PSN-L Email List Message

Subject: low-pass filters
From: sean@...........
Date: Mon, 2 Oct 2000 22:05:14 -0500 (CDT)

Jim, Carl, Chris, and co.

Regarding seismic low-pass filters vs urban noise:
(I have touched on this previously:)

For years we have found that an acceptable compromise for the
low-pass filter for a seismic station in an urban setting is
about 3 hz. We actually use 2.7 hz because it is the mechanical
period of the Wood-Andersion torsional optical seismometer that
has been used even recently in major stations because it is the
actual basis for the Richter scale, with a static magnification
of 2500 at 1 hz.

By urban setting, I mean the St. Louis Univ. campus vault under the
old gym on a pier that Macelwane built in 1923. It is within a km or so 
of freeways, railroads, and a foundry. And of course the gym and
new recreation center aren't exactly quiet settings.

Nonetheless, we are able to record M 3.0+ events at New Madrid,
200 km away.  As Chris pointed out, 60 hz AC noise is well attenuated
by a 3 hz filter. As for quake data, nearfield events (< 50 km), 
will arrive with higher frequencies, but still have a spectrum
rich in frequencies below 3 hz, so are well recorded. With the 
smooth roll-off of the phase-linear Bessel response, 10 hz is not
that far down.

For some time I have posted the schematic of the seismic preamp
and 4-pole filter that has been used in over 100 telemetry stations
from Alaska to Greece. The filter design is from the NASA filter 
handbook (details were posted). The schematic shows options for frequency 
scaling, as for a 2.7 second "Wood Anderson Seismometer" response 
that works well in an urban setting with a 15-second long period seis 
like a Lehman horizontal as input. 

stmmisc.html" PSN INFO ... SLU Seismic Network

For the "station" here in my basement, about 10 meters from a street,
I use a corner at about 2 hz for the broadband vertical just to
keep the car/truck noise to less than 2 mm on the drum record. I am
interested in teleseismic events (1 second to hundreds of seconds)
anyhow. Even so, the latest M3.9 in NE Arkansas recorded clear P 
and S phases, with Lg about 10 mm.

Unless you are running a tight (10 km spacing) network of stations to 
record micro- earthquakes (M<2), you don't need high frequencies. You 
also run into seismometer resonances (the famous L4-C gets very ringy 
at 16 and 22 hz), so even networks have peak responses at 10 to 20 hz, 
which still requires sample rates of 20 to 40 per second. The "standard" 
instrument for the new global network is the STS-1, which has a peak 
frequency response of 10 hz, so the VBB sample rate is 20 per second.


Public Seismic Network Mailing List (PSN-L)

[ Top ] [ Back ] [ Home Page ]

Larry Cochrane <cochrane@..............>