[gradsusr] Reading timesteps in Arpège grib2 file

Thu Dec 17 08:14:53 EST 2015

Brian,

I am being pedantic because I hope that the ARPEGE and AROME
people will get word and fix their grib2 files.

In order for GrADS to plot the ARPEGE file, GrADS needs to know the
time of the fields.  For Product Definition Template (PDT) 4.8 which is
used to describe averages and accumulations, there are 3 different times.

   Reference time:  start of the forecast
             Section 1, octets 13-19

   Start of the forecast time interval:
             reference time + forecast time (Section 3, octets 18-22)

   End of the forecast time interval:
              Section 3, octets 35-41

For your ARPEGE file, the only unique time should be the end of the forecast
time interval.   However when you examine the ARPEGE file.

ebis at linux-landing:~/Downloads$ wgrib2
ARPEGE_0.1_SP1_37H48H_201512150000.grib2 -s -start_ft -end_ft -match PRATE
133:32079786:d=2015121500:TPRATE:surface:0-37 hour acc
fcst::start_ft=2015121500:end_ft=2015121600
134:32391856:d=2015121500:TPRATE:surface:0-38 hour acc
fcst::start_ft=2015121500:end_ft=2015121600
135:32706662:d=2015121500:TPRATE:surface:0-39 hour acc
fcst::start_ft=2015121500:end_ft=2015121600
136:33024062:d=2015121500:TPRATE:surface:0-40 hour acc
fcst::start_ft=2015121500:end_ft=2015121600
137:33344053:d=2015121500:TPRATE:surface:0-41 hour acc
fcst::start_ft=2015121500:end_ft=2015121600
138:33666613:d=2015121500:TPRATE:surface:0-42 hour acc
fcst::start_ft=2015121500:end_ft=2015121600
139:33991684:d=2015121500:TPRATE:surface:0-43 hour acc
fcst::start_ft=2015121500:end_ft=2015121600
140:34319034:d=2015121500:TPRATE:surface:0-44 hour acc
fcst::start_ft=2015121500:end_ft=2015121600
141:34648323:d=2015121500:TPRATE:surface:0-45 hour acc
fcst::start_ft=2015121500:end_ft=2015121600
142:34979540:d=2015121500:TPRATE:surface:0-46 hour acc
fcst::start_ft=2015121500:end_ft=2015121600
143:35312524:d=2015121500:TPRATE:surface:0-47 hour acc
fcst::start_ft=2015121500:end_ft=2015121600
144:35647622:d=2015121500:TPRATE:surface:0-2 day acc
fcst::start_ft=2015121500:end_ft=2015121700

You see that the TPRATE (totat precipitation rate) has the end of
the forecast time interval of 2015121600 which is wrong.   This ARPEGE file
needs to be fixed.  A few weeks ago, someone sent me a AROME
forecast file which had the same problem.

So g2ctl and gribmap will not work until the grib file is fixed.

You can display the fields using

  alt_g2ctl -0t ARPEGE_0.1_SP1_37H48H_201512150000.grib2 >a.ctl
  alt_gmp -i a.ctl

This solution gives each field a unique name and doesn't use the
end of the forecast time.

Maybe someone will inform the ARPEGE and AROME people and they
will fix their files.

Wesley

On Thu, Dec 17, 2015 at 6:13 AM, Brian Gaze <brian.gaze at ntlworld.com> wrote:

> Wesley,
>
> You can ftp down the grib file here using anonymous:
>
> 137.135.248.188
>
> ARPEGE_0.1_SP1_37H48H_201512150000.grib2
>
> PS: It's about 40 meg and I tested the ftp download successfully from a
> command line rather than gui client
>
> Cheers
>
> Brian
>
>
> On 16 December 2015 at 15:03 Wesley Ebisuzaki - NOAA Federal <
> wesley.ebisuzaki at noaa.gov> wrote:
>
> Brian,
>
> It would be much easier if you made your grib file available
> by FTP. If the file is huge or is a proprietary forecast, you can
> zero out the data by
>
> wgrib2 file.grib -rpn 0 -grib_out file00.grb
>
> Wesley
>
>
>
> On Tue, Dec 15, 2015 at 4:32 PM, Brian Gaze < brian.gaze at ntlworld.com>
> wrote:
>
> Hi,
>
> I'm working with Arpège grib2 files. These should I think each contain up
> to 13 hourly forecast timesteps. If I run...
>
> g2ctl.pl -verf -b file.grib2 > file.ctl
> gribmap -i file.ctl
>
> ...I'm able to open the .ctl file in GrADS but it only has 2 timesteps. T1
> seems to be the time the model was initialized and T2 the time of the first
> forecast step.
>
>
> If I try:
>
> g2ctl.pl -b -ts1hr file.grib2 > file.ctl
> gribmap -i file.ctl
>
> It opens ok in GrADS and the correct number of timesteps are present but
> they begin from the model initialization hour rather than incrementing from
> the time of the first forecast step. For example:
>
> Model initialization: 0z 15/12
> Forecast time range contained in the file is +37 hours to +48 hours after
> initialization
>
> t2 = 1z 15/12
> t3 = 2z 15/12
> .. etc
>
> As a consequence, when I try to plot a variable the grid is undefined.
>
>
> What should I think be present is:
> t2 = 13z 16/12
> t3 = 14z 16/12
> ..etc
>
> Any suggestions / ideas on what I'm doing wrong?
>
> Thanks
>
> BWG
>
> _______________________________________________
> gradsusr mailing list
> gradsusr at gradsusr.org
> http://gradsusr.org/mailman/listinfo/gradsusr
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://gradsusr.org/pipermail/gradsusr/attachments/20151217/2b8480da/attachment.html 