, >1, >5, >10, >25 ];
}
];
Then when I reran, I got 37 matched pairs. You should figure out if
your point observations really do contain 24-hour accumulations, and
if so, modify the accumulation interval in the input to
ASCII2NC to indicate that: 24-hours = 86400 seconds.
Hope this helps get you going. Just let me know if more issues or
questions arise.
Thanks,
John
On 09/26/2013 04:14 AM, Michael Weston via RT wrote:
>
> Thu Sep 26 04:14:24 2013: Request 63192 was acted upon.
> Transaction: Ticket created by michael at escience.co.za
> Queue: met_help
> Subject: ascii2nc using 24 hour apcp
> Owner: Nobody
> Requestors: michael at escience.co.za
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63192 >
>
>
> Good Morning MET helpdesk
>
> I am trying to use point-stat to compare 24hr apcp from WRF to my
observed data.
>
> I used ascii2nc to create my observed netcdf file. The observed data
I entered had already been converted to 24hr totals.
> When I ncdump -v obs_arr observed.nc, I think it is still showing
the data as 1hr data (3600s).
>
> obs_arr =
> 0, 61, 3600, -9999, 0,
> 1, 61, 3600, -9999, 0,
> 2, 61, 3600, -9999, 0,
> 3, 61, 3600, -9999, 0,
> 4, 61, 3600, -9999, 0,
>
> Sooooo, when I run point-stat. I get this error message.
> NetCDF: Attribute not found
>
> My questions are:
> 1. How do I create 24hr observed netcdf file? (if this is in fact
the reason for my error)
> 2. In creating my observed file, I entered multiple stations. Does
MET handle this?
> I am asking because the manual says it only handles 1 point station,
yet ascii2nc ran with no problems and all the data is written to the
netcdf.
>
> FYI, this is the process following on from a previous inquiry.
> [rt.rap.ucar.edu #62959] POINT_STAT Error APCP
>
> I have placed my data on ftp.rap.ucar.edu fyr
> /incoming/irap/met_help/weston_data/
>
> Thanks guys!
> Michael Weston
>
> EScience Associates (Pty) Ltd
> E-mail: michael at escience.co.za
> Web: www.escience.co.za
> PO Box 2950, Saxonwold, 2132
> 9 Victoria Street, Oaklands, Johannesburg, 2192
> Tel: +27 (0)11 718 6380
> Fax: 086 527 5241
> VAT No: 473 025 4416
> Reg No: 2009/014472/07
>
------------------------------------------------
Subject: RE: [rt.rap.ucar.edu #63192] ascii2nc using 24 hour apcp
From: John Halley Gotway
Time: Thu Sep 26 10:58:10 2013
Mike,
I see the following new file on our ftp site:
ftp://ftp.rap.ucar.edu/incoming/irap/met_help/weston_data/ascii2nc.txt
Looking at that file, I see that the GRIB code, 61, is followed by an
accumulation interval of 1 hour. Is that correct, are you point
observations really hourly accumulations? I thought you'd indicated
in an
earlier email that they're really 24-hour accumulations. If so, you
should be using "24" there to indicate that it's a 24-hour
accumulation.
Thanks,
John
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63192 >
>
> Hi John
>
> Thanks for the feedback.
>
> 1.) Re: different versions.
> I am working on several machines. Some have 4.0, some don't.
> I will try use 3.1.
>
> 2.) I will adjust my Config file accordingly
>
> 3.) I have uploaded my text file used as input for ascii2nc. Could
you
> please check it for any obvious errors?
> If I get the conversion to netcdf right with ascii2nc, it should
solve the
> matched pairs issue.
>
> Thanks
> Mike
>
> EScience Associates (Pty) Ltd
> E-mail: michael at escience.co.za
> Web: www.escience.co.za
> PO Box 2950, Saxonwold, 2132
> 9 Victoria Street, Oaklands, Johannesburg, 2192
> Tel: +27 (0)11 718 6380
> Fax: 086 527 5241
> VAT No: 473 025 4416
> Reg No: 2009/014472/07
> ________________________________________
> From: John Halley Gotway via RT [met_help at ucar.edu]
> Sent: Thursday, September 26, 2013 6:15 PM
> To: Michael Weston
> Subject: Re: [rt.rap.ucar.edu #63192] ascii2nc using 24 hour apcp
>
> Michael,
>
> Thanks for sending your data. I grabbed it from the ftp site and
ran it.
> There's several things going on here...
>
> (1) I tried to figure out what version of MET you're using...
> ncdump -h pcp_subtract.nc | grep MET_version
> :MET_version = "V4.0" ;
> ncdump -h lieb_Oct2000_Apr2000_24hour.nc | grep MET_version
> :MET_version = "V3.1" ;
> grep version PointStatConfig_24hour
> version = "V4.0";
>
> So it looks like you're using a mixture of METv3.1 and METv4.0.
When I
> run METv3.1/bin/point_stat, I get the error you're seeing (NetCDF:
> Attribute not found). That's because METv3.1 doesn't know
> about the format of the data generated by later versions of MET.
When I
> run METv4.0/bin/point_stat, I do not get that error. While we
attempt to
> make things backwards compatible when possible, it's
> obviously preferable to use data that's all been generated by the
same
> version of the code.
>
> (2) When running METv4.0/bin/point_stat, I do get other errors. In
the
> Point-Stat config file, you need to adjust the "fcst" setting as
> follows....
> field = [
> {
> name = "APCP_24";
> level = "(*,*)";
> cat_thresh = [ >0, >1, >5, >10, >25 ];
> }
> The reason for this is that you're passing a NetCDF file to Point-
Stat.
> When using gridded NetCDF files, you set "name" to the NetCDF
variable
> name to be used. And you set "level" to indicate which
> dimensions contain the gridded data. For that output of
pcp_combine,
> there are only 2 dimensions, so "(*,*)" is what you use.
>
> (3) Even after I got Point-Stat to run, I got 0 matched pairs.
Running at
> debug level 3 (-v 3), I see the following information:
> DEBUG 2: Processing APCP_24(*,*) versus APCP/A24, for observation
type
> ADPSFC, over region FULL, for interpolation method UW_MEAN(1), using
0
> pairs.
> DEBUG 3: Number of matched pairs = 0
> DEBUG 3: Observations processed = 7955
> DEBUG 3: Rejected: GRIB code = 0
> DEBUG 3: Rejected: valid time = 7918
> DEBUG 3: Rejected: bad obs value = 0
> DEBUG 3: Rejected: off the grid = 0
> DEBUG 3: Rejected: level mismatch = 37
> DEBUG 3: Rejected: message type = 0
> DEBUG 3: Rejected: masking region = 0
> DEBUG 3: Rejected: bad fcst value = 0
>
> So most of the observations are rejected due to the valid time and
some
> are a "level mismatch". Since you're verifying accumulated precip,
that
> means that the observation accumulation interval
> doesn't match what you've requested in the config file. You
requested
> 24-hour accumulation intervals, but your observation data contains
hourly
> intervals (3600 seconds). So I modified the "obs"
> section of the config file like this:
> field = [
> {
> name = "APCP";
> level = "A01";
> cat_thresh = [ >0, >1, >5, >10, >25 ];
> }
> ];
>
> Then when I reran, I got 37 matched pairs. You should figure out if
your
> point observations really do contain 24-hour accumulations, and if
so,
> modify the accumulation interval in the input to
> ASCII2NC to indicate that: 24-hours = 86400 seconds.
>
> Hope this helps get you going. Just let me know if more issues or
> questions arise.
>
> Thanks,
> John
>
> On 09/26/2013 04:14 AM, Michael Weston via RT wrote:
>>
>> Thu Sep 26 04:14:24 2013: Request 63192 was acted upon.
>> Transaction: Ticket created by michael at escience.co.za
>> Queue: met_help
>> Subject: ascii2nc using 24 hour apcp
>> Owner: Nobody
>> Requestors: michael at escience.co.za
>> Status: new
>> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63192 >
>>
>>
>> Good Morning MET helpdesk
>>
>> I am trying to use point-stat to compare 24hr apcp from WRF to my
>> observed data.
>>
>> I used ascii2nc to create my observed netcdf file. The observed
data I
>> entered had already been converted to 24hr totals.
>> When I ncdump -v obs_arr observed.nc, I think it is still showing
the
>> data as 1hr data (3600s).
>>
>> obs_arr =
>> 0, 61, 3600, -9999, 0,
>> 1, 61, 3600, -9999, 0,
>> 2, 61, 3600, -9999, 0,
>> 3, 61, 3600, -9999, 0,
>> 4, 61, 3600, -9999, 0,
>>
>> Sooooo, when I run point-stat. I get this error message.
>> NetCDF: Attribute not found
>>
>> My questions are:
>> 1. How do I create 24hr observed netcdf file? (if this is in fact
the
>> reason for my error)
>> 2. In creating my observed file, I entered multiple stations. Does
MET
>> handle this?
>> I am asking because the manual says it only handles 1 point
station, yet
>> ascii2nc ran with no problems and all the data is written to the
netcdf.
>>
>> FYI, this is the process following on from a previous inquiry.
>> [rt.rap.ucar.edu #62959] POINT_STAT Error APCP
>>
>> I have placed my data on ftp.rap.ucar.edu fyr
>> /incoming/irap/met_help/weston_data/
>>
>> Thanks guys!
>> Michael Weston
>>
>> EScience Associates (Pty) Ltd
>> E-mail: michael at escience.co.za
>> Web: www.escience.co.za
>> PO Box 2950, Saxonwold, 2132
>> 9 Victoria Street, Oaklands, Johannesburg, 2192
>> Tel: +27 (0)11 718 6380
>> Fax: 086 527 5241
>> VAT No: 473 025 4416
>> Reg No: 2009/014472/07
>>
>
>
>
------------------------------------------------
Subject: RE: [rt.rap.ucar.edu #63192] ascii2nc using 24 hour apcp
From: Michael Weston
Time: Thu Sep 26 11:36:30 2013
Oh! Column 8 is accumulation period!
Apologies. I thought the time stamp would sort the timing out. I will
ammend.
Thank you for the other tips though with regards to versions and the
Config file.
Will see how it goes.
Regards
Mike
________________________________________
From: John Halley Gotway via RT [met_help at ucar.edu]
Sent: Thursday, September 26, 2013 6:58 PM
To: Michael Weston
Subject: RE: [rt.rap.ucar.edu #63192] ascii2nc using 24 hour apcp
Mike,
I see the following new file on our ftp site:
ftp://ftp.rap.ucar.edu/incoming/irap/met_help/weston_data/ascii2nc.txt
Looking at that file, I see that the GRIB code, 61, is followed by an
accumulation interval of 1 hour. Is that correct, are you point
observations really hourly accumulations? I thought you'd indicated
in an
earlier email that they're really 24-hour accumulations. If so, you
should be using "24" there to indicate that it's a 24-hour
accumulation.
Thanks,
John
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63192 >
>
> Hi John
>
> Thanks for the feedback.
>
> 1.) Re: different versions.
> I am working on several machines. Some have 4.0, some don't.
> I will try use 3.1.
>
> 2.) I will adjust my Config file accordingly
>
> 3.) I have uploaded my text file used as input for ascii2nc. Could
you
> please check it for any obvious errors?
> If I get the conversion to netcdf right with ascii2nc, it should
solve the
> matched pairs issue.
>
> Thanks
> Mike
>
> EScience Associates (Pty) Ltd
> E-mail: michael at escience.co.za
> Web: www.escience.co.za
> PO Box 2950, Saxonwold, 2132
> 9 Victoria Street, Oaklands, Johannesburg, 2192
> Tel: +27 (0)11 718 6380
> Fax: 086 527 5241
> VAT No: 473 025 4416
> Reg No: 2009/014472/07
> ________________________________________
> From: John Halley Gotway via RT [met_help at ucar.edu]
> Sent: Thursday, September 26, 2013 6:15 PM
> To: Michael Weston
> Subject: Re: [rt.rap.ucar.edu #63192] ascii2nc using 24 hour apcp
>
> Michael,
>
> Thanks for sending your data. I grabbed it from the ftp site and
ran it.
> There's several things going on here...
>
> (1) I tried to figure out what version of MET you're using...
> ncdump -h pcp_subtract.nc | grep MET_version
> :MET_version = "V4.0" ;
> ncdump -h lieb_Oct2000_Apr2000_24hour.nc | grep MET_version
> :MET_version = "V3.1" ;
> grep version PointStatConfig_24hour
> version = "V4.0";
>
> So it looks like you're using a mixture of METv3.1 and METv4.0.
When I
> run METv3.1/bin/point_stat, I get the error you're seeing (NetCDF:
> Attribute not found). That's because METv3.1 doesn't know
> about the format of the data generated by later versions of MET.
When I
> run METv4.0/bin/point_stat, I do not get that error. While we
attempt to
> make things backwards compatible when possible, it's
> obviously preferable to use data that's all been generated by the
same
> version of the code.
>
> (2) When running METv4.0/bin/point_stat, I do get other errors. In
the
> Point-Stat config file, you need to adjust the "fcst" setting as
> follows....
> field = [
> {
> name = "APCP_24";
> level = "(*,*)";
> cat_thresh = [ >0, >1, >5, >10, >25 ];
> }
> The reason for this is that you're passing a NetCDF file to Point-
Stat.
> When using gridded NetCDF files, you set "name" to the NetCDF
variable
> name to be used. And you set "level" to indicate which
> dimensions contain the gridded data. For that output of
pcp_combine,
> there are only 2 dimensions, so "(*,*)" is what you use.
>
> (3) Even after I got Point-Stat to run, I got 0 matched pairs.
Running at
> debug level 3 (-v 3), I see the following information:
> DEBUG 2: Processing APCP_24(*,*) versus APCP/A24, for observation
type
> ADPSFC, over region FULL, for interpolation method UW_MEAN(1), using
0
> pairs.
> DEBUG 3: Number of matched pairs = 0
> DEBUG 3: Observations processed = 7955
> DEBUG 3: Rejected: GRIB code = 0
> DEBUG 3: Rejected: valid time = 7918
> DEBUG 3: Rejected: bad obs value = 0
> DEBUG 3: Rejected: off the grid = 0
> DEBUG 3: Rejected: level mismatch = 37
> DEBUG 3: Rejected: message type = 0
> DEBUG 3: Rejected: masking region = 0
> DEBUG 3: Rejected: bad fcst value = 0
>
> So most of the observations are rejected due to the valid time and
some
> are a "level mismatch". Since you're verifying accumulated precip,
that
> means that the observation accumulation interval
> doesn't match what you've requested in the config file. You
requested
> 24-hour accumulation intervals, but your observation data contains
hourly
> intervals (3600 seconds). So I modified the "obs"
> section of the config file like this:
> field = [
> {
> name = "APCP";
> level = "A01";
> cat_thresh = [ >0, >1, >5, >10, >25 ];
> }
> ];
>
> Then when I reran, I got 37 matched pairs. You should figure out if
your
> point observations really do contain 24-hour accumulations, and if
so,
> modify the accumulation interval in the input to
> ASCII2NC to indicate that: 24-hours = 86400 seconds.
>
> Hope this helps get you going. Just let me know if more issues or
> questions arise.
>
> Thanks,
> John
>
> On 09/26/2013 04:14 AM, Michael Weston via RT wrote:
>>
>> Thu Sep 26 04:14:24 2013: Request 63192 was acted upon.
>> Transaction: Ticket created by michael at escience.co.za
>> Queue: met_help
>> Subject: ascii2nc using 24 hour apcp
>> Owner: Nobody
>> Requestors: michael at escience.co.za
>> Status: new
>> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63192 >
>>
>>
>> Good Morning MET helpdesk
>>
>> I am trying to use point-stat to compare 24hr apcp from WRF to my
>> observed data.
>>
>> I used ascii2nc to create my observed netcdf file. The observed
data I
>> entered had already been converted to 24hr totals.
>> When I ncdump -v obs_arr observed.nc, I think it is still showing
the
>> data as 1hr data (3600s).
>>
>> obs_arr =
>> 0, 61, 3600, -9999, 0,
>> 1, 61, 3600, -9999, 0,
>> 2, 61, 3600, -9999, 0,
>> 3, 61, 3600, -9999, 0,
>> 4, 61, 3600, -9999, 0,
>>
>> Sooooo, when I run point-stat. I get this error message.
>> NetCDF: Attribute not found
>>
>> My questions are:
>> 1. How do I create 24hr observed netcdf file? (if this is in fact
the
>> reason for my error)
>> 2. In creating my observed file, I entered multiple stations. Does
MET
>> handle this?
>> I am asking because the manual says it only handles 1 point
station, yet
>> ascii2nc ran with no problems and all the data is written to the
netcdf.
>>
>> FYI, this is the process following on from a previous inquiry.
>> [rt.rap.ucar.edu #62959] POINT_STAT Error APCP
>>
>> I have placed my data on ftp.rap.ucar.edu fyr
>> /incoming/irap/met_help/weston_data/
>>
>> Thanks guys!
>> Michael Weston
>>
>> EScience Associates (Pty) Ltd
>> E-mail: michael at escience.co.za
>> Web: www.escience.co.za
>> PO Box 2950, Saxonwold, 2132
>> 9 Victoria Street, Oaklands, Johannesburg, 2192
>> Tel: +27 (0)11 718 6380
>> Fax: 086 527 5241
>> VAT No: 473 025 4416
>> Reg No: 2009/014472/07
>>
>
>
>
------------------------------------------------
Subject: RE: [rt.rap.ucar.edu #63192] ascii2nc using 24 hour apcp
From: Michael Weston
Time: Fri Sep 27 08:24:53 2013
Working!
Ok, so point-stat seems to be using my multiple point ascii2nc file.
Thanks for the help John.
________________________________________
From: John Halley Gotway via RT [met_help at ucar.edu]
Sent: Thursday, September 26, 2013 6:58 PM
To: Michael Weston
Subject: RE: [rt.rap.ucar.edu #63192] ascii2nc using 24 hour apcp
Mike,
I see the following new file on our ftp site:
ftp://ftp.rap.ucar.edu/incoming/irap/met_help/weston_data/ascii2nc.txt
Looking at that file, I see that the GRIB code, 61, is followed by an
accumulation interval of 1 hour. Is that correct, are you point
observations really hourly accumulations? I thought you'd indicated
in an
earlier email that they're really 24-hour accumulations. If so, you
should be using "24" there to indicate that it's a 24-hour
accumulation.
Thanks,
John
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63192 >
>
> Hi John
>
> Thanks for the feedback.
>
> 1.) Re: different versions.
> I am working on several machines. Some have 4.0, some don't.
> I will try use 3.1.
>
> 2.) I will adjust my Config file accordingly
>
> 3.) I have uploaded my text file used as input for ascii2nc. Could
you
> please check it for any obvious errors?
> If I get the conversion to netcdf right with ascii2nc, it should
solve the
> matched pairs issue.
>
> Thanks
> Mike
>
> EScience Associates (Pty) Ltd
> E-mail: michael at escience.co.za
> Web: www.escience.co.za
> PO Box 2950, Saxonwold, 2132
> 9 Victoria Street, Oaklands, Johannesburg, 2192
> Tel: +27 (0)11 718 6380
> Fax: 086 527 5241
> VAT No: 473 025 4416
> Reg No: 2009/014472/07
> ________________________________________
> From: John Halley Gotway via RT [met_help at ucar.edu]
> Sent: Thursday, September 26, 2013 6:15 PM
> To: Michael Weston
> Subject: Re: [rt.rap.ucar.edu #63192] ascii2nc using 24 hour apcp
>
> Michael,
>
> Thanks for sending your data. I grabbed it from the ftp site and
ran it.
> There's several things going on here...
>
> (1) I tried to figure out what version of MET you're using...
> ncdump -h pcp_subtract.nc | grep MET_version
> :MET_version = "V4.0" ;
> ncdump -h lieb_Oct2000_Apr2000_24hour.nc | grep MET_version
> :MET_version = "V3.1" ;
> grep version PointStatConfig_24hour
> version = "V4.0";
>
> So it looks like you're using a mixture of METv3.1 and METv4.0.
When I
> run METv3.1/bin/point_stat, I get the error you're seeing (NetCDF:
> Attribute not found). That's because METv3.1 doesn't know
> about the format of the data generated by later versions of MET.
When I
> run METv4.0/bin/point_stat, I do not get that error. While we
attempt to
> make things backwards compatible when possible, it's
> obviously preferable to use data that's all been generated by the
same
> version of the code.
>
> (2) When running METv4.0/bin/point_stat, I do get other errors. In
the
> Point-Stat config file, you need to adjust the "fcst" setting as
> follows....
> field = [
> {
> name = "APCP_24";
> level = "(*,*)";
> cat_thresh = [ >0, >1, >5, >10, >25 ];
> }
> The reason for this is that you're passing a NetCDF file to Point-
Stat.
> When using gridded NetCDF files, you set "name" to the NetCDF
variable
> name to be used. And you set "level" to indicate which
> dimensions contain the gridded data. For that output of
pcp_combine,
> there are only 2 dimensions, so "(*,*)" is what you use.
>
> (3) Even after I got Point-Stat to run, I got 0 matched pairs.
Running at
> debug level 3 (-v 3), I see the following information:
> DEBUG 2: Processing APCP_24(*,*) versus APCP/A24, for observation
type
> ADPSFC, over region FULL, for interpolation method UW_MEAN(1), using
0
> pairs.
> DEBUG 3: Number of matched pairs = 0
> DEBUG 3: Observations processed = 7955
> DEBUG 3: Rejected: GRIB code = 0
> DEBUG 3: Rejected: valid time = 7918
> DEBUG 3: Rejected: bad obs value = 0
> DEBUG 3: Rejected: off the grid = 0
> DEBUG 3: Rejected: level mismatch = 37
> DEBUG 3: Rejected: message type = 0
> DEBUG 3: Rejected: masking region = 0
> DEBUG 3: Rejected: bad fcst value = 0
>
> So most of the observations are rejected due to the valid time and
some
> are a "level mismatch". Since you're verifying accumulated precip,
that
> means that the observation accumulation interval
> doesn't match what you've requested in the config file. You
requested
> 24-hour accumulation intervals, but your observation data contains
hourly
> intervals (3600 seconds). So I modified the "obs"
> section of the config file like this:
> field = [
> {
> name = "APCP";
> level = "A01";
> cat_thresh = [ >0, >1, >5, >10, >25 ];
> }
> ];
>
> Then when I reran, I got 37 matched pairs. You should figure out if
your
> point observations really do contain 24-hour accumulations, and if
so,
> modify the accumulation interval in the input to
> ASCII2NC to indicate that: 24-hours = 86400 seconds.
>
> Hope this helps get you going. Just let me know if more issues or
> questions arise.
>
> Thanks,
> John
>
> On 09/26/2013 04:14 AM, Michael Weston via RT wrote:
>>
>> Thu Sep 26 04:14:24 2013: Request 63192 was acted upon.
>> Transaction: Ticket created by michael at escience.co.za
>> Queue: met_help
>> Subject: ascii2nc using 24 hour apcp
>> Owner: Nobody
>> Requestors: michael at escience.co.za
>> Status: new
>> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63192 >
>>
>>
>> Good Morning MET helpdesk
>>
>> I am trying to use point-stat to compare 24hr apcp from WRF to my
>> observed data.
>>
>> I used ascii2nc to create my observed netcdf file. The observed
data I
>> entered had already been converted to 24hr totals.
>> When I ncdump -v obs_arr observed.nc, I think it is still showing
the
>> data as 1hr data (3600s).
>>
>> obs_arr =
>> 0, 61, 3600, -9999, 0,
>> 1, 61, 3600, -9999, 0,
>> 2, 61, 3600, -9999, 0,
>> 3, 61, 3600, -9999, 0,
>> 4, 61, 3600, -9999, 0,
>>
>> Sooooo, when I run point-stat. I get this error message.
>> NetCDF: Attribute not found
>>
>> My questions are:
>> 1. How do I create 24hr observed netcdf file? (if this is in fact
the
>> reason for my error)
>> 2. In creating my observed file, I entered multiple stations. Does
MET
>> handle this?
>> I am asking because the manual says it only handles 1 point
station, yet
>> ascii2nc ran with no problems and all the data is written to the
netcdf.
>>
>> FYI, this is the process following on from a previous inquiry.
>> [rt.rap.ucar.edu #62959] POINT_STAT Error APCP
>>
>> I have placed my data on ftp.rap.ucar.edu fyr
>> /incoming/irap/met_help/weston_data/
>>
>> Thanks guys!
>> Michael Weston
>>
>> EScience Associates (Pty) Ltd
>> E-mail: michael at escience.co.za
>> Web: www.escience.co.za
>> PO Box 2950, Saxonwold, 2132
>> 9 Victoria Street, Oaklands, Johannesburg, 2192
>> Tel: +27 (0)11 718 6380
>> Fax: 086 527 5241
>> VAT No: 473 025 4416
>> Reg No: 2009/014472/07
>>
>
>
>
------------------------------------------------
More information about the Met_help
mailing list