Tom, Chris, and company,

You are quite correct in surmising that very sensitive arrays were
deployed during the cold war to monitor Soviet nuclear tests. These
had as many as 100 sensors in deep boreholes, with sensor spacings of
kilometers or more. They could "beam" their sensitivity by summing
the output of sensors with a time delay appropriate to the velocity
of the propagation of the seismic wave across the array.  Most of the
science was top secret, run by the Air Force Geophysics Program (AFTAC),
with the university involvement through ARPA (Advanced Research Projects 
Agency of DOD), and AFOSR (Air Force Office of Scientific Research).
Some known installations were in Montana (the LASA: Large Aperture 
Seismic Array), at Yellowknife in central Canada, Grafenburg in Germany,
and NORSAR in Norway.

In a minimal example of how they worked, three sensors are aligned 
north , middle, and south , and a suspected test site in Kazakhstan is 
the source of a 1-second seismic pulse. The wavefront is lost in the noise
as it passes the north sensor, but the array sensitivity is "tuned"
to a velocity of a 1-second compressional wave (8 km/sec) propagating from 
the north. So the data from the north sensor is delayed by the appropriate
time and then summed with the center seismometer data. This is repeated
for the south sensor. Since the noise of each site is random, the summed
signals increase the seismic wave amplitude by three times over the noise.
Obviously much more improvement is made as more sensors are summed. And
sensors along an arc at a constant radial distance from the source can
be summed without any delays. So array geometry became a hot topic.
Fortunately nuclear tests sites are difficult to move around or hide
completely from satellites.

This method makes many assumptions about wave paths, etc., and in the
early '70s days of analog tape recorders, filters, and oscilloscopes, 
it was difficult to work with delays as large as a second. Digital
analysis with analog triggers followed, and of course it is all digital
now. As the arrays have been upgraded, the surplus scene has been flooded
with used HS-10 seismometers, which were originally installed in 30
to 100 meter boreholes. Now the whole world is an array with the
installation of several hundred broadband digital stations, with the
focus of a major part of the newest instrumentation being verification of 
compliance with the CTBT (Comprehensive Test Ban Treaty).

So arrays are not everyman's way to improve the signal to noise ratio.
But, as they say, location is everything. Short period noise, like traffic,
is generally outside our interest (is above 10hz), or attenuates rapidly,
so 100 meters can make a big difference. But cultural noise generally
does not interfere with 1 second and longer waves, so sharp filtering
at the seismometer and/or at the recorder is very beneficial. From the
VBB vertical seis in the basement here, I routinely see the 20 to 40 
second surface waves of anything above a 5.5 that NEIS lists. But even
with considerable filtering at 10 hz, I still recorded the Mblg 3.7
near Indianapolis on April 14, a distance of about 300 km from St. Louis.
It did help that it was at night, after the local noise subsided.

Regards,
Sean-Thomas
__________________________________________________________

Public Seismic Network Mailing List (PSN-L)