Crustal Attenuation in the region of the Maltese Islands … through the CODAQ subroutine in the seismic analysis package SEISAN 8.0 by applying the time domain coda-decay method

Crustal Attenuation in the region of the

Maltese Islands using Coda Wave Decay

By

Raffaella Bugeja

UNDER THE SUPERVISION OF

Dr. P. Galea

Department of Physics

University of Malta

May 2011

A dissertation presented to the Faculty of Science in part fulfillment of the requirements for

the degree of Bachelor of Science (Hons.) at the University of Malta

Statement of Authenticity

The undersigned declare that this dissertation is based on work carried out under the auspices

of the Department of Physics by the candidate as part fulfillment of the requirements of the

degree of B.Sc. (Hons.).

____________________ __________________

Candidate Supervisor

Abstract

The attenuation property of the region around the Maltese Islands was investigated by

analyzing coda waves from 43 local earthquakes in the south of the Islands and 6 local

earthquakes in the North-West of the Islands using the single back scattering model (Aki and

Chouet, 1975). These were digitally recorded by the WDD station at Wied Dalam, Malta

during the period January 2006 - January 2011. The frequency dependent coda Q values were

calculated through the CODAQ subroutine in the seismic analysis package SEISAN 8.0 by

applying the time domain coda-decay method. The coda Q was computed at central

frequencies from 2 to 12 Hz. Coda Q values obtained show a clear dependence on f according

to the relation . The relationship varies from in the North-West near

Pantelleria to in the South of the Maltese Islands. The average Q values vary

from 229 ± 93 at 2 Hz to 1984 ± 281 at 12 Hz in the south region and 117 ± 11 at 2 Hz to 4028

± 3073 at 12 Hz in the North-West region. The variation of Q with frequency reflects the

structural inhomogenity around the Maltese Islands. The subduction zone near Crete was

chosen as another area of study so that the attenuation results obtained for the Maltese Islands

could be compared to this region. The relationship obtained for this area is .

Dedicated to

Mum and Dad

Acknowledgements

Would like to thank my tutor Dr. Pauline Galea for her constant guidance and help with this

dissertation. My thanks also go to Dr Sebastiano D’Amico for his support shown regarding

programming. Finally I wish to thank my family and friends for their encouragement and

unlimited patience.

Contents

Chapter 1 Seismic Wave Attenuation ......................................................................................... 1

1.1 Seismic waves ................................................................................................................. 1

1.2 Introduction to Seismic Attenuation ............................................................................... 7

1.2.1 Geometrical spreading .............................................................................................. 7

1.2.2 Intrinsic Attenuation ............................................................................................... 10

1.2.3 Scattering attenuation ............................................................................................. 12

1.3 Coda Waves .................................................................................................................. 14

1.3.1 Coda Analysis ......................................................................................................... 15

1.3.2 Phenomenological Modeling of Coda wave excitation .......................................... 16

1.3.3 Scattering Characteristics ....................................................................................... 18

1.3.4 The AC (Aki and Chouet) Method: Single back-scattering model ........................ 21

1.4 Coda-Attenuation Measurements .................................................................................. 24

1.4.1 Tectonic dependence of Coda Attenuation ............................................................. 25

Chapter 2 Tectonics and Seismicity of the Maltese Islands Region ....................................... 27

2.1 History of the Mediterranean ........................................................................................ 27

2.2 Tectonics of the Mediterranean Region and the Maltese Islands ................................. 29

2.3 Pelagian Platform and the Pantelleria Rift System ....................................................... 31

2.4 Seismicity around the Maltese Islands .......................................................................... 32

Chapter 3 SEISAN - Earthquake Analysis Software .............................................................. 36

3.1 Structure of Seisan-Directories ..................................................................................... 36

3.2 Waveform Data ............................................................................................................. 38

3.2.1 Data Format ............................................................................................................ 38

3.3 Programs ....................................................................................................................... 40

3.4 Calculation of coda q, CODAQ .................................................................................... 41

3.4.1 Input ........................................................................................................................ 41

3.4.2 Operating CODAQ ................................................................................................. 44

3.4.3 Output ..................................................................................................................... 47

Chapter 4 Data Processing ...................................................................................................... 49

4.1 Seismic Recording in Malta .......................................................................................... 49

4.1.1 The Wied Dalam Station, WDD ............................................................................. 50

4.1.2 Aims of the Malta seismograph station .................................................................. 53

4.2 Seismic Monitoring and Research Unit at the University of Malta .............................. 54

4.2.1 Earthquake locations ............................................................................................... 54

4.2.2 The Website ............................................................................................................ 56

4.2.3 The Seismic Database, Online ................................................................................ 56

4.3 The Data Set .................................................................................................................. 60

4.3.1 South of Malta Events ............................................................................................ 60

4.3.2 North-West of Malta Events ................................................................................... 61

4.3.3 Crete Events ............................................................................................................ 62

4.3 Calculating the Qc - values ............................................................................................ 63

Chapter 5 Results .................................................................................................................... 66

5.1 The frequency dependence of Q relationship ................................................................ 66

5.2 Results for the South of Malta Earthquakes .................................................................. 67

5.2.1 The frequency dependence of Q relationship for the South of Malta Events ........ 71

5.3 Results for the North-West of Malta Earthquakes ........................................................ 72

5.3.1 The frequency dependence of Q relationship for the North-West of Malta Events 76

5.4 Results for the Subduction Zone near Crete Earthquakes ............................................. 77

5.4.1 The frequency dependence of Q relationship for the Crete Events ........................ 81

5.4.2 The variation of Q with depth for the Crete Events ............................................... 82

Chapter 6 Discussion .............................................................................................................. 89

6.1 Analyzing the results ..................................................................................................... 89

6.2 Comparing the South of Malta, North-West of Malta and Crete Results ..................... 90

6.3 Comparison with other Areas........................................................................................ 93

6.3 Further Work ................................................................................................................. 94

References ................................................................................................................................. 89

APPENDIX 1 Recorded Events

APPENDIX 2 Coda Q Values

List of Tables

Table 3.1: The main subdirectories of SEISAN ...................................................................... 36

Table 3.6: Abbreviations of SEISAN ...................................................................................... 45 Table 5.1: Results for the South of Malta Events .................................................................... 67 Table 5.2: Results for the North-West of Malta Events .......................................................... 72 Table 5.3: Results for the Crete Events ................................................................................... 77

Table 5.4: Estimated Q0 at Different Depths for Crete ............................................................ 82

Table 6.3: Frequency dependence of Qc for different tectonic areas……………………........95

List of Figures

Figure 1.1: Propagation of P and S waves ............................................................................... 3

Figure 1.2: Reflected and Refracted Seismic Waves .............................................................. 4

Figure 1.3: Reflection and Refraction of body waves through the Earth ............................... 4

Figure 1.4: A seismograph showing the different phases and the 3 earthquake components. 5

Figure 1.5: The forms of ground motion near the ground surface of a Rayleigh wave .......... 6

Figure 1.6: The forms of ground motion near the ground surface of Surface Love wave ...... 6

Figure 1.7: Cylindrical area showing the wave energy propagation ....................................... 8

Figure1.8: Geometrical spreading of body and surface waves ............................................. 10

Figure 1.9: A representation of seismic scattering. ............................................................... 13

Figure 1.10: Seismogram showing the P wave, S wave and the coda ..................................... 14

Figure 1.11: Modeling a random medium as a distribution of point-like scatte. ..................... 18

Figure 1.12: Differential scattering cross-section of a single scatteres ................................... 19

Figure 1.13: Geometry of the single backscattering model ..................................................... 21

Figure 1.14: Seismograms showing high and low coda attenuation ........................................ 24

Figure 2.1: Bathymetric Map of Central Mediterranean around the Maltese Islands ........... 29

Figure 2.2: Bathymetry of the Sicily Chanel………………………………………………..32

Figure 2.3: Seismicity in the Mediterranean region between 1980 and 2000 ....................... 33

Figure 2.4: More reliably located seismicity, 1990-2003 ...................................................... 34

Figure 2.5: A seismogram of an earthquake ......................................................................... 35

Figure 3.2: Structure of SEISAN ........................................................................................... 37

Figure 3.3: Example of an input file ...................................................................................... 41

Figure 3.4: Example of a parameter file ................................................................................ 42

Figure 3.5: Calculating Codaq ............................................................................................... 44

Figure 3.7: Type line 4 using Nordic Format ........................................................................ 46

Figure 3.8: An example of an output file ............................................................................... 47

Figure 3.8: A codaq plot for an earthquake .................................................................... …. 48

Figure 4.1: Location of the WDD station .............................................................................. 51

Figure 4.2: Wied Dalam Station WDD in the south of Malta. .............................................. 51

Figure 4.3: The MedNet Network .......................................................................................... 53

Figure 4.4: The main page of the Seismic Monitoring and Research Unit website .............. 55

Figure 4.5: The real-time plot for Februaury 2011 ................................................................ 57

Figure 4.6: The online database of seismic events ................................................................ 58

Figure 4.7: Single event displayed online .............................................................................. 59

Figure 4.8: A hybrid map showing the South of Malta earthquakes ..................................... 60

Figure 4.9: A hybrid map showing the North-West of Malta earthquakes ............................ 61

Figure 4.10: A hybrid map showing the Crete earthquakes .................................................... 61

Figure 4.11: Calcualtion of .......................................................................................... 64

Figure 4.12: Procedure in calculating Qc ................................................................................ 65

Figure 5.1: A graph of Qc against frequency for the South of Malta events ......................... 68

Figure 5.2: A graph of the Average Q values against frequency for the South of Malta ...... 69

Figure 5.3: A Graph of ln (Qc) against ln (f) for the South of Malta events .......................... 70

Figure 5.4: A graph of Qc against frequency for the North-West of Malta events ................ 73

Figure 5.5: A graph Average Q values against frequency for the North-West of Malta. ...... 74

Figure 5.6: A Graph of ln (Qc) against ln (f) for the North-West of Malta events ................ 75

Figure 5.7: A graph of Qc against frequency for the Crete events ......................................... 78

Figure 5.8: A graph of the Average Q values against frequency for the Crete events. ......... 79

Figure 5.9: A Graph of ln (Qc) against ln (f) for the Crete events ......................................... 80

Figure 5.10: A Graph of ln (Qc) against ln (f) for the Crete events having depth in the range

4-15 km ..................................................................................................................................... 83


17-25 km ................................................................................................................................... 84


26-35 km ................................................................................................................................... 85


36-45 km ................................................................................................................................... 86


46-53 km ................................................................................................................................... 87

Figure 5.15: A Graph of Q0 against depth for the Crete events ............................................... 88

Figure 6.1: A Google Map showing different types of crust between the Malta Escarpment

and Ionian Region ..................................................................................................................... 92

Figure 6.2: A graph of the Average Q values against frequency for the North-West of Malta,

the South of Malta and Crete evens .......................................................................................... 93

Figure 6.3: Comparing the South of Malta seismograms to the North-West of Malta

seismograms.. ............................................................................................................................ 94

Figure 6.4: Comparison of the Qc relations obtained in different tectonic and volcanic are. 97

Chapter 1: Seismic Wave Attenuation

1

Chapter 1 Seismic Wave Attenuation

In this chapter a brief introduction on seismic wave attenuation is given. A discussion

on coda waves and their properties is also included in this chapter. The back scattering

model (Aki and Chouet, 1975) is also discussed, which is a way to model coda wave

excitation.

1.1 Seismic waves

Seismic waves are waves of energy that travel through the earth for example after an

earthquake. There are two main types of seismic waves, the surface waves and the body

waves.

Body waves are waves that penetrate deeply thorough the interior of the Earth. These

waves represent short pulses of propagating energy. They follow refracted raypaths

determined by the elastic moduli and densities of different regions of the Earth’s interior.

There are two types of body waves generated, the P and S waves (Lay and Wallace, 1995).

The P waves are the fastest moving waves and are simply sound waves. The P wave is a

longitudinal wave made up of a series of compressions and rarefactions. This type of wave

forces the point in Earth from where it passes to vibrate back and forth in the direction in

which the wave is travelling. The equation of P waves is given by:

(1.1)

where

is the density.

is the cubic dilatation given by the sum of the longitudinal strains i.e. .


2

and are the Lame constants.

is the rigidity modulus and is defined as

where K is the bulk modulus.

This is a scalar wave equation representing the propagation of the dilatation This

means that it represents a disturbance in which the material expands and compresses

periodically. This is a case in which there is a change in volume that does not result in a

change in shear. The P-wave velocity is given by

(1.2)

where the constants are defined earlier.

The S waves unlike the P waves are transverse waves. They cause the particles of the

medium to move perpendicular to the path along which the wave is travelling. The S wave

being a transverse wave is polarized in two perpendicular planes, the vertically polarized

components, Sv and the horizontally polarized components, SH. These two components are

identical in the case of an isotropic medium but are separate components travelling at different

speeds if the medium not isotropic.

The equation of the S waves is given by:

(1.3)

The quantity that is propagating is . When the components of this term are

considered it can be shown that this quantity represents a rotational disturbance without a

change in volume. This means it is a shear wave called an S wave. The S-wave velocity is

given by:


3

(1.4)

In these equations β is always smaller than α. This means that the S waves always travel

slower than the P waves. It can be shown that in a liquid µ=0. This means that S waves do not

propagate in a liquid.

Figure 1.1: Propagation of P and S waves (Available online from:

http://science.jrank.org/pages/48108/seismic-body-waves.html)

P and S waves propagate independently of each other. The seismic body waves travel in

ray paths perpendicular to the wavefront. These are paths that small packets of seismic energy

follow as it travels throughout the Earth. The velocity of the wave changes as it propagates

and so the ray paths are bent according to Snell’s law (Snellius, 1621). This change in the

wave velocity is mainly due to strong discontinuities like changes in the type of rock along

which the wave is propagating. These discontinuities act as interfaces that reflect the seismic

body waves like a mirror and refract them like a lens. This is shown in figure 1.2.

http://science.jrank.org/pages/48108/seismic-body-waves.html


4

Figure 1.2: Reflected and Refracted Seismic Waves (Available online from:

http://science.jrank.org/pages/48108/seismic-body-waves.html)

Body waves generated by earthquakes travel from the core to the mantle and may be

refracted from the core-mantle interface. They can also travel through the core and emerge on

the other side of the Earth. Body waves were observed to travel along particular paths. These

paths are referred to as phases and are labeled as PcP, PKP, Pn, PmP in the case of P waves

and ScS ect for the case of S waves (Robertson, data unknown). These phases are shown in

figure 1.3.

Figure 1.3: Reflection and Refraction of body waves through the Earth (Available online

from: http://www.sciencebuddies.org/science-fair-projects/project_ideas/Geo_p018.shtml)

The phase SkS involves the conversion of the S-wave energy at the core-mantle

boundary to form a P wave which travels throughout the core before being converted to the

opposite side of the core. As a seismic disturbance reaches the surface of the earth from the

interior, the motion of the ground surface is a combination of both types of waves. A

seismograph usually records three components of ground motion. One vertical, Z component

and two horizontal components aligned NS and WE. A seismograph showing the phases and

these 3 components is shown in figure 1.4.


http://www.sciencebuddies.org/science-fair-projects/project_ideas/Geo_p018.shtml


5

Figure 1.4: A seismograph showing the different phases and the 3 earthquake

components. (15/02/2007, recorded by the Maltese Station, WDD)

Another type of Seismic waves is the surface waves. Surface waves are confined to a

surface layer of the earth usually crust and the upper mantle. The lower the frequency, the

larger the depth sampled by the waves. The amplitude of the waves decreases exponentially

with depth. There are two types of surface waves that are the Rayleigh waves and the Love

waves.

The Rayleigh waves, Lp are confined to a vertical plane containing the direction of

propagation (Rayleigh, 1887). They are a combination of P and SV displacements, in which

the particle motion is retrograde ellipse, with the major axis vertical. Retrograde ellipse

motion is a combination of a transverse and a longitudinal wave. This means it is a

combination of back and forth and up and down motion. Rayleigh waves can be thought of

arising from the constructive interference of multiple reflected P and S waves. As the depth

increases the size of this ellipse gets smaller until it decreases to zero. The amplitudes of these

waves diminish slowly with distance. Surface waves are usually the prominent feature on

seismograms, since they propagate as to encircle the Earth many times.


6

Figure 1.5: The forms of ground motion near the ground surface of a Rayleigh wave

(Earthquakes, B.A. Bolt, 1999)

Seismic surface waves of the second type are the Love waves, Lo are transverse waves

confined to the horizontal plane (Love, 1911).This means that particle motion is perpendicular

to the direction of propagation. Love waves are actually produced by SH waves guided in a

surface layer in which the S-wave velocity is smaller than the underlying medium.

Figure 1.6: The forms of ground motion near the ground surface of Surface Love wave.

(Earthquakes, B.A. Bolt, 1999)


7

1.2 Introduction to Seismic Attenuation

The main concern in discussions is usually the elastic properties of the earth. The

amplitude of a seismic pulse in an idealized, purely elastic earth is controlled by the reflection

and transmission of energy at the boundaries and by geometric spreading. These seismic

waves can propagate indefinitely once they are excited. But this would be true if the earth was

perfectly elastic. It is known that the real earth is not perfectly elastic. This causes the waves

that are propagating to attenuate with time as they travel. This attenuation in the propagating

waves is caused due to various energy loss mechanisms.

Due to the conservation of energy, it is known that the energy switches from being

potential to kinetic exactly without any losses. However this is only true if there is no other

form of energy involved. As the wave travels, its energy experiences a continuous conversion

between potential energy due to the particle position and kinetic energy due to the particle

velocity. This energy conversion is not perfectly reversible as the wave propagates. Apart

from these types of energies that are continuously being exchanged as the wave propagates,

there is also work being done. This work done can take many forms such as work done as the

wave travels along mineral dislocations. Work is also done as shear heating at the grain

boundaries. These processes are described collectively as internal friction. These will all affect

the energy of the wave as the wave travels away from the seismic source. The simplest way to

describe attenuation is by using an oscillating mass attached to a spring.

1.2.1 Geometrical spreading

Seismic wave amplitudes suffer changes as they travel across the earth. As the

wavefront moves out from the source, the initial energy released in the earthquake is spread

over an ever-increasing area and thus the intensity of the wave decreases with distance.


8

Energy intensity is the total energy flow through a unit area in a unit time. The wave

energy propagation direction coincides with the area of the cylinder

Figure 1.7: Cylindrical area showing the wave energy propagation

(1.5)

where v is the propagation velocity of the waves. By the conservation of energy the total

energy at any moment should be constant.

Consider two wave fronts. These form two spherical shells whose centers coincide at the

source. The greater radius of the outer shell is r2 and the radius of the inner shell is r1. The

surface areas of the outer and inner shells are and

respectively. By the

conservation of energy the total energy flowing through the inner and outer shell should be the

same and so

(1.6)

(1.7)

(1.8)


9

(1.9)

(1.10)

(1.11)

Generalizing thus gives that

(1.12)

The amplitude decays as

. This is the geometric spreading for spherical waves.

The same can be done for an infinitely long line source, the shape of the wavefront is a

cylinder and so this is referred to as the cylindrical wave.

The same is repeated and the conclusion is that

(1.13)

The amplitude decays as

for the waves generated by a line source.

This can be generalized to seismic waves. In body waves the energy spreads over a

hemisphere. The intensity therefore varies as:

(1.14)

Since the intensity is proportional to the square of the amplitude, then the amplitude of

body waves is proportional to 1/r. In the case of surface waves, the energy spreads out

approximately along the curved side of a cylinder, whose height is equivalent to the

penetration depth of the surface wave. Thus


10

(1.15)

and therefore the amplitude of the surface waves is proportional to

This shows that body

waves attenuate faster than surface waves. This is in fact shown on a seismogram where at

long distances from the earthquake the surface waves are the dominant feature.

The geometric spreading alone cannot describe the complete attenuation of seismic wave

energy. The decrease of the kinetic energy of seismic waves is also due to the energy

absorption caused the imperfections in the earth. This is the case when the elastic energy is

completely transferred to the mantle.

Figure1.8: Geometrical spreading of body and surface waves

1.2.2 Intrinsic Attenuation

There is another factor that affects seismic amplitudes. This is energy loss due to

anelastic processes or internal friction during wave propagation. This is called intrinsic

attenuation (Shearer, 1999). The strength of intrinsic attenuation is given by the dimensionless

quantity Q in terms of the frictional energy loss per cycle

(1.16)


11

where E is the peak strain energy and is the energy loss per cycle. Q is usually known as

the quality factor. It is often needed to talk about the inverse of the quality factor, Q-1

. Q is

inversely related to the strength of the attenuation. This means that in regions where Q is

found to be low are more attenuating than regions where Q is found to be high. An

approximation may be derived which is valid when considering that Q >> 1. This

approximation is better suited for seismic application:

(1.17)

where x is measured along the propagation direction and c is the velocity. This equation shows

that for a constant value of Q, the higher the frequency the higher the attenuation. This is

because for a given distance the high frequency wave will go through more oscillations than a

low frequency wave. As the wave travels away from the source, the pulse broadens at

successive distances. As the wave propagates, attenuation removes the high frequency

component of the pulse.

The constant c depends whether it is a P wave or an S wave. for P waves with

attenuation and c = for S waves with attenuation . The amplitude of harmonic waves

may then be written as a product of a real exponential and an imaginary exponential. The

amplitude decay due to attenuation is incorporated in the real exponential while the imaginary

exponential describes the oscillations. These two exponentials are brought together into one

equation that gives the amplitude of harmonic waves:

(1.18)

The exponentials can be combined together and the effect of Q is now found in

.

This is done by adding a small imaginary part to the velocity c.

This equation can also be written in terms of time. This is better suited when considering

a seismic application since the wave is propagating forward in time:

(1.19)


12

P waves and S waves have different values of Q with the values of the S waves usually being

larger than the values of Q for the P waves. This is because of the shear motion involved

between particles that lead to more frictional heating. Qα and Qβ are the values of Q for the P

waves and the S waves respectively. Intrinsic attenuation occurs mostly in shear wave motion.

In fact it is associated with lateral movements of lattice effects and grain boundaries. As the

density and the velocity of the material increases, Q increases. It has been found that if the

loses of a material are only due to shearing mechanisms then

(1.20)

For frequencies up to 1.0Hz the quality factor, Q for seismic waves is independent of

frequency. As the frequencies increase, Q becomes frequency dependent and in general it

increases with frequency. There are many ways to determine Q. A common way to determine

Q is by knowing the amplitude and frequency of the seismic wave at some point during its

propagation. A number of seismic rays that have travelled the same path or rather a similar

path are usually considered. Their amplitudes and frequencies are then compared.

1.2.3 Scattering attenuation

There is another different type of attenuation called scattering attenuation. This is the

effect of seismic amplitudes in the main seismic arrivals are reduced by scattering off small-

scale heterogeneities. This is different from other types of attenuation since the integrated

energy in the total wave field remains constant.

The region of the earth to about 100 km is known as the lithosphere. This refers to the

solid part of the earth and its thickness varies from one place depending on the tectonic setting

of the area. The heterogeneities of the Earth have been investigated using different methods

both geological and geophysical. Seismic velocities and density of rocks give a geophysical

characterization. On the other hand the evolution of rocks gives a geological interpretation of

such heterogeneities. They analyze the rocks from within the earth that gives a sign of

heterogeneity. There are many factors which contribute to the heterogeneity of the lithosphere.


13

These include tectonic processes such as faulting, and large scale crustal movements.

Scattering of high-frequency seismic waves shows the existence of such small scale

heterogeneities in the lithosphere. High frequency waves interact with discontinuities and

small-scale heterogeneities, so that the main arrivals are drawn out into a coda. Low frequency

meaning long wavelength waves are unaffected by small scale reflectors.

In figure 1.9 one can see seismic waves propagating after a seismic disturbance,

propagating away from the source. The S wave travels the shortest path and so arrives at the

seismic station before all the other waves which interact with the heterogeneities. The other

wave amplitudes are scattered off by the small-scale heterogeneities and so they arrive after

the S wave and have smaller amplitude than the S wave. These are in fact the coda waves.

This study is focused on scattering attenuation and will be explained in more detail in the next

section.

Figure 1.9: A representation of seismic scattering.

The quality factor representing the total attenuation, Qt is given by:

where

is the quality factor due to scattering losses

is the quality factor due to intrinsic absorption.


14

1.3 Coda Waves

One of the properties used to study the structure of the earth is the attenuation of seismic

waves in the lithosphere at high frequencies ranging from 1Hz up to 20 Hz. The most

important evidence that the earth is heterogeneous is that in seismograms of local earthquakes

there is the appearance of the coda waves. On seismograms this is seen as the direct S wave

being followed by wave trains whose amplitude decrease exponentially as the lapse time1

increases. These wave trains are called S coda waves or simply coda waves. Initially the word

coda didn’t refer to these wave trains but it used to refer to the oscillations of the ground as the

surface waves propagated through it or the tail portion in the seismogram. The definition of

the word coda has recently changed and now coda refers to all wave trains excluding the direct

waves that propagate after a seismic disturbance. This is shown in figure 1.10. There are two

different types of coda. P coda refers for waves between direct P and S waves and S coda

refers to the waves following the direct S waves. As the epicentral distance increases, the

direct S wave amplitude decreases. This is true if you take lapse times large enough. At small

times the average S coda amplitudes are nearly equal independent of epicentral distances. This

is taken into account when conducting experiments and usually twice the lapse time is taken.

Figure 1.10: Seismogram showing the P wave, S wave and the coda

1 The lapse time is the difference in time between the S wave starting time and the origin time


15

Rautian and Khalturin (1978) studied coda wave amplitude. In their study they studied

these coda amplitudes at different lapse time and frequency bands. In their study it was found

that the early portions of the coda are different from one station to another. If the data is taken

from a bandpass-filtered seismogram, coda shows no variation in shape from one station to

another after three times the S travel time from the source to the receiver. Coda is quite similar

at all stations when twice the S travel time, lapse time is taken.

The magnitude of local earthquakes can be determined if the amplitude of the direct

wave is known. The magnitude is determined from the average amplitudes of the direct waves

at many stations. This is done after correcting the distance from each station. The value for the

magnitude obtained from the amplitudes was found to be proportional to the logarithm of the

duration of a local seismogram. This duration is the time measured from the P wave arrival to

the time when the S coda amplitude decreases to the level of microseisms (Solov’ev, 1965). In

many studies all over the world, the logarithm of the duration time has been used to find the

magnitudes of each earthquake. A correlation has been found between the magnitude and the

duration time. This correlation is consistent with the similarity in shape of the portion of

seismograms. It was then concluded that coda portions of seismograms are composed of

scattered waves.

1.3.1 Coda Analysis

Many different methods have been developed to determine Q from coda waves (Aki and

Chouet, 1975; Rautian and Khalturin, 1978; Del Pezzo et al., 1983; Rovelli, 1984). As

discussed earlier, coda wave attenuation is caused by two types of effects scattering and

anelastic attenuation. These processes cannot be separated easily. Dainty (1981) has suggested

that in the frequency range 1 to 20Hz, the frequency dependence of coda Q is primarily due to

scattering while anelastic attenuation is almost frequency independent. A strong correlation

between the dependence of Q on frequency and the tectonics of the region was found by Aki

(1981), Roecker (1982) and Pulli (1984). In areas where there is strong tectonic heterogeneity,

a strong frequency dependence of coda Q was found as that compared to stable areas. This


16

relation shows that the attenuation of seismic waves as the distance from the source increase is

different for different frequencies. This means that the seismic data has to be bandpass-filtered

first before calculating the attenuation.

The Q factor increases with frequency (Mitchell, 1981) and it follows the following

relation

where is the quality factor at the reference frequency f0 (generally 1Hz) and is the

frequency parameter. vary as the region varies due to the heterogeneity of the

medium (Aki, 1981). This relation shows that attenuation as the wave propagates is different

for different frequencies. Hence seismic data are first bandpass-filtered when calculations of

attenuation are made.

1.3.2 Phenomenological Modeling of Coda wave excitation

The characteristics of high frequency S-coda waves of local earthquakes were

summarized by Aki and Chouet (1975). These characteristics are the following:

The S-coda of seismic waves observed at different stations are almost identical to each

other;

A reliable measure of an earthquake magnitude can be obtained from the total

duration2 of a seismogram;

S-coda traces of different local earthquakes that are first bandpass-filtered and are

recorded within the same region have a common envelope shape. Such traces are

independent of the epicentral distance;

2 defined as the length of time between the P-wave onset and the time when the coda amplitude equals the level

of microseisms


17

The temporal decay of S-coda amplitudes is independent of the earthquake magnitude,

for earthquakes having magnitude less than 6;

The S-coda amplitude depends on the tectonics of the area where the recording station

is.

Other studies show more different characteristics of coda waves. These include:

Array measurements show that S-coda waves are not regular plane waves coming

directly from the epicenter (Aki and Tsuijura, 1959).

(Tsujiura, 1978) found that the S-coda waves are composed primarily of S-waves. This

was confirmed as his studies show that S-coda waves have the same site amplification

factor as that of direct S-waves.

S coda waves have been first identified on seismograms which were recorded at the

bottom of deep boreholes drilled in hard rock beneath soft deposits (Sato,1978; Leary

and Abercrombie, 1994).

A phenomenological model has been proposed by Aki and Chouet (1975). This model is

for coda-wave generation and is based on a number of assumptions. The earth’s lithosphere is

viewed as composed of a random and uniform distribution of point-like scatters in a

homogeneous background medium. The wave velocity in the medium is assumed to be

constant. Aki and Chouet (1975) first presented this model for the case the source and the

receiver are at the same location. This model was then extended by Sato (1977) where the

source and the receiver were not collocated. Sato did this extension of the model for body

waves while Kopnichev (1975) did it for surface waves.

Many other phenomenological models have been proposed for the generation of S-coda

waves. Before Aki and Chouet (1975) presented their model, Wesley (1965) explained the

seismogram envelopes by using diffusion – like process. In studies conducted it was found

that the coda wave have long duration. These were studied using the diffusion model, Dainty

and Toksoz, (1981). The propagation in the lunar crust can be explained using the diffusion

model. This can be done since the lunar crust have low intrinsic attenuation and a large

amount of scattering. The energy flux model was developed by Frankel and Wennerberg


18

(1987). This model is base on the fact that the energy in the scattered wave is uniformly

distributed.

1.3.3 Scattering Characteristics

To model the randomly inhomogeneous media, homogenous background media with

propagation velocity Vo filled with distributed point- like scatters with number density n are

used. This is seen in the following figure 1.10.

Figure 1.11: Modeling a random medium as a distribution of point-like scatters. (Seismic

Wave Propagation and Scattering in the Heterogeneous Earth: P, 1998).

This distribution is taken to be randomly homogenous and isotropic. The scattering has a

scattering cross section

An incident wave with energy-flux density J

0 intersects a

scatterer. This is a stationary process. Due to this intersection of the incident wave with the

scatterer, spherical waves are generated having energy flux density3 J

1.

3 The energy flux density is defined as the amount of energy passing through a unit area perpendicular to the

propagation direction per unit time


19

Figure 1.12: Differential scattering cross-section of a single scatterer. (Seismic Wave

Propagation and Scattering in the Heterogeneous Earth: Sato, Fehler, 1998).

The amount of energy scattered per unit time into a given solid angle element dΩ is J1r2

dΩ

where r2

dΩ is the corresponding surface element. The differential scattering cross section is

defined as the ratio

(1.21)

The scattering coefficient is the scattering power per unit volume of a medium filled

with scatterers. This is given by the product of the number density and the differential

scattering cross section [Aki and Chouet, 1975]:

(1.22)

This product, g has dimension of reciprocal length. The scattering power may be

characterized using the scattering coefficient only. In this formula there is no distinction

between a small number of strong scatterers and a large number of weak scatterers. The

scattering coefficient may be in all directions and so the total scattering coefficient is the

average over all directions:


20

Sc (1.23)

Where

is the total scattering cross section4,

: is the mean free path5 ,

Sc is the scattering attenuation that represents the decrease in the incident wave energy due

to scattering as the distance travelled increases. This is defined for waves of wave number k.

The energy flux density at travel distance x decays exponentially for a plane wave. It decays as

Sc (1.24)

There are many models that are used to represent scattering. But the simplest of these models

is isotropic scattering:

(1.25)

g=g0 (1.26)

4 the integral of the differential scattering cross section over a solid angle

5 the reciprocal of the total scattering coefficient


21

The scattered waves are incoherent since the scatterers are considered to be randomly

distributed. Due to incoherence the phase may be neglected and the scattered wave power is

the summation of all the power from each scattered waves.

Figure 1.13: Geometry of the single backscattering model. (Seismic Wave Propagation and

Scattering in the Heterogeneous Earth: Sato, Fehler, 1998).

1.3.4 The AC (Aki and Chouet) Method: Single back-scattering model

This method was developed by Aki and Chouet in 1975. In this method the coda is

considered to be made up of single back scattered waves. These scattered waves are the result

of discrete randomly distributed heterogeneities. This method is a single backscattering model

that explains the coda waves as a superposition of secondary waves from randomly distributed

heterogeneities. In this method it is assumed that the distance between the source and the

receiver is negligible. Therefore this method is valid for signals that arrive long after the

primary waves. The coda wave amplitude decrease with lapse time at a particular frequency.

This is due to energy attenuation and geometrical spreading. It is independent of earthquake

source, path effect and site amplification (Aki, 1969).

Assuming single scattering from randomly distributed heterogeneities, Aki and Chouet

(1975) developed an equation for the coda wave amplitude at frequency f, and elapsed time, t

from the origin for a bandpass-filtered seismogram at central frequency f is related to the

attenuation parameter Q by the following equation:


22

(1.27)

Where

S(f) is the coda source factor at frequency f which is independent of time and radiation pattern,

f is the frequency,

Qc (f) is the quality factor of coda waves,

is the geometrical spreading parameter.

Body wave scattering has a value of 1, surface wave scattering has a value of 0.5 and diffusion

waves have a value of 0.75 (Sato and Fehler, 1998).

Studies by Aki (1981) show that the coda waves are S to S back scattered waves. This is

consistent with the observation that coda Q and Q of direct shear waves are often shown to be

identical (Aki, 1980: Kvamme, 1985). Since coda waves are body waves, in the analysis done

for the coda Q, a spreading parameter of is assumed. It was found by Rautian and

Khalturin (1978), that equation (1) is valid only for lapse time t, greater than 2 times for S

travel time, Taking the logarithm on both sides of equation (1.27) and arranging the

following equation is obtained:

(1.28)

(1.29)


23

The value of Q can be obtained by linear regression of on t at a constant

f. If the slope of the graph is assumed to be b, then Q is determined using:

(1.30)

A(f,t) is usually found by bandpass-filtering the signal with a narrow passband around f

and fitting a time decay envelope to the filtered signal (Rautian and Khalturin,1978). Equation

1.30 is valid only for lapse times that are chosen to be greater than twice the S-wave travel

time. This is done so there is no interference by the data from the direct S-wave.

This method is used by the program SEISAN (Havskov and Ottermoller, 2003) to

calculate the value of the coda Q in our analysis for a number of earthquakes that are recorded

at the same station.

One Q value for the same region can be obtained after inverting simultaneously all the

data available from the decay curves that are available for the same region (Aki and Chouet,

1975; Phillips, 1985). The same result can be obtained by first obtaining one Q value for each

decay curve and then averaging the Q-1

values (Kvamme, 1985). This latter method had faster

computation and the equation can be checked for each individual case.

In all coda Q studies done it has been shown that Q increases as the lapse time increases.

As the start time for the coda window was increased and longer windows were used, the value

of coda Q also increased (Kvamme, 1985; Lee et al., 1986). The sampling volume for the

back-scattered coda waves at lapse time t is an ellipsoid with source and station at the focal

points and semi – major axis equal to , where is the S-wave velocity (Pulli, 1984;

Scherbaum and Kisslinger, 1985). As coda Q increased with lapse time it has been interpreted

that it is increasing with depth (Roecker et al., 1982; Pulli, 1984). Coda Q values may also

increase due to other factors. These are multiple scattering (Gao et al., 1983). Another factor

is coda model parameters example the geometrical spreading, v. Therefore to obtain best

results, coda wave time windows of a constant and fixed length that start at about the same

lapse times in order to be able to compare results from different areas .


24

1.4 Coda-Attenuation Measurements

In a study conducted by Rautian and Khalturin (1978), it was found that the S coda has a

common amplitude decay curve. This is true if the lapse time is greater than twice the S-wave

travel-time. For a given region the shape of the decay curve is the same and this decay curve is

quantified using the parameter of coda attenuation.

The coda attenuation Qc-1

is an exponential decaying function. It is independent of the

source and the location of the station but it depends on the frequency band. Qc-1

can be

measured from records observed at a single station. This makes it possible to take

measurements of the coda attenuation even at locations where there aren’t more than one

station located. This is the case of Malta where the only station available to monitor the

seismicity around the Maltese Islands is WDD station.

On a seismogram, the coda amplitude decay with lapse time is characterized by this coda

attenuation Qc-1

. The larger the Qc-1

values means that the coda amplitude decay is more rapid.

This is schematically illustrated in the following figure 1.14.

Figure 1.14: Seismograms showing high and low coda attenuation


25

1.4.1 Tectonic dependence of Coda Attenuation

Values for the coda attenuation, Qc

have been obtained world wide. Since the

lithosphere is characteriszed by a heterogenety, studies have been conducted in the frequency

range between 1 to 30Hz. These measurements have been compared with seismtectonic

activity. The values of Qc vary by more than a factor of 10 from region to region. Different

regions have different tectonic activity. This is seen in the variation of measurements of Qc-1

with frequency from region to region. As a general trend in our study Qc is dependent on

frequency following the relationship (Mitchell, 1981). The frequency-dependence

relationship obtained indicated that attenuation at higher frequencies is less pronounced than at

lower frequency. This is because high frequency waves interact with discontinuities and small-

scale heterogeneities and then the main arrivals are drawn out into a coda. Low frequency

meaning long wavelength waves are unaffected by the small scale reflectors.

In studies done it has been shown that the values of Qc depend on the type of rocks

found in the region being studied. Regions which are characterized by hard, competent rocks

usually have high Qc values while areas characterized by soft, molten rocks such as volcanic

areas usually have low Qc values. The values of Qc also depend on the age of the rocks in the

area being studied. Sinn and Herrmann (1983) studied short period seismograms of local

earthquakes in the U.S.A. The highest values of Qc were found in central U.S.A. where the

type of rocks that are exposed are the oldest. This study shows that Qc is higher i.e. Qc-1

is

smaller in areas that are tectonically stable and the values are lower in areas that are active

where the lithosphere is highly heterogeneous. The frequency parameter α increases as the

tectonic activity of the region increase (Aki, 1981). An example is the Andaman Islands

(Parvez, Sutar, Mridula, Mishra & Rai, 2008). This is an active tectonic area where the

lithosphere is highly heterogeneous and so is characterized by low coda Q values.

Low-frequency dependence values have been obtained in seismically active areas

in different parts of the world (Japan, Yoshimoto et al. 1993; Northern Greece, Hatzidimitriou

1995; Turkey, Akinci & Eydogan 1996; and Horasan & Boztepe- Guney 2004). These low


26

values are the cause of processes such as faulting which are likely to introduce strong

heterogeneities. In general low-frequency-dependence of Q values are lower in volcanic

region and in the shallow crust. Keeping in mind that Q is inversely related to the strength of

attenuation, means that values of Qc-1

are higher in volcanic regions and so waves are more

attenuated in such regions (Tres Virgenes Volcanic Area Mexico: Wong, Cecilio, Munguia,

2001) and in Mt. Etna (Del Pezzo et al. 1995). This suggested that the presence of magma

under volcanic regions would contribute to the dominance of intrinsic attenuation due to

anelasticity over attenuation due to scattering losses. In particular in such tectonically active

areas the value of the frequency parameter α was found to increase up to a value of 1 (Rovelli,

1982; Kvamme and Havskov, 1989; Akinci et al., 1994; Gupta et al., 1998).

Chapter 2: Tectonics and Seismicity of the Maltese Islands Region

27

Chapter 2 Tectonics and Seismicity of the

Maltese Islands Region

2.1 History of the Mediterranean

In general it is accepted that the Earth was originally a hot gaseous mass. By time this

mass cooled down and changed from a gas state into a liquid state. It then formed a solid crust

on the surface. Evidence brought from studies show that the land masses are made of light

igneous rocks. These rocks by time have been covered by sedimentary and metamorphic

rocks. The Oceanic Crust consists mainly of rocks known as gabbro and basalt which is denser

than the igneous rock making up the Continental crust which is known as granite. The oceanic

crust areas are found under several kilometers of sea water and continental crust areas form

the main land areas but area also found in shallow seas.

As the depth and the temperature of the earth increases, the density of these rocks

increases as well. The core is the deepest part of the earth. The pressure found in the earth’s

core is enormous. Despite this pressure, the earth’s core is in a molten state. Therefore the

earth is made of a liquid core, a mantle and a crust. Oceans and continents have different crust

thickness and composition. A lot of convection currents are present in the mantle. These

currents force the crust to ride over the mantle since the crust tends to be lighter. It is thought

that in the past, these convection currents caused fractures in the crust. These fractures resulted

in a number of continental plates. These plates are moving with respect to each other and to

the Earth’s rotational axis. They are continuously pushing together and pulling apart

depending on the direction of the currents (Pedley et al. 2002). The seismicity of the world

gives us an indication of the active regions of Earth and also roughly pictures the plate

boundaries.


28

Around 540 to 250 million years ago, the continental crust underlying the Maltese

Islands formed a projecting corner of the African continent. At this time Africa was part of a

large continent. South America, India, Australia and Antarctica were all part of the southern

half continent while North America was part of the northern half continent. An east-west

ocean known as the Tethys separated these two continents. This ocean used to lie on the

southern edge of the Maltese segment (Pedley et.al, 2002).

This large continent started to break up into the continents we know today around 150

million years ago. Africa and southern America separated and formed the South Atlantic

Ocean. This resulted in the eastwards movement of North Africa to southern Europe. Around

100 million years ago the Atlantic began to open. This resulted in Europe being split eastward

away from North Africa. Such movement is still going on. The continents were then drifted

apart by plate movements. This resulted in the development of a narrowing zone between

northern Africa and Southern Europe. This has now developed into an area which is known as

the Mediterranean. The movements that resulted due to the slitting of the northern Africa from

southern Europe resulted in the building up of many stresses. This gave rise to the formation

of many islands like Sardinia and Corsica. Mountainous islands some of which are volcanic

are the result of compression stresses that arose in the Mediterranean sea-bed. Malta and

southern Sicily were part of the Pelagian Spur. Around 10 million years ago this started to tear

away from the main African land. This resulted in the opening of deep sub-sea rift valleys.

The stresses in the crust between Africa and Europe lead to an opening of another new

ocean basin. This ocean basin is found between Sardinia and Calabria and which is known as

the Tyrrhenian Sea. This Sea continued to expand and Calabria was then forced to move

eastwards. As the Tyrrhenian Sea continued to expand, the Calabrian continental block

between southern Italy and the Pelagian Spur broke away from the African continent. This

resulted in a series of NW-SE fractures that then produced rift valleys in the continental crust

across the shallow Pelagian Platform (Reuther, Eisbacher, 1985).


29

2.2 Tectonics of the Mediterranean Region and the Maltese

Islands

The Mediterranean Sea divides the continent of Europe and the continent of Africa. The

Maltese Islands being 316 square kilometers lie in the central part of the Mediterranean

Sea between Sicily and North African coast. The Maltese Islands are found in the Sicily

Channel. They lie on a stable plateau of the African foreland, the Pelagian Platform, about 200

km south the Europe-Africa plate boundary which is part of Sicily.

Figure 2.1: Bathymetric Map of Central Mediterranean around the Maltese Islands.

(Limestone Isles in a Crystal Sea-The Geology of the Maltese Islands: Pedley, Hughes Clarke,

Galea, 2002).


30

The area of sea that lies between the southern Sicily and the North African coast is

shallow with sea depth not exceeding the 200 meters. In this relatively shallow sea one can

find several important deep valleys running from the northwest to southeast. These are known

as the Pantelleria Rifts. In this great mass of shallow water the only lands presents are the

Maltese Islands, the Pelagian Islands: Lampedusa, Linosa, Lampone and the Pantelleria

(Pedley et.al, 2002). This large area of mainly shallow sea separates the Eastern and Western

Mediterranean. This shallow sea area is called the Sicilian-Tunisian Platform, scientifically

known as the Pelagian Platform. The difference in sea depth between the shallow seas of the

Sicilian-Tunisian Platform and the deeper areas of the Western and Eastern Mediterranean is

visible to the east of the Maltese Islands.

In a distance of only 15 kilometers from the Maltese Islands, the variation in depths is

large. Depths vary from 200 meters to over 3000 meters and even over 4000 meters across the

Ionian Abyssal Plain. Sea depths also vary widely to the northeast of the Maltese islands. The

depths vary similarly between the shallow Sicilian-Tunisian Platform and the Western

Mediterranean basin at the west end of the Sicily Channel. These changes in depths and

escarpments are the result of long-standing geological contrasts.

These wide variations in the topography and bathymetry around the Maltese islands are

the result of processes that have been occurring over millions of years. These processes

include sedimentary deposition and volcanism that are controlled by movements in the mantle

and in the crust. The massive tectonic movements tore the lithified sediments apart and lifted

the islands above sea level. Plate tectonics theory links the most large scale features of the

world’s geology to the effects of movements of the large plates of the earth’s crust across the

surface of the globe. This concept shows that the Maltese island form a long-standing conflict

between the crustal plates of Europe and North Africa.


31

2.3 Pelagian Platform and the Pantelleria Rift System

The Pelagian Platform, the platform on which the Maltese islands lie, forms a shallow

shelf that separates the deep Ionian Basin from the Western Mediterranean. The sea-bed

topography is characterized by the Sicily Channel Rift Zone (SCRZ). This rift zone is a young

tectonic feature made up of three grabens which are the Pantelleria Graben, Malta Graben and

the Linosa Graben in which the eater reaches a depth of 1700 meters (Reuther and Eisbacher,

1985). This is shown in figure 2.2 below. These grabens make up a fault system that extends

throughout the Sicily Chanel from Southern Sicily to Tunisia. It has been responsible for the

major tectonic development of the Maltese islands (Illies, 1981). The SCRZ was interpreted in

many different ways. It was thought to be a set of pull-apart grabens (Reuther and Eisbacher,

1985; Reuther, 1990). It was more simply thought to be the result of the N-S extension regime

related to Tyrrhenian back-arc spreading (Argani, 1990). The rift zone was also interpreted as

a part of the Medina Wrench, which is a dextral transform fault of more than 800 kilometers.

It extends from the Sicily Channel to the Eastern end of the Medina Ridge, which is located at

200 kilometers SE of Malta.

The North African Margin have been subjected to different stresses and resulted in a

complex horst and graben system. This shearing motion has been associated with the major

shearing between the African and Eurasian Plates (Dewey et al., 1973). At the moment this

system appears to be stable with minor vertical motion taking place along the sides of the horst

blocks. The most of the fault system occurs at latitude 35˚N in the region of the Pantelleria-

Linosa-Malta Troughs. The direction of these normal faults indicates a NE/SW directed

tensional stress. This was attributed to the early Miocene crustal extension (Illies, 1981).

The grabens are bounded by normal faults that extended NW-SE. The rift is extending and is

being controlled by the dextral transforms that are reactivated faults. The Malta

Escarpment separates the Hyblean-Malta plateau from the deep Ionian Basin. It exhibits

normal faulting with a minor sinistral strike slip component. (Grasso et.al., 1989; Reuther et

al., 1993).


32

Figure 2.2: Bathymetry of the Sicily Chanel and main tectonic features of the Sicily Chanel.

This shows the Sicily Chanel Rift Zone-bounding normal faults and strike-slip lineaments

(modified after Reuther and Eisbacher, 1985 and Reuther, 1990). Also shown are the

Calabrian Arc subduction zone and epicenter of the 11/01/1693 earthquake (Boschi et al.,

2000). Inset shows the Maltese islands.

2.4 Seismicity around the Maltese Islands

Seismicity in the Mediterranean region is caused by the Eurasian and African plates.

Such plate movement caused stress which in turn builds up energy. This energy must in some

way be released and is usually released through seismic activity. A seismic map of

earthquakes that occurs in the Mediterranean region between 1980 and 2000 is shown in figure

2.3. A number of active or dormant volcanoes are found in the area around the Maltese

islands. Mt. Etna (Sicily) is found to the North; Mt. Epomeo (Ischia, Bay of Naples); the

volcanic islands of Stromboli and the Lipari Islands; and Mt. Albani, Mt. Vesuvius and the

Phlegraean Camps (Italy). The submarine Graham volcano and the volcanic island of


33

Panteleria both lie to the northwest of the Maltese Islands; the volcanic islands of Linosa and

Lampione are found to the southwest and Santorin volcano much further away from the

Maltese Islands.

Figure 2.3: Seismicity in the Mediterranean region between 1980 and 2000 for events

magnitude greater than 4.0 (Limestone Isles in a Crystal Sea-The Geology of the Maltese

Islands: Pedley, Hughes Clarke, Galea, 2002).

These volcanoes in the vicinity of the Maltese Islands affect its seismicity. Many

earthquakes in the past which affected the Maltese Islands were accompanied by volcanic

eruptions. In January 1692 an earthquake was felt both in Sicily and Malta. This earthquake

was accompanied by an eruption from Mt. Etna. These are called volcanic quakes which are

due to the sudden release of steam or other volcanic gases which are under pressure. These

type of earthquakes lie at different depths under the sea. The earthquakes around the Maltese

islands are not usually due to the volcanoes as these volcanoes are situated at a fair distance

away from the Maltese Islands. It is more probable that earthquakes around the Maltese

Islands are tectonic in origin and the volcanic eruptions could have been the result of the

widespread earthquakes in the region. The disturbances arising from the earthquakes occurring

close to the Maltese Islands are more frequent than originally thought.


34

Figure 2.4: More reliably located seismicity, 1990-2003. (From Said, 1997; Zammit, 2003)

The Sicily-Tunisian platform is characterized by a small magnitude earthquake activity.

These earthquakes are being monitored by the seismic stations on Malta and on the Pelagian

islands of Pantelleria and Lampedusa. The Maltese Islands lie about 200m far away from the

Afro-Eurasian plate boundary and so are not affected by the large magnitude earthquakes that

occur near this boundary. But a large earthquake say magnitude 7 in Southern Sicily will be

strongly felt in Malta. A strong quake occurred in Eastern Sicily on 11th

January 1693. This

earthquake left several victims in the country and also many damages occurred in Malta. This

means that the seismic potential of the faults in the Sicily Chanel can cause earthquakes that in

fact can be large in magnitude. Earthquakes being recorded by stations in the Sicily Chanel

including the station in Malta show that a large seismic activity is occurring near the Maltese

Islands. This seismic activity surrounding the Maltese Islands shows that the tectonic activity

that resulted in the formation of the Maltese Islands as known today is still going on but does

not cause any damage.


35

Figure 2.5: The seismogram of an earthquake located about 130 km SW of Malta on 6 June

2006, as recorded on the broadband digital seismograph WDD on Malta.

Chapter 3: SEISAN – Earthquake Analysis Software

36

Chapter 3 SEISAN - Earthquake Analysis Software

In this chapter, a brief description is given on how the Earthquake Analysis Software, Seisan

(Havskov and Ottermoller, 2003) works. The main properties of this program are explained

mainly the part used to determine Codaq values for the set of given earthquakes. A more

detailed description of how the program works is given in the manual of the program, Seisan.

3.1 Structure of Seisan-Directories

The Earthquake Analysis Software, Seisan 8.0 (Havskov and Ottermoller, 2003) is used

to obtain Codaq values for a number of chosen earthquakes. The whole Seisan system is

located in subdirectories residing under a main directory called Seismo. The system contains

many subdirectories containing information that the program needs to run. The following are

the main subdirectories:

REA: This contains earthquake readings and full epicenter solutions in the database.

WOR: The users work directory, initially empty.

PRO: Programs, source code and executables

INC: Include files for programs and subroutines in PRO and LIB

COM: Command procedures

DAT: Default and parameter files, e.g. station coordinates

WAV: Digital waveform data files

CAL: System calibration files

INF: Documentation and information

SUP: Supplementary files and programs

Table 3.1: The main subdirectories of SEISAN


37

The database of Seisan contains two main directories REA and WAV. The REA

directory contains all the readings and information about the earthquakes that needs to be

analyzed while the WAV directory contains all the waveform data.

Figure 3.2: Structure of SEISAN

The REA directory contains all the phase reading and the derived source information

like hypocenters. The main directory REA is sub divided into a number of directories which

correspond to different databases. These sub directories are created by the user and are used to

store all the earthquake events that are going to be analyzed. The database names can have

between 3 to 5 characters. Each database has default storage of events. Each event is stored in

a single S file in yearly directories and monthly subdirectories. If new data is entered into the

database it is automatically saved as an individual event file. But when the interactive work

has finished, the single event flies are overwritten and stored in monthly files. The CAT-files

are these monthly files that serve as a backup data for the single files.

S file database structure

\REA\MALTA\2010\07\

This is a single S file representing one event. This S file corresponds to an event that happened

on the 9th

of July, 2010. This event is being stored in the sub directory MALTA under the

main directory REA. Each event is given an ID. The ID line contains a unique ID and it also


38

contains status information about the event like last action and the last time when it was

updated. An example of an S-file name is: 09-0019-24L.s201007.

The location program uses these S-files as an input and also as an output when a permanent

update is done to the event. The letter before the . indicates the event type. This can be L,R or

D for local, regional or distant event respectively. This is the same indicator as given in the

header line of the S-file.

3.2 Waveform Data

3.2.1 Data Format

SEED (Standard for the Exchange of Earthquake Data) is standard defined by the

FDSN6. A data format was needed since large amounts of seismic data needed to be

transferred within the network, analyzed and backed up. This type of data format contains a

header. This header contains the instrument name, location, sensitivity and the data selection

containing the waveform data.

SAC (Seismic Analysis Code) is a general program designed for studying time series

data. It was developed in ForTran and converted to C which was then renamed to SAC2000.

SAC2000 is the primary tool used by seismologists to analyze seismic signals.

Seisan works with various waveform formats including SEISAN, GSE,

SEED/MINISEED and SAC. The waveform data is usually kept in one format only mainly for

simplicity. There may be different arguments on which format to choose that depends on the

user’s requirements. SAC and GSE are widely used formats. SEISAN is a different format

which is a multi trace binary format. It makes it possible to access individual traces. GSE is a

multi-trace ASCII waveform format. The GSE format can keep a number of traces but it is

6

International Federation of Digital Seismograph Networks


39

usually recommended not to include more than 3 traces in a single file. Data centers mostly

use SEED and MINISEED formats.

The events available at the Seismic unit at the University of Malta are in SAC format

and so this is the format chosen to store the events in Seisan. SAC is a single trace binary or

ASCII format with a large number of header parameters. SAC format is widely used in

programs that are research oriented. SAC format is also recommended when a single file

include more than 3 traces.

The WAV directory contains the event files with digital waveform data. The analysis

system always uses the WAV directory to search for the files. Waveforms area automatically

transferred to WAV. The event files in the WAV directory are usually of the form: yyyy-mm-

dd-hhmm-ssT.NETWO_nnn with the abbreviations yyyy: year, mm: month, dd: day, hh: hour,

mm: minute, ss: second, T: file type indicator usually S, NETWO: maximum 5 letter network

code and nnn: number of channels. When storing events in the WAV database, it is required

that the waveform names start either yymm or yyyy-mm. Therefore the database consists of

single files with names corresponding to time down to second as well as the event type (L, R

or D). This means that two events can get the same name. A new event can therefore be over

written on an existing event. When using MULPLT to enter new events into the database, the

use will be prompted if a new event is to overwrite an existing one.

A directory MALTA is created under the REA directory. This is done using the program

MAKEREA. Under the MALTA directory, 4 sub directories one for each year from 2006 to

2010. Each year sub directory is again divided into months. Using another program called

MULPLT the data was accessed and the traces for each event where seen on screen and also

registered. Once registered, an S-file is created for the event and each event is saved in a sub

directory according to when it happened.

There are two ways to get digital data into the database. One method is by making the

individual S file directly in the REA directories using the editor. But this is rather slow. If the

original data available is a digital event waveform file another method is available. As already

explained the waveform can have different formats. These waveforms are stored in the WAV

directory and usually also in the WOR directory. The main aim is that the digital data is


40

transferred from the field station, demultiplexed and converted to SEISAN waveform format.

This is done using the program MULPLT. This program plots channels form a single

waveform file. The user can than decide whether or not to keep this event. If the event is

chosen to be kept then an S file is created in the database and the event is now moved in the

WAV directory.

3.3 Programs

Seisan organize incoming data from different sources into different directories. This is

using a simple time ordered database and also using a set of programs that are installed in

seisan. Some of the most important programs include:

EEV: This program is used when working with single events. This is used to find a given

event in the database. When this event is selected a large number of options can be applied ton

it such as phase picking and earthquake location.

MULPLT: This is the main program used for signal analysis and plotting. This can be used to

pick phases and amplitudes.

HYP: This is the general program used for hypocenter location. It can use all global and

crustal phases and can use all types of input data whether from single stations or arrays.

EPIMAP: This is the hypocenter plotting program and is used to make epicenter maps and

hypocenter profiles.

CODAQ: This program is used to determine the attenuation of local earthquakes using the

coda Q. Another program, SPEC determines Q by calculating the spectral ratios or else

calculates the near surface attenuation using the spectral decay method.

Other programs are also available to create a database, to input and output a large amount of

data into the database and also to manipulate the waveform data.


41

3.4 Calculation of coda q, CODAQ

The coda Q program calculates q for a series of events and stations at given frequencies.

Average values of q can then be calculated and a q versus f curve is then plotted using the

calculated values. This program plots the individual events and also plots the filtered coda

windows. The principle used to calculate the coda q is the standard coda q method where a

coda window is bandpass filtered. An envelope is then fitted. This envelop is a calculated

RMS value of the filtered signal using a 5 cycle window. Then the coda q at the corresponding

frequency is calculated (Havskov et al., 1989). This program can use all the waveform file

types that are accepted by seisan.

3.4.1 Input

The calculations are done using the parameter file codaq.par and the events lists to be used are

given in codaq.inp. An example of an input file and a parameter file are shown below:

Figure 3.3: Example of an input file


42

Figure 3.4: Example of a parameter file

Start in s-times: the coda window usually starts at twice the lapse time which is the S-

travel time from the origin. This factor can be chosen differently. The S-time is calculated

from the P-time and so the P-time is inputted in the parameter of each event.

Absolute start time: A 0.0 parameter is usually used. A time different from zero can be

chosen and the start of the coda window is put at an absolute time relative to the origin. This

would mean different lapse times and so different q-values may be produced. This parameter

must be chosen long enough.

Window length: This is the coda window measured in seconds and it must be chosen at

least to be 20secs for stable results.

Spreading parameter: It is the geometrical spreading parameter and the value of this

parameter is usually chosen to be 1.0.


43

Constant v in q=q0*f* *v: For all values of q(f), q0 is calculated keeping the value of v

fixed.

Minimum signal to noise ratio: When calculating an average value of q, the signal to

noise ratio must be chosen to be above this value. The signal to noise ratio is calculated using

the last tRMS secs of the filtered window and the first tRMS secs of the data file window. If

the data file starts with noise then this ratio will not be accurate. Usually a reasonably value of

5.0 is chosen.

Maximum counts to use: this is the maximum count in a coda window above which the

window is not used.

Noise window in front of signal and length of noise window, tnoise and tRMS: The first

number is the number plotted in front of the signal and gives the number of seconds of noise.

This is the number of noise found before P. The second number is the length of the noise

window that is then used to calculate the single to noise ratio.

Minimum correlation coefficient: In order that the average value of q calculated is

correct, the correlation coefficient must be larger than or equal to this value. This value

depends on the data being analyzed and a value higher than 0.5 is chosen. In reality the values

of this coefficient is negative.

Number of frequencies: The number of frequencies at which each value of q is

calculated. Maximum number of frequencies is usually 10.

Frequency and bands: These are the frequencies and each corresponding band. The

frequency band should increase as the frequency increase. E.g. 8, 3 mean that the signal is to

be filtered between 6.5 and 9.5Hz. It is important that each band has the same amount of

energy. This is done by using a constant relative bandwidth filtering. RBW is the relative

bandwidth and is defined by (fu-fl) / f0 where fu and fl are the upper and lower frequencies.

Such a filter would then be for example: 4±1. The energy in each filter band is represented by

the frequency. This frequency is the geometric center frequency and is given by

When calculating the coda Q at the given frequency, fu and fl are calculated such that the given

bandwidth is used. The actual values of fu and fl give the specified central frequency.


44

Figure 3.5: Calculating Codaq

Default stations: Stations be used are specified here or else in the codaq.inp file. In the

following line the components are specified. Then the event station information is obtained

from the codaq.inp file. In this case only one station is specified since many events recorded at

the same station MN_WDD are being analyzed.

The codaq.inp file will consist of al list of events. Each event has its own identity with

which it is identified. The program used default stations that are given in the codaq.par. An

example is given below:

\seismo\REA\MALTA\2010\07\13-2330-01L.s201007



3.4.2 Operating CODAQ

The program read the parameter file, codaq.par and also the input file, codaq.inp

containing the events to analyze. These files are both found in the current directory. In the S

file the name of the waveform is given. The program then searches for the station and the

components being used. The program searches in the WAV database and so the program can

work without moving the data from the database. The data header was adjusted for the correct


45

origin time of all events since the program uses the origin time and P arrival times from the S

files to calculate the S arrival time.

If no plot is chosen, one line will appear on the screen for each frequency. Each event is

recorded in a new page by the program. If the program is plotting the events on screen, the

next plot is obtained by hitting the return button.

A summary is given at the end. This information if found in the output file codaq.out. The

program has some abbreviations that are given below.

H: Focal depth

M: Magnitude

TP: P travel time

TC: Start time of coda window relative to origin time

F: Frequency

Q: Corresponding coda q, if 0 value is >1000 or negative

S/N: Signal to noise ratio AV

Q: Average q

SD: Standard deviation for average

NT: Total number of q values at all frequencies

N: Number of q values at all frequencies

q: Average of q values

1/q: q is calculated at 1/q averages

f:1/q: Q is calculated using the relation derived from the 1/q averages

cq0: Constant q0 obtained using the fixes user selected v

v: Constant v determined corr: Correlation coeffieicent in determining q vs f

Table 3.6: Abbrevaiations used in SEISAN


46

The coda q value is calculated by program by reading the P arrival time from the S file.

This P arrival time is written in the S file manually for each event or else using a progam. The

S file for each event is in Nordic format. This format uses free columns to obtain a readable

format. There are many ways by which data can be written in an S file using different line

commands. The line command chosen depends on the type of information to be inputted in the

S file. The line 4 command has been chosen since different events recorded at the same station

are being analyzed. The line 4 :

Figure 3.7: Type line 4 using Nordic Format


47

3.4.3 Output

When running the codaq program an output file codaq.output is generated. This is the

parameter file consisting of all the events each generated in a separate line. These are the list

of events that have been accepted by the program. The program accepted these events after

calculating the correlation and the signal to noise ratio of each event separately. Each line

event also has its average q value. The q values are averaged directly and the 1/q are averaged

separately. In this output file, there will also be the fits to the relation q=q0*f* *v.\

Another output file is generated codaq1.out that contains the same output as codaq.out but no

print for each event is generated. Below is an example of the codaq.out file:

Figure 3.8: An example of an output file

The following figure shows an example of a codaq plot. No options are available for the codaq

plots and the length of the window is always the first 200 secs from the original trace. If the

origin time or coda window is outside the 200 secs window, the coda window is not plotted.


48

Figure 3.9: A codaq plot for an earthquake recorded on 7th August 2009 by WDD station.

(SEISAN, Havskov and Ottermoller, 2003)

Shown here is an example of a coda Q plot. The trace shown on top is the original trace

and the coda windows shown below are the filtered ones. The selected filtered coda window

has 15 secs of noise in front. S/N ratio is calculated from the first 5 secs of noise shown. On

each filtered plot is given F: Center frequency, Q: Q-value, zero value means no Q-value

could be calculated, S/N: Signal to noise ratio.

Chapter 4: Data Processing

49

Chapter 4 Data Processing

In this chapter an overview of the seismic recording in Malta is given. A brief description on

the website where the seismic data is available online is also given. Further in this chapter the

data set chosen for the study of codaq is given.

4.1 Seismic Recording in Malta

The main aim of a seismic instrument is to record ground motion. This ground motion is

the result of both natural and man-made disturbances. Such seismic instrument is the

seismograph which is an instrument capable of making a seismic disturbance visible by

writing it as a continuous record of ground motion which is known as the seismogram. The

visible seismogram is the actual conversion that occurred between the signal that arrived at the

seismometer and a time record of the seismic event. The seismic ground motion that arrives at

a seismograph has the form of analogue data. This is then converted into electrical signals,

amplified, filtered and finally registered in a chart recorder. This is the acquisition of a

seismogram.

The first seismograph in Malta was installed at the beginning of the 20th

century at the

University of Malta. The seismograph that was installed was a Milne-Shaw horizontal

pendulum seismograph. At this time this was the main seismograph used world-wide due to its

high reliability. This instrument operated in Malta till around the 1950. Such recordings of this

seismograph are still available at the University of Malta.

The Milne-Shaw seismograph was replaced by a vertical component long period

Sprengnether seismograph in 1977. This was again installed at the University of Malta. Such

instrument having photographic recordings had a main disadvantage. This was that the


50

seismograms had to be developed each time and so were not available instantly after the

seismic disturbance occurs. Such instrument was capable of recording events having

frequencies of 0.01-0.1Hz and was only capable of recording teleseismic earthquakes only and

so few seismic recordings are available of this time.

In 1982, a 3-component, short period analogue seismograph was installed at the

University of Malta. This short period seismometer has a very short natural period and a high

resonant frequency. It is capable of responding to a seismic frequency of 1 to 10Hz and a

period range of 0.1 to 1s (Lowrie, 1997). It records 3 components of ground motion. These are

the vertical component (Z component), the North - South component (N component) and the

East – West Component (E component). This seismometer produced visible recordings unlike

the previous seismometers that used photographic recording. This was a huge advance is the

seismology of the Maltese Islands since now it was possible for local events to be detected.

All the events between 1983 and 1992 that were detected by this seismometer are chart

recorded and found at the University of Malta.

4.1.1 The Wied Dalam Station, WDD

In the early 1990’s the need to replace the existing analogue seismographs bye digital

one emerged. Digital broadband seismographs increase the dynamic range and the frequency

band of each event being recorded by the seismogram. A search started for a place having

minimum disturbances both natural and artificial that was appropriate for this seismograph to

be installed. In 1995 this digital seismograph was installed in Wied Dalam in the south of

Malta. The station is known as WDD. This seismic observatory station is located about 20m

below ground at 35.8374N, 14.5245E. The situation of this station in Malta is shown in figure

4.1.


51

Figure 4.1: Location of the WDD station. (Digital Seismic recording in Malta – 13 years on,

Galea, P; Aguis, M, 2008)

The seismometer at Wied Dalam is a digital broadband seismograph. This seismometer

is a Streckeiser Model STS-2 sensor triaxial component connected to a QUANTERRA 24-bit

integer data acquisition system. This is shown in figure 4.2.

Figure 4.2: Wied Dalam Station WDD in the south of Malta. (Digital Seismic recording in

Malta – 13 years on, Galea, P; Aguis, M, 2008)


52

It operates many channels

HH at 80 samples per second

BH at 20 samples per second

LH at 1 sample per second

VH at 0.1 samples per second

UH at 0.01 samples per second

In our case the HH component was chosen since seismic waves of local earthquakes are

predominant in the higher frequencies and more important is that they are attenuated at long

distances.

SeisComp is responsible for data transmission. It is a concept used within the

MEREDIAN7 project. This software is responsible for the acquisition, recording, monitoring

and controlling of seismic data. Once data is recorded it is transferred via the internet to

another computer at the University of Malta where another copy of the data is kept.

The WDD station forms part of the MEDNET, Mediterranean Network. This is a

network of the broadband seismographic stations that are installed in the countries of the

Mediterranean region shown in figure 4.3. This network having 14 stations is maintained by

the Instituto Nazionale Di Geofisica in Rome, the INGV with the help of other geophysical

institutes. Such network gives an instrumental coverage of the Mediterranean area which is an

area of high seismicity and a complex tectonic activity. Its aim is to improve the knowledge of

the structure of the Mediterranean region and so this will help to minimize earthquake losses.

Data can only be accessed from our station. The disadvantage of having one station is that the

epicentral point of a seismic event cannot be determined. This is because using the arrival

times of many seismic phases recorded at different stations, the earthquake hypocenter and

origin time can be determined.

7 An EU-funded project coordinated by the ORFEUS Data Centre in de Bilt, the Netherlands

(www.orfeus-eu.org).

http://www.orfeus-eu.org/


53

.

Figure 4.3: The MedNet Network (Mediterranean Very Broadband Seismographic Network,

I.N.G.V, 2011)

4.1.2 Aims of the Malta seismograph station

The primary aim of the seismograph station in Malta is to continuously monitor and analyze

the seismic activity in central Mediterranean. It focuses its analysis mainly in the seismicity

around the Maltese Island. Using such information seismologists can identify active faults in

the sea bed of the Sicily Channel. Such information will be useful in providing an assessment

of the seismic hazard in the Maltese Islands. This seismograph also improves the epicentral

location capability of the Mediterranean. Another aim is to contribute to the world-wide

gathering of seismic data done by the network of digital seismographs. This will provide

accurate information about the structure of the Earth.


54

4.2 Seismic Monitoring and Research Unit at the

University of Malta

All the seismic data that has been recorded at WDD since 2006 have been uploaded into

an online database. Such database can be accessed by the following link:

http://www.phys.um.edu.mt/seismic/. The main page of this webpage is shown in figure 4.4. This

website has many features and provides much information about the seismic events recorded by at the

WDD. Each event can be analyzed individually and each plot can be viewed using the program

Seisgram2k.

4.2.1 Earthquake locations

Various steps are carried out to generated earthquakes locations. Analysis is carried

daily at 4am local time and the earthquake is then verified within the next 24 hours.

Earthquakes recorded on WDD are located especially those occurring in the Sicily Channel.

Location of regional earthquakes is also done by the stations CEL and IDI. Single-Station

earthquake location is done. A list of events for a particular day is produced by LESSLA

(Aguis, 2006). Then events are added to the central database and grouped as either an

Earthquake or a Blast. P and S arrival times are checked manually. Each event is then

classified by the SMRU8 according to its quality. When the information available is reliable an

earthquake is verified and marked in red and placed in the database.

8 Seismic Monitoring and Research Unit

http://www.phys.um.edu.mt/seismic/


Figure 4.4: The main page of the Seismic Monitoring and Research Unit website

(http://www.phys.um.edu.mt/seismic/)



56

4.2.2 The Website

The website includes several press releases. These give details about seismic data and

maps of major earthquakes that have been recorded by the WDD and are here released to the

public. There are also many other links available on the website. These links give information

about the Seismic Monitoring and Research Unit and also projects, papers, posters and

presentations are available here. Apart from these links a real-time plot of seismic activity is

available. Such plot of 13 February 20011 is seen in figure 4.5. This displays all the data

being recorded during that day. Using such a plot the seismic activity of that day can be

analyzed in detail. Each day this data is stored into the database and a new active plot begins.

4.2.3 The Seismic Database, Online

All events recorded at the WDD since 2006 are stored in this database. A Google map is

available. This map displays the epicenters of all the events that have been verified by the

SMRU. A query is also available. This make it possible to the user to chose earthquakes

depending on a certain criteria. Such criteria can be the date when the earthquake occurred or

the latitude and the longitude of an earthquake. Such list of events is shown in figure 4.6.

The row in the table accounts for a single event, displaying all the relative information

about the event. Manual Attributes are displayed at the end of each row. These are labels used

reveling information about the event. Each event can be viewed separately. The window

displays a Google map showing the final location of the event. A blue circle is shown. This

has its center at the WDD station. The red line is equal to the radius of the circle. It extends to

the circumference and has an angle equal to the azimuth. This angle is calculated relative to

the geographical north of the map. The location of the earthquake is calculated by LESSLA

(Aguis, 2006). This is shown by the point where the red line intersects with the circumference

of the circle. Nine seismographs are displayed for each event, one for each of the nine

channels. The event can be seen by using the program Seisgram2k. A single event display is

shown in figure 4.7.


Figure 4.5: The real-time plot for 13 February 2011




58

Figure 4.6: The online database of seismic events. It is displaying the events that occurred

between 1st January 2006 and 31

st January 2011 having latitude between 32 and 36 and

longitude between 12 and 1 . (http://www.phys.um.edu.mt/seismic/)



Figure 4.7: Single event displayed online .Its displaying the main page of the January 1 earthquake located to the SW of the Maltese

Islands. This page displays all the relative information about the event. (http://www.phys.um.edu.mt/seismic/)



60

4.3 The Data Set

4.3.1 South of Malta Events

The earthquakes were chosen from the database available at the Seismic Monitoring and

Research Unit at the University of Malta. Out of 185 events, a total of 43 events are finally

selected for the determination of the Q factor. These are given in Table 1.1 in Appendix 1.

These occurred during 1st January, 2006 and 31

st January, 2011 in the latitude area from 32 to

36 and longitude area 12 to 1 . This area chosen for analysis of coda Q is shown in the

hybrid map in figure 4.8.

Figure 4.8: A hybrid map showing the South of Malta earthquakes between latitude area 32 and 36 and longitude 12 and 1 (http://www.phys.um.edu.mt/seismic/)



61

4.3.2 North-West of Malta Events

Another area of study was chosen for determination of the coda Q factor and so that an

analysis of the variation of codaq Q from one region to another could be done. These events

were chosen in the North-West of Malta area near Pantelleria. Out of 22 events, 6 events were

finally selected for the determination of the Q factor. These are given in Table 1.2 in Appendix

1. Such events again occurred between 1st January, 2006 and 31

st January, 2011 in the latitude

area from 36 to 37.5 and longitude area 8 to 1 . This area of study is shown in the hybrid

map in figure 4.9.

Figure 4.9: A hybrid map showing the North-West of Malta earthquakes between latitude area

36° and 37.5° and longitude 8° and 14°. (http://www.phys.um.edu.mt/seismic/)



62

4.3.3 Crete Events

For further comparison another area of study was chosen for determination of the coda

Q factor. This is the Subduction Zone near Crete. This will allow us to analyze the difference

between the Q values which were obtained around the Maltese Islands and the Q values

obtained in this highly active tectonic region. This could be done since real-time data is

received at the Seismic Monitoring and Research Unit at the University of Malta from the IDI

station on Crete. The Institute of Geodynamics’ database consisted of over 2000 events that

were recorded by IDI in this subduction zone. Out of these, the events that were available at

the Seismic Monitoring and Research Unit could be chosen. 42 events were finally selected

for the determination of the Q factor. These are given in Table 1.2 in Appendix 1. Such events

again occurred between 1st January, 2006 and 28

th March, 2011 in the latitude area from 33.5

to 35.5 and longitude area 23 to 27 . This area of study is shown in the hybrid map in figure

4.10.

Figure 4.10: A hybrid map showing the Crete earthquakes in the latitude area 33.5 and 35.5 and longitude 23 and 27 Shown also here is IDI station on Crete.




63

Each plot was checked manually for data quality. This includes duration, distortion,

spikes, saturation and signal-to-noise ratio. More than half the original data were rejected by

visual inspections mostly due to low signal-to-noise ratio. Other events were discarded due to

the fact that length of the coda wasn’t long enough for a time window of 20 seconds to be

taken after twice the lapse time. The final data set for this area consisted of 43 events in the

south of Malta area, 6 events in the North-West of Malta area near Pantelleria and 42 events in

the Subduction Zone near Crete.

4.3 Calculating the Qc - values

The Q values were calculated through the CODAQ in the seismic analysis package

SEISAN 8.0 (Havskov and Ottemoller, 2003). The lapse time portion of the coda wave used

in this is selected at where is the S- wave travel time. This is calculated using

the P-wave arrival time using the equation

The is used in this way so that direct and forward scattering waves are avoided

(Rautian and Khalturin, 1978). All the selected seismograms are then bandpass-filtered at

central frequencies of 2.0, 5.0, 7.0, 9.0 and 12.0 Hz with bandwidths of 1, 2.5, 3.5, 4.5 and 6,

respectively. An increasing frequency band is used for increasing central frequency. This is

done to avoid the ringing effect and to take constant relative bandwidths for getting an equal

amount of energy into each band (Havskov and Ottemoller, 2003). One window length was

taken at 13 sec.

The RMS amplitude of the last 5 sec cycle length of the lapse-time window is divided by

the noise data of the same length before the onset of the P wave to calculate the signal-to-noise

ratio. The Qc were accepted only when the correlation coefficient, C for the best-fit line for

coda decay slope with respect to lapse time were greater than 0.5. Initially the signal-to-noise

ratio was chosen to be 2. But the number of data reduces considerably and so events having


64

signal-to-noise ratio, S/N greater than 1.2 were chosen. Other events were rejected for reliable

values of Qc.

A description of the single back scattering model is given previously. The coda wave

amplitude at central frequency f and elapsed time t from the origin is found by band-

pass filtering the coda window trace data using a 6-pole Butterworth filter centered at

frequency f and calculating rms values using a sliding window of length 5/f sec. Then a time

decay envelope is fitted to this filtered signal. This is seen in figure 4.10.

Figure 4.11: found by band-pass filtering the signal and then fitting a

time-decay envelope to this filtered signal.

Finally, values of Qc are calculated using the slope of the linear regression of the

logarithm of product of RMS amplitude and lapse time against lapse time t. The

slope of such graph, b is given by

and so values of Qc can be calculated. Figure

4.11 shows the steps involved in the computation of the values of Qc with time. This is

procedure is carried out the program CODAQ in the seismic analysis package, SEISAN. The

Qc - values for such data sets were then averaged for each central frequency. Also the standard

deviation for each central frequency was calculated.


65

Figure 4.12: Procedure in calculating Qc (a) Unfiltered data trace with coda window, (b) and

(c) bandpass-filtered amplitudes of coda window at 1.5-2.5 Hz and 9.0-15.0 Hz respectively,

(d) and (e) the RMS Amplitude values multiplied with the lapse time along with the best

square fits of selected coda window at central frequencies 2 and 12 Hz respectively. The Qc is

determined from the slope of best square line.

(Coda Q Estimates in the Andaman Islands using Local Earthquakes: Parvez, A,. et al, 2008).

Chapter 5: Results

66

Chapter 5 Results

In this chapter the results of the Coda Q values calculated by SEISAN are given. Also given in

this chapter are the calculations done.

5.1 The frequency dependence of Q relationship

The Q factor increases with frequency (Mitchell, 1981) and it follows the following

relation

(5.1)

where is the quality factor at the reference frequency f0 (generally 1Hz) and is the

frequency parameter. This power law is fitted for Qc at each frequency.

This law is arranged in logarithmic form and is given by the following equation:

(5.2)

Therefore the value of the frequency parameter is obtained from the slope of a graph of

against and the value of is obtained from the intercept of such graph.

Chapter 5: Results

67

5.2 Results for the South of Malta Earthquakes

Table 5.1 shows the mean values of Qc at different central frequencies. Given also in this table

are the standard deviation and the number of observations for each central frequency. These

are used for the calculation of Qc. In Figure 5.1 all the values of Qc are plotted against

frequency. These values are given in Table 2.1 in Appendix 2. In figure 5.2 the mean Qc-

values against central frequencies are plotted. . A log is then plotted as shown in figure 5.3 so

that the relationship for the South of Malta events could then be obtained. It is observed that

Qc increases as the frequency increases.

Table 5.1

Average Quality factor, Qc and Estimated Standard Deviation at different frequencies

Frequency (Hz) Qc S.D N ln f ln Q

2 91 41.012 36 0.693 4.512

5 591 356.381 40 1.609 6.382

7 912 281.428 33 1.946 6.816

9 1349 911.461 36 2.197 7.207

12 1984 808.223 23 2.485 7.593

In the column heading, S.D. indicates the standard deviation and N is the number of observations made

for each central frequencies.

Chapter 5: Results

Figure 5.1: A graph of Qc against frequency for the South of Malta events

0

2000

4000

6000

8000

10000

12000

0 2 4 6 8 10 12 14

Qc

Frequancy (Hz)

Chapter 5: Results

Figure 5.2: A graph of the Average Q values against frequency for the South of Malta events. Vertical error bars are shown.

0

500

1000

1500

2000

2500

3000

0 2 4 6 8 10 12 14

Aver

age

Q

Frequency (Hz)

Chapter 5: Results

Figure 5.3: A Graph of ln (Qc) against ln (f) for the South of Malta events

4

4.5

5

5.5

6

6.5

7

7.5

8

0 0.5 1 1.5 2 2.5 3

Ln (

Q)

Ln (f)

Chapter 5: Results

71

5.2.1 The frequency dependence of Q relationship for the South of Malta

Events

The graph of against is plotted and the following calculations are made. This is


and

Therefore

Chapter 5: Results

72

5.3 Results for the North-West of Malta Earthquakes



are used for the calculation of Qc. In figure 5.3 all the values of Qc are plotted against


values against central frequencies are plotted. . A log is then plotted as shown in figure 5.6 so

that the relationship for the North-West of Malta events could then be obtained. It is observed

that Qc increases as the frequency increases.

Table 5.2



2 117 11.313 5 0.693 4.762

5 519.5 41.719 5 1.609 6.253

7 1489.5 1175.918 5 1.946 7.306

9 3453.5 2901.259 5 2.197 8.147

12 4028 3073.086 3 2.485 8.301



Chapter 5: Results

73

Figure 5.4: A graph of Qc against frequency for the North-West of Malta events

0

200

400

600

800

1000

1200

1400

1600

1800

2000

0 2 4 6 8 10 12 14

Qc

Frequency (Hz)

Chapter 5: Results

74

Figure 5.5: A graph of the Average Q values against frequency for the North- West of Malta events. Vertical error bars are shown.

0

1000

2000

3000

4000

5000

6000

0 2 4 6 8 10 12 14

Aver

age

Q

Frequency (Hz)

Chapter 5: Results

75

Figure 5.6: A Graph of ln (Qc) against ln (f) for the North-West of Malta events

4.5

5

5.5

6

6.5

7

7.5

8

8.5

9

0 0.5 1 1.5 2 2.5 3

ln (

Qc)

ln (f)

Chapter 5: Results

76

5.3.1 The frequency dependence of Q relationship for the North-West of

Malta Events



and

Therefore

Chapter 5: Results

77

5.4 Results for the Subduction Zone near Crete

Earthquakes



are used for the calculation of Qc. In figure 5.7 all the values of Qc are plotted against


values against central frequencies are plotted. A log is then plotted as shown in figure 5.9 so

that the relationship for the Crete events could then be obtained. It is observed that Qc

increases as the frequency increases.

Table 5.3



2 204.5 180.312 34 0.693 5.321

5 470.0 176.777 37 1.609 6.153

7 938.0 661.852 36 1.946 6.844

9 1520.0 185.212 32 2.197 7.326

12 2720.5 647.003 33 2.485 7.909



Chapter 5: Results

78

Figure 5.7: A graph of Qc against frequency for the Crete events

0

500

1000

1500

2000

2500

3000

3500

4000

4500

0 2 4 6 8 10 12 14

Qc

Frequency (Hz)

Chapter 5: Results

79

Figure 5.8: A graph of the Average Q values against frequency for the Crete events. Vertical error bars are shown.

0.0

500.0

1000.0

1500.0

2000.0

2500.0

3000.0

3500.0

0 2 4 6 8 10 12 14

Aver

age

Q

Frequency (Hz)

Chapter 5: Results

80

Figure 5.9: A Graph of ln (Qc) against ln (f) for the Crete events

5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Ln Q

Ln f

Chapter 5: Results

81

5.4.1 The frequency dependence of Q relationship for the Crete Events



and

Therefore

Chapter 5: Results

82

5.4.2 The variation of Q with depth for the Crete Events

The 42 events for the Crete Subduction Zone were divided into 5 groups having 15km,

25km, 35km, 45km and 53km as their maximum depth respectively. Table 5.4 shows the Q0

value and the frequency dependent α values calculated using the frequency dependence of Q

relationship (Mitchell, 1981) for each group of events.

Table 5.4

Estimated Q0 and Frequency Dependence α values at Different Depths

Depth Range(km) Q0-value α - value N

4-15 143.88 ± 1.63 0.93 ± 0.26 10

17-25 118.98 ± 1.15 0.71 ± 0.08 9

26-35 104.38 ± 1.06 0.76 ± 0.03 14

36-45 86.74 ± 1.36 1.25 ± 0.16 6

46-53 65.17 ± 1.49 1.33 ± 0.21 3

N is the number of observations made for each group having different maximum depth.

Chapter 5: Results

83

Figure 5.10: A Graph of ln (Qc) against ln (f) for the Crete events having depth in the range 4-15 km.

5

5.2

5.4

5.6

5.8

6

6.2

6.4

6.6

6.8

7

7.2

7.4

7.6

7.8

8

0 0.5 1 1.5 2 2.5 3 3.5

Ln Q

Ln f

Depth = 4 - 15km

Chapter 5: Results

84


5

5.2

5.4

5.6

5.8

6

6.2

6.4

6.6

6.8

0 0.5 1 1.5 2 2.5 3

Ln Q

Ln f

Depth = 17 - 25km

Chapter 5: Results

85


5

5.2

5.4

5.6

5.8

6

6.2

6.4

6.6

6.8

0 0.5 1 1.5 2 2.5 3

Ln Q

Ln f

Depth = 26 - 35km

Chapter 5: Results

86


5

5.2

5.4

5.6

5.8

6

6.2

6.4

6.6

6.8

7

7.2

7.4

7.6

7.8

0 0.5 1 1.5 2 2.5 3

ln Q

ln f

Depth = 36 - 45km

Chapter 5: Results

87


5

5.2

5.4

5.6

5.8

6

6.2

6.4

6.6

6.8

7

7.2

7.4

7.6

7.8

8

0 0.5 1 1.5 2 2.5 3

Ln Q

ln f

Depth = 46 - 53km

Chapter 5: Results

88

Figure 5.15: A Graph of Q0 against depth for the Crete events

60

70

80

90

100

110

120

130

140

150

0 10 20 30 40 50 60

Q0

Depth (km)

Chapter 6: Discussion

89

Chapter 6 Discussion

In this chapter the values obtained for the South and the North-West of Malta events are

discussed and compared together. These Qc values obtained for these two areas are then

compared to values obtained for the Subduction Zone near Crete in this study and to other Qc

obtained in other different regions of the world.

6.1 Analyzing the results

The relationship obtained for the North-West of Malta Region is 24.62f 2.1

. Such a low

value is normally associated with areas that are tectonically active. A low value in this area is

associated with the NW trending Pantelleria Rift or Sicily Channel Rift Zone (SCRZ) which is

a system that features three grabens of Miocene Pliocene age. These are the Pantelleria

Graben, Malta Graben and the Linosa Graben. Waves from the North-West of Malta travel

along the grabens and are usually more attenuated. In this region the crust is thinner and so

these waves could be sampling the mantle. As mentioned earlier, areas characterized by soft,

molten rocks usually have Qc values. Since the crust in a graben system is thinner, the waves

usually travel through the mantle which is characterized by soft and molten rocks. So a low

value of attenuation in this area could be attributed to this fact. In this area strong frequency

dependence was found. This was as expected since the frequency parameter increase as the

tectonic activity of the region increase. This could be related to the size of heterogeneities.

This active fault system is clearly studied using the accurate plotting of earthquakes.

Qc values were also obtained for the South of Malta region. The relationship obtained for

this region is 30.26f 1.73

. Again in this area the earthquakes recorded were small in magnitude.

This shows that energy along these faults is being released gradually but in small amounts.


90

The Qc values obtained in this area were higher than those obtained for the North-West of

Malta region and the frequency dependence is slightly higher in the North- West of Malta

region. A scattered activity in the Sicily Chanel is obtained when plotting the epicenters of

earthquakes correctly. This scattered activity roughly coincides with the trend of the

Pantelleria Rift. These earthquakes are shallow earthquakes since they occur within the upper

25km of the earth’s crust.

The relationship obtained for the Subduction Zone near Crete is 63.83f 1.43

. Such a low

value is associated with active tectonic areas and this is as was expected since this is a

subduction zone which is a highly active tectonic region. The variation of Q with depth was

also investigated in this region. This was done by plotting values of Q0 and investigated how

the attenuation characteristics vary with depth. It was found that Q0 decreases linearly with

depth. This means that attenuation of waves of 1Hz frequency increases with depth. This result

can be interpreted in many ways. One explanation is based on the observation that shear waves

are strongly attenuated as they travel through partially molten regions of the mantle. This

could be attributed to waves propagating across the mantle wedge. Such waves propagate

efficiently as they travel within the cold, high-strength lithosphere slab. Such variations with

depth at subduction zones have been used in past studies to identify the geometry of

subduction zones (Oliver and Isacks, 1967; Barazangi and Isacks, 1971; Barazangi et al.,

1972; Barazangi et al., 1973; Mele, 1998). The decrease of Q0 with depth can also be due to an

increase in the heterogeneity of the medium with depth beneath the study region.

6.2 Comparing the South of Malta, North-West of Malta

and Crete Results

The frequency relationship obtained for the North-West of Malta is 24.62f2.1

compared

to a 30.26f1.73

relationship obtained for the South of Malta and the 63.83f 1.43

relationship

obtained in the subduction zone near Crete. The value of Q0, the quality factor at 1Hz in the

North-West of Malta is lower than the value obtained in the South of Malta. This was the


91

expected case as the waves in the North-West of Malta were expected to be more attenuating

than those in the South of Malta. The frequency parameter also reflects this. The frequency

parameter increases as the tectonic activity of the region increase. This results shows that the

North-West of Malta is more tectonically active than the South of Malta.

Comparing values at the same frequency from the South of Malta and Crete one can

observe that Q values around Crete are higher. Taking for example the 2Hz frequency, the Q

value in the South of Malta are 91 compared to a Q value of 204.5 in the Crete region. This is

also observed at higher frequencies. Taking the 12 Hz frequency, the Q value for the South of

Malta region is 194 compared to a Q value of 2720.5 in the Crete region. This shows that

seismic waves from Crete are less attenuated as they travel towards Malta. Comparing the

values from to the South of Malta to those obtained from Crete one should keep in mind that

the crust around the Maltese Islands is different from that in the Crete area. There is evidence

that the low-frequency band dips down towards the edge of the Malta Escarpment, where

landward-dipping reflectors separate continental and oceanic crust lie in the central tract of the

Malta Escarpment. The crust in the Ionian region is Oceanic crust while that around the

Maltese Islands is Continental crust. This difference in crust type around the two regions is

shown in figure 6.1. Oceanic crust is composed of basalt, a dense rock while continental crust

consists of granite which is a less dense igneous rock. This explains why seismic waves

travelling from Crete are less attenuated since waves travelling through hard competent rock

are less attenuated than waves travelling trough soft, molten rocks. This could explain why

seismic waves travelling from Crete to Malta, a distance of around 814km are felt in Malta

while earthquakes occurring at the same distance in Italy or even at less distance are not felt in

Malta.


92

Figure 6.1: A Google Map showing different types of crust between the Malta Escarpment

and Ionian Region (Google Maps, 2011)

The results for the South, the North West of Malta together with those of Crete are plotted

together against frequency as shown in figure 6.2. In the three areas it can be shown that the

values of Q increases as the frequency increases. The Q values for the South of Malta events

vary linearly with frequency while the Q values for the North-West of Malta do not follow any

linearity. The value of frequency parameter, α is different for the three different regions

studied here. This will affect the Q values to vary differently from one region since they

variation of Q with frequency depend on the variation of α. For low frequencies up to 5Hz the

values of the North-West are linear but as frequency increases the Q values do not follow any

linearly but increase rapidly than the values at low frequencies. Taking for example Q values

at 5Hz, the values obtained for the North-West and Crete are lower than the values obtained

for the South of Malta. This means that at low frequencies, seismic waves at the North-West

of Malta and at the Subduction Zone in Crete are more attenuated than those in the South of

Malta. But as frequency increase, the Q values for the North-West and Crete are higher than

those for the South of Malta. The values for the North-West of Malta increase drastically. This

could be attributed to the number of problems that were encountered when analyzing the

North-West of Malta events.


93

South of Malta Crete North - West of Malta

Figure 6.2: A graph of the Average Q values against frequency for the North-West of Malta, the South of Malta and Crete evens

0

500

1000

1500

2000

2500

3000

3500

4000

0 2 4 6 8 10 12 14

Aver

age

Q

Frequency (Hz)


94

Inaccuracy in such a value could be due to the fact that very few events where recorded

in this study area. 22 events were available in this area compared to the 185 events that were

available in the South of Malta. These were then again down listed since not all events could

be used for the calculation of codaq. In fact only 6 events were finally chosen for the study of

codaq in this area. Another problem could also be that since the earthquakes recorded are

usually small in magnitude, the epicenter plotting is usually inaccurate. If the accuracy of

location is increased, then a correlation can be obtained between these events and the

individual active faults of the graben systems. The small size of the earthquakes shows that

energy along these faults is being released gradually but in small amounts. Many other small

earthquakes are occurring but having small magnitude does not allow an estimate for the

epicenter location to be made.

Another problem could be that the events in the North-West of Malta have a lower

signal to noise ratio than those in the South of Malta as seen in figure 6.2. Shown also here is

that the events from the South of Malta have a nicer waveform that the events from the North-

West of Malta. Both in both areas it can be shown that coda wave amplitude exponentially

decreases as the lapse time increases. The study for the North-West of Malta becomes

unreliable at high frequencies and so the study should be carried out again when a more

reliable dataset is available.

Figure 6.3: Comparing the South of Malta seismograms to the North-West of Malta

seismograms.


95

6.3 Comparison with other Areas

Table 6.3

Frequency dependence of Qc for different tectonic and volcanic areas in the world.

Zone Relation Qc= Q0 f α Authors

NW of Malta 25 f 2.1

This Study (2011)

Mt. Etna 29 f 0.9

Del Pezzo et al. (1995)

South of Malta 30 f 1.73

This Study (2011)

Tres Virgenes Volcanic Area (Mexico) 50 f 0.65

Wong, Rebollar, Munguia (2001)

Anaolian Highlands 51 f 1.01

Akinci et al. (1994)

State of Washington 63 f 0.97

Havskov, Malone, Mcclurg, Crosson (1989)

Crete Subduction Zone 63f1.43

This Study (2011)

Dead Sea Region 65 f 1.05

Van Eck (1988)

Charleviox Region 75 f 0.87

Woodgold (1994)

Friuli 80 f 1.1

Rovelli (1982)

South-eastern Canada 91 f 0.95

Woodgold (1990)

Konya Region (India) 96 f 1.09

Grupta et al. (1998)

Cerro Prieto Geothermal Field

(Mexico) 111.5 f 0.41

Minguez, Rebollar, Fabriol (1997)

Granada Basin (Spain) 126 f 0.95

Ibanez et al. (1990)

Table 6.3 shows the frequency dependence of Qc for different tectonic and volcanic

areas in the world. The relationships for the South of Malta, the North-West of Malta and

Crete found in this study are also inputted in this table.

Similar Qc like those found in the North West of Malta were obtained in the neighboring

volcanic region that of Mt. Etna (Del Pezzo et al., 1995). Values obtained in the neighboring

area that of South-Eastern Sicily show higher Qc than those obtained for the North West of

Malta. This could be explained with a higher degree of heterogeneity in the study area. On the

other hand, the presence of molten materials under the volcanic area of Mt. Etna may produce


96

higher frequency independent intrinsic attenuation (Dainty, 1981) respect to the North-West of

Malta area.

The Qc values obtained in the South of Malta area are similar to those obtained in the

South-Eastern Sicily with a frequency relationship being 49f 0.88

. The frequency parameter for

the South of Malta is higher than that for South-Eastern Sicily. This means that the tectonic

activity in the South of Malta is higher than that in South-Eastern Sicily. Higher Qc values

were obtained in Western Anatolia (Akinci et al., 1994), Konya Region (Grupta et al., 1998)

and those obtained in Eastern Canada (Woodgold, 1994). In the State of Washington

(Havskov; Malone; Mcclurg; Crosson, 1989), in the Dead Sea Region (Van Eck, 1988) and in

the Charleviox Region (Woodgold, 1994) all show a lower frequency dependence than that

obtained for the South of Malta Region in this study.

A similar study to that of Crete Subduction Zone was conducted in the Source Region of

the 1999 Chamoli Earthquake by Mukhopadhyay, S et al. (2008). In this study the values of

Q0 increased linearly with depth while in our investigation of Crete Subduction Zone the

values of Q0 decreased linearly with depth. This could be attributed to different tectonic

setting. The values obtained from different areas of the world are plotted with the values

obtained for the North-West of Malta, South of Malta and Crete as seen in figure 6.4. The

values plotted here for different areas of the world are brought from previous studies that have

been conducted worldwide. The study of attenuation for the Charleviox Quebec Region and

that for Southeastern Canada was conducted by Woodgold (1994). The frequency dependence

of Qc in the Mt.Etna region was conducted by Del Pezzo et al. (1995). Taking the 5-6 Hz one

can observe that the lowest values were obtained in the tectonically active region of Mt.Etna

(Del Pezzo et al., 1995). Following are the values obtained for the Crete subduction Zone in

this study and the Charleviox Region in Quebec (Woodgold, 1994). The values of this study

follow after these values with the values obtained for the North-West of Malta being lower

than the values obtained for the South of Malta. Highest values are obtained in South-Eastern

Canada (Woodgold, 1994). This means that the most attenuated region is that of Mt.Etna. The

values obtained in this study are not the lowest or the largest that have been obtained

worldwide. This means that the region around the Maltese Islands is not the most tectonically

active region in the world but is not a stable tectonic region either.


97

South of Malta Crete North - West of Malta Charleviox Mt.Etna South-East Canada

Figure 6.4: Comparison of the Qc relations obtained in different tectonic and volcanic areas in the world.

0

500

1000

1500

2000

2500

3000

3500

4000

0 2 4 6 8 10 12 14

Aver

age

Q

Frequency (Hz)


98

6.4 Further Work

This should be carried out again using a larger amount of events if time permits and

should be carried out for more areas of Study. This could now be done since the same

procedure adopted here could be used. Also the Study carried out for the North-West of Malta

area should be carried out again when a more reliable data set is available

References

Abela, M. (1969). Earthquakes in Malta, History Thesis (University of Malta).

Aguis, J. (2003). Discrimination between Quarry Blasts and Micro Earthquakes around

the Maltese Islands, B.Sc. Dissertation (University of Malta).

Aguis, M; Galea, P. (2008). Automated Single- Station Earthquake Location in the

Sicily Channel using WDD broadband station in the Maltese Islands. Available online:

http://193.188.45.245/downloads/Automated_Single-

Station_Earthquake_Location_in_the_Sicily_Channel_using_WDD_Broadband_Statio

n_on_the_Maltese_Islands.pdf, last accessed on 15 April 2011.

Aguis, M; Galea, P. (2008). Digital Seismic Recording in Malta. Available online:

http://193.188.45.245/downloads/Digital_Seismic_Recording_in_Malta_-

_13_Years_On.pdf, last accesses on 15 April 2011.

Akamatsu, J. (1991). Coda attenuation in the Lützow-Holm Bay, East Antarctica,

Phys. Earth Planet. Interiors, 67, 65-75.

Aki, K. (1980). Attenuation of shear waves in the lithosphere for frequencies from 0.05

to 25Hz, Phys. Earth Planet. Interiors, 74, 615-631.

Aki, K. (1980). Source and scattering effects on the spectra of small local earthquakes,

Bulletin of the Seismological Society of America, 71, 1687-1700.

Aki, K; Chouet, B; (1975). Origin of coda waves: source attenuation and scattering

effects, J. Geophys. Res. 80, 3322-3342.

Barazangi, M; Isacks, B. (1971). Lateral variations of seismic wave attenuation in the

upper mantle above the inclined earthquake zone of the Tonga island arc: deep

anomaly of the upper mantle, J. Geophys. Res, 76, 8493-8516.

Bolt, B.A. (1999). Earthquakes; W.H. Freeman, p.113-114.

Borg, I. (2006). Analysis of Seismic Activity around the Maltese Islands, and the use

of Waveform Cross-Correlation Technique, B.Sc. Dissertation (University of Malta).

Cello, G. (1987). Structure and deformation processes in the Strait of Sicily “rift zone”,

Tectonophysics, 141, 237-247.

http://193.188.45.245/downloads/Automated_Single-Station_Earthquake_Location_in_the_Sicily_Channel_using_WDD_Broadband_Station_on_the_Maltese_Islands.pdf



http://193.188.45.245/downloads/Digital_Seismic_Recording_in_Malta_-_13_Years_On.pdf

http://193.188.45.245/downloads/Digital_Seismic_Recording_in_Malta_-_13_Years_On.pdf

Corti, G; Cuffaro, M; Doglioni, C; Innocenti, F; Manetti, P. (2006). Coexisting

geodynamics processes in the Sicily Channel, Geological Society of America, 409, 83-

96.

Dainty, A. (1981). A scattering model to explain seismic Q observation in the

lithosphere between 1 and 30Hz, Geophys. Res. Lett. 11, 1126-11.

Dart, C.J; Bonsence, D.W.J; McClay, K.R. (1993). Stratigraphy and structure of the

Maltese graben system, Journal of the Geological Society, 150, 1153-1166.

Del Pezzo, E.; Giarreusso, A.; Ferualno, F; Martini, M.(1983). Seismic coda-Q and

scaling law of source spectra at Aeolian Islands, Southern Italy, Bulletin of the

Seismological Society of America, 73, 97-108.

Dominguez, T; Rebollar, C.J; Fabriol, H. (1997). Attenuation of Coda Waves at the

Cerro Prieto Geothermal Field, Baja California, Mexico, Bulletin of the Seismological

Society of America, 87, 5, 1368-1374.

Galea, P. (2007). Seismic History of the Maltese islands and considerations of seismic

risk, Annals of Geophysics, 50, 725-740.

Gao, L. S; Lee, L.C.; Biswas, N.N; Aki, K. (1983). Comparison of effects between

single and multiple scattering on coda waves from local earthquakes, Bulletin of the

Seismological Society of America, 73, 377-389

Giampiccolo, E; Tusa, G; Gresta L; Gresta, H.L. (2002). Attenuation in Southeastern

Sicily (Italy) by applying different coda methods, Journals of Seismology, 6, 487-501.

Giampiccolo, E; Tuve, T; Gresta, S; Patane, D. (2006). S-waves attenuation and

separation of scattering and intrinsic absorption of seismic energy in southeastern

Sicily (Italy), Geophys. J. Int, 165, 211-222.

Goutbeek, F.H; Dost, B; Van Eck, T. 2004). Intrinsic absorption and scattering

attenuation in the southern part of the Netherlands, Journal of Seismology, 8, 1, 11-23.

Grasso, M; Pedley, H.M. (1985). The Pelagian Islands: A new geological interpretation

from sedimentological and tectonic studies and its bearing on the evolution of the

central Mediterranean Sea (Pelagian block), Geologica Rom., 24, 13-34.

Grasso, M.; Pedley H. M; Reuther C.D. (1985). The geology of the Pelagian Islands

and their structural setting related to the Pantelleria rift (central Mediterranean Sea),

Centro, 2, 1-19.

Grasso, M; Reuther, C.D; Baumann, H; Becker, A. (1986). Shallow crustal stress and

neotectonic framework of the Malta platform and the Southeastern Pantelleria rift

(Central Mediterranean), Geologica Rom., 25, 191-212.

Gusev, A. A; I. R. Abubakirov. (1995). Simulated envelopes of anisotropically

scattered waves as compared to observed ones: New evidence for fractal heterogeneity,

Abstracts IUGG XXI General Assembly, B-408.

Havskov, J; Malone, S; Mcclurg, D; Crosson, R. (1989). Coda Q for the state of

Washington, Bulletin of the Seismological Society of America, 79, 4, 1024-1038.

Havskov, J; Ottemoller, L. (2003). SEISAN: The Earthquake analysis Softwares for

Windows, Solaris and Linux, Version 8.0. Institute of Solid Earth Physics, University

of Bergen, Norway.

Huang, J.W. (2009). Seismic Wave Attenuation due to Scattering and Leaky Mode

Mechanisms in Heterogeneous Reservoirs. Available online:

http://www.cseg.ca/conventions/abstracts/2009/2009abstracts/006.pdf.

Illies, J.H. (1981). Graben formation: the Maltese islands, a case history,

Tectonophysics, 73, 151-168.

Instituto Nazionale Di Geofisica. (2006). Mediterranean Very Broadband

Seismographic Network. Available online: http://mednet.rm.ingv.it/, last accessed on

11 February 2011.

Italian Seismological Instrumental and Parametric Database. (2010). Il Bollettino

Sismico Italiano. Available online: http://iside.rm.ingv.it/iside/standard/index.jsp

Jongsma, D; Van Hinte, J.E; Woodside, J.M. (1985). Geologic structure and

neotectonics of the North African Continental Margin south of Sicily, Marine and

Petroleum Geology, 2, 156-179.

Jongsma, D; Woodside, J.M; King, G.C.P; Van Hinte, J.E. (1987). The Medina

Wrench: a key to the kinematics of the central and the eastern Mediterranean over the

past 5MA, Earth and Planetary Science Letters, 82, 87-106.

Konstantinou, K.I; Melis, N.S. (2008). High-Frequency Shear-Wave Propagation

across the Hellenic Subduction Zone, Bulletin of the Seismological Society of America,

98, 2, 797-803.

http://www.cseg.ca/conventions/abstracts/2009/2009abstracts/006.pdf

http://mednet.rm.ingv.it/

http://iside.rm.ingv.it/iside/standard/index.jsp

Kvamme, L.B. (1985). Attenuation of seismic energy from local events in Norwegian

areas, M.Sc Thesis, (University of Bergen, Norway).

Kvamme, L.B; Havskov, J. (1989). Q in southern Norway, Bulletin of the


Lay, T.; Wallace, T.C. (1995). Modern Global Seismology; Academic Press: San

Diego, p. 104-115.

Locating an earthquake using a global seismic network. Available online:


Lowrie, W. (1997). Fundamentals of Geophysics; Cambridge University Press: New

York, p. 130-182.

Mele, G. (1998). High frequency wave propagation from mantle earthquakes in the

Tyrrhenian Sea: new constrains for the geometry of the south Tyrrhenian subduction

zone, Geophys. Res. Lett, 25, 2877-2880.

Mukhopadhyay, S; Sharma, J; Massey, R; Kayal, J.R. (2008). Short Note- Lapse-Time

Dependence of Coda Q in the Source Region of the 1999 Chamoli Earthquake, Bulletin

of the Seismological Society of America, 98, 4, 2080-2086.

Parvez, I.A; Sutar, A. K; Mridula, M; Mishra, S. K; Rai, S.S. (2008). Coda Q

Estimates in the Andaman Islands Using Local Earthquakes, Pure appl. Geophys., 165,

1861-1878.

Pedley, M.; Clarke, M. H; Galea, P. (2002). Limestone Isles in a Crystal Sea – The

Geology of the Maltese Islands; Publishers Enterprises Group (PEG) Ltd: Malta, p.13-

34.

Peng, J.Y. (1989). Spatial and temporal variations of coda Qc in California, Ph.D.

Thesis, (Univ. Southern California, Los Angeles).

Phillips, W.S. (1985). The separation of source, path and site effects on high frequency

seismic waves: an analysis using coda waves techniques, Ph.D Thesis, M.L.T., and

Cambridge, Massachusetts.

Physics Department, University of Malta (2009). Seismic Monitoring and Research

Unit Available online: http://www.phys.um.edu.mt/seismic/

Pulli, J.J. (1984). Attenuation of coda waves in New England, Bulletin of the




Rautian, T. G: Khalturin, V. (1978). The use of coda for determination of the

earthquake source spectrum, Bulletin of the Seismological Society of America, 68, 923-

948.

Reuther, C. D. (1984). Tectonics of the Maltese Islands, Centro, 1, 1-16

Reuther, C.D. (1990). Strike-slip generated rifting and recent tectonic stresses on the

African foreland (Central Mediterranean Region, Ann. Tectonicae, 4, 120-130.

Reuther, C.D; Eisbacher, G.H. (1985). Pantelleria Rift- crustal extension in a

convergent intraplate setting, Geologische Rundschau, 74, 3, 585-597.

Roecker, S. W; Tucker, B; King, J; Hartzfield, D. (1982). Estimates of Q in Central

Asia as a function of frequency and depth using the coda of locally recorded

earthquakes, Bulletin of the Seismological Society of America, 72, 129-149.

Rovelli, A. (1982). On the frequency dependence of Q in Friuli from short-period

digital records, Bulletin of the Seismological Society of America, 72, 6, 2369-2372.

Rovelli, A. (1984). Seismic Q for the lithosphere of the Montenegro region: frequency,

depth and time windowing effect, Phys. Earth Planet. Interiors, 34, 159-172.

Said, J. (1997). Relocation of earthquake epicenters in the Central Mediterranean,

B.Sc. Dissertation (University of Malta).

Sato, H. (1977). Energy propagation including scattering effects single isotropic

scattering approximation, J. Phys. Earth, 25, 27-41

Sato, H. (1988). Fractional interpretation of the linear relation between logarithms of

maximum amplitude and hypocentral distance, Geophys. Res. Lett. 15, 373-375.

Sato, H.; Fehler, M. C. (1998). Seismic Wave Propagation and Scattering in the

Heterogeneous Earth; Springer-Verlag, New York, p.1-308.

Scerri, T. (2001). Duration Magnitude Scale and Analysis of Seismicity around the

Maltese Islands, B.Sc. Dissertation (University of Malta).

Scherbaum, F.; Kisslinger, C. (1985). Coda Q in the Adak Seismic zone, Bulletin of the


Seismic Body Waves. Available online: http://science.jrank.org/pages/48108/seismic-

body-waves.html

Shearer, P.M. (1999). Introduction to Seismology; Cambridge University Press:

Cambridge, p.113-114.



Solov’ev, S.L. (1965). Seismicity of Sakhalin, Bull. Earthq. Res. Inst. 43, 95-102.

South Carolina Department of Natural Resources. (2005). Earthquakes in SC.

Available online: http://www.dnr.sc.gov/geology/images/Plate_Tectonics.jpg

Van Eck, T. (1988). Attenuation of Coda Waves in the Dead Sea Region, Bulletin of

the Seismological Society of America, 78, 2, 770-779.

Wong, V; Rebollar, C, J; Mungufa, L. (2001). Attenuation of Coda Waves at the Tres

Virgenes Volcanic Area, Baja California Sur, Mexico, Bulletin of the Seismological


Woodgold, C.R.D. (1990). Estimation of Q in Eastern Canada using Coda Waves,

Bulletin of the Seismological Society of America, 80, 2, 411-429.

Woodgold, C.R.D. (1994). Coda Q in the Charlevoix, Quebec, Region: Lapse-Time

Dependence and Spatial and Temporal Comparisons, Bulletin of the Seismological


Zammit, D. (2009). A Performance Evaluation of a Single-Station Earthquake

Location Algorithm (LESSLA) implemented at WDD Seismic Station, University of

Malta, B.Sc. Dissertation (University of Malta).

Zammit, S. (2003). A study of seismicity and earthquake swarms in the Central

Mediterranean, M.Sc. Thesis (University of Malta).

http://www.dnr.sc.gov/geology/images/Plate_Tectonics.jpg

Appendix 1: Recorded Events

Appendix 1

Recorded Events

Table A: The south of Malta earthquakes that were used for the calculation of coda Q. These

were recorded by the WDD station and obtained from the Seismic Monitoring and Research

Unit database at the University of Malta. In this table the date, origin time, P and S arrival

times, the SP time, the calculated lapse times, magnitude, distance, latitude and longitude for

each event are given

Event date Origin Time station P Time S Time SP Time

No. GMT GMT GMT GMT

1 05/01/2011 12:22:19 WDD 12:22:25 12:22:30 21:15:50

2 15/01/2011 05:14:47 WDD 05:14:50 05:14:53 07:35:02

3 03/12/2010 02:02:20 WDD 02:02:28 02:02:34 01:40:48

4 01/12/2010 11:41:52 WDD 11:42:20 11:42:40 11:15:22

5 18/11/2010 08:20:09 WDD 08:20:24 08:20:35 23:09:36

6 12/11/2010 04:43:35 WDD 04:43:50 04:44:01 04:22:05

7 21/09/2010 04:39:44 WDD 04:39:55 04:40:03 09:50:24

8 03/08/2010 08:06:49 WDD 08:06:56 08:07:01 01:16:19

9 15/07/2010 17:38:45 WDD 17:39:14 17:39:36 15:11:31

10 09/07/2010 00:19:24 WDD 00:19:44 00:19:58 12:12:58

11 15/06/2010 23:29:24 WDD 23:29:31 23:29:36 05:08:10

12 30/08/2009 02:10:57 WDD 02:11:12 02:11:23 01:03:22

13 25/08/2009 03:21:05 WDD 03:21:22 03:21:34 11:19:41

14 16/08/2009 07:49:42 WDD 07:49:59 07:50:12 16:16:19

15 07/08/2009 09:06:22 WDD 09:06:30 09:06:35 13:13:26

16 30/07/2009 02:58:36 WDD 02:58:54 02:59:07 06:20:10

17 05/07/2009 18:52:14 WDD 18:52:35 18:52:51 15:01:26

18 26/05/2009 21:37:21 WDD 21:37:38 21:37:51 00:10:05

19 16/05/2009 13:08:42 WDD 13:08:45 13:08:48 07:12:00

20 28/04/2009 02:56:40 WDD 02:56:45 02:56:49 04:03:22

21 25/04/2009 01:36:25 WDD 01:36:53 01:37:14 15:30:14



No. GMT GMT GMT GMT

22 23/03/2009 18:44:35 WDD 18:45:02 18:45:21 19:03:22

23 23/12/2008 23:47:24 WDD 23:47:31 23:47:36 05:03:50

24 30/11/2008 16:28:27 WDD 16:28:33 16:28:38 00:43:12

25 05/07/2008 00:59:41 WDD 00:59:45 00:59:48 22:27:50

26 15/10/2007 17:36:44 WDD 17:36:57 17:37:07 14:13:55

27 03/10/2007 02:15:08 WDD 02:15:14 02:15:18 08:34:05

28 05/09/2007 09:21:53 WDD 09:21:58 09:22:02 00:00:00

29 05/09/2007 08:02:48 WDD 08:02:54 08:02:59 07:32:10

30 05/09/2007 07:58:12 WDD 07:58:18 07:58:22 07:20:38

31 05/09/2007 06:36:03 WDD 06:36:09 06:36:13 08:26:53

32 16/08/2007 01:51:28 WDD 01:51:33 01:51:37 13:43:41

33 11/08/2007 02:19:05 WDD 02:19:12 02:19:17 19:06:14

34 24/06/2007 16:13:04 WDD 16:13:12 16:13:18 19:01:55

35 20/05/2007 22:09:31 WDD 22:09:39 22:09:44 14:54:14

36 11/05/2007 03:29:58 WDD 03:30:05 03:30:11 16:49:26

37 31/03/2007 13:08:01 WDD 13:08:25 13:08:43 18:33:07

38 31/03/2007 12:22:32 WDD 12:22:58 12:23:16 20:42:43

39 30/03/2007 13:59:47 WDD 13:59:54 13:59:59 01:30:43

40 22/03/2007 12:18:57 WDD 12:19:05 12:19:10 19:00:29

41 15/02/2007 04:26:22 WDD 04:26:30 04:26:36 23:05:17

42 26/01/2007 23:35:46 WDD 23:36:04 23:36:17 22:07:41

43 27/01/2007 23:03:16 WDD 23:03:33 23:03:45 10:40:48


Event Lapse Time: S 2S new time ( original +2S)

No. Seconds Seconds GMT

1 11 22 12.22.41

2 6 12 05.14.59

3 14 28 02.02.48

4 48 96 11.42.28

5 26 52 08.20.59

6 26 52 04.44.27

7 19 38 04:40:22

8 12 24 08:07:13

9 51 102 17:40:27

10 51 68 00:20:32

11 12 24 23:29:48

12 36 72 02:11:49

13 29 58 03:22:03

14 49 98 07:50:42

15 13 26 09:06:48

16 30 62 02:59:38

17 37 74 18:53:28

18 30 60 21:38:21

19 49 98 13:09:14

20 13 26 02:56:58

21 49 98 01:38:03

22 46 92 18:46:07

23 12 24 23:47:48

24 11 22 16:28:49

25 7 14 00:59:55

26 30 60 17:37:30

27 13 26 02:15:28

28 12 24 09:22:21

29 27 54 08:03:10

30 88 176 07:58:32

31 40 80 06:36:23

32 30 60 01:51:46

33 31 62 02:19:29

34 43 86 16:13:32

35 13 26 22:09:57

36 13 26 03:30:24

37 42 84 13:09:25

38 44 88 12:24:00


39 12 24 14:00:01

39 12 24 14:00:01

39 12 24 14:00:01

39 12 24 14:00:01

40 13 26 12:19:23

41 14 28 04:26:50

42 31 62 23:36:48

43 29 58 23:04:14


Event distance Ml Md latitude longitude

No. (km) (deg) (deg)

1 38.093 2.379 1.683 35.921 14.115

2 17.811 1.914 1.315 35.741 14.682

3 47.633 2.919 2.024 35.411 14.477

4 173.169 4.612 3.064 34.350 15.083

5 88.325 3.800 2.380 35.065 14.749

6 90.175 3.800 2.476 35.037 14.674

7 66.831 2.759 2.027 35.259 14.720

8 39.431 2.669 1.885 35.533 14.300

9 184.092 2.800 2.799 34.194 14.293

10 119.062 3.558 2.690 35.407 13.320

11 40.723 1.800 1.860 35.517 14.307

12 89.000 1.661 1.941 35.111 14.934

13 101.266 2.477 2.188 35.835 15.646

14 103.048 3.241 5.034 34.923 14.705

15 43.433 3.431 1.845 35.894 15.001

16 108.142 2.824 2.472 35.101 15.302

17 128.975 3.617 2.805 35.048 13.452

18 105.907 2.510 2.440 34.989 15.053

19 17.678 2.517 1.441 35.988 14.584

20 32.375 3.024 1.764 35.616 14.756

21 174.820 4.600 3.142 34.307 14.092

22 166.892 4.627 3.396 34.475 13.759

23 40.699 1.715 1.715 35.583 14.848

24 39.246 2.802 1.886 35.665 14.903

26 76.714 3.500 2.321 35.152 14.613

27 33.874 1.816 2.354 35.542 14.616

28 31.034 2.005 1.154 35.624 14.304

29 33.532 3.057 1.668 35.617 14.271

30 33.465 2.712 1.698 35.560 14.382

31 33.831 2.924 1.754 35.561 14.679

32 27.656 1.991 1.378 35.911 14.817

33 37.373 3.687 2.091 35.578 14.262

34 45.388 2.160 1.846 35.468 14.737

35 43.999 1.528 1.790 35.448 14.437

36 44.645 1.779 2.138 35.591 14.914

37 148.326 3.655 4.400 34.522 14.265


Event distance Ml Md latitude longitude

No. (km) (deg) (deg)

38 158.296 4.546 3.041 34.472 14.039

39 39.507 2.630 1.865 35.739 14.945

40 45.380 2.940 2.940 35.976 14.051

41 46.754 3.432 2.191 35.586 14.938

42 105.167 3.787 2.594 34.909 14.741

43 101.036 3.546 2.311 34.944852 14.726923


Table B: The events at the North-West of Malta area near Pantelleria that were used for the

calculation of coda Q. These were recorded by WDD station and obtained from the Seismic

Monitoring and Research Unit at the University of Malta. In this table the date, origin time, P

and S arrival times, the SP time, the calculated lapse times, magnitude, distance, latitude and

longitude for each event are given.


Lapse Time: S 2S

No. GMT GMT GMT GMT Seconds Seconds

1 19/03/2009 10:25:59 WDD 10:26:26 10:26:47 07:07:41 48 96

2 02/07/2008 09:17:52 WDD 09:18:19 09:18:38 14:19:41 45 90

3 11/02/2008 08:05:41 WDD 08:05:51 08:05:59 11:54:14 18 36

4 10/04/2007 19:17:30 WDD 19:18:01 19:18:24 21:33:07 54 108

5 11/02/2007 20:30:59 WDD 20:31:18 20:31:32 07:48:00 33 66

6 15/03/2006 20:33:13 WDD 20:33:35 20:33:51 05:03:50 38 76

7 13/03/2006 18:05:00 WDD 18:05:26 18:05:45 06:15:50 45 90

Event New time distance Ml Md latitude longitude

No. GMT (km) (deg) (deg)

1 10:27:35 171.56638 3.478 3.286 36.863 13.096

2 09:19:22 165.06616 4.498 3.096 36.456 12.856

3 08:06:17 59.27484 2.069 2.125 36.004 13.900

4 19:19:18 196.09474 4.411 3.578 37.028 12.911

5 20:32:05 117.44272 3.448 2.336 36.776 13.925

6 20:34:29 134.20744 3.365 2.954 36.505 13.281

7 18:06:30 161.96033 3.735 3.133 36.490 12.914


Table C: The earthquakes from the Subduction Zone near Crete that were used for the

calculation of coda Q. These were recorded by IDI station, Crete and obtained from the

Institute of Geodynamics Athens’s database that is available online. In this table the Origin

time, P time, magnitude, depth, latitude and longitude for each event are given.

Event Date Origin Time P time Station Latitude Longitude Depth Magnitude

No. GMT GMT (deg) (deg) (km)

1 28/02/2011 07:49:07 16:23:32 IDI 34.980 25.420 53 5.2

2 1/11/2010 16:23:11 14:48:16 IDI 34.260 24.500 5 3.6

3 1/11/2010 14:47:58 14:48:16 IDI 35.150 23.710 20 3.1

4 31/10/2010 21:29:47 21:30:08 IDI 36.000 23.820 35 3.5

5 25/10/2010 02:18:57 02:19:32 IDI 36.880 26.740 46 3.2

6 21/10/2010 09:48:44 09:49:41 IDI 38.870 26.010 20 3.5

7 21/10/2010 10:06:45 10:06:55 IDI 35.070 24.310 23 2.6

8 14/10/2010 09:08:22 09:09:21 IDI 36.270 29.650 49 4.3

9 11/10/2010 22:24:03 22:24:21 IDI 24.480 25.580 15 3.3

10 11/10/2010 19:49:09 19:49:29 IDI 34.170 25.060 4 2.9

11 11/10/2010 14:11:53 14:12:06 IDI 34.760 24.370 20 2.8

12 19/08/2010 21:34:03 21:34:33 IDI 34.210 26.340 21 3.6

13 17/08/2010 14:04:29 14:04:40 IDI 34.840 24.470 30 2.9

14 13/08/2010 15:37:54 15:38:03 IDI 34.840 24.470 15 3.0

15 05/01/2010 20:31:19 20:31:33 IDI 35.000 24.120 45 2.8

16 31/12/2009 00:12:41 00:13:02 IDI 34.200 25.180 36 3.6

17 24/12/2009 17:04:42 17:05:01 IDI 35.680 25.930 30 2.9

18 17/12/2009 14:53:44 14:53:49 IDI 35.040 24.980 17 2.7

19 03/11/2009 12:33:55 12:35:06 IDI 34.850 24.110 24 3.0

20 23/10/2009 17:59:48 18:00:29 IDI 37.480 26.690 26 3.7

21 06/10/2009 16:19:08 16:20:06 IDI 34.900 25.320 15 3.2

22 26/09/2009 02:52:01 02:52:30 IDI 33.750 25.490 41 4.1

23 22/09/2009 21:49:02 21:49:17 IDI 34.690 24.650 30 3.3

24 17/08/2009 08:36:36 08:37:02 IDI 33.950 25.360 7 3.3

25 08/08/2009 16:47:16 16:47:28 IDI 35.000 24.790 32 2.9

26 02/08/2009 08:49:42 08:49:53 IDI 35.290 24.370 43 3.3

27 27/07/2009 17:46:06 17:46:18 IDI 35.740 24.630 44 3.2

28 23/07/2009 03:03:45 22:03:57 IDI 34.760 24.970 29 3.1

29 14/07/2009 23:06:12 23:06:22 IDI 34.210 25.270 22 3.2

30 10/07/2009 07:29:02 07:29:12 IDI 34.890 24.910 30 3.2

31 05/06/2009 07:46:01 07:46:29 IDI 34.610 23.760 11 3.6

32 30/05/2009 13:29:32 13:29:44 IDI 34.930 24.770 35 3.2

33 22/05/2009 19:39:39 19:39:58 IDI 34.870 24.790 26 2.8


Event Date Origin time P time station Latitude Longitude Depth Magnitude

No.

GMT GMT

(deg) (deg) (km) 34 01/05/2009 22:42:25 22:42:32 IDI 35.600 24.770 33 2.8

35 27/04/2009 03:10:48 03:11:13 IDI 35.590 26.450 30 3.6

36 19/03/2009 14:15:13 14:15:41 IDI 35.100 23.440 37 4.8

37 17/03/2009 11:14:21 11:14:41 IDI 35.830 23.710 12 3.2

38 29/11/2008 21:18:07 21:18:18 IDI 34.81 25.02 10 3.3

39 18/11/2008 01:00:46 01:00:59 IDI 35.460 25.710 25 3.0

40 04/08/2008 19:38:23 19:38:56 IDI 33.890 26.560 32 5.0

41 12/06/2008 00:20:43 00:21:04 IDI 35.110 26.190 29 5.0

42 12/04/2008 07:58:31 07:58:56 IDI 34.070 25.310 9 4.1

Appendix 2- Codaq Q Values

Appendix 2

Coda Q Values - Data Tables

Table D: The values of the coda q, Qc that were calculated by SEISAN for the south of Malta

earthquakes.

Event date frequency Qc value

No. (Hz)

1 05/01/2011 2 120

5 993

7 -

9 4518

12 9920

2 15/01/2011 2 91

5 545

7 223

9 261

12 477

3 03/12/2010 2 53

5 331

7 -

9 870

12 860

4 01/12/2010 2 165

5 389

7 497

9 361

12 946

5 18/11/2010 2 226

5 566

7 576

9 673

12 -



No. (Hz)

6 12/11/2010 2 61

5 147

7 323

9 476

12 1036

7 21/09/2010 2 163

5 -

7 1111

9 705

12 -

8 03/08/2010 2 106

5 843

7 -

9 2414

12 1413

9 15/07/2010 2 -

5 780

7 -

9 -

12 -

10 09/07/2010 2 92

5 6747

7 -

9 -

12 -

11 15/06/2010 2 -

5 355

7 743

9 885

12 984

12 30/08/2009 2 187

5 429

7 -

9 -

12 -



No. (Hz)

13 25/08/2009 2 142

5 -

7 -

9 995

12 984

14 16/08/2009 2 172

5 1232

7 973

9 -

12 -

15 07/08/2009 2 -

5 121

7 192

9 337

12 468

16 30/07/2009 2 -

5 -

7 1016

9 2272

12 7895

17 05/07/2009 2 67

5 403

7 587

9 3313

12 -

18 26/05/2009 2 42

5 282

7 374

9 -

12 -

19 16/05/2009 2 187

5 231

7 251

9 222

12 -



No. (Hz)

20 28/04/2009 2 74

5 153

7 281

9 386

12 554

21 25/04/2009 2 95

5 898

7 570

9 831

12 1203

22 23/03/2009 2 305

5 422

7 -

9 744

12 1210

23 23/12/2008 2 132

5 261

7 320

9 376

12 603

24 30/11/2008 2 105

5 138

7 549

9 1109

12 872

25 05/07/2008 2 62

5 139

7 254

9 207

12 267

26 15/10/2007 2 144

5 265

7 502

9 866

12 1124



No. (Hz)

27 03/10/2007 2 78

5 269

7 248

9 929

12 -

28 05/09/2007 2 90

5 339

7 713

9 2156

12 2556

29 05/09/2007 2 50

5 171

7 402

9 777

12 899

30 05/09/2007 2 41

5 564

7 988

9 -

12 -

31 05/09/2007 2 54

5 266

7 622

9 3405

12 -

32 16/08/2007 2 70

5 298

7 881

9 6964

12 1282

33 11/08/2007 2 -

5 255

7 509

9 433

12 560



No. (Hz)

34 24/06/2007 2 239

5 659

7 2346

9 5296

12 7549

35 20/05/2007 2 -

5 1854

7 424

9 1978

12 799

36 11/05/2007 2 -

5 208

7 952

9 1235

12 2032

37 31/03/2007 2 -

5 3553

7 328

9 1043

12 -

38 31/03/2007 2 -

5 1430

7 1368

9 1328

12 6897

39 30/03/2007 2 59

5 345

7 348

9 694

12 1354

40 22/03/2007 2 409

5 258

7 756

9 2281

12 -



No. (Hz)

41 15/02/2007 2 -

5 193

7 357

9 366

12 606

42 26/01/2007 2 190

5 320

7 263

9 519

12 911

43 27/01/2007 2 62

5 -

7 494

9 1994

12 -


Table E: The values of the coda q, Qc that were calculated by SEISAN for the North-West of

Malta earthquakes.


No. (Hz)

1 19/03/2009 2 109

5 549

7 890

9 -

12 1855

2 02/07/2008 2 102

5 276

7 658

9 5505

12 -

3 11/02/2008 2 -

5 370

7 652

9 1656

12 -

4 10/04/2007 2 83

5 769

7 735

9 4238

12 -

5 11/02/2007 2 201

5 490

7 608

9 773

12 125

6 13/03/2006 2 125

5 5637

7 2321

9 1402

12 6204


Table F: The values of the coda q, Qc that were calculated by SEISAN for the Crete

earthquakes.


No. (Hz)

1 28/02/2011 2 77

5 595

7 -

9 -

12 3178

2 01/11/2010 2 184

5 -

7 1406

9 2501

12 -

3 01/11/2010 2 123

5 389

7 568

9 -

12 -

4 31/10/2010 2 283

5 341

7 428

9 574

12 742

5 25/10/2010 2 276

5 432

7 604

9 1065

12 1212

6 21/10/2010 2 327

5 281

7 366

9 434

12 689

7 21/10/2010 2 175

5 -

7 -

9 1909

12 1297


Event Date Frequency Qc value

No.

(Hz) 8 14/10/2010 2 139

5 1099

7 -

9 -

12 -

9 11/10/2010 2 -

5 435

7 858

9 1341

12 -

10 11/10/2010 2 148

5 601

7 605

9 1225

12 -

11 11/10/2010 2 -

5 315

7 401

9 1131

12 -

12 19/08/2010 2 117

5 528

7 618

9 981

12 -

13 17/08/2010 2 545

5 332

7 729

9 2420

12 -

14 13/08/2010 2 286

5 180

7 261

9 403

12 565



No.

(Hz) 15 05/01/2010 2 174

5 615

7 1072

9 -

12 -

16 31/12/2009 2 142

5 975

7 1108

9 1856

12 2390

17 24/12/2009 2 125

5 330

7 1035

9 774

12 950

18 17/12/2009 2 369

5 288

7 400

9 626

12 791

19 03/11/2009 2 105

5 664

7 857

9 946

12 2831

20 23/10/2009 2 -

5 343

7 1504

9 -

12 3931

21 06/10/2009 2 173

5 495

7 775

9 575

12 847



No.

Hz 22 26/09/2009 2 81

5 -

7 -

9 -

12 -

23 22/09/2009 2 191

5 295

7 697

9 521

12 700

24 17/08/2009 2 209

5 -

7 -

9 -

12 1576

25 08/08/2009 2 -

5 314

7 538

9 1302

12 4301

26 02/08/2009 2 225

5 614

7 427

9 827

12 1173

27 27/07/2009 2 300

5 390

7 486

9 633

12 680

28 23/07/2009 2 123

5 416

7 349

9 565

12 1925



No.

(Hz) 29 14/07/2009 2 89

5 243

7 358

9 538

12 911

30 10/07/2009 2 403

5 534

7 459

9 431

12 577

31 05/06/2009 2 -

5 468

7 847

9 1495

12 1684

32 30/05/2009 2 141

5 533

7 537

9 555

12 2606

33 22/05/2009 2 143

5 327

7 410

9 496

12 969

34 01/05/2009 2 112

5 563

7 449

9 512

12 560

35 27/04/2009 2 465

5 385

7 912

9 1052

12 1536



No.

(Hz) 36 19/03/2009 2

5 298

7 1007

9 937

12 1820

37 17/03/2009 2 -

5 775

7 794

9 729

12 3700

38 29/11/2008 2 -

5 478

7 448

9 546

12 623

39 18/11/2008 2 169

5 436

7 466

9 584

12 679

40 04/08/2008 2 -

5 289

7 643

9 916

12 740

41 12/06/2008 2 81

5 333

7 470

9 539

12 685

42 12/04/2008 2 332

5 345

7 -

9 -

12 2263

Department of Physics

Faculty of Science

Declaration on Plagiarism

Plagiarism is the unacknowledged use, as one’s own, of work of another person, whether

or not such work has been published (Regulations Governing Conduct at

Examinations, 1997).

This declaration is being made in accordance with the Faculties of Science and of

Information and Communications Technology Guidelines on Plagiarism, June

2007. These guidelines state that, where such declaration is required work shall

not be marked, and shall receive a mark of 0, if the declaration is missing or

incomplete.

The guidelines also require that all work submitted for assessment shall be

accompanied by a machine-readable copy that will be retained up to one year

following a student’s graduation. If the machine readable copy is not submitted,

or is found to be different from the physical copy of the submission, then the

physical copy shall not be assessed, and shall be awarded a mark of 0.

I/we, the undersigned, declare that the submitted work indicated below is my/our work,

except where duly acknowledged and referenced.

I/we, the undersigned, understand that plagiarism is a serious offense carrying penalties

as stipulated in Part 4 of the Faculties of Science and of Information and

Communications Technology Guidelines on Plagiarism (June 2007), and Article 6c

of the Regulations Governing Conduct at Examinations (1997).

Title of submitted work:

Submission date:

Submitting author/s Signature/s

1.

Documents

Crustal Attenuation in the region of the Maltese Islands … through the CODAQ subroutine in the seismic analysis package SEISAN 8.0 by applying the time domain coda-decay method