Sudheer,
I think you already found the correct solution, download
netcdf_check.f90 from my FTP directory and add that to your compile
line.
There are probably two different copies of the user subroutine
"check", your version and my version. They probably do exactly the
same thing. If you want to use your version only, simply delete one
line "use netcdf__check" in the file getnc_dims.f90, and do not
include netcdf_check.f90 on the compile line. This may be the
simplest thing for your purposes. It is also okay to use them both
in the same compile, because my version is inside a module.
Be careful about one other thing. Many fortran 90+ compilers depend
on the order of compilation of modules and the creation of *.mod
files. If a subroutine says "use netcdf__check", then the module
netcdf__check must have previously been compiled. This depends on
the order of source files on the command line.
So in your case, list the source files in order of increasing
dependency, main program last, for best results:
ifort -extend-source -O3 -o ncep2ascii.x -I/usr/local/include \
netcdf_check.f90 getnc_dims.f90 ncep2ascii.f90 \
-L/usr/local/lib/ -lnetcdf
See if that works for you. There are other methods to manage the
chain of dependencies, but this a simple method that I use.
--Dave
On 6/13/2010 11:41 PM, Sudheer Joseph wrote:
Thank you Dave& Pierce,
Sorry about that confusion, Yes it
is possible for allocating an array with 2 dimenstion as parameters
and last dimesion as variable. What I as facing was some error with
the library. I have a question regarding your modules "getnc_dims.f90
netcdf_check.f90". After seeing that they are very useful modules I
have added them to my include path ie under directory along with
netcdf.mod as below
sjo@sjo-desktop:~/hdd4/NRL_file_gen$ ll /usr/local/include/*.mod
-rw-r--r-- 1 sjo sjo 4272 2010-06-14 10:40
/usr/local/include/getnc__dims.mod
-rw-r--r-- 1 sjo sjo 562 2010-06-14 10:40
/usr/local/include/netcdf__check.mod
-rw-r--r-- 1 root root 277339 2010-06-03 18:57 /usr/local/include/netcdf.mod
-rw-r--r-- 1 root root 1510 2010-06-03 18:57 /usr/local/include/typesizes.mod
but when I complie by
ifort -extend-source -O3 -o ncep2ascii.x -I/usr/local/include
ncep2ascii.f90 getnc_dims.f90 -L/usr/local/lib/ -lnetcdf
it gives error as below
sjo@sjo-desktop:~/hdd4/NRL_file_gen$ ifort -extend-source -O3
-I/usr/local/include ncep2ascii.f90 getnc_dims.f90 -L/usr/local/lib/
-lnetcdf/tmp/ifortEEeYgg.o: In function `MAIN__':
ncep2ascii.f90:(.text+0x30c): undefined reference to `netcdf__check_mp_check_'
ncep2ascii.f90:(.text+0x55e): undefined reference to `netcdf__check_mp_check_'
ncep2ascii.f90:(.text+0x595): undefined reference to `netcdf__check_mp_check_'
ncep2ascii.f90:(.text+0x5cc): undefined reference to `netcdf__check_mp_check_'
ncep2ascii.f90:(.text+0x667): undefined reference to `netcdf__check_mp_check_'
/tmp/ifortEEeYgg.o:ncep2ascii.f90:(.text+0x6f8): more undefined
references to `netcdf__check_mp_check_' follow
but if i compile by
ifort -extend-source -O3 -o ncep2ascii.x -I/usr/local/include
ncep2ascii.f90 getnc_dims.f90 netcdf_check.f90 -L/usr/local/lib/
-lnetcdf
it works with out problem
So are these modules not supposed to be used similar way as netcdf.mod
? which can be used in any code by including its path during
compilation? or what change I need to make them to use them like
netcdf.mod?
I have attached here the code where I am trying to use these modules.
please advice.
with best regards,
Sudheer
On Fri, Jun 11, 2010 at 10:06 AM, Sudheer Joseph<sjo@xxxxxxxxxxxxx> wrote:
Thank you Dave,
The subroutine you mentioned is very
useful. But my question was slightly different. suppose if you get the
dimension size before reading the 3 dimensional variable. we cannot
use that information to read the netcdf 3rd dimension at a stretch.
I mean
if I use the
NTIM= nf_get_dim(time) ( just for illustration sake )
and use this for allocating the dimension for 3D variable it will not
accept as it is not defined as a parameter.
allocate(pres_in(NLONS, NLATS, NTIM))
count = (/ NLONS, NLATS,NTIM/)
start = (/ 1, 1, 1/)
One possibility came to my mind yesterday night was to read the data
as xy slabs and use a do loop to read the time dimension. It appears
there is no way to read the 3D variable at one call if we are not
defining the dimension as parameter.
with best regards,
Sudheer
On Fri, Jun 11, 2010 at 3:42 AM, Dave Allured<dave.allured@xxxxxxxx> wrote:
Sudheer,
If the names of the dimensions are always the same, then use
nf90_inq_dimid followed by nf90_inquire_dimension to get the length
of each dimension. See the example on this page:
http://www.unidata.ucar.edu/software/netcdf/docs/netcdf-f90/NF90_005fINQUIRE_005fDIMENSION.html
If all the dimensions have coordinate variables, then the following
subroutine will get several dimension lengths in one call:
http://www.esrl.noaa.gov/psd/people/dave.allured/data/f90/lib/getnc_dims.f90
If the dimension names change between Netcdf files, then the problem
is more complicated. We can address that if neither of the above
works for you.
--Dave A.
NOAA/PSD/CIRES
On 6/10/2010 7:55 AM, Sudheer Joseph wrote:
Dear Users,
I have been coding a program to read multiple
netcdf files having different time dimensions. I want to know if there
is a way to read netcdf varibale with out defining the dimension as
parameter. with in fortran 90 program.
normally we need to define like below
! timesteps of data.
integer, parameter :: NDIMS = 3, NRECS =1
integer, parameter :: NTIM=1464, NLATS = 94, NLONS = 192
allocate(pres_in(NLONS, NLATS, NTIM))count = (/ NLONS, NLATS,NTIM/)
start = (/ 1, 1, 1/)
! Read the surface pressure and temperature data from the file, one
! record at a time.
call check( nf90_get_var(ncid, pres_varid, pres_in, start =
start,count = count) )
In my case ntim may be different for different files. I wanted to use
the reading part as a subroutine where ntim is provided as a variable
and passed to subroutine
but if I declare ntim as varible then netcdf gives an error.
If any one have a solution please help. I am not looking for solutions
like replacing the parameter statement using shell script and re
compiling it etc.
_______________________________________________
netcdfgroup mailing list
netcdfgroup@xxxxxxxxxxxxxxxx
For list information or to unsubscribe, visit:
http://www.unidata.ucar.edu/mailing_lists/
