Understanding what sh_gamit does

1
Understanding what sh_gamit does
2
E-mail summary assessesover-all quality

• Number of stations
• Worst/best RMS scatter
• postfit normalized RMS from solve
• theoretically should be unity
• typically around 0.2
• (daily position scatter in accord with sigma)
• Large adjustments to site positions
• Any fatal messages

3
Output files

• Cleaning summary
• autcln.prefit.sum
• autcln.post.sum -- Final postfit run
• Solution files output by solve
• qexpta.doy
• human-readable
• oexpta.doy
• machine-readable
• hexpta.yyddd
• Full-precision for GLOBK suite

4
Status files common to all GAMIT/GLOBK runs

• File names, e.g, GAMIT.fatal
• One long line per event
• Three levels of severity
• Status
• Time of steps in processing
• Warning
• Items that might be in error
• Fatal
• Errors so large that the processing stopped

5
Example GAMIT.status

• STATUS 0201191201 6.0 ARC/aversn Started ARC,
  \Version 9.53 of 2001/01/03 103000 (Linux)
• STATUS 0201191201 6.0 ARC/arc Integrating \
• satellite 1 PRN 1
• STATUS 0201191201 6.0 ARC/arc
  Integrating\satellite 2 PRN 2
• ....
• To watch progress as it happens
• tail -f GAMIT.status

6
Analyzing a solution bias parameters

• Number of biases fixed
• Fixed bias have no sigmas in Q-file
• 608 B1L1 SMYC-TBLE 4- 2 0.0000000000
  10.0000 \
• 609B1L1 SMYC-TBLE 4- 5 0.0000000000
  2.3133 \ 0.0367 -2.31330402
• The means not fixed and sigma is given
• Adjustments
• parameters should not change much

7
Analyzing a solution adjustment sizes

• Adjustments should not be too large
• Convergence rate is 100 or 1000 to 1
• Suppose an adjustment of 1 m in position
• residual error could be 1-10 mm
• solution needs to be iterated
• Atmospheric delay variations
• Magnitude of gradients
• EOP adjustments
• Some examples (shown externally)

8
autcln output

• autcln.out.Z
• full output
• automatically compressed
• use uncompress or use zcat to view
• Lists actions taken by autcln
• autcln.prefit.sum
• autcln.post.sum
• Summaries
• Look at these

9
autcln summary files

• Clock range stats from range data
• Repeated twice since solution is iterated
• List of double difference bias flags
• Number of bias flags added to jumps in double
  differences
• If too many flags are added, then autcln removes
  the site/satellite combination completely. This
  can happen if a satellite burns. Rare if IGS
  orbits are used since satellite is removed from
  orbit files.
• Bad site coordinates can cause many bias flags

10
autcln summary file

• One-way post-fit residual statistics
• Only in postfit autcln summary
• Gives RMS in mm
• by station and each satellite for a station
• Normal values are 3-10 mm
• gt10 mm is not good
• sky plots show residuals
• cview views residual time series interactively
• reads C-files
• default is to delete these files during batch
  runs)

11
Autcln summary file

• AMS -- Averaged RMS of residuals
• statistics for 25-point averages of residuals
• theoretical white noise
• should be 5X smaller than for single-point value
• typically, though
• AMS is 2X smaller than single point RMS
• due to correlations in phase residuals
• ratio of RMS/AMS
• values to close to 1 mean systematic residuals

12
Autcln summary files

• NUM
• number of values used to compute RMS
• ATELV
• Elevation angle-dependence of RMS
• fit to data noise model of the form
• variance A2(B/sin(elev))2
• Sky-Plots
• residuals v. elevation angle
• show effects of multipath and antenna gain

13
Autcln summary data amounts

• Data counts by site and satellite
• N(good)
• N(deleted)
• N(bias flags)
• ignoring rising satellites

14
Bias flag report

• Reports number of bias flags added for
• GAP -- Due to gap in data
• JMP -- Large jump on phase-range difference
• ION -- Ionospheric delay jump
• DDS -- Double difference scanning
• WLS -- Wide lane scanning
• DDC -- Double difference jump while cleaning data