ESS 8 Earthquakes Week 7

1
ESS 8 - Earthquakes - Week 7
Prof. Didier Sornette TAs Sara Cina and Jelena
Tomic
http//www.ess.ucla.edu/academics/courses/web/fall
_2005/ess_8/index.asp
2
Video trip

• Saturday Nov. 19 from 10am to 1230pm in room
  CS76 of Young Hall.
• The "video trip" will consist in approximately
  two 1-hour videos followed by a 30 minute quiz
  with 25 multiple choice questions. You will be
  able to take notes during the video and use them
  during the quiz. You do not need to bring the
  scandrons.
• The videos will expose you to real-life
  situations associated with two recent important
  earthquakes and will help make more concrete the
  concepts and facts learned in the lectures and
  lab-sessions. This should be both fun and useful.
• The "video trip" accounts of 10 of the total
  grade.

3
http//earthquake.usgs.gov
4
Content for this week

• Earthquake sequences
• Number of earthquakes
• Foreshocks
• Aftershocks
• Quake Prediction

5
Types of Magnitude

• ML - Local or Richter magnitude
• Original magnitude, developed by Charles Richter
  in 1930s
• uses S wave recorded within 300 km of epicenter
• mb - Body-wave magnitude
• uses P wave recorded at 30 to 90 distance
• MS - Surface wave magnitude
• uses surface wave
• MW - Moment magnitude
• uses seismic moment M0 ? D S

6
Describing magnitude
M 7.8 or above Great

• Under M 5
• Small
• M 5 to 6
• Moderate
• M 6 to 7
• Large
• M 7 to 7.8
• Major

7
THE TEN LARGEST EARTHQUAKES IN THE UNITED STATES
Mag Date (UTC)
Location 1. 9.2 March 28, 1964
Prince William Sound, Alaska 2. 8.8
March 9, 1957 Andreanof Islands,
Alaska 3. 8.7 February 4, 1965
Rat Islands, Alaska 4. 8.3 November 10,
1938 East of Shumagin Islands, Alaska
8.3 July 10, 1958 Lituya Bay,
Alaska 6. 8.2 September 10, 1899
Yakutat Bay, Alaska 8.2 September 4,
1899 near Cape Yakataga, Alaska 8. 8.0
May 7, 1986 Andreanof Islands,
Alaska 9. 7.9 February 7, 1812
New Madrid, Missouri 7.9 January 9,
1857 Fort Tejon, California 7.9
April 3, 1868 Ka'u District, Island
of Hawaii 7.9 October 9, 1900
Kodiak Island, Alaska 7.9 November
30, 1987 Gulf of Alaska
8
THE TEN LARGEST EARTHQUAKES IN THE CONTIGUOUS
UNITED STATES Mag Date (UTC)
Location 1. 7.9 February 7,
1812 New Madrid, Missouri 7.9
January 9, 1857 Fort Tejon, California
3. 7.8 March 26, 1872 Owens
Valley, California 7.8 February 24,
1892 Imperial Valley, California 5.
7.7 December 16, 1811 New Madrid, Missouri
area 7.7 April 18, 1906 San
Francisco, California 7.7 October 3,
1915 Pleasant Valley, Nevada 8. 7.6
January 23, 1812 New Madrid, Missouri
9. 7.5 July 21, 1952 Kern
County, California 10. 7.3 November 4,
1927 west of Lompoc, California
7.3 December 16, 1954 Dixie Valley,
Nevada 7.3 August 18, 1959
Hebgen Lake, Montana 7.3 October 28,
1983 Borah Peak, Idaho
9
How manyquakes?
20,000/yr in California Just those measured, M gt
1 Most were small, M 1 or 2 Cant even be felt
LA Times 1/25/97 p. A21
10
How many earthquakes are there in a year?
Data from ANSS catalog for 2003
11
Data from ANSS catalog for 2003
12
How many earthquakes are there in a year?
Data from ANSS catalog for 2003
13
Data from ANSS catalog for 2003
14
Global counts of quakes per year
Gutenberg-Richter
1 magnitude 8 or bigger 10 magnitude 7 or
bigger 100 magnitude 6 or bigger 1000 magnitude 5
or bigger etc
Average data from 1904-1980 Kanamori and Brodsky,
2001
15
Gutenberg-Richter relation

• Relation between
• magnitude M and
• number of quakes N with magnitude greater than M
• Mathematically
• N 10a-bM C 10 -bM
• Typically b1 so 10 times more quakes greater
  than M-1 than greater than M
• Log N a-bM

16
California Seismicity
?
17
How many Magnitude 3 in CA?

• 3000 magnitude 6 world-wide in 15 years
• 200 magnitude 6 per year
• 200,000 magnitude 3 per year
• CA (1/2000) earth surface
• 100 magnitude 3 per year in CA
• gt about 2 per week

18
Spatial and temporal organization of seismicity
in California
Landers 28 june 1992 M7.3
Landers, M7.3 28/06 457
Big-Bear, M6.4 28/06 805
Joshua Tree, 22/04 M6.1
19
Foreshocks, Mainshocks, Aftershocks

• Seismicity rate in South California, 1978-2000

TRIGGERING Triggering is not due just to static
stress loads
Northridge
time (days)

• non-stationary seismicity rate at short time
  scales
• increase of the seismicity rate after large
  earthquakes aftershocks

20
Definitions

• Sequence
• Set of quakes that appear related
• Foreshock
• Quake followed by a bigger quake in same sequence
• Mainshock
• Biggest quake in a sequence
• Aftershock
• Quake after the biggest quake in a sequence
• Corollaries
• One never knows that an event is a foreshock
  until the mainshock comes along
• Aftershocks can turn into foreshocks

21
Differences between mainshocks, foreshocks and
aftershocks

• NONE! (we think at the time of this lecture)

22
Mainshock

• Largest earthquake in a sequence
• Larger mainshocks strain more volume of rock,
  have more aftershocks
• Foreshocks and aftershocks usually at least 1
  magnitude unit smaller than mainshock

23
Foreshocks

• Smaller earthquakes that precede the mainshock
• often by just hours
• Few in number
• only half of mainshocks have even one foreshocks
• Usually near mainshock hypocenter
• part of the nucleation process

24
Haicheng 1975 (China)
Magnitude of earthquakes over time
Foreshocks
(Days)
Great prediction success
25
Tangshan July 27, 1976 M 7.6-8.2no foreshocks

• 255,000 official death toll
• Maybe more than 650,000
• Worst casualties of the 20th century

26
Intensity up to XI
Occurred in large city (1 million) with
generally low seismicity
... like the East Coast?
http//c1.eq-igl.ac.cn/images/tangshan/10.jpg
27
Jan. 23, 1556 ShanXi Central China

• 850,000 people died
• Homes in caves that collapsed
• Worse loss of life ever documented from an
  earthquake

28
A blow for earthquake prediction

• Feb. 4, 1975 Haicheng
• A successful prediction based on foreshock
  activity
• 14 months later.
• The worst disaster in the 20th century

but
29
(No Transcript)
30
QINGLONG COUNTY and the 1976 TANGSHAN
EARTHQUAKE A MODEL INTEGRATING PUBLIC
ADMINISTRATION and the SCIENCE OF DISASTERS An
example of Best Practices in Public
Administrationlt
Qinglong District Damage Prevention One county in
the Tangshan area was warned about the earthquake
up to two years in advance (http//www.globalwatch
.org/ungp/doc69-tr.htm) . This allowed officials
to educate the people in the area about how to
deal with major earthquakes. After monitoring
stations detected the water was getting muddy the
whole county started holding classes outside and
preparing for a great quake. As a result of their
warning and public integration no one was killed
in the county by the earthquake (except for one
man who died of a heart attack). This is an
example of the integration of scientific
information and public administration. Long-term
and intermediate term prediction SUCCESS
(several years of advanced warning) "From all the
data and trends, we conclude that this area,
within 1-2 years, may have a M8 EQ the area
should therefore actively prepare, widely
circulate this EQ knowledge (especially to big
factories and mines) and make plans and proper
measures for EQ preparation. July 24, 1976 At
the CCP Standing Committee meeting, there is a
difference of opinion concerning the appropriate
response to the EQ situation, the possibility of
creating panic and loss of credibility, and the
degree of popular knowledge and/or presence of
superstitions about EQs. Ran Guangqi (absent at
another meeting but kept fully informed) and Yu
Shen, both of whom hold important posts, assert
the wisdom of community preparation in view of
Document No. 69 and information from Tangshan
Conference. July 25 The population is alerted
to the most recent EQ situation, on the dangers
of EQs, and of how to make themselves safer
before, during and after an EQ. Officials do
not rest during the emergency preparations,
working day and night
31
1- Most villagers believe the EQ broadcast
because county officials base their report on
data from scientists. 2 EQ warning corroborates
lay-monitoring evidence from two local sites
strange animal behavior and changes in water
level, color, chemistry, and gas release
observed. From July 20, 1976, villagers noticed
domestic animals behaving very strangely pigs
ran in circles and would not stay in pens
chickens refused to stay in chicken coops also,
yellow weasels left their hiding places and ran
around unafraid of the villagers. 3 Village
patrols (twice daily) are set up to prevent
people from sneaking back into their houses
(able-bodied people are fined if caught inside
houses). 4 Villagers learn lay techniques to
sense EQs, e.g., overturn empty glass bottle and
balance in metal wash basin, so as to hear the
bottle tip over in an EQ. 5. Every family
assigns one person to NOT sleep (in shifts),
i.e., families are instructed to take
responsibility for their own survival and
safety. 6. Heightened awareness that EQ could
"happen any day now !!" Widespread dissemination
of information on precursors. 7. Students are a
major resource in preparedness activities. 8.
Dong Wu, doctor at hospital in Qinglong County.
Goes to Tangshan on night of July 27, and stays
with relatives. Informs his relatives that
Qinglong County is prepared for an EQ and warns
them to prepare also they listen in disbelief
and tell him not to tell others to avoid panic
he puts his clothes by his bedside, to leave
house quickly should EQ begin. Relatives accept
advice to leave doors and windows open, sleep
lightly and stand an empty bottle upside-down on
edge of table.
32
July 28 342 am Qinglong County County residents
are aware and prepared by the time GTE
strikes. (Note Qinglong County is located 115 km
from Tangshan.) 1 DEATH ONLY (due to heart
condition). Animals are safe. County sustains
maximum damage of intensity VIII (8). Residents
at Wen Quan Village hear and experience the
destructive power of the EQ as large sections of
the historic 1000-year-old Great Wall split and
crash down from nearby hilltops. 7,000 buildings
collapse totally. 180,000 buildings are
damaged. At county middle school, roof shifts
and walls collapse, but no loss of life. EQs
continue to create havoc after mainshock close
cooperation between public administrators,
scientists and lay public is essential to
minimize damage and loss of life. Information
used in the chronology on Qinglong County is
drawn from interviews with the county during
field visits in September 1995 and July 1996.
Many of the statistics and some of the quotations
come from the following books and newspaper
articles Chen, Y., T. Kam-ling, F. Chen, Z.
Gao, Q. Zou and Z. Chen. 1988. THE GREAT TANGSHAN
EARTHQUAKE OF 1976 AN ANATOMY OF DISASTER.
Pergamon Press 153 pp. (in English).
33
Early warnings Many people in Tangshan reported
seeing strange lights so-called "earthquake
lights" the night before the earthquake. Well
water in a village outside of Tangshan reportedly
rose and fell three times the day before the
earthquake. Gas began to spout out of a well in
another village on July 12 and then increased on
July 25 and July 26. Other wells throughout the
area showed signs of cracking. There is also
evidence that animals in the area sensed the
earthquake before it struck. A thousand Chickens
reportedly refused to eat and acted wildly. There
were also reports that dogs would not stop
barking and goldfish jumped out of their bowls.
Qinglong story unearthed only recently,
perhaps for political reasons
34
Consequences The People's Republic of China
government refused to accept international aid,
and its own efforts were criticized as
inadequate. It was also criticized for having
ignored scientists' warnings of the need to
prepare for an earthquake. The earthquake came as
an event in the continuous "Curse of 1976" in
China it was preceded by the deaths of Zhou
Enlai (Premier from 1949 to 1976) and Zhu De
(Supreme commander) in earlier months and
followed two months later by the death of Mao
Zedong and the Gang of Four trying to grab
power. The political repercussions of the
disaster and its aftermath contributed to the end
of the Cultural Revolution in China 3
(http//discover.npr.org/rundowns/segment.jhtml?wf
Id4281429) . The Gang of Four accused Deng
Xiaoping of sabotaging relief efforts as part of
its "Criticize Rightist Deviationism" campaign.
Mao's chosen successor Hua Guofeng took the
opportunity to show concern, thereby solidifying
his status as China's paramount leader. He, with
Chen Yonggui, made a personal visit to Tangshan
on August 4 to survey the damage. This
visit earned him considerable prestige and two
months later, he staged what amounted to a coup
by arresting the Gang of Four.
35
Aftershocks

• smaller earthquakes following the largest
  earthquake of a sequence (the mainshock) near
  mainshock rupture zone
• follow almost all shallow earthquakes
• cover ruptured area
• can number in thousands
• can last for years or decades
• aftershocks of Northridge M 6.7 are still
  occurring

36
Aftershockfrequency

• Mathematically, rate of aftershocks follows
• N C/t
• where N is the number of earthquakes
• t is time
• C is a constant

Ignores foreshocks
37
Temporal decay of aftershocks

• the seismicity rate after a mainshock at time
  t0 follows the modified Omori law
• p is in the range 0.3, 2, often close to 1
• c is a small time 1 hour, delay between
  mainshock and aftershocks
• K increases with M (large earthquakes have more
  aftershocks)

Example for the Landers aftershock sequence
(1992, M7.3, California)
time (days)
38
Omoris Earthquake
The decay of aftershock activity following the
1891 Nobi, Japan, earthquake . for over 100
years!
Number of Earthquakes
Utsu (2002)
39
Magnitude
Landers Earthquake
Date
40
Numbers of aftershocks

• Northridge 13,523 aftershocks in 1994-1996
• Landers 65,380 aftershocks in 1992-1996
• 22 with M gt 5.0
• Most were M1 or 2
• There were many more too small to detect

41
Self-similarityNumber of aftershocks ????
(Helmstetter, 2002)
42
Distribution of sizes

• As for mainshocks, there are many more small
  aftershocks in a sequence than big aftershocks
• If mainshock has M 6
• 1 or 2 aftershocks with M 5 to 6
• 10s of M 4 to 5
• If mainshock has M 8, an M 7 aftershock is likely
• Likelihood of getting a big earthquake decreases
  with time
• Combine Gutenberg-Richter and Omoris Law

43
Cause of aftershocks

• Every time there is an earthquake, the volume of
  rock around the rupture is strained, that is,
  twisted or squeezed. (How, exactly?)
• Role of asperities that stopped the mainshock
• Sometimes, the strained rock breaks.
• Often, it takes a while for it to break, so the
  aftershocks may appear seconds to years after the
  causative quake.
• But we dont know for sure why there is a delay.
• Static fatigue
• Visco-elastic relaxation
• Diffusion processes (fluids?)

44
Aftershocks tell us about mainshock

• Seismologists estimate the area of rupture by
  mapping aftershock locations
• Aftershocks cover the rupture area and expand
  slightly outside of it
• Obtain length and width of faulted area gt
  magnitude of mainshock
• Obtain orientation of faulted area

45
LA aftershock examples

• 1971 San Fernando quake
• North-east dipping rupture plane
• Mainshock focus at bottom of rupture
• Then crack spread up and sideways
• 1994 Northridge quake
• Southwest dipping rupture plane
• Mainshock focus at bottom of rupture
• Then crack spread up and sideways

46
N
47
Cross Section
North East
48
Loma Prieta example

• 40-50 km long aftershock zone
• Extends to 12 km depth
• Slightly dipping to southwest
• Again, focus near middle of bottom of rupture
  zone
• Loma Prieta had two M 5 foreshocks 6 months
  earlier very near focus

49
Loma Prieta aftershocks
Cross-section
Along fault view
Map view
N
Ellsworth paper
50
Static stress triggering
Earth is stretched or compressed by fault
movement (like in elastic rebound simulations)
Some earthquakes near fault Some where
stress level was raised Few where stress level
dropped
51
Testing Static Stress triggering

• Calculate of where stress should be raised or
  lowered by a large rupture
• Do aftershocks occur preferentially in regions
  stressed by the mainshock?
• Yes, more or less
• But debatable

stress lowered
stress raised
52
(No Transcript)
53
Parsons, JGR (2002)
54
(No Transcript)
55
Long-Range Triggering
Earthquakes following the Mw 7.3 June 28, 1992
Landers earthquake
Hill et al. (1993)
56
Dynamic Triggering

• Shaking, rather than just long-term loading,
  triggers aftershocks
• Evidence from earthquakes with strong directivity

Gomberg et al. Nature (2004)
57
Sequences

• Whenever there is a quake, it becomes more likely
  that more quakes will come soon
• 10 chance that any quake will be followed by a
  bigger quake
• With passing time (and no quake), odds return to
  normal
• The longer the time till the last EQ, the longer
  the time till the next EQ?

58
Multiple triggering Epidemic Type Aftershock
Sequence (ETAS)

• Model proposed by Kagan and Knopoff 1981, 1987
  and Ogata 1988
• each earthquake can be both a mainshock, an
  aftershock and a foreshock
• each earthquake triggers aftershocks according to
  the Omori law, that in turn trigger their own
  aftershocks
• OMORI LAW 1/t
• the number of aftershocks triggered by a
  mainshock depends on the mainshock magnitude
• PRODUCTIVITY LAW
• aftershock magnitudes follow the
  Gutenberg-Richter distribution, independently of
  the time and of the mainshock magnitude
• GUTENBERG-RICHTER LAW

59
ETAS model and numerical simulations
Rate of aftershocks for a numerical simulation of
the ETAS model

• is the global law also an Omori law ?
• pglobal plocal ?

• particular case a0 studied by Sornette and
  Sornette,1999

60
Prediction of Mgt6 events n1, q0.2, b2/3,
c10-3, m03 and a?/21/3 In running window of
past 100 events, prediction based on max
magnitude, MLE of b-value and rate
Only 20 of large EQ are predictable in this model
ERROR DIAGRAM
Helmstetter and Sornette (2003)
61
Temporal variation of seismicity
Observations
62
Non-conventional sequences

• Swarms
• Long-range triggering
• Jumping faults

63
Swarms

• Definition Earthquake swarm is a sequence of
  earthquakes with no clear mainshock-aftershock
  relationship

Example Miyakejima, 2000
Magnitude
JMA, Earth Planets and Science, 2000
Date
64
Compare to aftershock sequence
Landers Earthquake
Magnitude
Date
65
Omori law in book sales, financial volatility,
internet downloads
D. Sornette et al., 2004
66
Earthquake predictionhttp//helix.nature.com/deba
tes/earthquake/equake_frameset.html

• Time-independent hazard. We assume that
  earthquakes are a random (Poisson) process in
  time, and use past locations of earthquakes,
  active faults, geological recurrence times and/or
  fault slip rates from plate tectonic or satellite
  data to constrain the future long-term seismic
  hazard. We then calculate the likely occurrence
  of ground-shaking from a combination of source
  magnitude probability with path and site effects,
  and include a calculation of the associated
  errors. Such calculations can also be used in
  building design and planning of land use, and for
  the estimation of earthquake insurance.
• Time-dependent hazard. Here we accept a degree of
  predictability in the process, in that the
  seismic hazard varies with time. We might include
  linear theories, where the hazard increases after
  the last previous event, or the idea of a
  'characteristic earthquake' with a relatively
  similar magnitude, location and approximate
  repeat time predicted from the geological dating
  of previous events. Surprisingly, the tendency of
  earthquakes to cluster in space and time include
  the possibility of a seismic hazard that actually
  decreases with time. This would allow the
  refinement of hazard to include the time and
  duration of a building's use as a variable in
  calculating the seismic risk.

67

• Earthquake forecasting. Here we would try to
  predict some of the features of an impending
  earthquake, usually on the basis of the
  observation of a precursory signal. The
  prediction would still be probabilistic, in the
  sense that the precise magnitude, time and
  location might not be given precisely or
  reliably, but that there is some physical
  connection above the level of chance between the
  observation of a precursor and the subsequent
  event. Forecasting would also have to include a
  precise statement of the probabilities and errors
  involved, and would have to demonstrate more
  predictability than the clustering referred to in
  time-dependent hazard. The practical utility of
  this would be to enable the relevant authorities
  to prepare for an impending event on a timescale
  of months to weeks. Practical difficulties
  include identifying reliable, unambiguous
  precursors, and the acceptance of an inherent
  proportion of missed events or false alarms,
  involving evacuation for up to several months at
  a time, resulting in a loss of public confidence.
  
• Deterministic prediction. Earthquakes are
  inherently predictable. We can reliably know in
  advance (all within narrow limits (again above
  the level of chance), so that a planned
  evacuation can take place)
• their location (latitude, longitude and depth),
• magnitude,
• time of occurrence.

68
To make an earthquake prediction need to state

• Time interval in which quake will occur
• Region in which quake will occur
• Magnitude range of predicted quake
• Small quakes occur more commonly
• Easy to predict there will be magnitude 3
  somewhere in Southern Ca. next month, but not
  useful

69
Impossible?

• Perhaps there is no information about exact time,
  place and size

70
Sandpiles and Critical States
Adding a grain of sand to a steep sandpile can
cause any avalanche size
If the Earth is always in a critical state,
there is no way to predict if the next
earthquake will be small or large
71
All known big faults in SoCal
Most faults have longer repeat times than San
Andreas
SCEC web page
72
Self-organized fault-earthquake model
P. Miltenberger, D. Sornette and
C.Vanneste "Fault self-organization as optimal
random paths selected by critical spatio-temporal
dynamics of earthquakes", Phys.Rev.Lett. 71,
3604-3607 (1993)
73
STRESS FIELD
74
FIELD OF STRESS/THRESHOLD
75
Fakes

• Richter in 1977
• "Journalists and the general public rush to any
  suggestion of earthquake prediction like hogs
  toward a full trough... Prediction provides a
  happy hunting ground for amateurs, cranks, and
  outright publicity-seeking fakers"

76
(No Transcript)
77
Myths to debunk

• Earthquake weather
• My dog barked at a strange pitch (or cat ran in
  circles or horse jumped through the fence, etc.)
• Big earthquakes always happen in the early
  morning

78
FALSIFIABILITY (Sir Karl Popper and Ernest
Gellner)
Falsifiability is an important concept in the
philosophy of science that amounts to the
apparently paradoxical idea that a proposition or
theory cannot be scientific if it does not admit
the possibility of being shown false.
Falsifiable does not mean false. For a
proposition to be falsifiable, it must be at
least in principle possible to make an
observation that would show the proposition to
be false, even if that observation had not been
made. For example, the proposition "All swans
are white" would be falsified by observing one
black swan.
79
Optimism

• Some success
• Haicheng 1975
• Many possible phenomena to study

80
Two strategies forearthquake prediction

• Find a specific precursor
• Water level changes, electromagnetic waves,
  foreshocks, etc.
• Forecast a general pattern
• Earthquakes are more likely when there are many
  earthquakes
• 62 probability of at least one magnitude 6.7 in
  the San Francisco Bay region before 2032.

81
Possible precursors

• Change (increase or decrease) in number of
  earthquakes
• For example, foreshocks
• Slow ground motion (geodetically measured)
• Radon emission
• Electrical resistivity
• Electromagnetic waves, ionospheric waves
• Water chemistry
• Seismic wave velocity

82
Post-dictions

• Retrospective predictions
• After an earthquake happens, look for strange
  signals before it
• VERY dangerous practice
• There is always something strange happening
• However, might be the only way to learn
• Seismology is an observational science

83
Example 1Electromagnetic fluctuations

• Oct. 17, 1989 Loma Prieta Earthquake
• Measurements 7 km from epicenter
• Increase in signal
• Oct. 5 and again Oct. 17

Still looking for another earthquake but NASA and
Los Alamos looking into it
84
Example 2The bottled water story
Jan. 17, 1995 Kobe Earthquake
85
Chemical precursor

• Analyzed dated, bottled water
• Found 6 month precursor in chloride
• and other chemicals

Rare case
(Tsunogai and Wakita, 1995)
86
Johansen et al.
87
Radon

• Radon is a radioactive gas emitted by rocks
• Radon gas concentration also increased before
  Kobe (Igarashi et al., 1995)
• A few
• days-long signal

Found for a few earthquakes
88
Radon expected?

• Increasing stress increases very small fractures
  in rocks before earthquake
• Releases small quantities of gas from rock
• The more fractures, the more gas released.

89
Critical Earthquake Concept

• Cooperative Behavior
• Correlation
• Susceptibility and response to external
  perturbations
• Fractal geometry
• D.D. Bowman, G. Ouillon, C.G. Sammis, A. Sornette
  and D. Sornette, An Observational test of the
  critical earthquake concept, J.Geophys. Res.
  103,(NB10), 24359-24372 (1998).
• D. Sornette, P. Miltenberger and C. Vanneste,
  Statistical physics of fault patterns
  self-organized by repeated earthquakes, Pure and
  Applied Geophysics 142, 491-527 (1994).

90
(No Transcript)
91
(No Transcript)
92
(No Transcript)
93
H. NECHAD, R. EL GUERJOUMA, A. HELMSTETTER and
D. SORNETTE (2004)
94
Our prediction system is now used in the
industrial phase as the standard testing
procedure.
J.-C. Anifrani, C. Le Floc'h, D. Sornette and B.
Souillard "Universal Log-periodic correction to
renormalization group scaling for rupture
stress prediction from acoustic emissions",
J.Phys.I France 5, n6, 631-638 (1995)
95
(No Transcript)
96
(No Transcript)
97
From Sykes and Jaumé, 1990
98
Bufe and Varnes, JGR, 1993
99
(No Transcript)
100
RMS Residuals
200 km
101
Kern County
102
(No Transcript)
103
D.D. Bowman, G. Ouillon, C.G. Sammis, A. Sornette
and D. Sornette An Observational test of the
critical earthquake concept, J.Geophys. Res. 103,
(NB10), 24359-24372 (1998).
104
(No Transcript)
105
Accelerating Seismicity
106
Future California Earthquakes?
No?
Yes?
Bowman et al. (2003)
107
(No Transcript)
108
Real predictions

• 1975 Haichung quake
• Predicted
• But there were many M6 foreshocks
• So it was a much easier than usual quake to
  predict
• Parkfield seemed to have a 22 year repeat time
• Was supposed to happen in 1988 or so
• Lots of equipment put out
• Quake has come in 2004

109
Haicheng 1975
Magnitude of earthquakes over time
Foreshocks
(Days)
110
Forward predictions

• Parkfield seemed to have 22 year repeat time
• Was supposed to happen in 1988 or so
• Lots of equipment put out

• Broke in 1857, 1881,
• 1901, 1922, 1934,
• 1966, 2004, ?

111
HISTORIC SEISMICITY OF THE PARKFIELD AREA

• 1857.
• 1881.
• 1901.
• 1922.
• 1934.
• 1966.
• 2004.
• ?
• (but direction of rupture and focus different)

24 20 21 12 32 38
112
Is the pattern real?

• Maybe historical data is bad
• Intensity in central California sparse in the
  19th century
• Maybe data is good, but random
• Not hard to find spurious patterns
• Like presidential deaths in office

113
American Presidents

• 1861-1865 A. Lincoln (elected 1860)
• 1865 April 14, Wounded by assassin John Wilkes
  Booth 1865 April 15, died early in the morning
  from wound in Washington, D.C.
• 1881-1884 J. Garfield (elected 1880)
• 1881 July 2, Wounded by assassin in Washington,
  D.C., 1881 September 19, died from wounds at
  Elberon, New Jersey
• 1897-1901 W. McKinley (re-elected 1900)
• 1901 September 6, Shot by an assassin in Buffalo,
  New York, September 14, died from wounds in
  Buffalo
• 1921-1924 W.G. Harding (elected 1920)
• 1923 August 2, died in San Francisco
• 1933-1945 F.D. Roosevelt (re-elected 1940)
• 1945 April 12, died at Warm Springs, Georgia
• 1961-1964 J.F. Kennedy (elected 1960)
• 1963 November 22, Assassinated in Dallas, Texas
• 1981-1984 R. Reagan (elected 1980)
• 1981 March 30, wounded in an attempted
  assassination
• 2001-2008 G.W. Bush? (elected 2000)

114
Quiescence

• Earthquakes are due if they have not happened in
  a while
• Are we due for a big one?
• Seismic gaps

115
Mexican coast
http//tlacaelel.igeofcu.unam.mx/vladimir/sismos/
100years.html
116
Guerrero gap
117
Slow Slip event
Our seismotectonic studies along the Pacific
coast of Mexico during the last few years have
resulted in the discovery an extremely "slow" or
"silent" earthquake on the plate interface of the
Guerrero seismic gap. Sudden rupture of this gap
may produce a catastrophic earthquake of
magnitude Mwgt8.0. The discovery of the "slow"
earthquake is based on very accurate long-time
geodetic measurements obtained with our new
network of global positioning system (GPS)
stations. The silent event began in October 2001
and lasted for about 6 months. The slow slip
occurred over an area of 250x200 km2. The
modeled average slip on the interface was about
15 cm and the equivalent magnitude, Mw, was 7.5.
In contrast, an ordinary Mw7.5 earthquake lasts
for about 15 seconds, ruptures an area of 55x55
km2, and involves an average slip of 200 cm. A
few silent earthquakes (also called transient
slips or "aseisms") have been recently reported
in other subduction zones but the last Guerrero
slow event is unique in that it is probably the
largest reliably detected so far anywhere in the
world and extends over an astonishingly large
area. The discovery of this slow event has far
reaching implications for our understanding of
the rheology of the plate interface, the
earthquake cycle and the source process. Our
study calls for a reassessment of the seismic
potential and a seismotectonic monitoring of the
Guerrero and other seismic gaps in Mexico. Our
discovery can be summarized by the 3D
hypsographic image of the subduction zone in
Guerrero, shown below, where the velocities of
the GPS sites for the interseismic steady-state
motion (yellow vectors) are compared with those
during the last 2001-2002 silent earthquake (red
vectors).
http//tlacaelel.igeofcu.unam.mx/vladimir/guerrer
o_level/deform0.html
118
(No Transcript)
119
Prof. Keilis-Borok

• UCLA Professor-in-residence
• Works at UCLA and in Moscow
• Working on earthquake prediction for 20 years

120
Short-term earthquake prediction
Research
A case history
A novel methodology Reverse Tracing of
Precursors (RTP), is developed for short-term
(months in advance) earthquake prediction
Advance prediction of San Simeon (M6.5, Dec. 22,
2003)

• Precursor detected May, 2003

• Precursor reported to the
• group of experts in June, 2003

The RTP methodology uses increase of earthquake
correlation range as a short-term premonitory
signal

• San Simeon earthquake, M6.5
• occurred on Dec. 22, 2003
• within the alarm.

An experiment in advance prediction has been
launched in four seismically active regions
around the World first results are encouraging
Broader Impact
Groups research was covered by National
Geographic Society, Nature, Time, LA Times, Wired
News, etc
Potential applications to other
geological, geotechical disasters. Collaboration
with experts in geodynamics, complex systems,
pattern recognition, and disaster management
from US, Russia, Japan, France, Italy and UN
Keilis-Borok's group
121
Rare event

• Only two earthquakes 6.4
• in the region in the 40 year
• training window
• Again, counting
• is tricky
• lt3 chance of random occurrence
• This is interesting!

122
Another Keilis-Borok prediction

• 50 likely Mgt6.4
• in shaded region
• Oct. 29, 2003
• Sept. 5, 2004
• Even probability throughout window
• Probability by standard methods of earthquake in
  region is 10

The earthquake did not occur false alarm
123
Japan, 2003

• Predicted M 7 earthquake in 9-month window
  covering March-December 2003
• Announced July 2003
• Tokachi-Oki (Hokkaido) earthquake M8.1 Sept. 25,
  2003

124
Hokkaido

• 9 earthquakes (by their count) in region in 27
  years
• Counting is tricky!
• Leave out aftershocks,
• deep earthquake and Mlt7
• 25 change of getting an earthquake in any
  9-month window by chance

125
How does he (they) do it?
126
Chains
127
Intermediate term

• Over 6-24 months before chain
• Combine
• Rate of earthquakes
• Area of ruptured faults
• Change in rate
• Number of earthquakes
• Magnitudes of earthquakes
• Spatial clustering
• Mainshocks
• Aftershocks
• Maximum distance between earthquakes
• Relative number of big and little earthquakes
  (b-value)

128
How to measure?

• Separate equation for each feature (8 equations)
• Some of these measures include some unknown
  parameters
• Each measure votes for or against prediction
• At least 8 (unknown) thresholds
• Add the votes and find (another) threshold to
  determine if intermediate-term pattern is there.

129
Training
Red are chains followed by earthquakes Blue are
false alarms
Use to find the best parameters Only 12 sequences
to train on
130
The Optimal Total Vote
131
M8 Kossobokovs method
132
Summary

• Scientific prediction
• This is NOT like listening to dogs bark
• Possible problems
• Training set is small
• Many parameters
• Transparency
• Hard to figure out method and data used
• Some information like 50 probability only
  announced after the window started

133
Forecast vs. Prediction

• Prediction Specific statements about when,
  where and how big
• Usually used for short-term (weeks- month) nearly
  deterministic
• Forecast General statement about likelihood of
  earthquakes
• Always has a probability
• Usually used for long-term (decades)

134
Example of a Forecast2002 Working Group Bay
area report

• Calculated 62 probability of a major (6.7)
  earthquake in the Bay Area in the next 30 years

Image courtesy USGS
135
30-yr probability of quakes in California
Parkfield
1906 repeat
1857 repeat
Yanev p. 39
Note absence of Northridge, Landers, San
Fernando, ...
136
Probability

• How often you expect something to happen
• ex) Flipping a coin lands on heads 50 of the
  time
• Reported as percent (50), decimal (0.5) or
  fraction (1/2)
• Must be between 0 and 100 (or equivalently
  between 0 and 1)

137
Mid-term example
22/1740.12612.6 gives me the probability that
a student who I pick at random has a score larger
than or equal to 90
138
What is Probability?

• Relative frequency of a given outcome when
  repeating the game (coin tossing,)
• Fraction of paths or trajectories leading to
  the outcome
• Belief (Bayesian theory) (prior to posterior by
  likelihood)
• Dutch book bet

80 probability of a Mgt7 in the next 30 years
139
Probability distributions

• Probabilities form curve
• Forecast reports most probable (like the mean of
  exam) and measure of error (width of the peak)

140
Confidence Interval

• Confidence interval measures the probability that
  a probability is correct
• Ex) Bay area group reports 62 probability of a
  major earthquake with a 95 confidence interval
  between 37 and 87
• 95 confidence interval says that 95 of the
  time, the probability will be between the bounds
• Lots of error
• Still useful for insurance companies, government
  planners, etc.

141
Probability of quake

• Find the faults
• Estimate how faults are segmented
• How does each segment behave
• Size of its quakes
• Time between quakes - recurrence interval
• Sum up risk from all segments of all faults
• (This exercise tells how much shaking)
• Then figure out expected damage

142
Fault zone segmentation

• Characteristic earthquake model
• Only one segment breaks at a time
• Segments defined by
• Ends of fault traces
• Fault intersections
• Changes in rock type along fault?
• Best guesses - segment defined from prior quakes
• Not clear whether the concept of fault
  segmentation is correct

143
Wasatch Faultsegmentation
1
1
2
3
4
5
6
Keller, 8-21
144
HistoryofWasatchsegments
Age of faulting events on the Wasatch Fault
Pinter workbook 10-4
Now
6000
(years)
145
Characteristic behavior of segments
Keller, Table 8-2
146

• Rupture displacement d and length of fault
  rupture L
• d a L
• d10m for L 1000km gt a 10-5
• Take L100km gt d1m
• V(tectonic) 1cm/y gt
• T d/V 1m/(1cm/y) 100 years

147
How does this apply to SoCal?

• Outline
• Segmentation of the San Andreas Fault
• Behavior of a segment on the San Andreas
• Probabilities for San Andreas segments
• Locations of all SoCal faults
• Total probability across SoCal

148
San Andreassegmentation
1906-type events
creep
Four major segments
1857-type events
Keller, 8-20
149
Pallett Creek A former marsh

• Very fast deposition
• Offset beds overlain
• by continuous beds
• Earthquake occurred between depositing bottom and
  top beds
• Dated by finding formerly living matter (plant
  leaves, etc.) and using Carbon-14

http//piru.alexandria.ucsb.edu/collections/atwate
r/saf/slide25.jpg
150
Combined result
151
How big?

• Amount of slip at a single site
• Bigger the slip, the bigger the earthquake
• Appearance at several sites
• The longer the rupture length, the bigger the
  earthquake

152
Trenching sites
153
Combining Sites
Grey bands are correlated events
Fumal et al., BSSA, 2002
154
Important questions for paleoseismologists

• Is a fault active? (Moved in lt10,000 years)
• How often do we get big earthquakes?
• How regularly spaced in time are they?
• Does a given segment always rupture all together?

155
Big Onehistoryin SoCal
1857-type segment
Keller, 8-23
156
From this history

• 10 events in 1300 years
• An event every 130 years, on average
• Last event 140 years ago
• We are overdue!
• But events are not regularly timed
• So better guess would be
• about 25 chance of this quake in next 30 years
• (thats 30 years / 130 year repeat time)

157
30-yr probability of quakes in California
Parkfield
1906 repeat
1857 repeat
Yanev p. 39
1857-type is given 30 chance in 30 years
158
Time-dependent forecastingAftershocks

• San Simeon statement of aftershock probabilities
  released 12 minutes after the earthquake and then
  again approximately 24 hours later
• Using Omoris Law and Gutenberg-Richter, there
  was a 90 chance of a magnitude 5 or greater
  quake in the first 7 days.
• The largest event to occur in that period was a
  4.9.
• USGS also forecasted approximately 120 to 200
  magnitude 3 to 5 quakes in the first week.
• Detected quakes in this range were 139.

159
(UCLA)
http//moho.ess.ucla.edu/helmstet/forecast.html
160
Time-dependent forecastingEarthquake
interactions

• Longer-range
• Predict continuation of a sequence
• Calculate how the stress from one earthquake
  changes probability of another

161
Some problems and complications

• Are magnitude 8 quakes possible on all faults, or
  just San Andreas?
• Do segments always break one at a time, or
  sometimes together? (same question, also a
  restatement of characteristic quake idea)
• Is seismicity uniform over time?
• How many faults dont we see?
• Effect of strong shaking on soil

162
Takes big quakes to test predictions

• Several natural biases
• Insurance companies like high rates
• Cities like perception of low risk
• Scientists like to make changes to status quo
• Real process is that everybody makes a guess,
  which is either verified or contradicted by real
  quakes
• Often decades later
• Is the reliable prediction of individual
  earthquakes a realistic scientific goal?
  http//helix.nature.com/debates/earthquake/equake_
  25.html

163
Product of a forecast (Output)

• When and where the earthquakes will be
• How big
• Magnitude
• How much shaking at a given spot

164
Predicted shaking from Hayward fault event
165
Combine earthquakes
From all known faults
and some model of unknown faults
166
(No Transcript)
167
A probability map for SoCal
Number of times that 20 g will be exceeded per
century
SCEC web page
168
World Hazard
169
(No Transcript)
170
Philippines, 1990
171
Europe,Middle East,and Africa
Turkey, 1999
172
Americas
Mexico City, 1985
173
Hazard and Risk

• Hazard probability that a given area will be
  affected by a given destructive process
• Risk Probability that a loss will occur

174
Hazard and Risk, continued

• Hazard is what seismologists predict
• Includes earthquake probability
• Risk is what insurance companies, the government,
  etc. need to know.
• How do we close the gap?
• Risk hazard vulnerability value

175
HAZUS Average Annualized EQ Losses

• Dr. Stuart Nishenko
• Senior Seismologist
• Building Sciences
• and Assessment
• Branch, FEMA
• WSSPC Conference
• Seattle, WA
• September 20, 2000

FEMA Federal Emergency Management Agency
176
FEMA Hazus results

• Average Annual Earthquake Loss by state

Nishenko, 2002
177
Average Annual Earthquake Loss per Capita for 35
Metropolitan Areas
178
Cost-Benefit Analysis

• Benefit-cost ratio
• Calculate annual benefits
• Multiply by lifetime
• Calculate projected cost of
• special earthquake construction
• Take ratio to get benefit/cost ratio
• Would it be better to spend this money on new
  schools, hospitals, etc.

179
Starquakes in neutron stars

• Flashes of soft g-rays repeaters
• Source of starquake fracture in the neutron star
  crust (1km thick)
• strain energy 1039 Joules gtgt 1019 Joules for
  largest earthquakes
• Star SGR1806-20 sun mass in 20km diameter,
  density 1014, rotation in 7.5s, magnetic field
  1015 Gauss.
• Crust made of solid lattice of heavy nuclei with
  free electron very heavy metal
• Stress loading magnetic forces

http//solomon.as.utexas.edu/duncan/magnetar.html
180
The Earths magnetic field, which deflects compass
needles
(measured at the N magnetic pole)
0.6 Gauss A
common iron magnet like those used to stick
papers on a refrigerator 100 Gauss
The strongest man-made fields achieved
so far in the laboratory
Sustained
(steady) 4 105 Gauss
The strongest man-made fields achieved so far
in the laboratory
Ephemeral (made using
explosives lasts only milliseconds)
107 Gauss The maximum field
observed on ordinary stars 106 Gauss
Typical magnetic field of radio pulsars
1012
Gauss (the ordinary, familiar kind of neutron
star (hundreds are known to astronomers)


Magnetars (soft gamma repeaters) 1014 -
1015 Gauss
181
(No Transcript)
182
(No Transcript)
183
(No Transcript)
184
Case agt b application to starquakes
Comparison of numerical simulations and a
starquake sequence
(Sornette and Helmstetter, 2002)