JPL DE406 ephemeris ASCII files

A few days ago, the JPL posted ASCII versions of the DE406 data files.
Finally, us non-Unix people should now have access to the DE406 long
span ephemeris.

ftp://ssd.jpl.nasa.gov/pub/eph/export/ascii/

But there's one big problem: I can't get it to work under Windows XP
Home. When I run the concatenated text file (header.406 + 100-year
ephemeris file) through asc2eph, it reports 0 records written (but no
error message), and the JPLEPH output file is only a few k. It seems
to have the header info only.

I have no trouble running the same process with DE200 and DE405 ASCII
files. My binary output files pass the testeph test routine
beautifully. (I compiled all the Fortran code with MinGW, a Windows
port of the free GCC GNU compiler.) So I don't understand why I can't
get DE406 to work. Has anyone successfully performed the DE406 ASCII
to binary conversion?

--
Paul Hirose
To reply by email remove INVALID

Paul Hirose wrote:

A few days ago, the JPL posted ASCII versions of the DE406 data files.
Finally, us non-Unix people should now have access to the DE406 long
span ephemeris.

ftp://ssd.jpl.nasa.gov/pub/eph/export/ascii/

Curious -- I downloaded the DE-406 from JPL many years ago. Converted,
it's a 200MB file, which was a mind-boggling amount of data at the
time. Just the other day, though, I pulled the Fedora Core 4 ISO
images ... 2.7GB in a few hours. Spooky.

Has anyone successfully performed the DE406 ASCII
to binary conversion?

I wrote my own conversion program, so it worked fine. How much is this
"asc2eph" program dependent on the presence of nutation/libration data
in a 'full' ephemeris? The DE-406 (or at least the version I
downloaded) doesn't have this data.

I wrote in message
ink.net...

But there's one big problem: I can't get it to work under Windows XP
Home.

I found the problem. It's the format of the ephemeris text files. The
column widths and number of decimal places in the DE406 files are
different from the DE200 and DE405 files. This formatting is hardwired
into the Fortran code, so the program fails when it doesn't find the
numbers in the expected columns.

The temporary solution is to change a few lines in ASC2EPH.

1. In the main routine, right before the big do-loop, change

READ (*,'(3D26.18)',IOSTAT =IN) (DB(K),K=1,NCOEFF)

to

READ (*,'(3D23.15)',IOSTAT =IN) (DB(K),K=1,NCOEFF)

The only difference is 3D23.15 instead of 3D26.18.


2. At the bottom of the do-loop, change

READ (*,'(2I6,3000(/3D26.18))',IOSTAT =IN)
.. NRW, NCOEFF, (DB(K),K=1,NCOEFF)

to

READ (*,'(2I6,3000(/3D23.15))',IOSTAT =IN)
.. NRW, NCOEFF, (DB(K),K=1,NCOEFF)


After those changes, I got a clean conversion from ASCII to binary.
The resulting ephemeris gave a perfect run with the TESTEPH program.

I think a better solution is for JPL to make the DE406 ASCII file
format compatible with ASC2EPH. Or, modify ASC2EPH so its input column
positions aren't hard-wired.

--
Paul Hirose
To reply by email remove INVALID

After successfully converting DE406 ASCII files to binary, I ran into
another difficulty. To merge the 1900-1999 and 2000-2099 binaries, I
needed the BINMERGE utility. But the GCC compiler said in this line of
the Fortran source code was illegal:

READ(nfil,REC=NREC,END=6,ERR=98)(DATA(K),K=1,NDATA )

Not knowing Fortran, I had to do some research on the web. It seems
the compiler is right. The END= isn't allowed when reading a
direct-access file. My solution was to delete that part, and use ERR=
to exit the loop:

READ(nfil,REC=NREC,ERR=6)(DATA(K),K=1,NDATA)

I commented out these lines:

98 write(*,'(/a18,i3)')'read error on unit',nfil
stop

After that, the program compiled and executed ok. The merged 1900-2099
binary output ran through the TESTEPH test program perfectly on my
Windows machine.

--
Paul Hirose
To reply by email remove INVALID

In a private communication, someone at JPL suggested a better fix for
the nonstandard Fortran in BINMERGE. Replace

READ(nfil,REC=NREC,END=6,ERR=98)(DATA(K),K=1,NDATA )

with

READ(nfil,REC=NREC,IOSTAT=ISTAT)(DATA(K),K=1,NDATA )
IF (ISTAT .EQ. -1) THEN
GO TO 6
ELSE IF (ISTAT .NE. 0) THEN
GO TO 98
END IF

Leave the rest of the program unchanged.

My correspondent said the BINMERGE source code online at the FTP site
will be rewritten to eliminate this problem.


The other problem, the modification needed by the ASCII conversion
program in order to process DE406 files, seems to be history. An
improved ASC2EPH now available for download seems to be a universal
converter. At least, it checks out ok with DE405 and DE406 on my
Windows system.

--
Paul Hirose
To reply by email remove INVALID