[gradsusr] Root Mean Square Error

Amulya Chevuturi amulya.chevuturi at gmail.com
Tue Dec 25 22:34:21 EST 2012


Hello Tahereh

First of all I made a slight mistake in my last mail. For RMSE calculations
the term "A" in the formula should be not the parameter you want to
calculate but the error of the parameter. That is A should be error value.
Something like model minus observation.

To answer your question, as I am not familiar with you data. I would like
you to download outgoing longwave radiation (OLR) data from this website.
http://www.esrl.noaa.gov/psd/data/gridded/data.interp_OLR.html. Download
the olr.daily.mean.nc. Now write the following lines in the terminal where
you have downloaded and stored your OLR data.

sdfopen olr.day.mean.nc
set t 1 4
set lon 85 100
set lat 0 10
define RMSE=sqrt(ave((olr*olr),t=1,t=4))
d RMSE

This should work. I have tried with the same dataset and the output has
come. If this is successful, try the same with you dataset and see what the
result is.
Now the mistake in this method is the term "olr" is not error. RMSE formula
is sqrt(ave(error*error),t=1,t=4)). So usually error must be added, not the
parameter itself. Hence, you write another command before defining RMSE,
where you define error by denoting "define error=(forecast-observation)".
You can read this short webpage to understand RMSE.
http://www.weatheroffice.gc.ca/verification/scores/rmse_e.html.

With regards
Amulya Chevuturi

On Tue, Dec 25, 2012 at 12:42 PM, T. haghroosta <haghroosta at hotmail.com>wrote:

>  Dear Amulya,
>
> I wrote these commands as you said in a .gs file (test.gs) and after that
> I tried to run the test.gs, I got nothing. I mean when I used the command
> "run test.gs", grads did nothin, niether error nor output!
>
>
> 'open tem.ctl'
> 'set t 1 4'
> 'set lon 85 100'
> 'set lat 0 10'
> 'define RMSE=sqrt(ave(T*T,t=1,t=4))'
> 'd RMSE'
>
>  and also in another .gs file :
>
>
> 'open tem.ctl'
> 'set t 1 4'
> 'set lon 85 100'
> 'set lat 0 10'
> 'define RMSE=sqrt(aave(lh*lh,lon=85,lon=100,lat=0,lat=10))'
> 'd RMSE'
>
> I also couldn't work with the sample scripts.
>
> Could you please lead me how to do?
>
> Regards,
>
> Tahereh
>
>
> > ------------------------------
> >
> > Message: 3
> > Date: Tue, 25 Dec 2012 04:07:21 +0000
> > From: "T.! haghroosta" <haghroosta at hotmail.com>
> > Subject: [gradsusr] Root Mean Square Error
> > To: <gradsusr at gradsusr.org>
> > Message-ID: <DUB121-W43F87358FE9C823BDD5BA2DD3A0 at phx.gbl>
> > Content-Type: text/plain; charset="windows-1256"
>
> >
> >
> >
> >
> >
> > Dear All, Could you please tell me how I can calculate RMSE (root mean
> square error) in Grads and draw relevant graphs? Regards, Tahereh
> > -------------- next part --------------
> > An HTML attachment was scrubbed...
> > URL:
> http://gradsusr.org/pipermail/gradsusr/attachments/20121225/e2532241/attachment-0001.html
> >
> > ------------------------------
> >
> > Message: 4
> > Date: Tue, 25 Dec 2012 09:55:21 +0530
> > From: Amulya Chevuturi <amulya.chevuturi at gmail.com>
> > Subject: Re: [gradsusr] Root Mean Square Error
> > To: GrADS Users Forum <gradsusr at gradsusr.org>
> > Message-ID:**>
> <CAOV67LkiGd7ayjps3XTMW-3FWq5V=Ao4R4-oz7BUowZN0gkhqA at mail.g mail.com>
> > Content-Type: text/plain; charset="iso-8859-1"
>
> >
> > Hello Tahereh,
> >
> > This is a website containing quite many scripts for various analysis.
> There
> > is a RMSE calculation script also called ltrend.gs. Please look it and
> see
> > if it is useful for your work.
> > http://www.atmos.umd.edu/~bguan/grads/GrADS_Scripts.html
> >
> > Usually the RMSE calculation is done through the commands:
> >
> > "define RMSE=sqrt(ave((A*A),t=t1,t=tn)"
> > Where,
> > RMSE=Root Mean Square Error
> > A=Is the parameter you want to calculate
> > t1=Time step 1 for you series
> > tn=Last time step for your series
> > This will give you the plot of a time period.
> >
> > OR
> >
> > "define RMSE=sqrt(aave((A*A),lon=lon1,lon=lonn,lat=lat1,lat=latn)"
> > Where
> > RMSE=Root Mean Square Error
> > A=Is the parameter you want to calculate
> > Lat1, Latn, Lon1 and Lonn = Tel! l the extent of your study
>
> > This will give you a time series of RMSE.
> >
> > Hopefully this clears your doubt.
> >
> > With regards,
> > Amulya Chevuturi
> > School of Environmental Sciences
> > Jawaharlal Nehru University
> > New Delhi
> >
> > On Tue, Dec 25, 2012 at 9:37 AM, T. haghroosta <haghroosta at hotmail.com
> >wrote:
> >
> > > Dear All,
> > >
> > > Could you please tell me how I can calculate RMSE (root mean square
> error)
> > > in Grads and draw relevant graphs?
> > >
> > >
> > > Regards,
> > >
> > > Tahereh
> > >
> > > _______________________________________________
> > > gradsusr mailing list
> > > gradsusr at gradsusr.org
> > > http://gradsusr.org/mailman/listinfo/gradsusr
> > >
> > >
> > -------------- next part --------------
> > An HTML attachment was scrubbed...
> > URL: http://grad!
> susr.org/pipermail/gradsusr/attachments/20121225/c861c614/attachment-0001.html
> >
> > ------------------------------
> >
> > Message: 5
> > Date: Tue, 25 Dec 2012 10:08:26 +0530
> > From: madhu latha <madhulatha11 at gmail.com>
> > Subject: Re: [gradsusr] problem for prebufr data
> > To: GrADS Users Forum <gradsusr at gradsusr.org>
> > Message-ID:
> > <CAAhFJCSWRsxZMmPfNuZosfDgbzfJETwccC+sJxoSgMroxmWmTw at mail.gmail.com>
> > Content-Type: text/plain; charset="windows-1252"
> >
> > hi
> >
> > Thank you
> > First itself i have converted prebufr file to litle endian prebufr file
> > using ssrc.exe
> > then i have assimilated data using 3dvar.
> > Then mentioned problem came
> >
> > thank you.
> >
> > On Mon, Dec 24, 2012 at 11:56 PM, Rabah Hachelaf <r.hachelaf at gmail.com
> >wrote:
> >
> > > Hello,
> > >
> > > try to convert your prepbufr file from big endian to littele endian buy
> > > the ssrc.c too! l.
> > >
> > > ssrc.exe <Big Endian prepbufr file> Little Endian prepbufr file
> > >
> > >
> > > 2012/12/24 madhu latha <madhulatha11 at gmail.com>
> > >
> > >> hi
> > >> earlier i have tried the suggestions in the link by setting the
> > >> environment
> > >> setenv F_UFMTENDIAN "little:94-99"
> > >> but no use
> > >>
> > >> then itried to do reblock.c mentioned in the same link.
> > >> but i did not understand how to use the utility reblock.c
> > >>
> > >> kindly suggest.
> > >>
> > >>
> > >> On Mon, Dec 24, 2012 at 10:15 PM, Rabah Hachelaf <
> r.hachelaf at gmail.com>wrote:
> > >>
> > >>> Hello,
> > >>> see this link , it may be help you.
> > >>>
> > >>> http://www.mmm.ucar.edu/wrf/users/wrfda/Docs/readBufr.htm
> > >>>**> >>> --
> > >>> Cordialement,
> > &gt ;>> Best regards,
> > >>> Rabah Hachelaf
> > >>>
> > >>> ____
> > >>> ( )
> > >>> ( )
> > >>> (___ __)
> > >>> /////////
> > >>>
> > >>>
> > >>> 2012/12/24 madhu latha <madhulatha11 at gmail.com>
> > >>>
> > >>>> Dear all
> > >>>>
> > >>>>
> > >>>> Iam trying to run 3dvar on intel compiler
> > >>>> i have successfully installed wrf3dvar
> > >>>> but when iam trying to assimilate prepbufr data it is showing the
> > >>>> following error
> > >>>>
> > >>>>
> > >>>> Error message
> > >>>> **************BUFR ARCHIVE LIBRARY ABORT*****************
> > >>>> BUFRLIB: OPENBF - ERROR READING INPUT FILE CONNECTED TO ! UNIT 96
> WHEN
> > >>>> CHECKING
> > >>>> FOR ?BUFR? IN FIRST 4 BYTES OF RECORD
> > >>>> **************BUFR ARCHIVE LIBRARY ABORT*****************
> > >>>>
> > >>>>
> > >>>> can anyone know how to fix the problem
> > >>>> Any help will be appreciated.
> > >>>>
> > >>>> Thank you.
> > >>>>
> > >>>>
> > >>>> --
> > >>>> *With Regards
> > >>>> Madhu Latha Akkisetti
> > >>>> SRF
> > >>>> NARL <http://www.narl.gov.in>
> > >>>> Department of space
> > >>>> Gadanki*
>
> > >>>>
> > >>>> _______________________________________________
> > >>>> gradsusr mailing list
> > >>>> gradsusr at gradsusr.org
> > >>>> http://gradsusr! .org/mailman/listinfo/gradsusr
> > >>>>
> > >&g t;>>
> > >>>
> > >>>
> > >>> --
> > >>> Cordialement,
> > >>> Best regards,
> > >>> Rabah Hachelaf
> > >>>
> > >>> ____
> > >>> ( )
> > >>> ( )
> > >>> (___ __)
> > >>> /////////
>
> > >>>
> > >>>
> > >>> _______________________________________________
> > >>> gradsusr mailing list
> > >>> gradsusr at gradsusr.org
> > >>> http://gradsusr.org/mailman/listinfo/gradsusr
> > >>>
> > >>>
> > >>
> > >>
> > >> --
> > >> *With Regards
> > >> Madhu Latha Akkisetti
> > >> SRF
> > >> NARL <http://www.narl.gov.in>
> > >> Department of space
> > >> Gadanki*
> > >>
> > >> ________! _______________________________________
>
> > >> gradsusr mailing list
> > >> gradsusr at gradsusr.org
> > >> http://gradsusr.org/mailman/listinfo/gradsusr
> > >>
> > >>
> > >
> > >
> > > --
> > > Cordialement,
> > > Best regards,
> > > Rabah Hachelaf
> > >
> > > ____
> > > ( )
> > > ( )
> > > (___ __)
> > > /////////
>
> > >
> > >
> > > _______________________________________________
> > > gradsusr mailing list
> > > gradsusr at gradsusr.org
> > > http://gradsusr.org/mailman/listinfo/gradsusr
> > >
> > >
> >
> >
> > --
> > *With Regards
> > Madhu Latha Akkisetti
> > SRF
> > NARL <http://www.narl.gov.in>
> > Department of space
> > Gadanki*
> > -------------- next part --------------
> > An HTML attachment ! was scrubbed...
> > URL: http://gradsusr.org/pipermail/gradsusr/attachments/20121225/0f3c0bbd/attachment.html
> >
> > ------------------------------
>
> >
> > _______________________________________________
> > gradsusr mailing list
> > gradsusr at gradsusr.org
> > http://gradsusr.org/mailman/listinfo/gradsusr
> >
> >
> > End of gradsusr Digest, Vol 34, Issue 74
> > ****************************************
>
> _______________________________________________
> 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/20121226/4061926d/attachment-0003.html 


More information about the gradsusr mailing list