Edgetech tow fish datum corrections

    This example shows applying a datum correction to some Edgetech
towed fish chirp data where the fish depth nor the wireout distance
are known.  The datum correction is calculated by identifying and
picking different events or wave paths on the data.
    The first method involves identifying and picking the surface
reflection (the upward pulse from the fish, reflected off the water
surface, then back down to the fish).  This is simply the fish depth
in two way travel time. See figure 1 and figure 2.
    The second method requires picking the event caused by the
water bottom reflection of the upgoing surface reflection and picking
the water bottom reflection of the downgoing pulse.

     The first step in Edgetech data processing is to convert the data
into SEG-Y.

sioseis << eof
procs diskin xstar diskoa end
diskin
   format edgetech ipath line164.sgy end end
xstar
  type 2 end end
diskoa
  opath line164.segy end end
end
eof

*****    Note   *****
There's a new XSTAR parameter, TYPE, that's needed to differentiate
between 1 trace and two trace XSTAR system.


METHOD 1          Pick the surface reflection
------ -


METHOD 2:  Pick the water bottom reflection of the surface reflection
------ -
    A quick QC plot (created with this script) shows
unrealistic geology caused by the variation in the depth of the
towed fish.
    In an earlier Edgetech project I was able to correct the
data to a datum by an automated pick of the water bottom multiple.
In a different project I was able to do the datum correction 
using the water depth from Seabeam center depth.
    Unfortunately, neither of those methods worked on this
dataset.  The recorded data length of this dataset is not
long enough to include the water bottom multiple.  The corrections
using the Seabeam (from an xyz file), didn't work well either
because the wireout distance was changed frequently and was not
recorded, thus the fish depth is not known.
    While looking at the above plot, I noticed a nice, geologically
realistic, event at the bottom.  If we assume that this event
is the water bottom, relative to the sea surface, and shift the
water bottom of the data to be at this time, then the plot might
be correct!  Unfortunately, I could not find an automatic picking
algorithm that would pick the even.  An automated pick would also
introduce the "heave" (wobble) I noticed on the event.  Picking the
event "by hand" was too tedious.
    The script was used to produce a SIOSEIS rasterfile (plot)and
header file suitable for the picking program SIOPLT.
    SIOPLT was executed as:
sioplt -if siofil -wbt wbt -hf hdrfil
and file  siopltrc had:   width 1400 height 800
    The picks were loaded into another script which automatically picks
the water bottom time after doing a 3 trace mix, computes the time
shift between the water bottom time and the event picked as where the
water bottom is supposed to be, then applies the shift.  The data are
save to disk (in IBM floating point so SU can read it) and then
filtered and plotted for quality control.
    Notice that the water bottom has several peaks and troughs.  This
is caused by the filter being too narrow and or too steep slope.
With experimentation I found dbdrop 12 made the water bottom into
a single cycle.  The end user also wanted to data compressed more,
so I decimated the data by throwing out 2 of every 3 traces (this was
also the reason why I did a three trace mix earlier; average 3 traces
and then keep only 1 of the 3).  It was also decided to get as
much dynamic range as possible without apply any gains, so the
plot is with gray scale.
    The script created the final plot.