EARTHQUAKE RISK ASSESSMENT FOR
ISTANBUL METROPOLITAN AREA
EXECUTIVE SUMMARY
DEPARTMENT OF EARTHQUAKE ENGINEERING
Boğaziçi University
Kandilli Observatory and Earthquake Research Institute
Istanbul, May 2002
1
EARTHQUAKE RISK ASSESSMENT
FOR ISTANBUL METROPOLITAN AREA
EXECUTIVE SUMMARY
Background
Using funds entrusted to the American Red Cross (ARC) by the American public for
assistance to Turkey, ARC together with Turkish Red Crescent Society (KIZILAY) sponsored
the Department of Earthquake Engineering to develop a sub-district level earthquake risk
assessment for the Istanbul metropolitan area, under protocol agreement with Bogazici
University Foundation. The 400-page long full report will be placed on the Internet with all
figures and appendices.
The project is focused on the following objectives:
1. Development of an Istanbul metropolitan area risk model.
2. Development of estimates for human and social impacts as a result of such a scenario
earthquake, including: Building damage, Casualties and People left homeless. Damage to
infrastructure and lifelines are only dealt in general terms based on the intensity levels.
3. Public provision of the results via the Internet to enable disaster response planning.
Department of Earthquake Engineering of the Kandilli Observatory and Earthquake
Reasearch Institute of Bogazici University in Istanbul, Turkey is the first and only academic
unit in Turkey that provides graduate level training on Earthquake Engineering leading to
M.S. and Ph.D. degrees on “Earthquake Engineering”. The department also conducts research
and implementation on earthquake engineering that contributes to seismically safer structures,
systems and environment. The emphasis of activities are placed on: Earthquake hazard and
risk analysis; Development of urban earthquake damage scenarios; Characteristics of strong
earthquake ground motion; Site and soil response analysis; Earthquake response of buildings,
industrial facilities, bridges and dams; Soil-structure interaction; Dynamic testing of small-
scale model and prototype structures; Retrofitting and post earthquake strengthening of
structures; Damage evaluation and earthquake insurance; and the Development of earthquake
resistant design codes. The faculty and graduate students of the department have conducted
their first study on the urban earthquake risk in Istanbul for a World Bank commissioned
study in 1994. They are involved in a NATO advanced study institute that resulted on the
publication entitled “Issues in Urban Earthquake Risk, Kluwer, 1994”. The preparation of the
Earthquake Loss Assessment and Earthquake Master Plan for the City of Izmir has also been
prepared mainly by the staff of the department in connection with the RADIUS project of the
UN-IDNDR (International Decade for Natural Disaster Reduction) in 1999.
The B.U.-Department of Earthquake Engineering acknowledges and appreciates the
contributions of the following individuals who participated in the development of this report:
Prof. Dr. Nuray Aydınoğlu, Prof.Dr. Aykut Barka, Prof. Dr. Mustafa Erdik (Project Director),
Prof.Dr. Özal Yüzügüllü, Assoc. Prof. Dr. Bilge Siyahi, Assoc. Prof. Dr. Eser Durukal,
Assist. Prof. Dr. Yasin Fahjan, Res.Asst. Hakan Akman, Gülüm Birgören, Yeşim Alpay-Biro,
Mine B. Demircioğlu, Res.Asst. Cem Özbey and Res.Asst. Karin Şeşetyan. The report was
also reviewed by: Prof.Dr. Atilla Ansal.
The B.U.-Department of Earthquake Engineering would like to acknowledge the generous
support of the following organizations and individuals: American Red Cross, Turkish Red
Crescent, Boğaziçi University, Boğaziçi University Foundation, State Statistics Institute (SSI)
2
Istanbul Governorate–Disaster Management Center-E. Akol), Istanbul Metropolitan
Municipality (Disaster Coordination Center-M. Pektaş, Directorate of Soil and Earthquake
Investigations-M. Baş, IGDAS, ISKI), TEDAS, Turk Telekom, KGM – Directorate of 17th
Division-Dr. N. Apaydın, USGS-Golden Office-Dr. M. Petersen and Pasadena Office-Dr. E.
Şafak, RMS-Dr. F. Bendimerad, Munich-Re-Dr. A. Smolka, Zetas Inc.-Prof. T. Durgunoğlu
and Basar Inc.-A. Küçükpehlivan.
Introduction
Following the losses suffered during the two major earthquakes that struck Turkey in 1999,
there has been a broad recognition among Turkey's governmental, non-governmental and
academic organizations of the need for extensive response planning based on detailed risk
analyses of likely seismic hazards in Turkey general and, Istanbul, in particular.
In recent decades earthquake disaster risks in urban centers of Turkey have increased mainly
due to very high rate of urbanization, faulty land-use planning and construction, inadequate
infrastructure and services, and environmental degradation. The other important source of the
increased risk in Istanbul is the unprecedented increase of the probability of occurrence of a
large earthquake (which stands at about 65% during the coming 30 years). The inevitability of
the occurrence of such a large earthquake in Istanbul makes it imperative that certain
preparedness and emergency procedures be contrived in the event of and prior to an
earthquake disaster, which in turn requires the quantification of effects of the earthquake on
physical and social environment.
The specific study reported herein is essentially intended for the quantification of building
losses, which is directly related to casualties, planning of emergency response, first aid and
emergency shelter needs. The first ingredient of this loss scenario is the assessment of the
earthquake hazard, quantified in terms of spatial distribution of site-specific intensities or
spectral accelerations. The vulnerabilities and the damage statistics of buildings, lives and
other facilities constitute the second ingredient. Urban earthquake loss scenarios are based on
the intelligent consideration and combination of the hazard and these vulnerabilities.
Earthquake Hazard
For Istanbul a scenario earthquake is determined to take place on the Main Marmara Fault.
Compilation and interpretation of propagation path characteristics, topographical, geological
and geotechnical data, and the identification of the proper attenuation and site response
analysis models constitute other important ingredients of earthquake hazard assessment.
The selected attenuation relationships provide earthquake intensities, peak ground
acceleration, velocity and displacement and, spectral acceleration, at specific frequencies and
damping ratios, for given earthquake magnitude, distance, fault mechanism and local geology.
For the site-specific modifications of ground motion intensity change degrees empirically
correlated with the geological ground conditions and average horizontal spectral amplification
factors are used.
Figure 1 provides a map of earthquake intensities that would result from the scenario
earthquake. Using the damage definitions of 1998 European Macroseismic Scale (EMS-1998)
a general understanding of damage under exposure to these intensity levels can be gained. For
the vulnerability class where the general reinforced concrete multi-story building stock in
Istanbul is located, EMS-1998 provides the following damage definitions.
Intensity VII: A few buildings sustain moderate damage
3
Intensity VIII: Many buildings suffer moderate damage; a few substantial to heavy damage
Intensity IX: Many buildings suffer substantial to heavy damage; a few very heavy damage
Where “Few” describes less than 20% and “Many” describes between 20% and 60%.
For computational purposes, earthquake hazard data are aggregated in 0.005 x 0.005 degree
geo-cells. The same geo-cells are also used for aggregation of the geotechnical data and the
building inventories. In addition to the vibratory ground motion, the urban seismic hazard
assessment study encompassed herein incorporates soil failures and terrain movements, as
appropriate.
Vulnerabilities
Vulnerability is defined as the degree of loss to a given element at risk, or a set of such
elements, resulting from the occurrence of a hazard. Vulnerability functions (or Fragility
curves) of an element at risk represent the probability that its response to earthquake
excitation exceeds its various performance limit states based on physical and socio-economic
considerations. The vulnerabilities of lives, structures, systems, and the socioeconomic
structure are the main factors influencing the earthquake risk and losses in urban areas.
There are two main approaches for generating vulnerability relationships. The first approach
is based on damage data obtained from field observations after an earthquake or from
experiments. The second approach is based on numerical analysis of the structure, either
through detailed time-history analysis or through simplified methods.
In the earthquake loss scenario developed for Istanbul have used building vulnerability
relationships that express the percentage damage for each typified building group and in
certain damage classes against the EMS-1996 intensity ranges and Spectral Displacements.
The intensity–based vulnerability relationships are empirical in nature and are based on
damage data from local earthquakes. The spectral displacement-based building vulnerabilities
(also called as “fragility curves”) relate the probability of being in, or exceeding, a building
damage state to for a given spectral displacement demand parameter using the methodology
developed in HAZUS (1999).
Death tolls in earthquakes arise mostly from structural collapses and to a lesser degree from
collateral hazards. In this study casualties per damaged building for a given class of buildings
is obtained on the basis of data obtained from local earthquakes.
The vulnerabilities of Turkish building stock are at least an order of magnitude higher than
their counterparts in California. The reasons for this high vulnerability can be traced back to
several reasons. Essentially the building development system was conducive to poor
construction due to high (chronic) rate of inflation (consequently very limited mortgage and
insurance, impediment to large scale development and industrialization of the
construction sector), high rate of urbanization (which created the demand for inexpensive
housing), ineffective control/supervision of design/construction, regulations with limited
enforcement, no accountability and government acting as a free insurer of earthquake risk.
Observations acquired from past urban earthquakes in Turkey, supplemented by the
worldwide experience have been used as a guide to assess the physical vulnerability of
lifelines, which includes the transportation, telecommunication, natural gas, power, water and
waster water distribution systems.
4
Inventory of Elements at Risk
In urban areas, buildings, population, lifeline systems and socio-economic activities constitute
the “elements at risk”. Buildings and lifeline systems are generally termed “Built
Environment”. The physical losses to elements at risk that would result from a specified
earthquake scenario necessitate an extensive and comprehensive collection of their
inventories. The inventories of the built environment will be studied broadly in the following
categories: buildings, transportation systems, telecommunication systems, electrical
distribution systems, water and waste water system, natural gas systems.
Building inventory data is compiled from Istanbul Metropolitan Municipality and State
Statistical Institute complemented by Turk Telekom analog maps, imagery from helicopter
flights and aerial and satellite imagery.
Classification of buildings in Istanbul is essential to ensure a uniform interpretation of data
and results. The building inventory is classified using three basic categories on Structural
Systems, Number of Stories and Year of Construction. Each category is further subdivided
into groups to yield 24 different building classes.
Istanbul Metropolitan Area was divided into grids as 0.005° x 0.005° (approximately 400 m x
600 m) cells for aggregation of hazard and physical inventory data.
The maps of Turkish Telecommunication Association (Turk Telekom) have been an
invaluable source of information for the compilation of data regarding various facility stocks
in Istanbul. All the maps have been visually screened and data have been complied for:
Educational institutions, Hospitals, Polis stations, Fire Stations, Historical buildings,
Religious places, Governmental buildings, Military Compounds, Transformer buildings, Gas
stations, Stadiums and Free fields.
Inventory data concerning lifelines were available through the respective offices and archives
of the Istanbul Metropolitan Municipality.
The day and night time population of 28 districts were determined, and then assigned to the
geo-cells in order to calculate the human losses in Istanbul due to a major earthquake. The
population and building data for Istanbul have been obtained from the State Statistics
Institute. For the determination of day population, Istanbul Transportation Master Plan,
which is prepared by Municipality of Istanbul Metropolitan in 1997, is been utilized.
Lifeline data has been compiled from the concerned offices of the Government and Istanbul
Metropolitan Municipality for Power distribution system, Water Aqueducts and Supply
System, Transportation system, Natural Gas Network and the Telecommunication System.
Methodology and Software Developed
Several methodologies and attendant software (such as, HAZUS, EPEDAT and NHEMATIS)
exist for computation of urban earthquake risk using hazard, inventory and vulnerability
inputs through a GIS engine for the manipulation data and display of results. Owing to the
closed-form nature of these software with built in computational routines we have decided to
develop our own software built on the internationally accepted state-of-the art methodology
used in the study. The methodology will generate an estimate of the consequences "loss
estimate" to a city under probabilistic earthquake hazard or exposure to a "scenario
earthquake", that is, an earthquake with a specified magnitude and location. Although the
5
study is carried out for Istanbul, if it is extended to other cities in Turkey the study will help
guide the allocation of national resources to stimulate risk mitigation efforts and to plan for
national earthquake response.
In simplified form, the steps for using the methodology are:
Select the city area to be studied.
Specify the scenario earthquake with potential fault breaks.
Collect/compile information for delineating local soil conditions.
Collect/compile information of inventory of building classes and other lifelines.
Compute earthquake hazard information in the form of site-specific ground motion.
Using vulnerability data embedded in the software (or externally provided) assess damage to
different classes of buildings and casualties.
These systems, methods, and data have been coded into user-friendly software that operates
through a geographic information system (GIS); MapInfo is the software used. Use of a GIS
makes possible the convenient manipulation of loss estimation data concerning building
stock, population, and economy. The software permits losses and consequences to be
portrayed on spreadsheets and maps. The level of analysis generally corresponds to Level 2
type analysis in HAZUS (1999).
KOERILoss is software developed by the Earthquake Engineering Department of Bogazici
University, Kandilli Observatory and Earthquake Research Institute (KOERI). The software
applies a loss estimation methodology (Probabilistic vs Deterministic) developed by KOERI
to perform analyses for estimating potential losses from earthquakes. KOERILoss Version 1.0
in its current form is capable to perform building damage estimation analysis using both
intensity and spectral displacement based methodology. It is also able to estimate the direct
economic losses and casualties related to building damages. KOERILoss is an user-friendly
software that operates through Geo-cells systems. Geo-cells (Grids) facilitate the
manipulation of data on building stock, population and, earthquake hazards. The software is
developed using the MapBasic language and runs efficiently under MapInfo software.
Therefore, the software is fully integrated with MapInfo and capable to utilize the powerful
features in displaying, querying, manipulating and mapping inventory databases. KOERILoss
provides a great flexibility in displaying the outputs. Tables of building damages, social and
economic losses can be easily mapped and displayed on the screen, printed or pasted into
electronic documents.
In order to perform building damage and loss analysis, the building inventory stock database
should be provided for each geo-cell. The seismic hazard information in terms of: Intensities
for intensity based-analysis and Spectral accelerations for spectral displacement-based
analysis should also be aggregated at the center of geo-cell. To compute the damage
probability ratios intensity based vulnerabilities and/or spectral displacement based fragility
curves for each building class type vulnerabilities should be specified.
Economic losses associated with the general building stock is estimated using the building
damage losses and costs for different structural damages of each building group.
In order to estimate the number and the severity of the casualties, the day and night time
population for each building type in geo-cells should be provided together with the casualty
rates for each building class and damage grade.
6
The software provides the building damage loss, economic loss and the number of casualties
in terms of Geo-cells, which can then be integrated into sub-districts (Mahalle) and districts
(İlçe), as MapInfo Output tables.
It is believed that the software (KOERI-LOSS) developed has the potential of gaining
widespread use not only in Turkey but internationally as well due to its open architecture and
the robust methodology that it incorporates.
Results
The study culminates with maps depicting the distribution of building losses, lifeline facilities
overlaid on intensity maps, distribution of casualties, temporary shelter needs and expected
financial losses due to building damage. The resolution of these maps is the geo-cell (400m
by 600m) scale. All the digital data, some subject to security classification, will be made
available to professional users that can build any information they need at any scale and with
any overlay desired using the GIS software of their choice.
The expected building damages, calculated both by the intensity based and the spectral
displacement based deterministic approaches, are provided in Figure 2 and 3. Results indicate
that, on the basis of two independent approaches (intensity based and spectral displacement
based approaches), a total of about 35,000 to 40,000 buildings (about 5% of the total building
stock) were estimated to be damaged beyond repair (complete damage). Most of casualties are
expected in this damage group, especially in a subset of this group where the collapse will be
of the worst “pancake” form. In pancaked buildings the floors pile up on top of each other
rendering very difficult conditions for search and rescue. Our estimate for pancaked buildings
will be in the vicinity of 5,000 to 6,000.
Furthermore about 70,000 buildings will receive extensive damage and about 200,000
buildings will be moderately damaged. Both of these damage groups are repairable. The total
monetary losses due to building damages caused by the scenario earthquake are estimated to
be in the range of about USD 11 Billion (Figure 4).
Building damages are mostly concentrated in the densely populated districts located in the
southwestern part of the city, such as Eminönü, Fatih, Zeytinburnu, Bakırköy, Bahçelievler,
sourhern part of Küçükçekmece and Avcılar, and to a lesser degree, in districts such as
Kadıköy, Maltepe and Kartal located on the southestern part of Istanbul. Although situated in
relatively farther locations from the causative fault, due to the building density and
vulnerability conditions, districts of Beyoğlu, Eyüp and Bayrampasa are also expected to
undergo high levels of damage. As the result of the analysis conducted for various structure
types, the mid-rise reinforced concrete structures constructed before 1980 are found to
constitute the most vulnerable building class.
The expected casualties were calculated using both the intensity based and the spectral
displacement based approaches. The results for both the nighttime and daytime population are
obtained for the four levels of casualty severities from the spectral displacement based
approach. Thus, on the basis of these two independent approaches it is estimated that the
death would vary between 30000 and 40000. How many of the casualties in serious injury
class would be lost is difficult to estimate since it would depend on the post earthquake
emergency services. If we assumed that about 1/3 would be lost, that would bring the total
estimate of deaths to about 40,000-50,000 level. Death and hospitalized injury distributions
7
based on spectral displacement approaches are presented in Figure 5 and Figure 6
respectively.
As the result of the intensity based analysis a total of about 600,000 households and from the
spectral based analysis a total of about 430,000 households were assessed to be in need of
shelters following the scenario earthquake (Figure 7). This amount constitutes an upper bound
limit for the expected shelter needs, since the conversion of this number to total number of
households is subject to some discussion and the vacancy of some dwelling units should also
be considered.
Limitations of the Study
History has taught that the next major earthquake to affect Istanbul will likely be somewhat
different from the “scenario earthquake” used in this study. Thus, the results the earthquake
loss study for Istanbul should not be interpreted as a prediction but rather as an indication of
what type of losses may take place. Due to this heterogeneity and lack of sufficient data from
past earthquakes, relatively little is known concerning the earthquake vulnerability of
buildings and other structures. To deal with this complexity the loss methodology used in this
study, groups buildings and components of lifelines into categories, based upon key
vulnerability characteristics. Under these uncertainties and lumping process the estimated
losses depend upon the “law of averages”, in other words are applicable to a population of
buildings rather than individuals.
The scenario earthquake that has been selected qualifies for the so-called “Credible Worst
Case Scenario”. As such the losses should be regarded as upper bounds within their
uncertainty. On the other hand in the deterministic hazard assessment we have used the
median (50-percentile) of the attenuation relationships. Thus, statistically there exist a 50 %
chance that the hazard results will be less or greater than the hazard value used in the loss
assessments.
While building related losses serve as a reasonable indicator of relative urban earthquake risk,
it is important to recognize that these estimates are not absolute determinations of the total
risk from earthquakes. The loss parameters used in this study address direct economic losses
to the building inventory. Seismic risk also depends on other parameters, which have not been
comprehensively included in this study, such as consequential physical losses (fire etc.) and
indirect economic losses.
Conclusions
What will be needed, in the context of planning for or mitigation of these indicated losses, is
the dissemination of this urban earthquake loss data in understandable formats to increase the
awareness of the general public, to inform technical personnel in charge of disaster planning
and mitigation and to sensitize the top-level decision makers. We are afraid that, if this
dissemination activity is not properly and adequately conducted, the return of this
comprehensive state-of-the art study will only academic and marginal.
The effectiveness of using and disregarding earthquake risk mitigation techniques - building
rehabilitation, the adoption and enforcement of seismic provisions in building codes, and
prudent community growth – are well known has been demonstrated in Turkish cities and also
throughout the world. The focal element of these mitigation activities is a technically sound
state-of –the-art assessment of the level of losses that needs to be addressed.
8
Only with a comprehension of the consequences of the earthquake losses in Istanbul can
decisions be made at any governmental level about mitigation priorities and optimal
approaches. The results of the study can and should be used for the development of
earthquake hazard mitigation strategies (such as rehabilitation of existing vulnerable
buildings, development of appropriate zoning ordinances, and enforcement of earthquake
building codes) as a countermeasure against earthquake losses indicated in the loss estimation
study and for the planning for contingency measures (such as identification of alternate
transportation routes), emergency response and recovery including, identification of
emergency sheltering means and locations, availability of medical services, food, water and
other essential utilities.
The results of this study strongly support the need for strategies to reduce the current seismic
risk by focusing on rehabilitation of the existing building stock in the most vulnerable
communities of Istanbul to whom the substance and findings of this work are dedicated.
9
Figure 1. Site dependent deterministic intensity distribution.
10
Figure 2. Intensity based distribution of all buildings damaged beyond repair (complete damage).
11
Figure 3. Spectral displacement based complete (beyond repair) damage distribution.
12
Figure 4. Distribution of Direct Financial Losses due to Building Damage
13
Figure 5. Distribution of Deaths for a Night-Time Scenario Earthquake.
14
Figure 6. Distribution of Injuries that Needs to be Hospitalized for a Night-Time Scenario Earthquake.
15
Figure 7. Distribution for Families in Need of Emergency Sheltering.
16