[gradsusr] ensemble reforecast runs

Jennifer Adams jma at cola.iges.org
Mon Aug 30 10:09:26 EDT 2010


Daniele M. --
There is no way to fudge the time metadata with GRIB data in GrADS to  
make all your ensembles line up in the time dimension. Each forecast  
must be described as the metadata dictates in the data files. Your  
ensembles don't overlap at all, and the only way to merge them into  
one data set is to set TDEF to span all 21 years.

Your descriptor should have these components:

tdef 14677 linear 12z29aug1989 12hr
edef 21
e1 67 12z29aug1989
e2 67 12z29aug1990
...
e21 67 12z29aug2009
endedef

I'm not sure this is the most efficient way to handle this data.
--Jennifer




On Aug 27, 2010, at 7:25 AM, Daniele Mastrangelo wrote:

> Hi Daniele
>
> your method works: the "set time" command is the crucial point.
> By defining a new variable for the same field of each new opened  
> file (or by using the .#of-ctl-file at the end of the variable  
> name), one can create the original ensemble set.
>
> Grazie!
>
> D.
>
> Da: Daniele Gandini <daniele.gandini at arpa.piemonte.it>
> A: GrADS Users Forum <gradsusr at gradsusr.org>
> Inviato: Mar 24 agosto 2010, 09:33:53
> Oggetto: Re: [gradsusr] ensemble reforecast runs
>
> Dear Daniele Mastrangelo,
>
> you can do so: put each ensemble member in a different
> ctl file and then:
>
> 1) open file 1
> 2) display var.1
> 3) open file 2
> 4) set time 12z29Aug1990
> 5) display var.2
> 6) open file 3
> 7) set time 12z29Aug1991
> 8) display var.3
>
> and so on.
>
> It looks boring but in a script file you can use
> a while cycle on ensemble files, i.e.
> display var.i (i=1,21)
>
> Buona fortuna.
>
> Daniele Gandini
>
>
>
> ----- Messaggio originale -----
> Da: "Daniele Mastrangelo" <daniele_mastrangelo at yahoo.it>
> A: "GrADS Users Forum" <gradsusr at gradsusr.org>
> Inviato: Lunedì, 23 agosto 2010 18:02:59
> Oggetto: Re: [gradsusr] ensemble reforecast runs
>
>
>
>
> Hi Wesley,
>
> thanks for your reply.
>
> My previous email was not totally clear. Actually, I have 21  
> ensemble members with the same forecast length; these members are  
> produced initializing the model at the same hour, day, and month  
> (but) of 21 different years (12 UTC 29 aug 1989, 12 UTC 29 aug  
> 1990, ... , 12 UTC 29 aug 2009).
>
> I tried, however, to apply something similar to your point 2): I  
> aggregated 2 members (e1, e2) in a single ctl file and changed by  
> hand the year of the second member in the EDEF section, making it  
> the same as the first member
> ...
> tdef 67 linear 12Z29aug1989 12hr
> zdef 1 linear 1 1
> vars 8
> ...
> ENDVARS
> EDEF 2
> e1 67 12Z29aug1989
> e2 67 12Z29aug1989 ---1989 instead of 1990---
> ENDEDEF
>
> Gribmap ran with no error messages but then, in GrADS, I could not  
> use the e2 fields: "Cannot contour grid - all undefined values"
>
> Is there some other way I could try?
>
> Best regards,
> Daniele
>
>
>
>
> Da: Wesley Ebisuzaki <Wesley.Ebisuzaki at noaa.gov>
> A: GrADS Users Forum <gradsusr at gradsusr.org>
> Inviato: Lun 23 agosto 2010, 16:43:08
> Oggetto: Re: [gradsusr] ensemble reforecast runs
>
> Daniele,
>
> g2ctl assumes that the length of the forecast is the same for all
> ensemble members.
> If not, you can do either
>
> (1) You could try using the longest ensemble as the 1st ensemble.
> (It may work.)
>
> (2) You have to set the TDEF by hand.
>
> Wesley Ebisuzaki
>
>
> Daniele Mastrangelo wrote:
> > Dear GrADS users,
> >
> > I'd like to use GrADS v 2.0.a8 to evaluate a set of 21 reforecast
> > ensemble runs. These runs are obtained running the same global model
> > initialized at the same hour and date of 21 different years.
> >
> > I obtained a single ctl file through "g2ctl.pl -ens" (g2ctl.pl v
> > 0.0.4o) but, when I used gribmap to produce the corresponding idx
> > file, I got the following error message:
> >
> > Open Error: ensemble 1 (e2) extends beyond the time axis limits
> > The data file was not opened.
> >
> > I see that the problem is that these runs cover different periods;
> > moreover, I get the same error when I try to open two of the 21
> > different reforecast runs in a normal GrADS session (as file 1 and
> > file 2). Therefore, my question is: can GrADS treat a set reforecast
> > runs used as ensemble runs?
> > If it cannot directly do that, is there a way to bypass the problem?
> >
> > Thank you,
> > Daniele
> >
> >  
> ------------------------------------------------------------------------
> >
> > _______________________________________________
> > gradsusr mailing list
> > gradsusr at gradsusr.org
> > http://gradsusr.org/mailman/listinfo/gradsusr
> >
>
> _______________________________________________
> gradsusr mailing list
> gradsusr at gradsusr.org
> http://gradsusr.org/mailman/listinfo/gradsusr
>
>
> _______________________________________________
> gradsusr mailing list
> gradsusr at gradsusr.org
> http://gradsusr.org/mailman/listinfo/gradsusr
>
> _______________________________________________
> gradsusr mailing list
> gradsusr at gradsusr.org
> http://gradsusr.org/mailman/listinfo/gradsusr
>
> _______________________________________________
> gradsusr mailing list
> gradsusr at gradsusr.org
> http://gradsusr.org/mailman/listinfo/gradsusr

--
Jennifer M. Adams
IGES/COLA
4041 Powder Mill Road, Suite 302
Calverton, MD 20705
jma at cola.iges.org



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://gradsusr.org/pipermail/gradsusr/attachments/20100830/1ddee80f/attachment-0003.html 


More information about the gradsusr mailing list