Keck Expedition 2004
August 5, 2004 Day 7
Timing is Everything—by Doug Toomey and Paul McGill
Update for August 5, 2005 (by Debra Stakes)
By Debra Stakes from the Western
Flyer
(pictured here with Chief ROV Pilot Buck Reynolds)
Today we had gorgeous weather and an enthusiastic team that
completed three dives in one long day to the Endeavour Segment. Dive T713 went
to site KESQ on the northern Endeavour spreading segment. This corehole site was
established in 2002 by Tiburon drilling into the side of a small volcanic
collapse pit that marks the spreading axis just west of the Sasquatch
Hydrothermal Field. Thus the KESQ and KEMO sites encompass the hydrothermal
fields AND the fringes of the axial magma chamber recently imaged by a
multichannel seismic survey. 
The
sensor is tucked into the wall between two lobate flows and you have to look to
find it. The existing data logger and acoustic beacon were left on the top of
the collapse pit. 
Tiburon
put the replacement equipment there also while the sensor was extracted from the
wall. These collapse pits mark the eruptive fissure at the center of the
spreading axis where lava extrudes so rapidly it ponds. The sides are the
portions that cool and harden, but then the wall breaks and the still liquid
inner pond drains away. The center of the Endeavour spreading axis is delineated
by a series of such collapse pits or drainbacks. We put the data logger out of
harm’s way so even if there was another eruption, our data could be recovered
(even if the sensor got buried in new lava).
Dives
714 and 715 replaced “seismonuments” on the outer flanks of the Endeavour
segment. The seismonument is basically a portable borehole that we use when
there are no viable drilling targets for the coreholes. These cement blocks may
look simple (or even primitive) but they have careful research behind their
design. Early work in assessing the oceanographic noise on seafloor seismometers
suggested that the coupling between tidally driven bottom currents and the
instruments sitting on the seafloor was the greatest source of noise. Basically
the problem is that the bottom currents tend to rock the sensors, which is then
picked up and recorded, increasing the background noise. This noise is worse for
the horizontal components of the seismometer sensor package and reduces our
ability to study the shear waves generated by true seismic events. Traditional,
ocean-floor seismometers are deployed by dropping them over the rail. from a
surface ship to free-fall to the seafloor.
Such traditional seismometers may have a separate sensor package but it
is on a swing arm that permits them to sit directly on the seafloor. However,
the entire package, including the flotation required to recover the instrument
without an ROV, sits directly in the zone of the bottom current noise. In
contrast, the simple seismonument sits as low as possible with angled sides to
reduce the current’s impact. There is no flotation or leveling hardware as the
ROV does that part of the deployment task. Also, we separate the mechanical and
electrical noise of the logger from the sensitive sensor package. So in this
case, the KISS technique (keep it
simple stupid)
seems to work well.
The
KESE seismonument site is on the eastern side of the Endeavour Segment,
approximately 5 km from the spreading axis. Surprisingly the terrain here is not
completely buried in deep sediment. Rather it is the same type of sheet flows
that characterize the spreading axis, except that everything has a thin (a few
inches) of sediment. We can see the folds of an amazingly fresh sheet flow right
next to our existing instrument. This is a perfect site for a seismonument—a
flat surface with just enough sediment so that the block can sink down an inch
or so and get
well coupled.
It
is actually so flat that when the ROV manipulator placed the new seismonument
right beside the existing instrument, both LED’s (pitch and roll) began
flashing the news that they were well within their 5-7 degrees of horizontal
requirement.
The
third dive of the day is T715 to replace the 
seismonument on the western side of
the spreading axis at site KESW. This site is nearer to the flank slope and is
characterized by long skinny pillow tubes pouring downhill toward the west.
There is still plenty of sediment between the pillows however. We
put out the last new seismonument and then grab the one that has 
been out for a
year. The manipulator has to pull on it fairly hard because by this time it is
somewhat stuck in the mud (see top image to the right).
The last sensor and datalogger get stowed into the ROV’s equipment
drawer (bottom image to the right) and we are done with Endeavour for another 12
months.
At this point, the science party sits back to consider what has been accomplished. All seven short period systems left on the Endeavour spreading segment a year ago have been recovered. They were all alive with good timing and storage disks bulging with new data. The broadband system on Endeavour checked out also, with a valid in situ time synchronization from the ROV. The first three months of broadband data are good and we hope to extract the rest from the disks when we return to shore.
The hard-working data backup and quality check team (Andrew
Barclay, Will Wilcock and Doug Toomey) have a few comments on the data so far:
From Andrew: Of the four short-period seismometers we've looked at so far, the quality of the data has been consistently good. All four instruments recorded a year's worth of local and regional events; the signal to noise of known events has been very good. We won't really know how low the detection threshold of the network is until we start to calculate hypocenters and magnitudes, but we do know that there are very many events in the data that are below the threshold of land networks and SOSUS.
We've still to look at the three remaining short-period seismometers. Two were embedded in concrete blocks (seismonuments); one was in a corehole. It'll be interesting to see how the seismonument and corehole seismometers compare.
From Doug: These are some of the best S wave I have seen for microearthquakes recorded on ocean bottom seismometers. The corehole instrument is obviously well coupled, as the S wave is a nice clean pulse without a ringing coda. In my experience, instruments that free fall to the seafloor yield S waveforms that are more 'ringy', suggesting that once the seismometer is hit by the arrival, it rings. It looks like a real improvement in data quality to have the sensor installed in a corehole.
Great job!
And from Will: This morning I spent about 2 hours looking at data from two of the short period instruments. I scanned through about 3 days of data for an interval following the recent M6.4 on the Nootka fault. I found about 30 local events that must lie within or very near the network, several events that are within ~50 km of the network and many aftershocks of the recent Nootka fault earthquake. I looked at a few other 1-day periods scattered through the data and while the earthquake count was not as high, I was able to find local and regional earthquakes in each interval. The data quality is very high - I do not think I have ever seen shear waves that are so well recorded for local mid-ocean ridge earthquakes. (Editor’s note—also heard in passing was Will Wilcock commenting that Paul should make the sensors less sensitive because there are just so many darn events in the data record. I think he was joking…)
Our work is not over, but everyone else has gone to get some well-earned sleep. We still need to install the Explorer broadband, try again to level and synch the Nootka instrument and deploy three seismonuments around the Nootka seep site. It is now 2:45 am and we are steaming to the Explorer Plate site in a heavy rain. This will be our last broadband installation. It will not begin until the afternoon after engineer Paul McGill tries to resolve the communications problems with the ROV serial line.
But we are satisfied and looking forward to finishing up.
Timing is Everything—by Doug Toomey and Paul McGill
Accurate timing is essential to seismology.
The reasons for this are easily understood if we consider the nominal
speed of seismic waves near the Earth's surface, which are typically 6.5 to 8
km/s. Using 8 km/s, if the timing
of seismic data is accurate to only 1 s, then from a single station the distance
to an earthquake can only be measured to within 8 km (i.e., 8 km/s times 1 s).
This is a very large and unacceptable error.
If timing is accurate to 0.1s, the uncertainty in measuring a
distance is about 800 m, or nearly a kilometer.
Although this is an improvement, it is still somewhat large
as many geologic features can be found within any given kilometer.
A typical accuracy of 0.01 s (or 10 milliseconds or ms) yields distance
estimates accurate to 80 m, which is clearly an improvement and for many (but
not all) applications, acceptable to the average seismologist.
As a first cut, it seems pretty obvious that we would like to have a
'time base' that is accurate at any given time to within 10 ms.
Although the above is illustrative, it is also a bit simplistic. For one, errors in the location of an earthquake can also be reduced by using many observations since mathematicians tells us that the uncertainty in a measurement decreases as more observations are available. So one way to improve the accuracy of earthquake location is simply to observe it with many stations. It is also important to remember that to get such an improved location all the stations need to be networked either to land via a cable or some sort of seafloor network. The results from all the stations need to be combined to get a location solution.
There are other things seismologists do besides locate earthquakes.
One active area of investigation is using seismic data, particularly the
arrival times of seismic waves, to image the three-dimensional structure of the
Earth's interior. Much like CAT-scan in the medical sciences, where physicians
use acoustic energy to image the interior of a person, seismologists use seismic
energy to generate pictures of the Earth's interior. Seismic images provide, for example, a glimpse of
the plumbing system beneath volcanically active regions.
Skipping over the details and calculations, it is generally true that
seismic imaging also require a time base that is accurate to 10 ms.
Most seismic stations on land use the Global Positioning System
(GPS) to provide accurate timing. A
receiver costing just a few hundred dollars can provide timing accuracy to
better than 1 us (microsecond, a millionth of a second).
The problem for marine geophysicists is that GPS signals can't penetrate
seawater, so autonomous ocean-bottom seismometers must use crystal oscillators.
The best crystal oscillators are accurate to about 1 second per year.
Ocean bottom seismometers can still utilize GPS if they are connected to shore via a cable that provides a GPS-like timing signal.. Autonomous, un-networked instruments put in place by a Remotely Operated Vehicle (ROV) can also take advantage of GPS. The ROV has a connection to the ship on the surface, and the GPS signals can be sent down the ROV tether to provide a time reference on the ocean floor. If the time error of a seismometer is measured at both the beginning and the end of a deployment, then all of the recorded data can be time corrected to better than 10 ms.
Attached
is a photograph of University of Oregon geophysics grad student Troy Durant
installing the four GPS antennas used on the Western Flyer to synchronize our
data logger both at the surface and on the seafloor via the ROV Tiburon.