Tuning AUTCLN for editing

1
Tuning AUTCLN for editing
2
Autcln basics

• Controlled by the autcln.cmd file
• The defaults in the program are set for data from
  modern receivers
• These default values need not be included in the
  command file
• Post-fit cleaning is primary exception
• Can be "tuned" extensively
• Very useful for data from older receivers

3
Basic aims and operation

• Aims
• Generate a set of one-way phase time series in
  which all cycle slips are either
• corrected (with integer step functions)
• flagged (to introduce real-valued offset)
• Bias flags are used for cycle slips that cannot
  be resolved to integer values
• Generate one-way phase residuals that can be used
  for data weighting

4
How autcln works

• (1) Generate models for receiver and satellite
  clocks
• Measure clock offset epoch-by-epoch using
  pseudorange data
• Align offsets with the polynomial estimated from
  the K- and J-files.
• (2) Identify and flag possible jumps in phase
  data
• using wide-lane (WL) when available
• and/or ionospheric-delay (LG) phase

5
How autcln works(II)

• (3) Align the phase data with the range clock
  solution
• look for large slips in the phase data
• look for jumps in the ionospheric delay
• (4) Scan selectively double differenced phase
• Look for jumps
• Attribute each DD jump to 1 sat and 1 station
• (5) Iterate steps 1-4 to improve estimates
• Phase data set now bias flags at each possible
  slip and after gaps

6
How autcln works(III)

• (6) Remove short segments
• (Actually, they are assigned a flag to unweight
  them.)
• (7) Estimate the number of cycles missing
  (jumped) at each bias flag
• by polynomial fits to surrounding data
• use good data from other stations and satellites.
  
• determine integer values and their reliability

7
How autcln works(IV)

• (7) Iterate step 7 using previously-cleaned data
• (8) Estimate receiver and satellite clock offsets
  again
• use the phase data in an iterative process
• generate new one-way phase residuals
• (sum for each station and satellite is zero)
• (9) Write clean C-files
• preserve input C-files, OR
• over-write intermediate C-files

8
autcln command file

• Go to your computer to examine the distributed
  autcln.cmd file
• Note the use of the POST option
• Which options would you change for older
  code-less data?
• Using sh_gamit makes 2 summaries
• autcln.cmd.prefit AND
• autcln.cmd.posfit

9
Checking autcln.sum files

• Postfit scatter
• Usually RMS lt 10 mm
• Discarded data
• Usually lt 10 data discarded
• More than 20 discarded indicates problem
• Missed cycle slips are now very rare
• Most problematic slip is
• 1 cycle in L1 AND 1 cycle in L2
• does not change WL

10
Why are ALL the data from my favorite station
deleted?

• This question is number 1 FAQ
• Bias flags too close together
• Post-fit phase RMS too large
• Check editing reports in autcln.sum
• Bad apriori coordinates
• Example /dell3/tah/swis/gps00/1999_149/autcln.post
  .sum

11
Autcln commands

• Read the HELP_DIR/autcln.hlp file
• Also output by just typing autcln

12
Tuning autcln

• dd_fit_tol
• sets the sizes of jumps to be flagged as
• ratio to rms
• min size
• max size
• ion_jump
• similar parameters for ionspheric jumps
• trim_oneway
• sets minimum sizes for segments of data as
• time
• number of epochs
• ratio to good length
• min at end

13
Summary

• Default autcln.cmd usually works fine
• Tuning is useful for
• anomalous data
• some old data
• If you need to tune
• delete the link in the day dir -gt tables dir
• copy the autcln.cmd file to the day dir
• modify the autcln.cmd file in the day dir
• re-run sh_gamit (or bexpt?.bat) for that day