Chris,

I think some of your confusion may arise from the=20
fact that I had chosen to look at much lower=20
frequencies than you may be used to seeing.  A=20
good discussion of tilt effects at those=20
frequencies is given in Prof. Wielandt's Chapter=20
5 of the=20
NMSOP-2=20
http://ebooks.gfz-potsdam.de/pubman/item/escidoc:56076:3/component/escidoc:6=
1055/Chapter_5_rev1.pdf
where he observes that a tilt oscillation of=20
10-minute period (1um over 3km in his example)=20
will affect the horizontal trace 100x more than=20
the vertical.  At shorter periods the difference=20
will be less dramatic--possibly closer to the 10x number you reference.

Tilt creates an apparent acceleration.  On a=20
vertical channel the acceleration effect is as=20
cos(tilt) which for small tilts is nearly=20
constant, i.e., tilt has little effect.  On a=20
horizontal channel it creates an acceleration=20
proportional to sin(tilt) which for small tilts=20
is proportional to 'tilt'.  Being an=20
acceleration, the effect of tilt gets integrated=20
in a velocity recording, so as the frequency=20
decreases its effect on apparent velocity increases as 1/F.

I don't claim that the noise difference must be=20
due to pressure changes, just that it is most=20
likely related to oscillations in ground tilt,=20
which is generally considered to be a result of=20
pressure changes.  I guess I'm saying that I=20
don't really care what causes the (tilt?) effect,=20
but simply that it exists and that it=20
predominately affects the horizontal channels.

Dave's pier is on bedrock.  For more detail see=20
http://bnordgren.org/seismo/gif_images.htm

I don't see any obvious correlation of LP noise=20
with the time of day as we do at high frequencies.

Since the instruments are well-insulated=20
thermally their internal temperatures generally=20
show a more or less smooth 24-hour oscillation,=20
though differential heating of the ground and=20
even of the pier might be possible.  We believe=20
that we could be seeing something of the sort=20
when doing vertical 3-instrument coherence=20
measurements, due to various points on the pier=20
not moving vertically precisely together, but, if=20
it exists at all, it is a very, very small effect.

The digitizer channels had just been checked=20
out.  In fact the traces I showed were recorded=20
with a 24-bit G=FCralp CMG-DM24S6 that Dave was evaluating.

Dave's micro barograph uses the usual=20
configuration of a differential pressure sensor=20
referenced to a Dewar which has a controlled leak=20
to the atmosphere, essentially the configuration=20
shown in http://www.belljar.net/microbar.htm
The leak's low corner period is roughly 1 hour,=20
which effectively creates a 3600-second 1-pole=20
high pass in the sensor response.  This filters=20
out long-term barometric pressure changes that=20
could saturate the sensor while still allowing it=20
to display relatively small long-period=20
fluctuations.  His pressure setup is shown in=20
http://bnordgren.org/seismo/vault02.zip
The general arrangement is pictured=20
in  IMG_1694.jpg, though he now uses the MKS Type=20
223 sensor shown in DSCF1120.JPG  which is=20
providing a pressure sensitivity of ~196 Pa/V

Regarding horizontal noise, Adam Ringler of the ASL recently commented:
"For horizontals the dominant noise source is=20
tilt. Even on a granite slab there is incoherent=20
tilt from the block bending in various ways which=20
gets attributed to the instrument's "self-noise."=20
You essentially eliminate this with depth so you=20
have less of a signal (hopefully coherent) to=20
remove.  When we try to test vault style=20
horizontals at really long periods [in ASL's=20
testing tunnel, carved into granite bedrock], we=20
usually have to wait until we don't have a windy=20
day to get a good estimate of the self-noise.
Attached is a plot which shows the theoretical=20
improvement in horizontal noise with depth at=20
various periods. For reference you can see that=20
at 100 meters depth you could expect as much as a=20
20 dB improvement at 500 seconds period! However,=20
getting a 100 meter borehole drilled can be a bit of a task!"
http://bnordgren.org/seismo/ImprovenoiseGranite.pdf
In summary, for ground velocity, at long-periods,=20
horizontal noise is much larger than=20
vertical.  Whatever its source, seismologists=20
have generally concluded that there is little you=20
can do to avoid it other than to operate in a=20
deep borehole or a deep mine, not something we amateurs can normally=
 consider.

Regards,
Brett

__________________________________________________________

Public Seismic Network Mailing List (PSNLIST)

To leave this list email PSNLIST-REQUEST@.............. with 
the body of the message (first line only): unsubscribe
See http://www.seismicnet.com/maillist.html for more information.