Where Do Large Earthquakes Occur in the Eastern United States? or Reflections on the

Preview:

DESCRIPTION

Where Do Large Earthquakes Occur in the Eastern United States? or Reflections on the Holy Grail of Earthquake Prediction. Alan Kafka, November 2009. “… she gets her premonitions from something in the air…”. - PowerPoint PPT Presentation

Citation preview

Where Do Large Earthquakes Occur in

the Eastern United States?

or

Reflections on the Holy Grail of Earthquake Prediction

Alan Kafka, November 200909:06 AM 1

2

“… she gets her premonitions from something in the air…”

“Since my first attachment to seismology, I have had a horror of predictions and of predictors. Journalists and the general public rush to any suggestion of earthquake prediction like hogs toward a full trough.”

- Charles Richter (1977)

09:06 AM 3

My current opinion on earthquake prediction:

Given the current state of knowledge in seismology, most (probably all?) specific predictions that are publicly announced are misleading and are often scientifically irresponsible.

However, earthquake forecasting (at some level of detail) is quite realistic, and can be scientifically responsible.

09:06 AM 4

09:06 AM 5

Earthquake Predictionvs.

Earthquake Forecasting

Earthquake Prediction = a short-term (hours to days) statement that an earthquake will occur at a given location.

Earthquake Forecast = a long term (years to decades) statement of the probability of an earthquake in a region (or the probability of one or more earthquakes in a region).

Where Do Large Earthquakes Occur

in the Eastern United States?

1700-2008 (Magnitude 5.0-7.4)09:06 AM 6

Earthquakes

Plate Tectonics

Earthquakes correlated with:

* Mid-Ocean Ridges

* Fracture Zones

* Ocean Trenches

1980-2001 (NEIC, M5)

09:06 AM 7

09:06 AM 8

Eastern United States Northeastern United States

Weston Observatory1568 – November 2009

USGS1700-2008, Magnitude 5.0-7.4

09:06 AM 9

1975-2004(M3)

New EnglandSeismic Network

Weston ObservatoryBoston College

Magnitude 4.2 earthquake (10/03/06) in Bar Harbor, ME generated ground shaking that caused rockfalls in Acadia National Park. From Ebel et al. (2007)

09:06 AM 10

* MLg 6.2, M 5.9

Cape Ann, MAEarthquake

November 18, 1755

Magnitude 5.9-6.2*

From: Ebel (2006)

VIVII

V

IV

VII

09:06 AM 11

Woodcut illustration showing damage to Boston, MA, during the Cape Ann earthquake of November 18, 1755. Shaking was strongest northeast of Boston. The earthquake caused considerable damage in Boston, knocking down or damaging as many as 1,600 chimneys and collapsing brick walls of several buildings.

Image from National Information Service for Earthquake Engineering, University of California, Berkeley.

12

Magnitude has been estimated to be as high as 8.1 and as low as 7.2

From Hough et al. (2000) and Hough (2002)

1811 - New Madrid, MO

09:06 AM 13

Earthquakes

Plate Tectonics

Earthquakes correlated with:

* Mid-Ocean Ridges

* Fracture Zones

* Ocean Trenches

1980-2001 (NEIC, M5)

09:06 AM 14

Dear Professor Kafka:

I work upstairs, in the Fine Arts Dept. My husband and I recently purchased a home in the western part of Framingham, just south of the intersection of the Mass. Pike and route 9. We have always carried earthquake insurance in the past, but our new house is masonry, and the cost of earthquake coverage is high - close to $400/year! Of course, because it's masonry, earthquake damage could be catastrophic.

We are trying to make an informed decision. I went down to the third floor and looked at the Geological Map of Massachusetts; I saw plenty of fault lines in our area (as well as everywhere!), but was not sure of how to interpret them. I noticed your name on a couple of reports next to the map. While I do not want to put you in the position of making a recommendation to us, I was wondering if you had a moment to share your thoughts regarding the likelihood of potentially destructive seismic activity in western Framingham.

Thanks so much!09:06 AM 15

1906 San Francisco Earthquake

09:06 AM 16

Using land surveys conducted several years apart, H.F. Reid discovered that during the 50 years prior to the 1906 earthquake the land at distant points on both sides of the San Andreas fault showed relative displacement of slightly more than 3 meters.

09:06 AM 17

Based on this information, H.F. Reid developed the:

“Elastic Rebound” Theory of

Earthquakes

09:06 AM 18

Satellite Geodesy

09:06 AM 19

Space-Based Geodesy

Using satellite observations: Can measure relative positions of points on the Earth’s surface with a precision of a few mm.

20

Earthquake Probabilities ???

09:06 AM 21

1980 (M≥1.5)

1906

1857

www.data.scec.org

Earthquakes and Faults in Southern

California

M 6.501/01/88 – 11/25/09

09:06 AM 22

01/01/88 – 11/25/09

Large Earthquakes in California Since 1988

M 6.5

09:06 AM 23

The colors on this California map represent the UCERF probabilities of having a nearby earthquake rupture (within 3 or 4 miles) of magnitude 6.7 or larger in the next 30 years. As shown in the table, the chance of having such an event somewhere in California exceeds 99%. The 30-year probability of an even more powerful quake of magnitude 7.5 or larger is about 46%.

Forecast for California Earthquakes (2008)

Uniform California Earthquake Rupture Forecast (UCERF)—the first comprehensive framework for comparing earthquake likelihoods throughout all of California.

You must unlearn what you have learned.09:06 AM 25

Any claim that Eastern United States earthquakes are correlated with some geological or geophysical feature has to (as a minimum) be based on a correlation of earthquakes with that feature.

Correlated in the Strong Sense: Correlation based on a formal statistical analysis. High (95%?) level of confidence that r>0.

Correlated in the Weak Sense: It “sort of looks like” there is an association between earthquakes and the feature in question.

09:06 AM 26

Sykes (1978)

Extensions of Fracture Zones Hypothesis

09:06 AM 27

Sykes (1978)

Extensions of Fracture Zones Hypothesis

09:06 AM 28

NEIC (1973-2008, M≥4.5)09:06 AM 29

From:

“NATURAL DISASTERS”

by Patrick Abbott (2002)

09:06 AM 30

Ramapo Fault

Aggarwal and Sykes (1978)09:06 AM 31

Kafka, Winslow, and Barstow (1989)

USGS (1924-2001, M≥3.0)

NEIC (01/01/75-11/26/09, M≥3)09:06 AM 32

"The keys to all this local rumbling are the Ramapo fault in North Jersey and the so-called 125th Street fault across Manhattan."

From:

N.J. IS NOT IMMUNE TO QUAKES

The Bergen Record

March 2, 2001

09:06 AM 33

09:06 AM 34

09:06 AM 35

NY Nuclear Plant Likely a Quake Risk: Study

“The Indian Point nuclear reactor is within a mile or two of both a seismic zone running from Stamford, Connecticut, to Peekskill, New York, and the Ramapo seismic line, said the study by seismologists at Columbia University's Lamont-Doherty Earth Observatory.”

~ Reuters, New York, Aug 22, 2008

Earthquakes May Endanger New York More Than Thought, Says Study: Indian Point Nuclear Power Plant Seen as Particular Risk

“One major previously known feature, the Ramapo Seismic Zone, runs from eastern Pennsylvania to the mid-Hudson Valley, passing within a mile or two northwest of Indian Point.”

~ The Earth Institute at Columbia University, Aug 21, 2008

09:06 AM 36

Entire Catalog1677-2006

M ≥ 1.61974-2006

M ≥ 2.11974-2006

M ≥ 3.01928-2006

M ≥ 3.51840-2006

M ≥ 5.01737-2006

Seismicity of Greater New York City Area

37

Ramapo Seismic Zone (RSZ) ?

Entire Catalog1677-2006

M ≥ 1.61974-2006

M ≥ 2.11974-2006

M ≥ 3.01928-2006

M ≥ 3.51840-2006

M ≥ 5.01737-2006

09:06 AM 38

“Any claim that Eastern United States earthquakes are correlated with some geological or geophysical feature has to (as a minimum) be based on a correlation of earthquakes with that feature.”

Entire Catalog1677-2006

M ≥ 1.61974-2006

M ≥ 2.11974-2006

M ≥ 3.01928-2006

M ≥ 3.51840-2006

M ≥ 5.01737-2006

09:06 AM 39

USGS

09:06 AM 40

USGS National Seismic Hazard MapsPast Seismicity Future Earthquakes

Petersen, et al. (2008)10% Probability, 50 years

09:06 AM 41

Johnston and Kanter (1990)

M 7

“It appears that on a global scale, Mesozoic rifts and continental margins are of premier importance as crustal features that together account for nearly three-fourths of all SCI seismicity.”

- Johnston (1989)

09:06 AM 42

Johnston and Kanter (1990)

M 7

“Our results suggest that on a global scale, the correlation of seismicity within SCRs and ancient rifts has been overestimated in the past.”

- Schulte and Mooney (2005)

09:06 AM 43

“Pre-existing Zones of Weakness”09:06 AM 44

World Stress Map Project

09:06 AM 45

Cross-section of Atlantic Ocean

09:06 AM 46

Magnitude has been estimated to be as high as 8.1 and as low as 7.2

From Hough et al. (2000) and Hough (2002)

1811 - New Madrid, MO

09:06 AM 47

From USGS

09:06 AM 48

From Braile and others (1982)

Ancient, buried rift currently reactivated by ~east-west compressional stress generated by plate-tectonic processes.

09:06 AM

09:06 AM 50

2008 National Seismic Hazard Map

09:06 AM 51

USGS

09:06 AM 52

USGS

09:06 AM 53

Questions:

(1) Is the “tendency for future earthquakes to occur near past earthquakes” a real, measurable, physical phenomenon?

(2) Do we have samples that are representative of this phenomenon?

(3) Can we measure it?

“If you can't measure it, it isn't science.” - Lord Kelvin

USGS National Seismic Hazard MapsPast Seismicity Future Earthquakes

Petersen, et al. (2008)10% Probability, 50 years

09:06 AM 54

09:06 AM 55

Choose a radius such that circles fill P percentage of map area.

= 6/8 = 75% = sample of binomial random variable, .

= Probability(“success”)

success = red circle occurs within one of the green circles.

^

From Kafka (2002, 2007)

past(before)

future(after)

“Cellular Seismology” (analogous to a cellular phone system)

09:06 AM 56

78% Hits 79% Hits

Northeastern United States

M ≥ 3.0 (1984-1987)M ≥ 5.0 (1988-2001)

Southern California

M ≥ 2.0 (1975-1987)M ≥ 4.0 (1988-2001)

33% Map Area

Central and Eastern United States

M ≥ 3.0 (1924-1987)M ≥ 4.5 (1988-2003)

90% Hits

Kafka (2002, 2007)09:06 AM 57

74%Hits

33%Area

Regions Studied

Results(1988-2001)

NEUS = Northeastern USSEUS = Southeastern USNM = New MadridCEUS = Central and Eastern USSCA = Southern CaliforniaNCA = Northern CaliforniaPNW = Pacific NorthwestISR = IsraelTKY = Turkey

From Kafka (2002, 2007)09:06 AM 58

From Kafka (2007)

INAP

RIDGE

CONTINENT

%Hits for 33% Map Area

SUBDUCTION80%

39%

98%

100%

INTERIOR OF NORTH AMERICAN PLATE (INAP): 80%CONTINENT: 39% RIDGE: 98% SUBDUCTION: 100%

09:06 AM

Central and Eastern United States

M ≥ 3.0 (1924-1987)M ≥ 4.0 (1988-2003)

-65-105

25

50

Future large earthquakes in the CEUS have about 86% probability of occurring within 36 km of past earthquakes.

- Kafka (2007)

green zones = 33% map area

95% Confidence Interval

79% 86% 93%

09:06 AM 60

=ˆ ρ ±1.96ˆ ρ (1− ˆ ρ )

n

= sample of binomial random variable, .

= Probability(“success”)

success = red circle occurs within one of the green circles.

^

From Kafka (2002, 2007)

past(before)

future(after)

“Cellular Seismology”

95% Confidence Interval(given n “after” earthquakes)

09:06 AM 61

Central and Eastern United States

green zones = 33% map area

95% Confidence Interval(from previous slide)

79% 86% 93%

M ≥ 3.0 (1924-1987)

M ≥ 4.0 (2004-2008)

09:06 AM 62

Central and Eastern United StatesM ≥ 3.0 (1924-1987)

M ≥ 4.0 (2004-2008)

green zones = 33% map area

95% Confidence Interval(from previous slide)

79% 86% 93%

67% hits

“Market gurus predict stock rebound but won’t rule out extreme move up - or down.”

- USA Today, January 2, 2009

09:06 AM 63

Central and Eastern United States

M ≥ 3.0 (1924-1987)M ≥ 4.0 (1988-2008)

-65-105

25

50

green zones = 33% map area

95% Confidence Interval

74% 81% 88%

Updated Forecast

09:06 AM 64

Before: 1924-1987, M3.0After: 1988-2008, M4.0

Before: 1924-1987, M3.0After: 1988-2008, M4.0

Rates

SmoothedRates

CEUS - RIRates

CEUS - RISmoothed Rates

% map area

% h

its

% h

its

09:06 AM 65

Cellular Seismology (CS)

E07 RELM Forecast

33% Area

37% Area E07 = RELM Forecast of Ebel, Chambers, Kafka and Baglivo (2007), based on rates of activity in cells.

% map area

% h

its

Before: 1932-2004 M4.0

After: 2005-2008 M4.5

CS

E07

96% Hits

92% Hits

09:06 AM 66

E07 - RELM

Rundle et al. (PI)

CEUS - RI

Rates

Rates + Changes in Rates

Rates

Cellular Seismology

Other Methods

% area

% h

its

Before: 1932-2004, M4.0After: 2005-2008, M4.5

After: 2000-2008, M5.0

Before: 1932-1999M3.0

Before: 1984-1987M3.0

Before: 1924-1987, M3.0After: 1988-2008, M4.0

From: Kafka and Ebel (2009), under review.

% area

09:06 AM

Before: 1964-1999, M>5.0After: 2000-2008, M>5.0

33% Area55% Hits

Before: 1924-1987, M>3.0After: 1988-2008, M>4.0

33% Area81% Hits

Tectonic regions analyzed by: Kafka (2002, 2007), Kafka and Ebel (2009), and This Study.

CS for Stable Continental Regions (SCRs) on a global scale

CS for CEUS

Blue: Intraplate (SCR)Red: Plate BoundaryGreen: Continental Collision

Observed %Hits (33% Map Area) for the Different

Tectonic Regions Analyzed

09:06 AM 68

09:06 AM 69

Kafka and Ebel (2007)

09:06 AM 70

NASA WebsiteOctober 2004:

“Earthquake Forecasting Program Has Amazing Success Rate”

“Amazing Success” or “Remarkably Unremarkable”?

Rundle, et al. (2002)09:06 AM 71

Choose a radius such that circles fill P percentage of map area.

= 6/8 = 75% = sample of binomial random variable, .

= Probability(“success”)

success = red circle occurs within one of the green circles.

^

From Kafka (2002, 2007)

past(before)

future(after)

“Cellular Seismology” (analogous to a cellular phone system)

09:06 AM 72

Allowing 11 km “Margin of Error”:37% area

14/15 = 93% hits

Before: 1932-1999, After: 2000-2007

4.9 km Radius:14% area

12/15 = 80% hits

M≥4.2 M≥4.2

Is this difference “statistically significant”?09:06 AM 73

Kafka and Ebel (2007)

09:06 AM 74

Rundle et al., EOS, 88(24), 2007

09:06 AM 75

Rundle et al. 14/15 = 93% hits, for either 14% map area or 44% map area

Earthquake on Boundary of Box

09:06 AM 76

-115-12332

37

-115-12332

37

(a) (b)

Before: 1932-1999, M3.0Before: 1984-1987, M3.0

Cellular Seismology Rundle et al. (PI)

After: 2000-2008, M5.0

Is this difference statistically significant?

19/21 Hits 17/21 Hits

09:06 AM 77

Rates +

Changes in Rates

After: 2000-2008, M5.0

Before: 1932-1999, M3.0

Before: 1984-1987, M3.0

% area

% h

its

Cellular Seismology

Rundle et al. (PI)Is this difference

statistically significant?

17/21

19/21

09:06 AM 78

09:06 AM 79

Boston CollegeForecasts

January 2007

09:06 AM 80

Collaboratory for the Study of Earthquake Predictability

(CSEP) University of Southern California

Establish rigorous procedures for registering and evaluating prediction experiments.

Construct community standards and protocols for comparative testing of predictions.

09:06 AM 81

R08_m41

Cellular Seismology ForecastForecast Issued on August 26, 2007

84% probability that future earthquakes will occur in green zones.

95% Confidence Interval:

73% ≤ 84% ≤ 94%

09:06 AM 82

Before: 1932-1996, M≥4After: 1997-2007, M≥5

R08_m41

Cellular Seismology ForecastAfter: 01/01/2008-11/27/09

84% probability that future earthquakes will occur in green zones.

82% hits.

95% Confidence Interval:

73% ≤ 84% ≤ 94%

09:06 AM 83

Before: 1932-1996, M≥4After: 1997-2007, M≥5

My current opinion on earthquake prediction:

Given the current state of knowledge in seismology, most (probably all?) specific predictions that are publicly announced are misleading and are often scientifically irresponsible.

However, earthquake forecasting (at some level of detail) is quite realistic, and can be scientifically responsible.

09:06 AM 84

85

“… she gets her premonitions from something in the air…”

09:06 AM 86

09:06 AM 87

Recommended