Interactive DIMAS program for processing seismic signals

Abstract

The DIMAS program is designed for detailed processing and visual analysis of digital seismic signals from digital seismic
stations. It is specially designed for users whose task is to quickly determine the parameters of an earthquake from one or
a group of seismic stations in the system of seismic monitoring and tsunami warning services. The program allows for a variety
of operations with seismic signals in the temporal and spectral ranges. It has a simple and functional user interface.

KeywordsDIMAS–processing of seismic signals–seismic monitoring–tsunami watch

Full text

ISSN 07479239, Seismic Instruments, 2011, Vol. 47, No. 3, pp. 215–224. © Allerton Press, Inc., 2011.
Original Russian Text © D.V. Droznin, S.Ya. Droznina, 2010, published in Seismicheskie Pribory, 2010, pp. 22–34.
215
INTRODUCTION
The DIMAS program (Display, Interactive Manipula
tion, and Analysis of Seismograms) was developed in 1997.
Since that time it has been used in the Kamchatka Branch
of the Geophysical Service of the Russian Academy of Sci
ences for processing seismic records in near realtime
mode. In subsequent years, the program was repeatedly
upgraded, taking into account the wishes of users and with
regard to new opportunities of networks of digital seismic
stations. In the course of the project under the name
“Development of a Network of Seismological Observa
tions and Instruments for Processing and Transmission of
Data for Tsunami Warning” in 2006–2010, the DIMAS
program was adapted to the challenges of realtime predic
tion of tsunamis on the basis of seismological data. It has
no practical limit to the number of recording channels of
seismic signals and the number of readings for each chan
nel. The program allows for analysis and processing of a
signal in the frequency and time domain, as well as for the
analysis of the threecomponent record of the seismic sig
nal and evaluation of parameters of an earthquake.
INITIAL DATA
Initial data for the DIMAS program are seismic signals
in the form of readings, evenly sampled in time, the arrays
of which are stored on a disk in the form of files. Data series
can be obtained in different ways, i.e., (1) as a result of
recording of seismic signals conducted by means of ana
log–digital converters connected to computers, (2) as a
result of sampling of seismic signals from remote systems
of digital recording, and (3) as a result of sampling of seis
mic signals from archived media. The disk file can contain
data files from one or several seismic channels. Each chan
nel must have the following service information: the name
of the seismic station, the code of the network of seismic
stations, the abbreviation of the seismic channel, the start
date and time of the first reading of the seismic trace, the
sampling frequency, and the number of readings in the file.
To use all the features of the program, additional informa
tion is required on each of the seismic channels. This is a
characteristic of the recording equipment, which includes
(1) the transfer function of the seismometer provided in the
form of poles and zeros, (2) digital filter coefficients and
the sensitivity of the equipment to bring digital readings to
the actual ground motion, (3) the geographical coordi
nates of observation points (latitude, longitude, and alti
tude), and (4) the seismometer orientation towards the
cardinal points and the vertical. This information can be
stored in files with the original data or automatically added
by the program from a specially designed metrological
database. The results of calculating the characteristics of
the data recording equipment by calibration pulses are a
part of this database. The results of the previous processing
also serve as additional data for each channel. These
include (1) the time at the earthquake origin, (2) the geo
graphical position of the hypocenter, (3) energy estimates
of the event (class and magnitude), and (4) times of arrival
and types of seismic waves, their periods, and amplitudes.
The program focuses on the standard data format
SEED (Standard for the Exchange of Earthquake Data).
It was specifically designed for the storage of digital seis
mic data and provides an opportunity to present all the
initial parameters required for the subsequent analysis.
This format is widely used in acquisition processors of the
GSN global network and in digital data collection system
of the Kamchatka regional network of seismic stations. It
is possible to enter data in other formats from other digital
data collection systems into the program and handle them
simultaneously. Currently, the program has processed the
following representation formats of digital seismic data:
SEED, SEGY, DATAMARK, POSEIDON, WAV, SAC,
GSE, VISEIS, ASCII, BINARY, GEOSIG (GSR,
GBR), GURALP (GCF), CSS 3.0, and REFTEK.
To determine the spatial and temporal parameters
of the earthquake origin, it is necessary to have the
Interactive DIMAS Program for Processing Seismic Signals
D. V. Droznin and S. Ya. Droznina
Kamchatka Branch, Geophysical Service, Russian Academy of Sciences, PetropavlovskKamchatskii, Russia
email: ddv@emsd.ru, dsv@emsd.ru
Abstract
—The DIMAS program is designed for detailed processing and visual analysis of digital seismic sig
nals from digital seismic stations. It is specially designed for users whose task is to quickly determine the
parameters of an earthquake from one or a group of seismic stations in the system of seismic monitoring and
tsunami warning services. The program allows for a variety of operations with seismic signals in the temporal
and spectral ranges. It has a simple and functional user interface.
Keywords:
DIMAS, processing of seismic signals, seismic monitoring, tsunami watch.
DOI:
10.3103/S0747923911030054

216
SEISMIC INSTRUMENTS Vol. 47 No. 3 2011
DROZNIN, DROZNINA
hodograph file of seismic waves in a binary format on
the hard drive. The hodograph includes a table of
arrival times of seismic waves from different depths
and distances (ASPI91.DAT file) and the table of first
arrivals of the longitudinal wave, obtained by combin
ing data from the ASPI91.DAT file with the theoreti
cal propagation time from the flat velocity model of
the environment for the analysis of the position of the
origin for near and far events in relation to the station
(FIRSTP.BIN file). Binary files of contours of the
shoreline and rivers are used to construct a geographi
cal map. Files of arrays of heights and depths are used
to construct the color maps of the terrain.
ANALYSIS AND PROCESSING
OF THE SIGNAL IN THE FREQUENCY
AND TIME DOMAINS
To analyze signals in the time domain, it is possible to
correct the baseline or delete a permanent component of
the signal, as well as to differentiate and integrate the sig
nal by different methods. The program allows for recur
sive and nonrecursive filtration, design of Butterworth
and Bessel infiniteimpulseresponse (IIR) filters, and
finiteimpulseresponse (FIR) filters such as low pass,
high pass, bandpass, and bandstop filters. In this case, the
design of filters is reduced to the calculation of digital fil
ter coefficients based on information received from the
user. These are values of cutoff frequencies, filter order,
and the type of weighting windows for calculating the
FIR filter. The software allows for the construction of the
signal envelope and smoothing of the signal in the time
domain by a moving time window.
The data processing in the frequency domain is per
formed using the fast Fourier transform and includes the
following operations on the original signal: (1) the differ
entiation and integration of the signal in the frequency
domain, (2) the computation of the envelope, and (3) the
estimation of the spectral power density of the signal
[Embree, Kimble, 1991]. For the visual analysis of the
power spectrum, a graph is constructed in a logarithmic
scale of frequencies, where along the vertical axis the
magnitude of the power spectral density of the signal is
plotted in decibels (Fig. 1).
If there is information about the characteristics of the
recording equipment at the user’s request, the power
spectrum can be corrected for the characteristics of the
device. The availability of information about the charac
teristics of the device is necessary for the restoration of
true ground motion (displacement, velocity, and acceler
ation). The program also allows for the emulation of the
recording of standard seismometers (Fig. 2). This feature
–100
–120
–140
–160
–180
1.0
0.1
40
20
0
–20
–40
100
50 150 200
Signal power, dB Ground motion,
µ
m/s
f
, Hz
t
, s
(a) (b)
PSD: 10 log(m
2
/s
4
/HZ) PET IU 00BHZ
Fig. 1.
Power spectrum of the (a) original signal, corrected for the characteristics of the device and the experimental data on min
imum and maximum levels of microseism, the original signal (b).

SEISMIC INSTRUMENTS Vol. 47 No. 3 2011
INTERACTIVE DIMAS PROGRAM FOR PROCESSING SEISMIC SIGNALS 217
0
–0.01
–0.02
0.00002
0.00001
0
–0.00001
–0.00020
–0.00003
50
40
10
0
–60
–120
–180
1010.10.010
30
20
0.01
–0.0004
–0.0002
0
0.0002
0.0004
–0.0004
–0.0002
0
0.0002
0.0004
–0.00004
–0.00003
–0.00020
–0.00001
0
0.00001
0.00002
–0.00003
–0.00020
–0.00001
0
0.00001
0.00002
Signal power, dB Ground motion, m/s
t
, s
f
, Hz
PET IU 00BHZ ? ??/?
PET IU 10EHZ ? ??/?
PET IU 20HNZ ? ??/?
2
PET IU 00BHZ [EMU WOODANDERSON 20:1] m
PET IU 10EHZ [EMU_ WOODANDERSON 20:1]m
PET IU 20HNZ [EMU_WOODANDERSON 20:2] m
1
2
3
1a
2a
3a
1
2
3
1a
2a
3a
(b)(a)
Fig. 2.
Amplitudefrequency characteristics of channels (a) and the type of re cord ing o f th e ear thqu ake on dif fer ent c hann els (b): broa dband (
1
), shortperiod (
2
), accelerometer
(
3
). The emulation of a recording of the standard WOODANDERSON seismometer from the respective channels (
1a
,
2a
,
3a
).

218
SEISMIC INSTRUMENTS Vol. 47 No. 3 2011
DROZNIN, DROZNINA
is used for energy estimates of earthquakes, since nomo
grams, which are required for this, exist for standard
types of recording equipment.
The analysis of the signal in the frequency time
domain is reduced to study of changes in its frequency
over time. The signal is passed through a set of band
pass filters [Stearns, Ruth, 1993]; then the signal enve
lope in each frequency band is calculated. As a result
we have a level of change in the signal amplitude over
time at different frequencies, by which colored iso
lines, i.e., a spectrogram, are constructed.
By analyzing the spectral components, the optimal
filter can be selected for better separation of the signal
at the noise level in order to isolate the signal with a
weak signaltonoise ratio.
The program can carry out arithmetic operations
between channels and assess the instrumental seismic
intensity at the point of recording.
ANALYSIS OF THE THREECOMPONENT
RECORD OF A SEISMIC SIGNAL
The spatial analysis of waveforms is carried out
using threecomponent digital recordings of a signal.
For selected regions in the threecomponent record
ing, the program constructs a threedimensional
graph of the trajectory of particle motion and projec
tions of the motion trajectory to planes NE, NZ, and
EZ. In this case, the effect of the bulk trajectory is gen
erated by rotation by means of function keys, and the
projections on the corresponding planes are obtained
by turning the solid figure on the corresponding angles
(Fig. 3).
Using the program, it is possible to study the polar
ization characteristics of the signal [Hutton et al.,
1989], as well as to obtain a graphical display of the
azimuth and exit angles of the polarization axes of
seismic waves on the Wulff net for a specified period of
time (Fig. 4). The direction of the longest axis of the
P
wave gives the azimuth of the source. The study of
polarization characteristics of the signal makes it pos
sible to estimate the position of the epicenter of the
earthquake by the record from one station (Fig. 5).
EVALUATION OF PARAMETERS
OF EARTHQUAKES
The realtime evaluation of the main earthquake
parameters is one of the main advantages of the pro
gram. For convenience of work with seismic records, it
is possible to display the signal in graphs in arbitrary
time and amplitude scales. The determination of
arrival times of longitudinal waves and measurement
of periods and amplitudes of seismic waves is imple
mented in the manual and automatic modes. To con
trol the quality of the first arrivals of seismic waves and
estimate the time at the origin, the difference depen
dence is constructed of arrivals of longitudinal and
transverse waves (
t
S
–
t
P
) at the moments of arrival of
P

waves (
t
p
) (Fig. 6). The Wadati diagram is constructed
for three variants of velocity ratios
V
P
/
V
S
: (1) test line
is for regional events with the ratio close to 1.73, (2)
the line drawn as best as possible according to these
points, and (3) the line drawn with respect to time at
the origin.
For the known hypocenter of the earthquake on the
original seismograms, it is possible to display theoreti
cal arrivals of different waves, according to the
hodograph, in addition to those taken by the operator.
All the measured parameters can be displayed on
the monitor as a list. This list can be adjusted by the
user and stored in a specific format.
The program implements its own algorithm for cal
culating the earthquake hypocenter. In general, the
task of finding the hypocenter is reduced to finding a
space–time position of the origin, for which differ
ences between the observed moments of arrival of seis
mic waves and the theoretical moments of arrivals,
according to the hodograph, are minimal [Bullen,
Bolt, 1985]. The search for the hypocenter is con
ducted by a threedimensional Cartesian coordinate
system with the origin at the center of the Earth. For
this purpose, we introduce adjustments to the observed
arrival times at stations for the station’s altitude above
sea level The correction is entered as
Δ
t
i
=
on the grounds that the upper layer of the crust has a
low velocity of
V
top
, which leads to small exit angles of
the seismic wave. Further, geographical coordinates of
seismic stations (latitude and longitude) are converted
in the geocentric latitude and longitude, and then in
Cartesian coordinates.
Let us write down the relationship of coordinates of
the hypocenter with the coordinates in a threedimen
sional Cartesian system of coordinates.
(1)
where ( ) are coordinates of the seismic
station that recorded the event, (
x
o
,
y
o
,
z
o
) are coordi
nates of the origin of the seismic events;
R
(
T
o
,
H
o
,
D
o
)
is the originstation distance, dependent on the time at
the origin, the focal depth, and the epicentral origin
station distance.
We use a spherical model of the environment. Tak
ing into account that the coordinates of seismic sta
tions are reduced to the level of the Earth’s surface
(
H
= 0 km) and with corresponding adjustments of the
observed arrival times of longitudinal waves, (1) can be
rewritten as follows:
(2)
Hi
sta.
Hi
sta
Vtop
 ,
xi
sta xo
–()
2yi
sta yo
–()
2zi
sta zo
–()
2
++
=
R2ToHoDi
o
,,(),
xi
sta,
yi
sta,
zi
sta
REarth
2REarth Ho
–()
22xi
staxo2yi
stayo
–2zi
stazo
––+
=
R2ToHoDi
o
,,(),

SEISMIC INSTRUMENTS Vol. 47 No. 3 2011
INTERACTIVE DIMAS PROGRAM FOR PROCESSING SEISMIC SIGNALS 219
0.0002
0.0001
0
–0.0001
–0.0002
0.4
0.3
0.2
0.1
0.0002
0.0001
0
–0.0001
–0.0002
0.0002
0.0001
0
–0.0001
–0.0002
0.450.30.20.1
0.30.20.1 0.45
Press F1 or F2 to rotate
Ground motion,
µ
m/s
t
, s
PET BHE
Max = 0.000255685
Min = 0.000243989
Min = 0.000243989
Max = 0.000255685
Min = 0.000243989
PET BHN
PET BHZ
+ N
+ E
+ Z
Fig. 3.
Threedimensional ground motion.
Max = 0.000255685

220
SEISMIC INSTRUMENTS Vol. 47 No. 3 2011
DROZNIN, DROZNINA
0
–0.00008
–0.00016
–0.00024
2.41.60.80
0
–0.00008
–0.00016
–0.00024
0
–0.00008
–0.00016
–0.00024
2.41.60.8
2.41.60.8
Ground motion,
µ
m/s
t
, s
Pick = 0.000113412
PET BHE
PET BHN
PET BHZ
Pick = 0.000238004
Pick = 0.000298733
Azimuth = 201.6, exit angle = 38.3, ratio of axes 74.9%
Azimuth = 21.5, exit angle = 51.7, the ratio of the axes 18.7%
Azimuth = 111.5, exit angle = 90.0, the ratio of axes of 6.4%
1
2
3
1
2
3
Fig. 4.
An example of determining the exit angle of the seismic wave and the azimuth of the seismic source. Three main axes of the virtual ellipsoid of polarization are displayed
on the Wulff net.
P
P
P

SEISMIC INSTRUMENTS Vol. 47 No. 3 2011
INTERACTIVE DIMAS PROGRAM FOR PROCESSING SEISMIC SIGNALS 221
0.0004
0.0002
0
–0.0002
–0.0004
40302010
–0.0004
–0.0002
0
0.0002
0.0004
0.0004
0.0002
–0.0002
–0.0004
50
50
50
40302010
40302010
PET IU 00BHE 2007.09.12 11:13:38.316
PET IU 00BHN 2007.09.12 11:13:38.316
PET IU 00BHZ 2007.09.12 11:13:38.316
t
, min
PET
Ground motion,
µ
m/s
0
Fig. 5.
The realtime determination of the epicenter location of a distant earthquake by the record at the PET station.

222
SEISMIC INSTRUMENTS Vol. 47 No. 3 2011
DROZNIN, DROZNINA
–4E005
4E005
2E005
–1E005
4E005
2E005
0
–2E005
–4E–005
80604020
28
24
20
16
12
21147
0
4E005
8E005
–4E005
–2E005
2E005
4E005
–4E005
4E005
8E005
–8E005
0
8E005
0.0016
–2E006.
–1E005
1E005
2E005
–6E005
–3E005
3E005
6E005
–0.00012
–6E005
0
6E005
0.00012
–5E005
0
5E005
–8E005
–4E005
0
4E005
8E005
–0.0001
–5E005
0
5E005
0.0001
–8E005
–4E005
0
4E005
8E005
0
–3E005
0
0
80s
60s
40s
20s
80s
60s
40s
20s
80s
60s
40s
20s
80s
60s
40s
20s
80s
60s
40s
20s
80s
60s
40s
20s
80s
60s
40s
20s
80s
60s
40s
20s
80s
60s
40s
20s
80s
60s
40s
20s
80s
60s
40s
20s
80s
60s
40s
20s
SPN SHZ
NLC SHZ
RUS SHZ
PET SHZ
UGL SHZ
SDL SHZ
SMA SHZ
AVH SHZ
KRK SHZ
GRL SHZ
KRY SHZ
GNL SHZ
MKZ SHZ
Max = 8.37896e005
Min = –9.03407e05
Max = 0.000109797
Min = –0.000106706
Max = 9.88922e005
Min = –9.81813e005
Max = 0.000101573
Min = 0.000101475
Max = 0.000135182
Min = –0.000141403
Max = 7.58379e005
Min = –7.17285e005
Max = 2.95108e005
Min = –2.75324e005
Max = 0.000170572
Min = –0.000161791
Max = 0.00010983
Min = –7.15503e005
Max = 5.42578e005
Min = –4.60654e005
Max = 0.000111418
Min = –8.09664e005
Max = 4.45467e005
Min = –3.24774e005
Max = 6.07047e005
Min = –5.16421e005
P
S
S
S
S
S
P
P
P
P
P
P
P
P
P
P
P
PS
T
s
–
T
p
, s
MKZ
GRL
SDL
PET
RUS
NLC
Max = 28.9836
t
n
, s
1
2
3
Fig. 6.
The Wadati diagram: (1) a test line, for regional events the VP/VS ratio is close to 1.73, (2) a line drawn as best as possible according to these points, (3) a line drawn with
respect to time at the origin.
V
p
/
V
s
= 1.730
V
p
/
V
s
= 1.708
V
p
/
V
s
= 1.698
t
, s

SEISMIC INSTRUMENTS Vol. 47 No. 3 2011
INTERACTIVE DIMAS PROGRAM FOR PROCESSING SEISMIC SIGNALS 223
–0.004
–0.008
0.0004
0
0.0002
15M10M
5M
0
0.004
0.008
15M10M
15M
10M
10M 15M
5M 10M 15M
5M 10M 15M
5M 10M 15M
0.006
0.003
–0.003
–0.006
0
0
0
0.0016
0.009
–0.009
–0.018
–0.008
0.008
0.0016
Station ADK
Station BILL
Station Ma2
Station PET
Station YAK
Station TIXI
LR
LRP
P
P
P
P
ADK IU 008HZ m/s
°
1 Start: at 2006.04.20 23:23:50.273
BILL IU 008HZ m/s
°
1 Start: at 2006.04.20 23:23:50.273
MA2 IU 008HZ m/s
°
1 Start: at 2006.04.20 23:23:50.273
PET IU 008HZ m/s
°
1 Start: at 2006.04.20 23:23:50.273
BILL
TIXI
YAK
MA2
PET
ADK
YSS
Fig. 7.
A graphical representation of the result obtained by the DIMAS program. The epicenter of the Olyutorsk earthquake on April 20(21), 2006, according to the IRIS global
network. Arrivals of
P
waves are marked on seismic records.
23.50.273
23.50.273
23.50.273

224
SEISMIC INSTRUMENTS Vol. 47 No. 3 2011
DROZNIN, DROZNINA
where
R
Earth
is the Earth’s radius. After determining
the origin time
T
o
and the focal depth
H
o
, equation (2)
becomes
(3)
where is the epicentral originstation distance. For
a fixed depth and time at the origin, it is determined by
the table of arrival times of longitudinal waves.
Taking into account that equations similar in form
to (3) are created for each station with index
i
, the task
of finding the coordinates of the hypocenter for a fixed
time and focal depth can be reduced to the minimiza
tion of the following functional, produced by the
method of least squares
(4)
It is easy to see that finding the minimum of the func
tional (4) is a system of linear equations of the form of
where the matrix
A
has dimensions 3
×
3:
(5)
(6)
= (
x
o
,
y
o
,
z
o
) is the desired position vector of the
hypocenter, on which an additional condition (
x
o
)
2
+
(
y
o
)
2
+ (
z
o
)
2
= is imposed.
Thus, the search for the hypocenter is carried out as
follows. The interval of probable values
T
o
,
H
o
and the
search step are set. For each of the possible values, the
equation of the form of (4)–(6) is solved as =
A
–1
.
For each step in the process of enumeration, an array
is filled of differences between the station times of
observed arrivals and the resulting calculations. The
time values at the origin, focal depths, and coordinates
of the origin, when this difference takes the minimum
value, correspond to the most probable position of the
hypocenter. To determine the energy class
K
S
, a
nomogram of S.A. Fedotov is used [1972].
The program makes it possible to display graphi
cally the obtained parameters on the map (Fig. 7). Ini
tial data for the map are data on the shoreline, river
contours, and heights of the specified geographical
area. The map is plotted with locations of seismic sta
tions, the epicenter of the earthquake, stationepicen
ter raysdirections, determined by the polarization of
the seismic wave, necessary estimations of the origin,
if data are available only from one station, circles with
the stationepicenter radius, determined by the
hodograph, incorporated into the program.
CONCLUSIONS
(1) An easytouse program has been developed
that allows the user to perform complex processing
and analysis of a seismic signal, as well as to evaluate
the basic parameters of earthquakes.
(2) The DIMAS program meets high realtime
performance requirements and allows rapid process
ing of earthquakes for the Alert and Tsunami Warning
Service in an interactive mode.
REFERENCES
Bullen, K.E. and Bolt, B.A.,
An Introduction To the Theory
of Seismology
, New York : Cambridge University Press,
1985.
Embree, P.M. and Kimble, B.,
C Language Algorithms for
Digital Signal Processing
, Upper Saddle River: Prentice
Hall, 1991.
Fedotov, S.A.,
Energeticheskaya klassifikatsiya KuriloKam
chatskikh zemletryasenii i problema magnitud
(Energy Clas
sification of the KurilKamchatka Earthquakes and the
Problem of Magnitudes), Moscow: Nauka, 1972.
Hatton, L., Worthington, M.H., and Makin, J.,
Seismic
Data Processing
, Hoboken : WileyBlackwell, 1991.
Stearns, S.D. and David, R.A.,
Signal Processing Algorithms
Using Fortran and C
, Upper Saddle River: Prentice Hall,
1993.
xi
staxoyi
stayozi
stazo
++
=
REarth REarth Ho
–()
Di
o
REarth

⎝⎠
⎛⎞
,cos
Di
o
xi
staxoyi
oyozi
stazo
++(
i1=
N
∑
–
REarth REarth Ho
–()
Di
o
REarth

⎝⎠
⎛⎞
cos ⎠
⎞2
min.=
Ax b,=
A
xi
sta
()
2
i1=
N
∑xi
stayi
sta
()
i1=
N
∑xi
stazi
sta
()
i1=
N
∑
xi
stayi
sta
()
i1=
N
∑yi
sta
()
2
i1=
N
∑yi
stazi
sta
()
i1=
N
∑
xi
stazi
sta
()
i1=
N
∑yi
stazi
sta
()
i1=
N
∑zi
sta
()
2
i1=
N
∑
,=
b
REarth REarth Ho
–()
Di
o
REarth

⎝⎠
⎛⎞
xi
sta
cos
i1=
N
∑
REarth REarth Ho
–()
Di
o
REarth

⎝⎠
⎛⎞
yi
sta
cos
i1=
N
∑
REarth REarth Ho
–()
Di
o
REarth

⎝⎠
⎛⎞
zi
sta
cos
i1=
N
∑
⎝⎠
⎜⎟
⎜⎟
⎜⎟
⎜⎟
⎜⎟
⎜⎟
⎜⎟
⎜⎟
⎜⎟
⎛⎞
,=
x
REarth Ho
–()
2
x
b