Thursday, April 10, 2008

Sad at CMONOC Rerun V2 result.


Platform:
  • GPSDC - Dell Precision 390, 1G RAM, Solaris 10 X86, GAMIT/GLOBK 10.32 (????) with Sun Studio 12;
  • GPSAC - Dell Optiplex 755, 4G RAM, RHEL-5_X64, GAMIT/GLOBK 10.32 (????) with Intel Fortran/C Compiler 10.
*Note: The version of GAMIT/GLOBK is different in two systems. The parameters may vary in different sessions.

The data are split into four sessions.
  1. 2000-2001 (gpsac);
  2. 2002-2003 (gpsac);
  3. 2004-2005 (gpsdc);
  4. 2006-2007 (gpsdc).
It took those tow computers one week to complete the solutions.

The resulting time series (combined with SIO igs1/igs2/igs3) have large offsets at the year boundaries of 2002.0, 2004.0, and 2006.0 epochs, especially for North component.

Why those jumps?

5 comments:

Enod said...

The year-boundary offsets have nothing to do with globk/glorg processing step. I re-ran all the combination by using the same software and parameters, there are still obvious large jumps. And strange, the other solutions in igs1/igs2/igs3 (say, jplm/usud) also has the same problems. It must be an error in the combination step.

Enod said...

Yeah. I made a mistake again. The newly generated time series is in GPSDC, and I looked for it in GPSAC, where the old solutions were stored. Thun, the re-run time series are OK now. There are no more jumps between some years.

God, I can't believe I lost another 30 min.

Enod said...

Large jumps:
2005.5959
2006.0370
The other parts seem normal.

Enod said...

CMONOC: there are no data for the above time span: 2005.5959-2006.0370.

Enod said...

The cause is: I accidentally modified the tables/sites.defaults when there was an error during the reprocessing. But forgot to revise it back after I solved the problem. Thus, those CMONOC sites are missing from sites.defaults for most of the days of year 2005.