Recipe: Fortran program to read data from downloaded file

From GEOS-5
Revision as of 12:09, 28 October 2014 by Pchakrab (talk | contribs) (Discussions)
Jump to navigation Jump to search

Problem

We want to read a downloaded data file using Fortran.

Solution

For the purpose of this example, we assume that we have already downloaded the file c1440_NR.inst01hr_3d_T_Cv.20060918_0900z.nc4 from the ftp server. For more information about file naming conventions, and how to download a file from the ftp server, please follow the links in the #See Also section.

Here we use the NetCDF-4 library to read the downloaded file (since NetCDF-4 uses HDF-5 as the underlying data format, HDF-5 library can also be used directly). First, we ensure that out NetCDF-4 library has been built with Fortran and HDF5 support. If both the queries

> nc-config --has-f90
> nc-config --has-hdf5

return "yes", we have a compatible NetCDF-4 library. Here nc-config is a utility bundled with NetCDF-4 package.

Code
program g5nr_reader_global

  use netcdf           ! for reading the NR files

  implicit none

  ! File name
  ! ---------
  character(len=256) :: T_file

  ! 4D array: (lon,lat,lev,time)
  ! ----------------------------
  real, allocatable :: T(:,:,:,:)    ! global data

  ! Miscellaneous
  ! -------------
  integer :: ierr
  integer :: im, jm, lm
  integer :: ncid, varid
  integer :: start(4), count(4)

  ! file name and dimensions
  ! ------------------------
  im = 720
  jm = 361
  lm = 72
  T_file = "c1440_NR.inst01hr_3d_T_Cv.20060918_0900z.nc4"

  ! Open file and get var id
  ! ------------------------
  call check(nf90_open(T_file,NF90_NOWRITE,ncid), "opening T file")
  call check(nf90_inq_varid(ncid,"T",varid), "getting T varid")

  ! Read global temperature data
  ! ----------------------------
  allocate(T(im,jm,lm,1))            ! global 4D array with 1 time level
  start = [1, 1, 1, 1]
  count = [im, jm, lm, 1]            ! 1 time level, 3D (lon,lat,lev) array
  write(*,*) 'Reading global T...'
  call check(nf90_get_var(ncid,varid,T,start=start,count=count), "reading T")
  write(*,*)'T: ', maxval(T),minval(T)

  ! close file, release memory
  call check(nf90_close(ncid), "closing T file")
  deallocate(T)

  ! All done
  ! --------

contains

  subroutine check(status, loc)

    integer, intent(in) :: status
    character(len=*), intent(in) :: loc

    if(status /= NF90_NOERR) then
       write (*,*) "Error at ", loc
       write (*,*) NF90_STRERROR(status)
    end if

  end subroutine check

end program g5nr_reader_global
Compile and link

We use the utility nf-config (included in the NetCDF-4 installation) to identify the linking rules.

For a typical NetCDF-4 installation, the above code, g5nr_reader_global.f90 can be compiled and linked to the NetCDF-4 library via

gfortran -o g5nr_reader_global.x `nf-config --fflags` g5nr_reader_global.f90 `nf-config --flibs`

creating the executable g5nr_reader_global.x.

NOTE:

  1. You can use your favorite Fortran compiler instead of gfortran.
  2. If the NetCDF-4 library was built with parallel I/O support, you will need to use mpif90 to link, even if your code does not use the MPI library.
Run

Running the executable

> ./g5nr_reader_global.x

produces the output

Reading global T...
T:    315.6512       180.3667

Discussions

The above Fortran code can be easily modified to read a subset of the temperature data instead. If we want to read temperature inside the box bounded by latitudes 25oN, 50oN and longitudes -130oN, -65oN, we need to modify the values of the dimensions im, jm and the argument start of the call to nf90_get_var.

Our new dimensions are

! array indices from boundary lat/lon
imin = nint((-130. + 180.)/0.5)
imax = nint((- 65. + 180.)/0.5)
jmin = nint((  25. +  90.)/0.5)
jmax = nint((  50. +  90.)/0.5)

! array sizes
im = imax-imin+1
jm = jmax-jmin+1

The new start argument to nf90_get_var defining the first indices of the slice of array to be read, is

start = [imin, jmin, 1, 1]

Note that, although

See Also

  1. Recipe: File naming conventions
  2. Recipe: Retrieve (global) data from FTP server

No Warranty

Copyright