Thailand Earthquake

1
Thailand Earthquake Tsunami Warning Training
Program
Session III.2 Instrumentation, Recording
systems Data transmission Archiving
Presented by R.F. Mereu Department of Earth
Sciences University of Western Ontario May 17,
2006 Bangkok, Thailand
Sponsored by TMD, ATT, USGS, USAID
2
Topics covered

• Introduction
• Fourier Analysis
• Filter theory
• The seismometer
• Types of seismic instruments
• Other equipment needed
• How sensors are installed
• Data transmission (telemetry)
• Data archiving (storage)
• Data display and analysis

3
The Life History of an Earthquake Signal
Earthquake -----gt Source Filter -----gt Earth
transmission filter -----gt Station site response
filter -----gt Seismometer -----gt Amplifier
------gt Anti-aliasing filter -----gt Digitizer
filter -----gt Transmitter -----gt Station
recorder -----gt Data storage and Analysis
-------gt Digital filters ----gt Graphical
Outputs (Seismic drum, computer terminals,
seismograms etc) ----gt Eye ball filter
----gt Brain filter
4
Fourier representation of a seismic waveform
A(f)The Amplitude Spectrum the amplitude
values of the cosine waves q(f) The Phase
Spectrum the phase angle values of the
cosine waves X(f) The complex spectrum
X(f) A(f) exp(- i2pq(f))
When you filter a signal you remove some of the
cosine wave components
5
Local earthquake recorded by SOSN-PKRO
station, Distance 0.65 deg.

Plotted using the PLTSEK display and analysis
program
6
Teleseism recorded by SOSN-PKRO station,
Distance 42.61 deg.

Plotted using the PLTSEK display and analysis
program
Plotted using the PLTSEK display and analysis
program
7
Teleseism recorded by SOSN-PKRO station,
Distance 132.71 deg.
Plotted using the PLTSEK display and analysis
program
8
Distance 0.65 deg.

Distance 42.61 deg
Three earthquakes recorded by SOSN-PKRO
station. Because of absorption of energy, the
earth acts as a low pass filter which removes the
high frequency components of the wave forms.
Plotted using the PLTSEK display and analysis
program
Distance 132.71 deg
9
(No Transcript)
10
Theorems from Fourier Transform Analysis
Time Domain lt----gt Frequency
Domain a(t)
lt------gt
A(f) b(t)
lt------gt
B(f) Multiplication in the
time domain lt------gt Convolution in the
frequency domain
a(t).b(t) lt------gt
A(f)B(f) Convolution in the time
domain lt------gt Multiplication in the
frequency domain
a(t)b(t) lt------gt A(f).B(f)
Filtering is equivalent to multiplication in the
frequency domain
a(t)f(t) lt------gt A(f).F(f)
where f(t)
F(f)
impulse response of filter frequency
domain response of filter Sampling a
waveform is equivalent to multiplying the
waveform in the time domain with a spike
sequence s(t). S(f) is also a spike sequence
a(t).s(t)
lt-------gt A(f)S(f) Truncating
a signal is equivalent to multiplying the signal
with a rectangular window function w(t).
Here W(f) the sync function (sinx/x)
a(t).w(t)
lt------gt A(f)W(f)
11
Digitizing or Sampling a Waveform Sampling a
waveform is equivalent to multiplying the
waveform in the time domain with a spike sequence
s(t). The spike separation T sampling
interval The spectrum of the time signal is
convolved with another spike sequence S(f)
spike separation 1/T. T the
sample interval The maximum frequency value of
the spectrum 1/(2T) SR/2 This
is called the Nyquist Frequency If the sampling
rate (SR) 100 The Nyquist frequency 50
Hz


From Brigham ,
The Fast Fourier Transform, 1974
12
Aliasing Errors If you are going to sample
a time series at sampling
rate SR the maximum frequency value of the
spectrum
SR / 2 Hz. If the original analog signal
coming out of the seismometer has frequencies
above SR/2, you will encounter aliasing errors
which show up as false spectral values. To avoid
aliasing errors an analogue anti-aliasing low
pass filter is applied to the signal before it is
digitized. Note If a sine wave is
insufficiently sampled such as once per cycle
at the same place, the computer could treat it
as a DC offset signal.
Aliased Fourier Transform From Brigham ,The
Fast Fourier Transform, 1974
13
The Digital Filter The digital filter can
be designed either in the time domain or the
frequency domain. In the time domain, it is a
sequence of numbers of different weights. To
filter a signal in the time domain, one convolves
the time series with the digital filter
Time series ---- a(i), i1,la
------gt a(1), a(2), etc
Filter series ---- f(i), i1,lf ------gt f(1),
f(2), etc Output series
b(i),i1,lb -----gt b(1),b(2), etc
where lb
lalf-1 Numerical example The simple
smoothing filter a(i) 2, 5, 1, -3,
-8, -2, 5, 9, 10, 3, 8 f(i) ( 1,
1, 1 )/ 3 --------gt Note this filter has
the shape of a rectangular pulse b(i)
(2,7,8,3,-10,-13,-5,12,24,22,21,11,8)/3
0.66, 2.33, 2.66, 1.0, 3.33, 4.33, 1.67, 4.0,
8.0, 7.33, 7.0, 3.66 , 2.66 The frequency
domain amplitude response of a square pulse is an
sync function (sinx/x) This not a good
filter. It is better to use a Butterworth filter .
14
The Butterworth Digital Filter The
Butterworth filter is normally designed in the
frequency domain using a simple formula of the
form High-pass
Low-pass
F(f) 1.0/(1.0 - fc/f)2n0.5
F(f) 1.0/(1.0 -f/fc)2n0.5
Here F(f) the amplitude response spectrum
f frequency values fc
cut-off frequency n the order
of the filter and determines the steepness of
the slopes By manipulating the phase
spectrum, one can design either a realizable or
non- realizable Butterworth filter. The
non-realizable filter requires future values of
the signal and hence cannot be used when the
application is in real-time. The
non-realizable filter operates only on past
values of the signal. It behaves in a manner
similar to any analogue filter. This is used in
real- time processing.
15
The example above shows a Butterworth realiazable
band-pass filter (0.5-20Hz) The top figure
shows the time domain impulse response of the
filter. The lower figure shows the frequency
domain response of the filter.
Plotted from PLTSEK Analysis and Display program
16
PkiKP
PP
PPP
Three component traces from 2 stations of the
SOSN network in near
Toronto, Ontario, Canada Data was filtered
with a Butterworth realizable band-pass filter
(0.01-1.00 Hz) Note How PP has much more high
frequency energy compared to PKiKP and PPP.
Plotted from PLTSEK Analysis and Display program
17
PkiKP
PP
PPP
Time-Frequency display of seismic trace from
Sumatra earthquake recorded near Toronto, Canada
Each trace is the output from a different
Butterworth Band-Pass filter Note How PP
has much more high frequency energy compared to
PKiKP and PPP. Red trace at the
top sum of all the traces in lower part of
graph. Plotted from PLTSEK Analysis and
Display program
18
The Seismometer when properly
damped acts as a simple high-pass filter
The figure shows Manufacturer's Specs for the
1 Hz short-period ELECTROTECH
seismometer. A large cylindrical magnetic
mass is suspended by springs over an electrical
coil. The coil is attached to the base of the
seismometer. The slightest relative motion
between the mass and the coil generates a voltage
in the coil which is proportional to
the ground velocity. The natural frequency
of the seismometer is 1 Hz When properly damped
as shown in the response curve (see bottom
graph), the seismometer acts as a simple
high-pass filter with a cut-off frequency of 1
Hz. This seismometer will not be sensitive
to long-period earthquake waves with frequencies
below 0.5 Hz.
19
A simple seismometer
20
(No Transcript)
21
A large seismometer 1200kg, 2m
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
Instrumentation by Nanometrics
Trillium 240 Very broad band seismometer. Flat
response fro 240 secs to 35 Hz
OBS
Trillium 40 All purpose Broadband seismometer
Indonesia Tsunami Warning System
Taurus fields seismograph Records 600 days
continuous Power 680 mW
31
Where is a good place to put a seismometer?

• Far from human-generated noise (roads and
  machinery)
• Far from the ocean
• On solid (competent) rock if possible
• Otherwise hard clay or gravel
• Avoid soft swampy areas
• In a temperature-stable environment
• Far from trees

32
Seismic vault under construction in Madagascar
33
(No Transcript)
34
(No Transcript)
35
(No Transcript)
36
(No Transcript)
37
Recording room
38
(No Transcript)
39
(No Transcript)
40
Data are then sent to the IRIS Data Management
Center in Seattle, USA where they are available
to anyone who requests copies. For more
information, see http//www.iris.edu.
41
The Display and Analysis of Seismic Data
There are many computer programs used by various
institutions which were written to display and
analyze seismic data. Requirements of such
a program Select stations.
Select earthquake data sets. Display
the data with appropriate digital filters and
gain controls. Provide an easy method
to pick arrival times and identify phases.
Perform spectral analysis of data traces.
Display particle motion diagrams.
Provide outputs for magnitude determination
programs and fault-plane
determination programs The PLTSEK Display and
Analysis program was used in the laboratory part
of the course. The first few slides in this
presentation were obtained from this program.

42
PLTSEK-Display --- A batch program designed to
display sets of past earthquakes using a
simulated multipen recorder. It also simulates a
mechanical drum recorder for viewing live
data. The old analog drum recorders cannot
display digital data
43
The Life History of an Earthquake Signal
Earthquake -----gt Source Filter -----gt Earth
transmission filter -----gt Station site response
filter -----gt Seismometer -----gt Amplifier
------gt Anti-aliasing filter -----gt Digitizer
filter -----gt Transmitter -----gt Station
recorder -----gt Data storage and Analysis
-------gt Digital filters ----gt Graphical
Outputs (Seismic drum, computer terminals,
seismograms etc) ----gt Eye ball filter
----gt Brain filter
44
THE END Thank You!!