Go to the list of seismic processes.      Go to SIOSEIS introduction.
         CATS: Controlled Archeology Test Site

     The objective of this project is to image and locate buried
objects 1 to 2 meters deep.  Three Geometrics Strataviews were used
to record several different seismic experiments with several different
seismic sources.

photographs of the data acquistion

               The Reflection Survey
               --- ---------- ------

     The three Geometrics had a total of 108 recording channels which
in turn had three different type of Mark Product geophones;
48 40Hz. phones, 48 12Hz. phones, and 12 horizontal phones.
     The 48 trace Stratview (the unit was rented from Geometrics, so the
log book calls it the "Geometrics") was set with recording parameters of
-10 mil delay, 3200 samples per secord sample rate (the highest since
the vibrator would sweep from 20Hz. to 1600Hz.). It was also set to
record 4096 samples per trace ( from -10mils to .127 seconds)

     The first round of data collection was a combined reflection,
refraction, surface wave experiment.  The first 48 channels, with the
40Hz. geophones, were a conventional split spread reflection survey.
The phones were spaced 5cm apart (yes, .05m) with a 5cm offset from
the shot to phones 24 and 25.  i.e. phones 1-24, shot, 25-48.
     Each shot location had three "shots" of different sources.  A
sledge hammer was used to drive a 1in. diameter pipe into the ground.
Typically, 4 sledge pounders were stacked and recorded.  A small hand
hammer was then used to create a higher frequency source.  Nine small
hammer taps were stacked.  A small vibrator was then attached to the
pipe and driven by a non-linear sweep of frequencies.
     The shot location and geophones were then advanced 20cm in order
to obtain 6 CMP (Common Mid Point) coverage.

     After examining a few shots  with all the geophones and source
types for quality control, it was decided to concentrate on the
small hammer reflection data first.

     I thought the bookkeeping and data management would be simpler
if all the small hammer SEG2 shots were in a single SEG-Y file and
the shots were renumbered.
  • script #1 - Create a SEG-Y file.
        Files 1102, 1106, 1110, 1114, 1118, 1122, 1126, 1130, 1134, 1138
    had errors while reading.  Upon examination (hex dump), the Geometrics
    appears to have dropped a single byte on several traces of each shot.
    SIOSEIS was changed to recover from the error.
        Since there are only 38 shots, I decided to plot every shot on the
    laser printer.  I plotted the first shot by itself and tried various
    parameters until I found what I wanted.  AGC is needed to examine the
    whole trace.  I also decided to plot all shots and all traces with the
    same scalar, but this wouldn't really show the amplitude variation
    from shot to shot or trace to trace because of the agc.
        The Geometrics was set to start recording slightly before the
    actual trigger, so the plot parameter stime 0. is used.  The time of
    the first sample is -.020 seconds
         Yikes, there are bad traces scattered throughout the line.  Some
    traces are bad probably due to a bad connection between the geophone
    on the cable.  All the shots with read errors have multiple bad traces.
  • Script to plot every shot to the laser printer.
  • plot of shot 29
         Before applying the geometry and doing CMP gathers, I'll kill the
    bad traces by either weighting the whole trace to zero using PROCESS 
    WEIGHT or by muting the back end of the trace using PROCESS SMUTE.
  • Script to weight smute plot shot 29.
  • plot of shot 29 after weight and smute
         The above C shell script is executed by using the first shot
    number at the first argument and the last shot number to plot as
    the second argument.  e.g.  plt 29 29  plots just shot 29.
         I plotted each shot on the screen, one at a time,  to check that
    I got all the bad traces.  Good thing too, because I'd missed a few.
  • Script to weight smute plot all shots.
         Note that xloadimage is terminated by typing the letter q
    in the plot window.  Also note that the pause between plots is
    caused by xloadimage being in the foreground.  Continuous plots
    (without a pause between shots) may be obtained by doing xloadimage
    in background (by placing the ampersand sign (&) on the xloadimage
    command line.
         Each SEG2 shot is a separate file, so I callected them all
    together into one SEG-Y file using this script.
         The next step shown in script is to:
    1) edit the data (kill bad traces, invert traces with reverse polarity
       and surgically mute noise bursts).
    2) Assign the shooting and geophone geometry as well as assigning
       the topographic corrections.
    3) Do a CMP gather (collect traces according to common mid points)
         The final script is short, but took a great deal of analysis to derive.
    1)  Only the traces near the shot were retained.  Traces further than
    40cm from the shot don't appear to have reflections and have some
    other mode (surface waves or shear waves or ???); so ranges 45cm and
    greater are eliminated from the stack by weighting them to zero.
    2)  Filter was applied before NMO because the low frequencies didn't
    move-out correctly.  I never determined why.
    3)  The datum or topographic corrections are applied via process SHIFT.
    4) A 100x800 Hz bandpass filter was applied.  
    5)  The plot is done in reverse direction since the first shot was on
    the southe end of the line.  The plot annotation is by meters with
    the top of the mound being zero.
    final plot

    User Discussion

                   New (2001) Analysis
                   --- ------ --------
         The recording unit clipped some of the traces on some of the shots,
    so it was decided to mute the clipped data first thing.  Script is
    the sioplt script used to pick the mutes.  The tsets were then used to
    create a new set of cmp gathers using (script).
    The cmp gathers can be plotted with script.
    The cmp gathers were moved out and stacked with script.
    The stack was fk migrated with script.
    The final plot was generated with script.
    The plot of the filtered agced shot directly over the mound was generated
    by script.
                   Tomography Survey
                   ---------- ------
        The tomography survey geophone layout has "Geometrics" channel 2-48,
    LeRoy channel 1-32, Alistair channels 1-34.
    1)  Create a SEGY file of all 72 shots
  • Script to combine all three Geometrics recorders
    2)  Create a file with the "closest" trace to the vibrator.
        Note: There is confusion about the polarity between the two series
    of Geometrics.  After step 1 above, traces 1-47 are from the rental
    Geometrics and traces 48-105 are from LeRoy and Alistair.  The polarity
    of the pilot traces of all "closest" traces of the Geometrics are 
    reversed in this script.
  • Script to collect the trace "closest" to each vibrator point.
    3)  Pick and print the time of the largest positive amplitude after
    correlating with the trace closest to the vibrator.  SIOSEIS does not
    permit specifying a different pilot trace on every shot when the pilot
    is in a different file, so each shot is correlated in a new sioseis
    job.  The first sioseis job created the output file and subsequent
    jobs positioned after the last shot in the output file.
        The special lprint value of 4 was created in process WBT to print
    the picks and a perl script was run to write the pick in the "Cornuelle"
    format and to only print if the time of the pick was greater than .05
  • Script to correlate and pick.
  • Perl script to reformat the picks
  • Plot of the first shot after correlation. (62KB)
            GPR (Ground Penetrating Radar) Survey
            ---                            ------
        Larry Conyers of the University of Denver collected and processed
    a bunch of GPR data.  I will try some seismic processing techniques on
    these data.  The digital data were recorded in the GSSI DZT format.
    GPR uses nanosecond sampling and SEG-Y can do no better than micro-
    seconds, so all processing units have been scaled by 10**3.  Individual
    GPR traces are displayed here.  A typical GPR line is displayed here
           GPR line 59, meters 26 to 33 corresponds to the location
    of the seismic reflection line.  A simple plot doesn't reveal
    anything useful. (plot script).
           A common GPR processing technique is to use a "background"
    filter, which subtracting the average trace from all the
    traces.  SIOSEIS found the average trace through process STACK
    and process UADD then subtracted it (script).  
    The plot does help!
           A plot with a bandpass filter of 500 2000.
           A plot with a bandpass filter of 500 1000.
           A plot with the background filter and 500x1000 bandpass.
           A plot and script for filter and migration.
           A plot with background filter, filter and migration.
           A plot with filter, migration and topography corrections.
           A plot with background filter, filter, migration and topography corrections.
          The units used in GPR processing in SIOSEIS are confusing:
    1)  DZT2SEGY uses a sample interval in microseconds rather than nanoseconds
        since the SEG-Y standard (and SIOSEIS) expect it that way.  DZT2SEGY
        sets the sample interval to 59, where it should be .059.
    2)  The elevations in SEG-Y are also integers, but they need to be in
        millimeters, so the elevations are multiplied by 1000.  Thus GEOM
        parameters datume 600 datumv 200000 are in millimeters and mm/sec.
    3)  The FK migration was done using a sample interval of .001 so the
        program wouldn't bomb.  The migration velocity was 2500.  The
        distance between traces in migration was set to 16 rather than the
        .016m actually recorded.
    Go to the list of seismic processes.      Go to SIOSEIS introduction.