Real-time Seismics
A
real-time seismic system was established not only to perform quality control,
but to provide instant backup of the data. The two Geometrics Geodes were connected to laptop computers
which in turn were connected to the shipÕs gigabit network (which failed and
became 100Mb). The laptop
computers were Windows based and ÒexportedÓ each seismic line as a unique
directory using the Samba protocol.
HopperÕs iMac mounted each directory as it became available (a manual
process). One Geometrics/laptop
was to record the MCS data and the other one for the two sonobuoy receivers.
The
Bergen acquisition group setup in the Òaft conÓ (science lab overlooking the
fantail along with all the winch controls). Hopper & Henkart doing the processing setup down the
hall in the ÒFuture LabÓ (a room which will be made into a lab someday).
The
HealyÕs computer system broadcasts a variety of information via the ÒUDPÓ
protocol; the posmv position is broadcast on port 33101. The Geometrics laptop read port 33101
and inserted the contents along with a time tag into the normal Geometrics log
file. SIOSEIS read the log file
and used this fix as the position of the shot in processing. There was no check that the time of the
fix was close to the time of the shot.
(The fix was the last fix received at the time of the shot).
The
laptop computers were not time synchronized initially, but were set with NTP
(network time protocol) after the first week of shooting. Relating the time of shot to the
navigation before NTP was installed is a problem, so using the navigation from
the log is the best that can be done.
The
Bergen group wanted to record in SEG-Y rather than the SEG-D format. The existing real-time input in SIOSEIS was using the SEG-D format, so
process DISKIN was modified to keep track of the SEG-Y file disk address, then
close the input file, sleep for a few seconds, then open and position the SEG-Y
file before reading again.
Initially,
the Òlatest_shotÓ file was plotted on demand on the iMac as the quality
control. A script to plot a single
trace from the entire line was developed, including converting the SIOSEIS
raster file to an HP RTL file and plotting on the large format HP plotter in
the computer lab. Steve Roberts
had a freeware program ÒdisplayÓ
(part of ImageMagick) that displays Sun rasterfiles using the X11
standard. A plot of each
Òlatest_shotÓ and a plot of the last 500 traces of a selected trace were
displayed on a Bergen laptop using the ÒdisplayÓ program from the iMac.
The
shot number or trigger count was also written the Geometrics log file along
with the nav. The Geometrics
numbers each shot recorded with the SEG-D FFID; a file id. The FFID is supposed to increment for
each shot written to disk. A
subroutine was written for SIOSEISÕs process GEOM that moves the FFID from
SEG-Y word 3 to word 5 and places the shot number into word 3. This routine also converted the shot
long/lat into 100ths of an arcsecond and wrote them into SEG-Y x/y coordinates
in words 19 and 20 (along with the scalar in short word 36). Complicating this procedure was that
the Geometrics produced duplicate FFIDs occasionally. The Geometrics also does not ÒflushÓ the log write buffers
after each write so there can be a large delay before the log entry is actually
written to disk. This latter
problem didnÕt affect us too much since we never got to the real-time stack
stage of processing.
Many
SIOSEIS processes use parameters that are given relative to the water bottom,
so the same subroutine searched and read the SeaBeam centerbeam file for the
water depth and put it ine SEG-Y word 16.
The routine was modified later to use the Knudsen depth file because
SeaBeam was inoperable for a while.
The last good depth is used whenever SeaBeam was unable to determine a
depth.
Other
changes to SIOSEIS while on HLY0503 were:
1)
Change process stack to
discard zero amplitudes from the stack average.
2)
Real-time geometry (type
9 or 14) computed the DFLS (distance from last shot) incorrectly after a
missing shot.
3)
Parameter ENDMUTE was
added to process despike to zero the remainder of the trace after a spike is
detected.
4)
The maximum number of
semblance velocities was increased in process VELAN.
5)
The maximum number of
samples per traces was increase from 16384 to 32768 (signed 16 bit integer to
unsigned) so that long Knudsen traces (22222 samples) can be convert to
envelope using a 32K fft.
Unfortunately,
the SIOPLT program for trace edited did not work on any computers on the Healy
since SIOPLT was written for 8bit displays and all modern displays are 24 bit.
The distance from the GPS antenna to the mid-point for the first trace is 89m. (offset = 88m). The Posmv is ~38m from the fantail (Consolidated survey: Posmv=-49.57, aft fantail survey point = -86.74, eyeball survey point to the end of the fantail = -.8m). Eyeball fantail to guns = 5m, guns to trace 1 = 90m. (38 + .8 + 5 + 45 = 88.8)
Script
1 determines the most recent file on the Bergen Geometrics laptop. It writes every shot to a SEG-Y file on
the MacIntosh. It writes every
second shot to a file Òlatest_shotÓ, from the beginning each time Ð a circular
file. Every 15th shot is also written to a circular disk file after
being filtered and amplitude balanced.
Script
set 2 displays the Òlatest_shotÓ and the Ònear trace: on the Bergen groupÕs
laptop computers.