Lecture 6

LECTURE 6
MODELING EARTHQUAKES
AS TSUNAMI SOURCES
PRINCIPLES of HYDRODYNAMIC SIMULATIONS

CLASSICAL APPROACH
1.
Obtain model of Earthquake Rupture
2.
Compute Static Deformation of Ocean Bottom
3.
Use as Initial Conditions of

Vertical Surface Displacement with Zero Initial Velocity
4.
Run Hydrodynamic Model (e.g., MOST)
5.
Propagate, up to and including

INUNDATION of Receiving Shore
(a)
CLASSICAL
APPROACH
t = 0
(b)
(d)
t = 0+
(c)
TSUNAMI
(e)
t = 0++
t =+
GENERIC
EARTHQUAKE
DISLOCATION
Involves MANY parameters
Earthquake moment M 0
Earthquake geometry , ,
Earthquake depth h
Water depth H
Epicentral distance to shore L
Beach slope
M0 :
Fault Length L F
Fault width W
Slip on Fault u
FIRST STEP
Position a point force F in an infinite homogeneous elastic medium

F = {F j }
u = {ui }
r = r{ k }
Obtain the Dynamic displacement field of the

deformation
[Aki and Richards, 1980; p. 73, Eqn. (4.23)]
The STATIC displacement is simply obtained

by putting t .
[This expression is known as the Somigliana Tensor]
SECOND STEP
Replace Single Force by Double-Couple
Simply use Somiglianas tensor as a Greens function and take appropriate

derivatives.
NEAR FIELD
Note that these are

the P and S waves
of the near [and far]
field[s].
NEAR FIELD
NEAR FIELD
[Far Field]
[Aki and Richards, 1980; p. 79; Eqn. (4.29)]
THIRD STEP
Include effect of free surface
Integrate over finite area of faulting
(Combine with "reflection" of

equivalent P and S waves)
Reflected P
Incident P
Reflected
S
The problem has an analytical solution
TWO equivalent algorithms
[Stein and Wysession, 2002]
Mansinha and Smylie [1971]

Okada [1985]
Only difference: Okada allows for
tensile crack
(non-double-couple solution).
STATIC DEFORMATION OF OCEAN BOTTOM

Straightforward, if somewhat arcane analytical formul
[Mansinha and Smylie, 1971; Okada, 1985]
STATIC DEFORMATION OF OCEAN BOTTOM

EXAMPLE: VALPARAISO, CHILE
1906 CHILEAN EVENT
17 AUGUST 1906
km
-300 -250 -200 -150 -100 -50
300
M 0 = 2. 8 1028 dyn-cm
50
100 150 200 250 300
-31
250
200
-32
ZAPALLAR
150
100
-33
km
50
0
-34
-50
-100
PICHILEMU
LLICO
-150
-35
-200
f = 3; = 15; = 117
-250
-36
-300
-74
L F = 200 km; W = 75km;

u = 5. 3 m
-100
-65
-45
-25
-15
-72
-2
15
25
-70
40
60
AMPLITUDE (cm)
[Okal, 2005]
80
100
120
150
500
Use this static deformation field (limited to its oceanic portion) as

the initial condition (t = 0+ ) of the hydrodynamic calculation.
Justification: The seismic source is generally MUCH FASTER

than any tsunami process, hence it can be taken as instantaneous.
(even in the case of SLOW, so-called "Tsunami" earhtquakes)
PRODUCTS OF SIMULATION
CHILE 1906 +10 hr.
1. Snapshots of Sea Height at Given Times

CHILE 1906 + 5 hr.
130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290
50
50
40
40
30
30
20
20
10
130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290
0
50
50
-10
CHILE 1906 + 1 hr. 45 mn
40
40
-20
30
30
-30
20
20
-40
10
130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290
-50
0 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290
50
50
-10
40
40
-20
30
30
-30
20
20
-40
10
10
-50
0
0 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290
-10
-10
-20
-20
-30
-30
-40
-40
-50
-50
130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290
-20.0 -5.0 -2.0 -1.5 -0.5 -0.1 -0.05 0.05 0.1
0.5
AMPLITUDE (m)
1.5
3.0
6.0
PRODUCTS OF SIMULATION
2. Map of Maximum Amplitude of Tsunami Wave
130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290
50
50
40
40
30
30
20
20
10
10
-10
-10
-20
-20
-30
-30
-40
-40
-50
-50
130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290
0.05
0.10
0.20
0.30
0.40
0.50
AMPLITUDE (m)
1.00
HOW ROBUST IS THIS PROCEDURE ?
It is worth exploring the robustness of our results in the far

field, with respect to detailed parameters of our sources, a fortiori unknown in the context of many simulations.
We study simulated amplitudes of the 2004 Sumatra-Andaman

tsunami in the far field under fluctuations of source parameters, while keeping the seismic moment of the source constant.
We conclude that our results are indeed robust.
The primary parameters controlling the far field tsunami

amplitudes are the size (moment) of the parent earthquake and
the depth of the water column in the epicentral area.
40
1. MOVE SOURCE
LATERALLY
Move 1 West
40
50
60
70
80
90
100
50
60
70
80
90
100
110
20
20
10
10
-10
-10
-20
-20
-30
-30
-40
-40
-50
-50
-60
-60
-70
110
120
120
40
50
60
70
80
90
100
110
-70
120
40
50
60
70
80
90
100
110
120
Move 1 North
Move 1 East
40
50
60
70
80
90
100
110
120
20
20
20
20
20
20
10
10
10
10
10
10
-10
-10
-10
-10
-10
-10
-20
-20
-20
-20
-20
-20
-30
-30
-30
-30
-30
-30
-40
-40
-40
-40
-40
-40
-50
-50
-50
-50
-50
-50
-60
-60
-60
-60
-60
-60
-70
120
-70
-70
-70
40
50
60
70
80
90
100
110
40
50
60
70
80
90
100
110
-70
120
40
50
60
70
80
90
100
110
120
20
20
10
10
-10
-10
-20
-20
-30
-30
-40
-40
-50
-50
-60
-60
40
50
0.05
0.10
60
0.20
70
0.30
0.40
80
0.50
90
1.00
2.00
100
110
3.00
5.00 50.00
-70
120
AMPLITUDE (m)
Move 1 South
-70
40
50
60
70
80
90
100
110
-70
120
NO MAJOR EFFECT !!
2. CHANGE SOURCE PARAMETERS

Heterogeneous Slip
SUMATRA 2004 Original

40
50
60
70
80
90
100
110
70
80
90
100
110
120
20
10
10
10
10
-10
-10
-10
-10
-20
-20
-20
-20
-30
-30
-30
-30
-40
-40
-40
-40
-50
-50
-50
-50
-60
-60
-60
-60
-70
120
-70
50
60
70
80
90
100
110
80
90
100
40
50
60
70
80
90
Fault Dip
SUMATRA 2004; D = 20 km
70
60
20
Depth
60
50
20
40
50
40
20
-70
40
120
110
40
50
60
70
80
90
100
110
-70
120
Strain Released
SUMATRA 2004 Dip = 12 deg.

120
100
110
SUMATRA 2004 Large Strain
120
40
50
60
70
80
90
100
110
120
20
20
20
20
20
20
10
10
10
10
10
10
-10
-10
-10
-10
-10
-10
-20
-20
-20
-20
-20
-20
-30
-30
-30
-30
-30
-30
-40
-40
-40
-40
-40
-40
-50
-50
-50
-50
-50
-50
-60
-60
-60
-60
-60
-60
-70
120
-70
-70
120
-70
-70
40
50
0.05
0.10
60
0.20
70
0.30
0.40
80
0.50
90
1.00
2.00
AMPLITUDE (m)
100
3.00
110
5.00
40
50
60
70
80
NOTE: M 0 sin
kept constant
90
100
110
40
50
60
70
80
90
100
110
-70
120
NO MAJOR EFFECT !!
By CONTRAST, WATER DEPTH at the SOURCE PLAYS a CRUCIAL ROLE

NOTE: This explains the much smaller tsunami during the 2005 Nias earthquake.
EPICENTRAL BATHYMETRY
DIVIDED BY 4.0
UNPERTURBED
EPICENTRAL BATHYMETRY
40
50
60
70
80
90
40
100
50
60
70
80
90
100
20
20
10
10
-10
-10
-20
-20
-30
-30
40
50
60
70
80
0.05
90
0.10
0.20
40
100
0.30
0.40
0.50
1.00
50
2.00
AMPLITUDE (m)
3.00
60
5.00
70
80
90
100
NORMAL MODE FORMALISM: A different approach

[Ward, 1980]
At very long periods (typically 15 to 54 minutes), the Earth, because of its finite
size, can ring like a bell.
Such FREE OSCILLATIONS are equivalent to the superposition of two progressive waves travelling in opposite directions along the surface of the Earth.
T = 54 minutes
T = 21.5 minutes
"FOOTBALL
Mode"
[After Lay and

Wallace, 1995]
"BREATHING
Mode"
Ward [1980] has shown that Tsunamis come naturally as a special branch of
the normal modes of the Earth, provided it is bounded by an ocean, and gravity is included in the formulation of its vibrations.
In the normal mode formalism, the solution of the vertical displacement (both in the
water and solid Earth) is sought as
u z (x; t) = u z (r, , ; t) = y1 (r) Y lm ( , ) exp(i t) = y1 (r) P lm ( , ) ei m exp(i t)
where Y lm is a spherical harmonic of order l and degree m; P lm is the Legendre polynomial of order l and degree m; and {r, , } is a system of spherical polar coordinates.
This allows for the separation of the variables {r, , }.
The problem is complemented by similar expressions for the overpressure p = y2 in the
tsunami wave, the horizontal displacement u x = l y3 , and the change in the gravity
potential y5 .
Under the linear approximation, the equations of hydrodynamics transform into a system
of linear differential equations of the first order.
For any given l, i.e., wavenumber k = (l + 1/2) (a radius of the Earth), the system has
non trivial solutions for only one value of . The relationship between l and is the
Disppersion Relation of the Tsunami.
SPHEROIDAL MODE HAS 6COMPONENT EIGENFUNCTION SATISFYING:

2
( + 2 ) r
1
( + 2 )
L2
( + 2 ) r
y1
4
( + 2 ) r
g 2 (3 + 2 )
L2
r
( + 2 ) r 2
L2
r
y2
1
r
1
r
y3
2 (3 + 2 )
( + 2 ) r 2
r
( + 2 ) r
4 L 2 ( + ) 2
2
2 +
( + 2 ) r 2
r
3
r
y4
4 G
y5
4 L 2 G
r
L2
r2
2
r
y6
dy1
dr
dy2
dr
2 +
dy3
dr
dy4
dr
4 (3 + 2 ) 4 g
r
( + 2 ) r 2
dy5
dr
dy6
dr
y 1 : Vertical displacement
y 4 : Tangential stress
y 3 : Horizontal displacement
y 5 : Gravity potential
y 2 : Normal stress
y 6 : Auxiliary gravity
EASILY SOLVED WITH APPROPRIATE BOUNDARY CONDITIONS
EIGENFUNCTIONS of SPHEROIDAL MODES

Rayleigh Mode
l = 200; T = 52 s
Tsunami Mode
l = 200; T = 908 s
y 1 Vertical Displacement
y 3 Horizontal Displacement
0
y 2 Pressure
5 km
y1 ; y3
100
in solid !!
200 km
TSUNAMI EIGENFUNCTION is CONTINUED (SMALL) into SOLID EARTH
EXCITATION OF TSUNAMI in NORMAL MODE FORMALISM
Gilbert [1970] has shown that the response of the Earth to a

point source consisting of a single force f can be expressed
as a summation over all of its normal modes
1 cos n t exp ( n t/2Q n )

u(r, t) = sn (r) s*n (rs ) f(rs )
,
2
n
N
the EXCITATION of each mode being proportional to the scalar

product of the force f by the eigen-displacement s at location rs .
Now, an EARTHQUAKE is represented by a system of

forces called a double couple
couple:
Normal to Fault Plane
Direction of Slip
The response of the Earth to an earthquake is thus
1 cos n t exp ( n t/2Q n )

u(r, t) = sn (r) n* (rs ) : M (rs )
2n
N
where the EXCITATION is the scalar product of the earthquakes MOMENT M with the local eigenstrain at the source
rs .
This formula is directly applicable to the case of a tsunami
represented by normal modes of the Earth.
ADVANTAGES of NORMAL MODE FORMALISM
Handles any Ocean-Solid Earth Coupling

Including Sedimentary Layers
Works well at Higher Frequencies

No need to assume Shallow-Water Approximation
DRAWBACKS of NORMAL MODE FORMALISM
Must assume Laterally Homogeneous Structure
Linear Theory -- Does not allow for Large Amplitudes
NOTE: Energy scales as L 4 , i.e., as M 04/3 .
ENERGY of a TSUNAMI -- STATIC THEORY [Kajiura, 1981]

E =
1 w g 2/3
1 w g 2/3
4/3
4/3
F(
,
h,
R)
M
=
M
max
0
0
2 2
24/3 4/3
= invariant ratio of M 0 to S 3/2
F : dimensionless factor expressing geometry of faulting, and aspect ratio R of fault rupture
area.
NOTE: Energy of Tsunami grows faster than Seismic Moment

Energy released by rupture, proportional to M0 : grows like moment.
Hence, Fraction of Earthquake Energy transferred to Tsunami Grows with Earthquake Size
Fortunately, it remains VERY SMALL
(max. 1.3% for Chile, 1960)
TSUNAMI ENERGY COMPUTED from NORMAL MODE THEORY

[Okal, 2003]
Compute Kinetic Energy of water in Normal Mode Formalism

Note that most energy is carried by HORIZONTAL FLOW
Weigh by excitation function for each mode for given seismic moment
(averaged over focal geometry)
M0 .
Sum over individual modes (equivalent to integrating over frequency)

Account for source spectrum (according to seismic scaling laws)
Account for Finite extent of source depth.
E = 0. 219
w g
4/3
2/3
M 04/3
max
Essentially Equivalent to Kajiuras.
E grows as
M 04/3
Sumatra 2004: E 7. 5 1023 erg

(100 times Hiroshima)
WHAT ABOUT THE ATMOSPHERE ?

If the tsunami eigenfunction is prolonged into the Solid Earth which is
not totally rigid,
It should be possible to prolong it into the atmosphere, which is not a perfect

vacuum.
(The sea surface is not a totally "free" boundary)
This idea, hinted at by Yuen et al. [1970], was proposed by Peltier [1976].
<<<<<< STAY TUNED >>>>>>
M TSU
Use high seas tsunami waveforms recorded by
DART system
Consider tsunami as free oscillation branch of
Earths normal modes [Ward, 1980]
Recall Magnitude M m for seismic mantle waves;
Define
M TSU = log10 X( ) + C D + C S + C 0
Then, log10 M 0 = M TSU + 20
IT WORKS !!
[Okal and Titov, 2006]
RECALL MANTLE MAGNITUDE

[Okal and Talandier, 1989]
M m = X( ) + C D + C S + C 0
Applied to mantle Rayleigh waves; typically, T = 50 to

300 seconds.
X( ) is spectral amplitude in m*s
C D is distance correction
C S is source (frequency) correction
C 0 = 0. 90 is locking constant (predicted theoretically)

THEN,
M m is directly related to seismic moment M 0 :

M m = log10 M 0 20
M m combines simple "quick-and-dirty" concept of one-station magnitude with modern analytical approach (measuring a
bona fide physical quantity, the seismic moment, using physical units). It does not saturate.
Valid even for 1960 Chilean earthquake.
A tsunami wave on the high seas is a branch of normal
modes of the Earth [Ward, 1980].
QUESTION: Can we extend the concept of M m to a
tsunami wave measured on the high seas -- and call it
M TSU ?
DEVELOPING A FORMULA FOR M TSU

The basic formula for the spectral amplitude of a spheroidal wave
by a dislocation remains applicable:
X( ) = M 0 a
1
sin
2U Q
1
s R l 1/2 K 0 p R l 3/2 K 2 i q R l 1/2 K 1
M m = log10 M 0 20 = log10 X( ) + C D + C S + C 0
Need only adjust the corrections C D and C S and the constant C 0 .
THE DISTANCE CORRECTION C D
CD
1
=
log10 sin
2
THE SOURCE (FREQUENCY) CORRECTION C S

< s R p R > 1/2 g3/4
3/4
C S = log10
H
3/2
a
8
C S = 0. 087 3 0. 069 2 + 0. 508 + 2. 299

( = log10 T 3. 122).
(The latter formula uses Okals [2003] asymptotic expressions of the tsunami eigenfunction to compute the various
excitation coefficients for a shallow source in the limit
0).
DEVELOPING A FORMULA FOR M TSU

The basic formula for the spectral amplitude of a spheroidal wave
by a dislocation remains applicable:
X( ) = M 0 a
1
sin
2U Q
1
s R l 1/2 K 0 p R l 3/2 K 2 i q R l 1/2 K 1
M m = log10 M 0 20 = log10 X( ) + C D + C S + C 0
Need only adjust the corrections C D and C S and the constant C 0 .
THE LOCKING CONSTANT C 0
If X( ) is the spectrum of the wave height at the surface in

cm*s, then
C 0 = 3. 10
If one uses the bottom pressure p(t) recorded in dyn/cm2 on

the ocean bottom, then use P( ) rather than X( );
[P( ) = w g X( )], and
C 0 = 0. 11
If p(t) is recorded in pounds[-force] per square inch, then

C 0 = 4. 95
Case Study: KURIL ISLANDS, 04-OCT-1994

140
160
180
200
220
240
60
60
AK59
AK60
50
50
WC61
WC62
40
M 0 = 3 10
28
40
dyn cm
30
30
140
160
180
-160
-140
-120
To date, Largest Event Recorded by DART
Equivalent wave height at surface (cm)
AK 59
Time (hours) in Julian Day 277
WC 61
AK 60
WC 62
M TSU = 8. 23 0. 37
Published (CMT): 8.48
Case Study: KURIL ISLANDS, 04-OCT-1994

140
160
180
200
220
240
60
60
AK59
AK60
50
50
WC61
WC62
40
40
30
30
140
160
180
-160
-140
-120
CHILE -- 30 JUL 1995

210
240
270
60
60
AK64
45
45
WC67
WC68
WC69
30
30
15
15
-15
-15
210
240
270
Works despite UNFAVORABLE GEOMETRY

requiring NON-GEOMETRICAL propagation !!
SUCCESSFUL OPERATIONAL USE

17 NOV 2003
This is a smaller earthquake which was not recorded at the
Alaskan and West Coast DART gauges.
However, a new station, D-171, is only 900 km from the epicenter, and clearly recorded the tsunami, although at a very coarse
sampling (1 minute).
Despite this limitation, the event can be successfully processed.
160
60
170
180
190
200
210
60
M 0 = 5. 2 1027 dyn-cm
(CMT)
55
55
50
50
D-171
45
160
45
170
180
Rayleigh
(aliased)
190
TSUNAMI
200
210
Published (CMT): 7.71
M TSU = 7. 70 0. 22
This estimate was used in real-time to call off

an alert for Hawaii.
APPLICATION of M TSU to JASON SATELLITE TRACE

DETECTION by SATELLITE ALTIMETRY gives first
definitive measurement of MAJOR tsunami on HIGH SEAS
(previous detection by Okal et al. [1999] during 1992 Nicaragua
tsunami -- 8 cm -- at the limit of noise).
Satellite at the right place at the right time!
measures 70 cm across Bay of Bengal
cm
QUESTION: Can we quantify the JASON trace, i.e.,

recover from it the source of the tsunami ?
PROBLEM : JASON is neither a time series nor a

space series.
SOLUTION : Rebuild an approximate times series

from the JASON trace, then process through M TSU .
75
90
105
Original Jason Trace
03:02
15
-15
(b)
00:59
02:52
(a)
Equivalent Time Series
(c)
(d)
CONCLUSION: IT WORKS !!
M TSU: CONCLUSION
The algorithm succesfully retrieves the seismic

moment of the parent earthquake.
The examples tested suggest that the precision is

sufficient to avoid false alarms and failures to warn.

Lecture 6

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Lecture 6

Загружено:

Авторское право:

Доступные форматы

LECTURE 6

PRINCIPLES of HYDRODYNAMIC SIMULATIONS

Obtain model of Earthquake Rupture

Compute Static Deformation of Ocean Bottom

Use as Initial Conditions of

Run Hydrodynamic Model (e.g., MOST)

Propagate, up to and including

Involves MANY parameters

Position a point force F in an infinite homogeneous elastic medium

Obtain the Dynamic displacement field of the

[Aki and Richards, 1980; p. 73, Eqn. (4.23)]

The STATIC displacement is simply obtained

Replace Single Force by Double-Couple

Simply use Somiglianas tensor as a Greens function and take appropriate

Note that these are

[Aki and Richards, 1980; p. 79; Eqn. (4.29)]

Include effect of free surface

Integrate over finite area of faulting

(Combine with "reflection" of

[Stein and Wysession, 2002]

Mansinha and Smylie [1971]

STATIC DEFORMATION OF OCEAN BOTTOM

STATIC DEFORMATION OF OCEAN BOTTOM

100 150 200 250 300

L F = 200 km; W = 75km;

Use this static deformation field (limited to its oceanic portion) as

Justification: The seismic source is generally MUCH FASTER

(even in the case of SLOW, so-called "Tsunami" earhtquakes)

1. Snapshots of Sea Height at Given Times

CHILE 1906 + 1 hr. 45 mn

-20.0 -5.0 -2.0 -1.5 -0.5 -0.1 -0.05 0.05 0.1

HOW ROBUST IS THIS PROCEDURE ?

It is worth exploring the robustness of our results in the far

We study simulated amplitudes of the 2004 Sumatra-Andaman

We conclude that our results are indeed robust.

The primary parameters controlling the far field tsunami

2. CHANGE SOURCE PARAMETERS

SUMATRA 2004 Original

SUMATRA 2004 Dip = 12 deg.

SUMATRA 2004 Large Strain

By CONTRAST, WATER DEPTH at the SOURCE PLAYS a CRUCIAL ROLE

NORMAL MODE FORMALISM: A different approach

[After Lay and

SPHEROIDAL MODE HAS 6COMPONENT EIGENFUNCTION SATISFYING:

EASILY SOLVED WITH APPROPRIATE BOUNDARY CONDITIONS

EIGENFUNCTIONS of SPHEROIDAL MODES

EXCITATION OF TSUNAMI in NORMAL MODE FORMALISM

Gilbert [1970] has shown that the response of the Earth to a

1 cos n t exp ( n t/2Q n )

the EXCITATION of each mode being proportional to the scalar

Now, an EARTHQUAKE is represented by a system of

The response of the Earth to an earthquake is thus

1 cos n t exp ( n t/2Q n )

ADVANTAGES of NORMAL MODE FORMALISM

Handles any Ocean-Solid Earth Coupling

Works well at Higher Frequencies

DRAWBACKS of NORMAL MODE FORMALISM

Must assume Laterally Homogeneous Structure

Linear Theory -- Does not allow for Large Amplitudes

NOTE: Energy scales as L 4 , i.e., as M 04/3 .

ENERGY of a TSUNAMI -- STATIC THEORY [Kajiura, 1981]

= invariant ratio of M 0 to S 3/2

NOTE: Energy of Tsunami grows faster than Seismic Moment

TSUNAMI ENERGY COMPUTED from NORMAL MODE THEORY