Re: [ferret_users] reading FORTRAN sequential access data - simplequestion

["William S. Kessler" <William.S.Kessler@xxxxxxxx>]

Just to establish the more elementary piece that may be useful to  
other users: Ferret easily and straightforwardly reads fortran binary  
writes if the file is NOT 'sequential access'.

In the fortran code:

	parameter (nx=154, ny=35)		! 141E-66W, 34S-0 by 1 deg

	real uav(nx,ny),vav(nx,ny)		! input data
	real curl(nx,ny),psi(nx,ny)		! output
.....
	open(30,file='ir4.dat',status='new',form='unformatted')
.....
	write(30) ((psi(i,j),curl(i,j),uav(i,j),vav 
(i,j),i=1,nx),j=1,ny)    ! write 4 grids on one record

In Ferret:

define axis/x=141e:66w:1/unit=lon xpsi
define axis/y=34s:0:1/unit=lat ypsi
define grid/x=xpsi/y=ypsi gpsi
file/format=unf/var="psi,curl,uav,vav"/g=gpsi/col=`4*154*35`  
ir4.dat   ! read 4x(154,35) grids on one record
save/file=ir4.cdf psi,curl,uav,vav    ! Ferret converts the fortran  
binary to netcdf

The main reason to do this is that Ferret is such a good netcdf  
converter. I could write a subroutine to write the output of my  
fortran code in netcdf directly. But it is much easier to write the  
above 5 lines of extremely transparent Ferret code to accomplish the  
same thing. Then I would delete the fortran output ir4.dat.

Thus, without the complication of sequential access, there is no  
issue of counting record information or anything else. It is very  
easy. This is also very well tested as I have done this sequence  
(fortran=>binary output=>Ferret conversion to netcdf) many, many  
times over a decade.

Billy K