Observations on Instrument Response
Robert B. Herrmann
Email; rbh@eas.slu.edu
May 2, 2008
Introduction
following the Mw=5.2 southeastern Illinois Earthquake of April 18,
2008, it became necessary to ensure that the important data streams
collected were properly archived. The first check, was to focus
on the main event to determine if we actually know the responses of the
CMG5TD accelerometers.
We verified the responses by a detailed side-by-side comparison of
recorded signals with a trusted broadband data stream:
http://www.eas.slu.edu/Earthquake_Center/MECH.NA/ILLINOISEQ.REPORT/
Archiving Questions and QC
The specific questions are
1. Do we archive continuous accelerometer data streams, which requires an answer as to whether there is anythink worthwhile to archive
2. How do we ensure that the sensors are functioning correctly, e.g., whether individual channels are working and whether the timing is correct. In a region of low level seismicity, we cannot rely on local earthquakes. An alternative to to see if large teleseisms can be used. It seems as if the CMG5-TD's are not up to this purpose. Since these instruments are usually not co-located with broadbands, their lack of usefulness in looking a teleseisms indicates that continuous data should not be archived. However the lack of sensitivity does not assist with the need to verify that these are operating tot he proper standards.
This is what I tried to address in following:
CMG5-TD's
There are several aspects of the installed system.
First, because of initial firmware problems that gave problems with
transmission of the continuous 100 Hz data stream, all sensors in the
region (SLU and CERI) were set to 50 Hz sampling.
Action:
Change data acquisition to 100 Hz
As part of this history the 50 Hz channels were named HNZ HNE and
HNE. We also should have 100 and 200 Hz triggered traces in the
flash memory.
Since we have little historical data for archive, we will rename the
components of the 50 HZ now for the Illinois earthquake and
aftershoicks according to the
following convention:
Hello, Eric-
Thanks for the note. I think you are right on track with this
channel naming.
Use EN[ZNE] for those pesky 200hz hummingbird seismos.
Thanks for taking on the HN sex change and making them BN, too,
if that becomes necessary. We are completely fine with accepting these
strong motions into the archive if you would like to have them
co-managed.
The USArray is making it's way toward you, and besides, should a
large
New Madrid event (God forbid) come along, these would be very useful
research data.
Cheers,
Rick
Action: Use the above naming in the future and
retroactively for recent significant data
Second, we had not succeeded in creating a validated dataless SEED
because one of the FIR's had over 1200 coefficients. This broke most
dataless creators. We are currently trying the latest version of
Antelope. There is a question of the ability of the current evalresp to
handle the FIR continuation.
Action:
Manually code up the complete response including FIRS in
FORTRAN and compare to amplitude and phase output of evalresp
Sensor QC Checks
Since the system must work perfectly in order to catch the rare events,
the question arises as to whether the field systems are actually
working. This is especially difficult if there are few
significant local earthquakes.
If the accelerometers are sensitive enough, then a teleseismic signal
can be used to check timing and channel sensitivities.
As a quick example, consider the May 2, Mw=6.6 Andreanof Island
Earthquake:
The following figures compare the raw data stream as well as a 0.05 -
0.10 Hz bandpass [hp c 0.05 n 3 ; lp c 0.10 n 3] of the raw digital
counts. The instrument response has not been removed. The purpose is to
determine whether the accelerometers can actually see teleseisms.
If so, then further effort can be directed toward using the teleseisms
to chack polarities, timing and amplitudes. If such a test passed, then
the instrument is definitely not malfunctioning.
Figure 1. Comparison of recordings at SLM: BH traces STS-2/Q380; HN
traces - CMG-5TD - note noise level on CMG
Figure 2: 0.05 - 0.10 Hz band pass of traces in Figure 1.
Unsurprisingly there is little similarity given the low digital counts
Figure 3. OLIL: BH: CMG3-ESP/Q330; BL: Episensor/Q330 - note higher
digital couonts on acceleerometer
Figure 4. 0.05 - 0.10 bandpass of traces in Figure 3. Note
similarity of Z and N channels
Figure 5. CMG5TD raw traces from HAIL WVIL - both free field. Note
glitches
Figure 6. Bandpass version of Figure 5. Note the teleseism is not
seen
Questions:
What is the reason for the success fo the Q330/Episensor
versus the CMG5TD's. Is this due to inherent different sensitivities or
perhaps to fundamentally difference sensor noise and digitizer
resolution
Question:
Are the glitches in the raw data stream a problem of all
accelerometers?
Action:
What are the noise characteristics of the ANSS
acceleerometer systems installed. Can Albuquerque provide such
information on a timely basis.
Summary:
I question whether we have the ability to determine if ouor sensors are
running according to specifications.
This could be addressed in the future by bi-level recording, using 6
channels instead of 3.
What should be acrhived? Everything including garbage. Note
the Q330/Episensor combination can provide valuable data for archive.