
           libUN : 'User level' NetCDF READ / WRITE routines
--------------------------------------------------------------------------

  Puropose : Simplify access to NetCDF (No need to manage 'identifiers')
  ---------- + enable writting of clear output routines.
             + no need for 'include' statements for NetCDF outside this.

  Main routines provided :
  ------------------------
    UNscreate : General file creation routine,
                   defining multiple dimensions + attributes

    UNcreate  : Simple file creation routine,
                   non-staggered (x,y,lev,time) grid

    UNwrite   : General variables writting routine
                   (also updates 'range' attribute and variable if present)
                   (Note: Use UNlwrite to write 2D planes in 3D variables)
                                -
    UNwratt   : Real attributes writting
                   (Not necessary for 'actual_range' attribute
                    updated by UNwrite)
    UNwcatt   : Characters attributes creation & writing
                   (Not necessary for 'units', written by UNscreate)
                   WARNING: this routine uses a temporary disk space
                            equal to the file length (duplicate the file)
                            (its use is NOT recommended)

    UNread    : Reading routine (grid coordinates + variable)

    UNsread   : Reading routine (variable only)
 

  Complementary routines :
  ------------------------
    UNwopen   : re-open file for writting
    UNropen   : open file for reading
    UNgtime   : get information about recorded
                  time steps -> easy reading

----------------------------------------------------------------------------
  For more INFORMATIONS, see the header of each routine in the file 
  ----------------------
                     libUN.f

  To compile the library, use one of files the
  -----------------------
                     complib_...

       or create your own complib_... AND NetCDF.inc_...
       using the given files as examples.

  Double precision variables in memory
  ------------------------------------
       If your code uses REAL*8 variables, you must use a version
       of libUN with REAL*8 arguments (at least for f77 compilation).
       This is available as libUN_dbl.f
       To obtain libUN_dbl from the standard libUN, type the following:
      
          cdfdouble
          fdouble libUN_dbl.f

       The first script creates or updates libUN_dbl with the 
       appropriate NetCDF lib calls, but still includes REAL
       declarations: the compiler must then specify that REAL should
       be *8 by default. The second script (fdouble) replace all REAL
       declarations by REAL*8. You may use this scripts with your
       other FORTRAN codes, but be aware that any 'REAL' or 'real'
       word preceded by 6 blanks or more will be replaced by
       REAL*8.

       IMPORTANT:
       This does NOT change the type of the variables in the file.
       UNscreate and UNcreate are always creating REAL*4 variables.

  Some small example(s) is (are) given in the directory /examples.
  ---------------------
    Those examples are NOT ready to use: you must adapt it to your needs.
    (but the library is, of course, ready to use once it is compiled)

