[Met_help] [rt.rap.ucar.edu #79028] History for problem locating matching times
John Halley Gotway via RT
met_help at ucar.edu
Wed Jan 18 09:55:18 MST 2017
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Dear Met Help,
I am having difficulties matching observations vs forecast at the same
valid times in the output of point_stat when using gdas files and providing
a particular list of stations. I am interested in surface stations (ADPSFC)
data, like UGRD,VGRD,RH and TMP at levels Z10,Z10,Z2,Z2.
I have beem using observations from gdas in prebuffr format, which are
produced daily every 6 hours
gdas1.t00z.prepbufr.nr
gdas1.t06z.prepbufr.nr
gdas1.t12z.prepbufr.nr
gdas1.t18z.prepbufr.nr
I then convert them to nc format using the pb2nc tool.
In the configuration file of pb2ncI set the following
observation range
obs_window = {
beg = -10800;
end = 10800;
}
so that the output nc file will pressumably contain all the observations
+/- 3 hours around the lead time of the prebuffr file.
Then I use point_stat with the output from a 24 hours (UPP-postprocessed to
grib files) WRF simulation with forecast files produced every hour.
I use a particular list of stations in the search.
The problem is that the program does not seem to find any observations
at times 00,03,06,12,15,18,21 (ie, every 3 hours). I do find a match at any
other times.
I am not sure why is that. For example, for forecast hour 3
I have tried using the pre-processed observation file at 00
point_stat WRF_00.GrbF03 pb2nc.2016032900.nc pstat_config -outdir . -log
log_03
or at 06
point_stat WRF_00.GrbF03 pb2nc.2016032906.nc pstat_config -outdir . -log
log_03
in case the time "03" is not contained in the 00 but in the 06 file.
But I am unable to find any matches in the mpr or stat files.
I also tried narrowing or widening the pb2nc output with obs_window, but
the result was the same.
Am I allowed to compare observations versus forecast at times
00, 06, 12 and 18 UTC, or do the prebuffr files only contain observations
*around* those times but not at those times?
I was also interested in comparing some global models with the gdas
observation data, but the output from these models comes every 3 hours, and
that is precisely at the times where I have difficulty finding any matches
For example, I tried using files from GFS, and I don't get
any matches over my domain.
My station list looks like
StationList
10501
10954
10091
10425
10460
10575
10618
....
and they are all located over Germany.
In order to check if the converted gdas files indeed contain these stations
at the selected times I used the -dump option in pb2nc. However, I don't
know how to interpret the output. For example, I see
10618 7.33 49.70 -2.000 377.0 281.0 511.0 ******
--------------------------------------------------------------------------------
lev 1 ob qm pc rc fc an cat
--------------------------------------------------------------------------------
P 958.8 2.0 1.0 6.0
U 3.8 9.0 4.0 3.0 10.5 6.0
U 3.8 2.0 1.0
V 3.2 9.0 4.0 3.0 4.8 6.0
V 3.2 2.0 1.0
Would that be station 10618, with coordinates lon/lat 7.33/49.70 ... etc?
But what are the rest of the variables?
I could not find much info on this topic in the MET manual.
Could you point me to some other source where I can find
what those variables mean?
I have been using the stat_analysis tool and some python scripts
to analyze the data output. With stat_analysis I used commands like
stat_analysis -job filter -lookin
./point_stat_150000L_20160329_150000V.stat -line_type MPR -fcst_var RH
-fcst_lev Z2 -column_str OBS_SID
10954,10852,10238,10505,10519,10613,10335,10343,10321,10200,10246,10803,10815,10439,10500,10628,10616,10038,10476,10618,10860,10142,10771,10172,10805,10857,10837,10856,10022,10471,10418,10067,10304,10827,10743,10502,10136,10506,10146,10306,10765,10037,10788,10281,10268,10863,10055,10126,10334
-column_str FCST_VALID_END 20160329_150000 -column_str FCST_VALID_END
20160329_150000 -column_str OBS_VALID_BEG 20160329_150000 -column_str
OBS_VALID_END 20160329_150000 -column_str VX_MASK StationList -dump_row
pstat_20160329_RH_15.txt
Could you please tell me if I am doing something wrong here?
Thanks heaps for any help.
Carlos
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: problem locating matching times
From: John Halley Gotway
Time: Fri Dec 23 09:06:54 2016
Hello,
I see you are having trouble using GDAS point observations for surface
verification.
Please take a look at the note about GDAS in bold at the bottom of
this
page:
http://www.dtcenter.org/met/users/downloads/observation_data.php
I suspect that will fix your problem.
Thanks,
John Halley Gotway
On Fri, Dec 23, 2016 at 2:07 AM Carlos Peralta via RT
<met_help at ucar.edu>
wrote:
>
>
> Fri Dec 23 02:07:17 2016: Request 79028 was acted upon.
>
> Transaction: Ticket created by carlos9917 at gmail.com
>
> Queue: met_help
>
> Subject: problem locating matching times
>
> Owner: Nobody
>
> Requestors: carlos9917 at gmail.com
>
> Status: new
>
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=79028 >
>
>
>
>
>
> Dear Met Help,
>
>
>
> I am having difficulties matching observations vs forecast at the
same
>
> valid times in the output of point_stat when using gdas files and
providing
>
> a particular list of stations. I am interested in surface stations
(ADPSFC)
>
> data, like UGRD,VGRD,RH and TMP at levels Z10,Z10,Z2,Z2.
>
>
>
> I have beem using observations from gdas in prebuffr format, which
are
>
> produced daily every 6 hours
>
> gdas1.t00z.prepbufr.nr
>
> gdas1.t06z.prepbufr.nr
>
> gdas1.t12z.prepbufr.nr
>
> gdas1.t18z.prepbufr.nr
>
>
>
> I then convert them to nc format using the pb2nc tool.
>
> In the configuration file of pb2ncI set the following
>
> observation range
>
>
>
> obs_window = {
>
> beg = -10800;
>
> end = 10800;
>
> }
>
>
>
> so that the output nc file will pressumably contain all the
observations
>
> +/- 3 hours around the lead time of the prebuffr file.
>
>
>
> Then I use point_stat with the output from a 24 hours (UPP-
postprocessed to
>
> grib files) WRF simulation with forecast files produced every hour.
>
> I use a particular list of stations in the search.
>
> The problem is that the program does not seem to find any
observations
>
> at times 00,03,06,12,15,18,21 (ie, every 3 hours). I do find a match
at any
>
> other times.
>
> I am not sure why is that. For example, for forecast hour 3
>
> I have tried using the pre-processed observation file at 00
>
> point_stat WRF_00.GrbF03 pb2nc.2016032900.nc pstat_config -outdir .
-log
>
> log_03
>
>
>
> or at 06
>
> point_stat WRF_00.GrbF03 pb2nc.2016032906.nc pstat_config -outdir .
-log
>
> log_03
>
>
>
> in case the time "03" is not contained in the 00 but in the 06 file.
>
> But I am unable to find any matches in the mpr or stat files.
>
> I also tried narrowing or widening the pb2nc output with
obs_window, but
>
> the result was the same.
>
>
>
> Am I allowed to compare observations versus forecast at times
>
> 00, 06, 12 and 18 UTC, or do the prebuffr files only contain
observations
>
> *around* those times but not at those times?
>
>
>
> I was also interested in comparing some global models with the gdas
>
> observation data, but the output from these models comes every 3
hours, and
>
> that is precisely at the times where I have difficulty finding any
matches
>
> For example, I tried using files from GFS, and I don't get
>
> any matches over my domain.
>
>
>
> My station list looks like
>
>
>
> StationList
>
> 10501
>
> 10954
>
> 10091
>
> 10425
>
> 10460
>
> 10575
>
> 10618
>
> ....
>
> and they are all located over Germany.
>
>
>
> In order to check if the converted gdas files indeed contain these
stations
>
> at the selected times I used the -dump option in pb2nc. However, I
don't
>
> know how to interpret the output. For example, I see
>
>
>
>
>
> 10618 7.33 49.70 -2.000 377.0 281.0 511.0 ******
>
>
>
--------------------------------------------------------------------------------
>
> lev 1 ob qm pc rc fc an
cat
>
>
>
--------------------------------------------------------------------------------
>
> P 958.8 2.0 1.0
6.0
>
> U 3.8 9.0 4.0 3.0 10.5
6.0
>
> U 3.8 2.0 1.0
>
> V 3.2 9.0 4.0 3.0 4.8
6.0
>
> V 3.2 2.0 1.0
>
> Would that be station 10618, with coordinates lon/lat 7.33/49.70 ...
etc?
>
> But what are the rest of the variables?
>
> I could not find much info on this topic in the MET manual.
>
> Could you point me to some other source where I can find
>
> what those variables mean?
>
>
>
> I have been using the stat_analysis tool and some python scripts
>
> to analyze the data output. With stat_analysis I used commands like
>
>
>
>
>
> stat_analysis -job filter -lookin
>
> ./point_stat_150000L_20160329_150000V.stat -line_type MPR -fcst_var
RH
>
> -fcst_lev Z2 -column_str OBS_SID
>
>
>
10954,10852,10238,10505,10519,10613,10335,10343,10321,10200,10246,10803,10815,10439,10500,10628,10616,10038,10476,10618,10860,10142,10771,10172,10805,10857,10837,10856,10022,10471,10418,10067,10304,10827,10743,10502,10136,10506,10146,10306,10765,10037,10788,10281,10268,10863,10055,10126,10334
>
> -column_str FCST_VALID_END 20160329_150000 -column_str
FCST_VALID_END
>
> 20160329_150000 -column_str OBS_VALID_BEG 20160329_150000
-column_str
>
> OBS_VALID_END 20160329_150000 -column_str VX_MASK StationList
-dump_row
>
> pstat_20160329_RH_15.txt
>
>
>
>
>
> Could you please tell me if I am doing something wrong here?
>
>
>
> Thanks heaps for any help.
>
>
>
> Carlos
>
>
>
>
------------------------------------------------
More information about the Met_help
mailing list