[RegCNET] Re: Clarification about last phase... cru.param
Moetasim
mashfaq at purdue.edu
Wed Feb 22 03:00:25 CET 2006
Hi Svetlana,
Well CRU code actually regrids CRU data which is at half degree lat-lon grid
to your RegCM grid/resolution (which uses map projection). While doing
comaprisons between observations and model output, you need both datasets at
same grid.
Ok lets look at any of cru.param file:
This portion requires the idatecru1 and idatecru2 from user. Leave idatecru0
as it is and replace rest of two dates with the dates for which you have
processed RegCM output in postproc step. The cru data is monthly so it would
require you to input only year and month (no day and hour).
--------
parameter(idatecru0=190101)
parameter(idatecru1=198901)
parameter(idatecru2=198912)
--------
This is the second portion which requires user input. You need to provide the
names of the output files, which you will get after running CRU script.
---
c NAME OF THE CRU FILES ON THE RegCM GRID.
character outfile(nfld)*50
DATA (outfile(ifld),ifld=1,nfld)
& /'PRE1989.nc'
& ,'TMP1989.nc'
& ,'DTR1989.nc'
& ,'VAP1989.nc'
& ,'CLD1989.nc'/
----
You may give any name to your output files as in this case, its PRE1989.nc
etc... Its not necessary that you process all five CRU files. If you are only
interested in processing CRU Precp file then you may remove rest of the names.
This portion would look like somthing as under if you choose 'PRE1997.nc' name
as your PRECP output file:
---
c NAME OF THE CRU FILES ON THE RegCM GRID.
character outfile(nfld)*50
DATA (outfile(ifld),ifld=1,nfld)
& /'PRE1997.nc'/
----
Also remember that there is one variable defined in the portion before this
one as:
PARAMETER (nfld=5)
This actually tells that how many fields/files you are going to process. In
this case, you have only one file so you should change nfld from nfld=5 to
nfld=1
The third portion which requires user input is as following:
----
c NAME OF THE GLOBAL INPUT CRU FILES
character infile(nfld)*50
DATA (infile(ifld),ifld=1,nfld)
& /'/home/RAID2-D1/CRU/TS2.02/CRUPRE.CDF'
& ,'/home/RAID2-D1/CRU/TS2.02/CRUTMP.CDF'
& ,'/home/RAID2-D1/CRU/TS2.02/CRUDTR.CDF'
& ,'/home/RAID2-D1/CRU/TS2.02/CRUVAP.CDF'
& ,'/home/RAID2-D1/CRU/TS2.02/CRUCLD.CDF'/
----
This actaully needs information of the path where you have your CRU data. If
its in same folder where you are running CRU script then you should just
mention the file names but if its somewhere else then you should set the paths
accordingly. e.g. in this case when you are only processing CRU PRECP, you
only need to provide its file path. if you have CRU data
in /home/svetlana/RegCM/
then you should replace
'/home/RAID2-D1/CRU/TS2.02/CRUPRE.CDF' with
'/home/svetlana/RegCM/CRUPRE.CDF'
The overall changed portion will look like:
c NAME OF THE GLOBAL INPUT CRU FILES
character infile(nfld)*50
DATA (infile(ifld),ifld=1,nfld)
& /'/home/svetlana/RegCM/CRUPRE.CDF'/
---
The last potion which may or may not need your modification is
--
c NAME OF THE CRU VARIABLES
character vnam(nfld)*10
DATA (vnam(ifld),ifld=1,nfld)
& /'PRE'
& ,'TMP'
& ,'DTR'
& ,'VAP'
& ,'CLD'/
---
In current example, we are only processing 'PRE' variable so you should change
this portion as follwoing
c NAME OF THE CRU VARIABLES
character vnam(nfld)*10
DATA (vnam(ifld),ifld=1,nfld)
& /'PRE'/
You should ignore the portion of the file, which talks about CRU grid.
Moet
Quoting ?????? <dolgikh at meteo.kz>:
> Dear Moet.
>
> Thanks a lot for previous answer. It really helped a lot. The problem was in
>
> 'end date' param.
>
> But for now we face some new misunderstandings.
> 1) We ran postproc.x and received 4 output files: ATM199703AVG.nc,
> ICBC199703AVG.nc, RAD199703AVG.nc, SRF199703AVG.nc.
> 2) As described in manual we should change two params in cru.param (dates),
> and we have changed them.
> 3) But after that we should change input "ORIGINAL CRU GRID" (does it mean 5
>
> files which located on web-site???), and how do we do it?
> 4) NAME OF THE CRU FILES ON THE RegCM GRID - does it mean the output files
> which we received from previous stage? And we only had 3 files (ATM, RAD,
> SRF) but in cru.param there are 5 (PRE, TMP and so on). Question is how we
> adjust output file names.
> 5) And I would really really appreciate it if you could explain a little
> bit about contents of cru.param (ORIGINAL CRU GRID, NAME OF THE CRU FILES ON
>
> THE RegCM GRID, NAME OF THE GLOBAL INPUT CRU FILES, NAME OF THE CRU
> VARIABLES)
>
> THANK YOU IN ADVANCE
> look forward to hearing from you
> best regards
> Svetlana
>
> ----- Original Message -----
> From: "Moetasim" <mashfaq at purdue.edu>
> To: "Ð"олгиÑ." <dolgikh at meteo.kz>
> Cc: <regcnet at lists.ictp.it>
> Sent: Tuesday, February 21, 2006 10:57 AM
> Subject: Re: [RegCNET] Problems with poswt processing
>
>
> >
> > Hi,
> > I am not exactly sure that which error you are refering me to..
> > If its this error :
> > ncvarid: ncid 5: Variable not found
> > ncdimid: ncid 5: Invalid dimension id or name
> > ncvarid: ncid 5: Variable not found
> > ncvarid: ncid 5: Variable not found
> > ncvarid: ncid 5: Variable not found
> > then just ignore it. Postproced data wont have any problem. Jason recently
> > replied for such an email over RegCNET.. Postproc uses old netcdf
> > fucntions
> > and needs replacement of these functions with new ones...
> > And if you are refering to this error
> > --
> > *ERROR* in putdef in grbtst
> > ---
> > then in case you have reached to your end date of postprocessing then
> > ignore
> > it as well. I guess that output file will be alright. But, if you are
> > still in
> > middle of postprocessing then I just advise you to re-run postproc or run
> > it
> > by slightly changing initial dates..it should disappear.. (BTW, there is
> > no
> > logic behind this).
> >
> > If you are still unable to get a successful postprocessing of data then I
> > would suggest you to send the complete output from the beginning
> > of ./postproc.x
> >
> > Moet
> >
> >
> >
> >
> > Quoting Ð"олгиÑ. <dolgikh at meteo.kz>:
> >
> >> Dear Moet,
> >> thank you for previous answers, but now we have a little problem with
> >> post
> >> processing.
> >> We calculated the data through XRegCM and completed the process
> >> successfully.
> >> But after that we tried to average the data through the post processing
> >> script and recieved following errors (attached error.txt have a look
> >> please).
> >>
> >> Looking forward to hearing from you.
> >> Thanks in advance
> >> Svetlana
> >>
> >>
> >>
> >> ----- Original Message -----
> >> From: moetasim
> >> To: ??????
> >> Sent: Thursday, August 25, 2005 9:09 PM
> >> Subject: [RegCNET]Re: Problems with RegCM3
> >>
> >>
> >> Hi Svetlana,
> >>
> >> Well its not problem from RegCM3 but due to some issue of IFC9 at some
> >> linux plateforms. Let me refer you to the weblink:
> >>
> >>
> >> http://softwareforums.intel.com/ids/board/message?
> > board.id=11&message.id=2693&view=by_date_ascending&page=1
> >>
> >>
> >> My suggestion is that: its better to use a previous version of IFC
> >> (IFC7 or
> >> IFC8) compiler instead of going for any solution written at this weblink
> >> until INTEL people fix it by themselves.
> >>
> >> MOet
> >>
> >>
> >>
> >>
> >> > Dear Moet,
> >> > thank you for quick reply. It was very useful for us.
> >> > $1
> >> > We have run scripts in csh shell, setting ibyte=1, ntypec=60 in
> >> domain.param and using Makefile_IFC8 for compilation, but we still find
> >> error:
> >> > > IPO link: can not find "("
> >> > > ifort: error: problem during multi-file optimization compilation
> >> (code
> >> 1)
> >> > > make: *** [terrain] Error 1
> >> > > ./terrain.x: line 16: ./terrain: No such file or directory
> >> >
> >> > We hope for your further help,
> >> >
> >> > Svetlana
> >>
> >> > ----- Original Message -----
> >> > From: moetasim
> >> > To: äÃÃOÃ?Ã?Ã^ ; regcnet at lists.ictp.it
> >> > Sent: Wednesday, August 24, 2005 8:34 PM
> >> > Subject: Re: [RegCNET] Problems with RegCM3
> >> >
> >> > Hi,
> >> > >
> >> > > Could you please inform us with contact info of persons who can
> >> help
> >> with running of RegCM3.
> >> > -----
> >> >
> >> > I CAN HELP!
> >> > -----
> >> > >
> >> > > We have installed RegCM3 (downloaded from your web-site:
> >> http://www.ictp.trieste.it/~pubregcm/RegCM3/) and have some problems with
> >> starting the MODEL, particularly with Pre-Processing phase (Terrain.x for
> >> example). When we running the script terrain.x we have some errors:
> >> > > the system couldn't find some commands like "setenv", "ifc".
> >> >
> >> > -----
> >> > Well, RegCM3 scripts are written for csh shell so whenever you try
> >> to
> >> run these in bash or anyother shell, such errors are encountered. Its
> >> advisable that you change to csh before running these scripts. This can
> >> be
> >> done by typing 'csh' in terminal.
> >> > ------
> >> >
> >> > >
> >> > > Model installed on system:
> >> > > Operating System: Red Hat Linux 10 (Fedora Core 4)
> >> > > Intel Fortran Compiler 9
> >> > >
> >> > -----
> >> > I guess that ibyte setting for IFC-8 and IFC-9 should be same so
> >> set
> >> ibyte=1 in domain.param if you have not done it already.
> >> > ------
> >> >
> >> > > Questions:
> >> > > 1) Could you provide us some info about system configuration
> >> (necessary software, LINUX ver ???, Intel Fortran Compiler ver ??? or
> >> something more)
> >> > > 2) and what DATA was needed to be downloaded from the web site to
> >> successful work of the model. We are downloaded only 8 file for NNRP1
> >> (T,H,RH, etc) for 1997 and for SURFACE - GTOPO30_60MIN, GLCC60MIN_BATS,
> >> and
> >> OIIST - OISST_198112_200302, OISST_198112_200302.
> >> > ----------
> >> >
> >> > If you have IFC/PGI compiler then normally you need not to have
> >> anyother special software to run RegCM3. When you have downloaded 60 min
> >> dataset then I hope you have made adjustments accordingly in
> >> domain.param.
> >> Set ntypec=60 if you have not already done so.
> >> > ------------
> >> > >
> >> > > listing with errors in Terminal:
> >> > >
> >> > > ./terrain.x
> >> > > ./terrain.x: line 1: setenv: command not found
> >> > > rm -f *.o terrain
> >> > > ifort -tpp7 -g -cm -w -w90 -w95 -L../../Commons/env/liblinuxold
> >> -lnetcdf -Vaxlib -c -o terrain.o terrain.f
> >> > > ifort -tpp7 -g -cm -w -w90 -w95 -o terrain terrain.o
> >> -L../../Commons/env/liblinuxold -lnetcdf -Vaxlib
> >> > > IPO link: can not find "("
> >> > > ifort: error: problem during multi-file optimization compilation
> >> (code 1)
> >> > > make: *** [terrain] Error 1
> >> > > ./terrain.x: line 16: ./terrain: No such file or directory
> >> >
> >> > ----
> >> > Now come to this problem:
> >> > You have downloaded IFC-9... Remember that Makefile_IFC7 can't work
> >> with this compiler version. You must use Makefile_IFC8.
> >> >
> >> > ----
> >> >
> >> > >
> >> > > we have changed "ifc" to "ifort" command, because IFC 9.0 is
> >> worked
> >> with this command right now to load a IFC driver.
> >> >
> >> > ---------
> >> > If even after running scripts in csh shell, setting ibyte=1,
> >> ntypec=60
> >> in domain.param and using Makefile_IFC8 for compilation, you still find
> >> error
> >> then don't hesitate to contact again.
> >> > ------
> >> > MOet
> >> >
> >> >
>
More information about the RegCNET
mailing list