Turkey Earthquake Risk Model

Financing the Risks of Natural Disasters

World Bank

Washington, DC , June 2-3, 2003

Dennis E. Kuzak

Senior Vice President, EQECAT, Inc

Overall Model Components

Portfolio of Insured TCIP ExposuresPortfolio of Insured TCIP ExposuresTurkish Census and building stock data, plus TCIP information

Turkish Census and building stock data, plus TCIP information

Seismotectonic Model - fault/source locations, magnitudes, frequenciesSeismotectonic Model - fault/source locations, magnitudes, frequencies

Professor Erdik, USGS Research Papers, historical catalogsProfessor Erdik, USGS Research Papers, historical catalogs

Seismic Hazard - Ground ShakingSeismic Hazard - Ground Shaking Attenuation Functions,Local Soil ConditionsAttenuation Functions,Local Soil Conditions

VulnerabilityVulnerabilityHistorical Building Performance, Engineering Analysis, Turkish construction practices

Historical Building Performance, Engineering Analysis, Turkish construction practices

Damage and Financial LossDamage and Financial Loss Policy Deductibles, LimitsPolicy Deductibles, Limits

Seismic Hazard Modeling Methodology

Source zonation model (location) Recurrence model (size and frequency) Ground motion model (ground shaking) Hazard curve (ground shaking vs.

probability) Soil model (ground shaking amplification)

The location, size, and frequency of rare catastrophic earthquakes and their associated ground shaking are modeled using a Probabilistic Seismic Hazard Analysis (PSHA), which is composed of the following elements:

Tectonic Setting of Turkey

White

Quaternary Faults of Turkey

Legend: M 5.0-5.9 (purple); M 6.0-6.9 (blue); M 7.0+ (red)

Earthquakes in Turkey, 1904-1992

Earthquake Sources of Turkey

White

Source: Erdik et al. (1999), “Assessment of earthquake hazard in Turkeyand neighboring regions,” Annali di Geofisica, Vol. 42, pp. 1125-1138.

Seismic Hazard Map of Turkey

PGA (g) with a 10% Probability of Exceedance in 50 Years(475-year Return Period) on soft rock

White

Istanbul

Ankara

Black Sea

Mediterranean Sea

Source: Erdik et al. (1999), “Assessment of earthquake hazard in Turkeyand neighboring regions,” Annali di Geofisica, Vol. 42, pp. 1125-1138.

EQECAT Source Zonation Model

Basic area-source model from Erdik et al. (1999)

Large characteristic earthquakes (M 6.5): North Anatolian Fault (NAF) Northern NAF (NNAF; Marmara Sea) Southern NAF (SNAF; south of Marmara Sea)

Distributed faulting on NNAF and SNAF Smaller earthquakes included as NAF,

NNAF and SNAF background seismicity zones

Marmara Sea Region

NAF

SNAF

NNAF

North Anatolian Fault System (NAF)

Potentially the largest contributor to seismic hazard in Istanbul and northwestern Turkey

Ruptures in a sequence of large earthquakes Previously known rupture sequences

occurred in 967-1050 and 1254-1784 Latest sequence began in 1939 and is still in

progress Latest events in current sequence are the

1999M 7.4 Izmit and M 7.2 Duzce earthquakes

Next event in sequence is likely to be in Marmara Sea (Yalova and/or Prince’s Islands Faults?)

Rupture History of the NAF to 1992

Note: Mean repeat time is now estimated to be 225 years east of NNAF/SNAF split and 280 years west of this split (Stein, personal comm., 2000; Parsons et al., 2000)

EQECAT Recurrence Model

Area and background sources Fault sources:

North Anatolian Fault (NAF) Northern NAF (NNAF; Marmara Sea) Southern NAF (SNAF; south of Marmara

Sea)

Recurrence: Area and Background Sources

Earthquake catalog from regional sources: All magnitudes converted to moment

magnitude Foreshocks, aftershocks and swarms

removed Corrected for incompleteness

Exponential (Gutenberg-Richter) recurrence law

Maximum magnitude from seismicity/tectonics

Recurrence: Fault Sources

Characteristic earthquake recurrence law Recurrence rates from previous historic

events Magnitudes on NAF from last historic

event Magnitudes on NNAF from length/historic

events Magnitudes on SNAF from largest historic

event Time-dependent probability on NAF and

NNAF

EQECAT Time-Dependent Model

Definition of Time-Dependent Probability

Time-dependent probability is the conditional probability that an earthquake will occur on a fault within some specified period of time in the future (i.e., 2000 to 2030), given that a similar earthquake has occurred on this fault at some known time in the past

Causes of Time-Dependent Behavior

Secular stress increase due to long-term strain accumulation

Permanent stress increase or decrease due to stress interaction from an earthquake on a nearby fault

Transient stress increase or decrease due to stress interaction from an earthquake on a nearby fault (decays with time)

Time-Dependence from Secular Stress

Lognormal Probability Model

Last Earthquake

PARAMETERS

Te = Elapsed Time

T = Exposure Period

Conditional Probability = A/(A + B)

A

B

Note: We use recurrence intervals of 225 years east of NNAF/SNAF split and 280 years west of split; exposure period of 1 year; elapsed time from last historic event

Time-Dependence from Secular Strain

Sensitivity to Total Variability

Note: We use = 0.5

Time-Dependence from Fault Interaction

Transient Stress for NAF

Source: Parsons et al. (2000)

Time-Dependenc

e from Fault

Interaction

Progressive Effect of

1939-1992 Earthquake Sequence

Source: Stein et al. (1997)

Time-Dependenc

e from Fault

Interaction

Effect of 1999 Izmit Earthquake

Source: Parsons et al. (2000)

Time-Dependence from All Effects

Faults within 50 km of Istanbul

Source: Parsons et al. (2000); Our results give 50% smaller probability

Time-Dependence from All Effects

Istanbul

30-year (%) 10-year (%) 1-year (%)Fault

Interaction Secular Interaction Secular Interaction Secular

Yalova 33 21 22 18 14 11 7 7 1.7 1.7 0.8 0.8

Prince’s Is. 35 15 26 12 16 9 10 6 2.1 1.6 1.1 0.7

Marmara 13 9 11 8 5 5 4 4 0.6 0.7 0.5 1.0

Combined 62 15 49 15 32 12 20 9 4.4 2.4 2.3 1.5

Earthquake Probabilities of M 7 for Faultswithin 50 km of Istanbul Beginning in May 2000

(Interaction includes Secular and Interaction Stress Effects)

Note: 30-year time-averaged (Poisson) probability is 20 10%

Source: Parsons et al. (2000); Our results give 50% smaller 1-year combined probabilities

EQECAT Ground Motion Model

Four attenuation relationships for shallow crustal earthquakes: Abrahamson and Silva (1997) Boore et al. (1997) Campbell (1997) Sadigh et al. (1997)

One attenuation relationship for subduction earthquakes: Youngs et al. (1997)

1999 Izmit earthquake validation

PGA Estimate for 1999 Izmit Event

Note: PGA in g; Values are consistent with limited strong-motion recordings

EQECAT Soil Model

NEHRP soil classifications (A-E) were derived from a 1:500,000-scale geologic map of Turkey

Correlation with geology was based on extensive studies relating shear-wave velocity and geologic units in California

PGA amplification factors were based on empirical and theoretical studies relating shear-wave velocity and amplification

Turkish Soil Map

EQECAT Hazard Results

Seismic hazard curves: Istanbul (Bakirkoy) Izmit Ankara

Seismic hazard maps: All of Turkey Marmara Sea Region

Seismic Hazard Curves

Lowest hazard in Ankara, located in more stable central region of Turkey

Hazard in Izmit is approximately the same as that in Istanbul (Bakirkoy) because:

Time-dependent probability effects have increased hazard in Istanbul

Time-dependent probability effects have decreased hazard in Izmit0.01 0.1 1

Peak Ground Acceleration (g)

10

100

1000R

etu

rn P

eri

od

(ye

ars

)

AnkaraIstanbulIzmit

Soft Rock

Seismic Hazard Map for Turkey

0.05 - 0.100.10 - 0.150.15 - 0.200.20 - 0.250.25 - 0.300.30 - 0.350.35 - 0.400.40 - 0.450.45 - 0.500.50 - 0.550.55 - 0.600.60 - 0.650.65 - 0.700.70 - 0.750.75 - 0.800.80 - 0.850.85 - 0.90

PGA (g)

PGA with 10% Probability of Exceedance in 50 Years(475-year Return Period)

Soft Rock

Seismic Hazard Map for Marmara Sea

PGA with 10% Probability of Exceedance in 50 Years(475-year Return Period)

0.05 - 0.100.10 - 0.150.15 - 0.200.20 - 0.250.25 - 0.300.30 - 0.350.35 - 0.400.40 - 0.450.45 - 0.500.50 - 0.550.55 - 0.600.60 - 0.650.65 - 0.700.70 - 0.750.75 - 0.800.80 - 0.850.85 - 0.90

PGA (g)Soft Rock

Building Vulnerability

Good building codes- poor construction practices Non ductile concrete frames, soft story Limited/No claims data for residential multi-family EQECAT approach- use aggregate residential

damage data from Izmit, select damage curve consistent with Izmit data.

Summary

Turkish Model incorporates current research on time-dependent studies of North Anatolian Fault Raises hazard in Istanbul Lowers hazard east of Izmit epicenter

Recognizes uncertainty in NAF location in the Marmara Sea

Incorporates poor building practices into vulnerability assumptions

Fully probabilistic damage and loss model consistent with other EQECAT earthquake models for the US, Japan, Taiwan, etc.

TCIP RISK ANALYSIS RESULTS

Portfolio exposures estimated as of 31-10-2002

Estimated Annual Damage = .138% (1.38 per mill)

Exceedance Frequency

Return Period (Years)

Damage as a % Replacement

Value10.0% 10 0.18 9.0% 11 0.21 8.0% 13 0.25 7.0% 14 0.30 6.0% 17 0.36 5.0% 20 0.46 4.0% 25 0.63 3.0% 33 0.92 2.0% 50 1.38 1.0% 100 2.28 0.9% 111 2.43 0.8% 125 2.62 0.7% 143 2.89 0.6% 167 3.04 0.5% 200 3.38 0.4% 250 3.87 0.3% 333 4.61 0.2% 500 5.61 0.1% 1000 7.26