WRITE4D() and write4dc()

Fortran version:

    LOGICAL FUNCTION WRITE3( FNAME, VNAME, JDATE, JTIME, TSTEP, NRECS, BUFFER)
        CHARACTER*(*), INTENT(IN   ) :: FNAME     !  file name for query
        CHARACTER*(*), INTENT(IN   ) :: VNAME     !  vble  name   (or ALLVAR3 (='ALL'))
        INTEGER      , INTENT(IN   ) :: JDATE     !  starting date, formatted YYYYDDD
        INTEGER      , INTENT(IN   ) :: JTIME     !  starting time, formatted HHMMSS
        INTEGER      , INTENT(IN   ) :: TSTEP     !  time step,     formatted HHMMSS
        INTEGER      , INTENT(IN   ) :: NRECS     !  number of time step records to write
        <type>       , INTENT(IN   ) :: BUFFER(*) !  array holding output data

C version:

write3c() is a C wrapper calling the Fortran WRITE3()

    int write4dc( const char * fname ,
                  const char * vname ,
                  int          jdate ,
                  int          jtime ,
                  int          tstep ,
                  int          nrecs ,
                  void       * buffer )

Summary:

Returns .TRUE. (or 1) if the operation succeeds, .FALSE. (or 0) if it fails. For failure, writes a log message indicating the nature of the failure.

Preconditions:

FNAME must have one of the file types CUSTOM , GRIDDED , or BOUNDARY .

INCLUDE 'IODECL3.EXT' for Fortran, or #include "iodecl3.h" for C.

I/O API must already be initialized by a call to INIT3() .

FNAME and VNAME must have length at most 16.

FNAME must already have been opened by OPEN3() or open3c() .

FNAME must have one of the file types CUSTOM , GRIDDED , or BOUNDARY .

JDATE, JTIME, and TSTEP must be expressed in terms of Models-3 date and time conventions. JDATE:JTIME must be an exact positive integer multiple of the file's time step from the file's starting date and time. TSTEP must be an exact positive integer multiple of the file's time step.

Dimensionality of the BUFFER argument should agree with I/O API conventions for the dimensionality of the data being written. In particular,he time step record subscript is the "slowest" subscript in the buffer. (I.e., Fortran subscripting for BUFFER should be BUFFER( ..., [LAYER,], TIMESTEP-RECORD ).)

See Also:

• READ3
• READ4D
• WRITE3

Fortran Usage:

    INCLUDE 'PARMS3.EXT'
    PARAMETER ( NCOLS = ??, NROWS = ??, NLAYS = ??, NSTEPS = 10 )
    ...
    REAL  TGD( NCOLS, NROWS, NSTEPS )
    REAL  TA ( NCOLS, NROWS, NLAYS, NSTEPS )

C            NOTE:  It isn't required that the name of the Fortran
C            variable match the name of the file-variable, but it
C            can often help the maintainability of the code if it does.
    ...
    IF ( WRITE4D( 'AFILE', 'TGD', 1988021, 123000, 10000, NSTEPS, SO4 ) ) THEN
C            TGD time step sequence starting 12:30:00 Jan 21, 1988 
C            and having 10 steps at hourly intervals written to AFILE
        ...
    ELSE
C           Error:  see program log for further info.
        ...
    END IF
    ...
    IF ( WRITE3( 'BFILE', 'TA', 1988021, 123000, 10000, NSTEPS, METS ) ) THEN
C            TA time step sequence starting 12:30:00 Jan 21, 1988  
C            and having 10 steps at hourly intervals written to BFILE
        ...
    ELSE
C           Error:  see program log for further info.
        ...
    END IF
    ...

C Usage:

    ...
    #include "iodecl3.h"
    ...
    float  ta[ NSTEPS ][ NLAYS ][ NROWS ][ NCOLS ] ;
    ...
    if ( write4dc( "BFILE", "ALL", ALLAYS3, 
                  1988021, 123000, 10000, NSTEPS, ta ) )
        {
            /*  TA time step sequence starting 12:30:00 Jan 21, 1988
                and having 10 steps at hourly intervals written to BFILE */                         */
        ...
        }
    else
        {
            /*  Error:  see program log for further info.  */
        ...
        }
    ...

Up: I/O API: Public Routines

To: Models-3/EDSS I/O API: The Help Pages