Dissertation2009 Daniell

Istituto Universitario
di Studi Superiori

Universit degli
Studi di Pavia

EUROPEAN SCHOOL FOR ADVANCED STUDIES IN
REDUCTION OF SEISMIC RISK

ROSE SCHOOL

COMPARISON AND PRODUCTION OF OPEN SOURCE
EARTHQUAKE LOSS ASSESSMENT PACKAGES

A Dissertation Submitted in Partial
Fulfilment of the Requirements for the Masters Degree in

EARTHQUAKE ENGINEERING AND ENGINEERING SEISMOLOGY

by

JAMES EDWARD DANIELL

Supervisors: Dr HELEN CROWLEY, Dr RUI PINHO

April, 2009

The dissertation entitled, Comparison and Production of Open Source Earthquake Loss
Assessment Packages, by James Edward Daniell, has been approved in partial fulfilment of
the requirements for the Masters Degree in Earthquake Engineering.

Name of Reviewer 1

Name of Reviewer 2

Abstract
Daniell, April 2009 i

ABSTRACT

The aim of this thesis is to provide a comparative in-depth view of current Earthquake Loss Estimation
(ELE) and other earthquake software packages using an Open source Procedure for Assessment of
Loss using Global Earthquake Modelling software (OPAL-GEM1) with the view of creating a truly
Open source Program for Assessment of Loss for Global Earthquake Modelling (OPAL-GEM2).
The OPAL procedure has been developed to provide a framework for selection of optimal
components:
1) Overview of current and new components of earthquake loss assessment (vulnerability, hazard,
exposure, specific cost and technology) was made, identifying the disadvantages and advantages of
methods used;
2) Preliminary research, acquisition and familiarisation with all available ELE software packages was
carried out;
3) Assessment of these software packages was undertaken in order to identify the ELE methods used;
and
4) Loss analysis was undertaken for a deterministic earthquake (Mw7.2) for the Zeytinburnu district,
Istanbul, Turkey, by production and use of 2 software packages: displacement-based MDBELA
(Matlab-based DBELA); and CSM-based MHAZUS (Matlab-based HAZUS). Also SELENA was
adapted for use in order to gain an understanding of the different processes needed for the production
of damage, economic and social loss estimates. MDBELA was found to be more computationally
expensive. Other mediums and optimisation techniques have also been presented.
Optimisation of the software and ELE components needed for OPAL-GEM2 were identified through a
multi-criteria analysis applied to all ELE software packages using the knowledge gained through the
OPAL-GEM1 process. Future improvements to the step 4 in the OPAL procedure have been
recognised and will be undertaken in future work, including conversion to Octave and Fortran.
OPAL-GEM2 will be a dynamic, open-source, multiple vulnerability level, Fortran-based (Java-
compatible) ELE Software Package integrated with an open source GIS package, and it will be
produced to provide a solution for both forecasting and rapid loss estimation.

Keywords: OPAL, displacement-based, DBELA, earthquake loss estimation, earthquake loss
assessment, GEM, OPAL-GEM, open source, HAZUS
Acknowledgements
Daniell, April 2009 ii

ACKNOWLEDGEMENTS

I would like to extend thanks to many people who have helped and supported me during my Masters
program and during this thesis.
To Maren, it is Pavia that brought us together and I would like to thank you for the support, help,
organisation, love and companionship throughout this phase of my life and also looking to the future.
To my Mum (Anne) and Dad (Trevor), I want to thank you for the love, constant support, and advice
not only during my Masters, but also throughout my life. To my sister, Katherine, and my brother-in-
law, Quentin, it has been wonderful to have family in Europe during this period of my life, away from
the southern hemisphere.
To my supervisors, Rui and Helen, thankyou for the constant advice, problem solving and discussions.
To the wonderful people of the office of the ROSE School, Saverio, Elena, Luisa et al., thanks for all
the good times and help!
I would like to also thank everyone individually from LGIT Grenoble and also ROSE School, Pavia,
and to all the new French, Italian and worldwide friends I have made along the way, but special thanks
go to those who were with me in both Grenoble and Pavia, Tim, Jorge, Manuela, Jagadish, William
and Maria-Angela.
I would also like to thank Tim again as well as Manuela and Rita, who helped in the case study for
bouncing ideas back and forth.
Last but not least, I would like to thank Marens parents, Hans and Marianne, for their help and
support in looking after me for the last 3 weeks while I have been writing up the final parts of this
thesis.

Table of Contents
Daniell, April 2009 iii

TABLE OF CONTENTS

..........................................................................................................................................................Page
ABSTRACT ............................................................................................................................................i
ACKNOWLEDGEMENTS....................................................................................................................ii
TABLE OF CONTENTS...................................................................................................................... iii
LIST OF FIGURES..............................................................................................................................vii
LIST OF TABLES.................................................................................................................................. x
LIST OF SYMBOLS AND ABBREVIATIONS .................................................................................xii
1. INTRODUCTION..................................................................................................................... 1
2. LITERATURE REVIEW EARTHQUAKE LOSS ASSESSMENT AND ESTIMATION... 6
2.1 Introduction................................................................................................................................6
2.2 What constitutes an ELE?..........................................................................................................7
2.3 Exposure ....................................................................................................................................8
2.4 Vulnerability ............................................................................................................................12
2.4.1 Why is it so important to define large-scale vulnerability methods? ...................... 12
2.4.2 Empirical Methods of Vulnerability Assessment.................................................... 13
2.4.3 Analytical and Hybrid Methods of Vulnerability Assessment ................................ 20
2.5 Hazard......................................................................................................................................34
2.5.1 Potential Hazards analysed in Loss Models ............................................................ 34
2.5.2 Methods of Seismic Hazard Assessment................................................................. 36
2.5.3 Modelling of Uncertainty ........................................................................................ 38
2.5.4 Other Ground Motion Issues for ELEs.................................................................... 41
2.6 Damage Loss Conversion, Economic and Social Costs...........................................................42
2.6.1 Social and Economic Vulnerability......................................................................... 43
2.6.2 Social Costs ............................................................................................................. 46
2.6.3 Economic Costs....................................................................................................... 47
Table of Contents
Daniell, April 2009 iv
2.7 Conclusion ...............................................................................................................................49
3. CURRENT ELE SOFTWARE PACKAGES AVAILABLE.................................................. 50
3.1 Open Source vs. Closed Source...............................................................................................50
3.2 Overview of Worldwide Earthquake Loss Estimation Packages.............................................51
3.2.1 CAPRA.................................................................................................................... 52
3.2.2 CATS....................................................................................................................... 53
3.2.3 DBELA.................................................................................................................... 53
3.2.4 ELER/NERIES........................................................................................................ 54
3.2.5 EmerGeo (previously NHEMATIS)........................................................................ 54
3.2.6 EPEDAT.................................................................................................................. 55
3.2.7 EQRM..................................................................................................................... 56
3.2.8 EQSIM..................................................................................................................... 57
3.2.9 Extremum................................................................................................................ 57
3.2.10 HAZ-Taiwan ........................................................................................................... 58
3.2.11 HAZUS.................................................................................................................... 58
3.2.12 InLET ...................................................................................................................... 59
3.2.13 LNECLOSS............................................................................................................. 59
3.2.14 MAEviz ................................................................................................................... 60
3.2.15 OPENRisk ............................................................................................................... 61
3.2.16 OSRE via MIRISK.................................................................................................. 61
3.2.17 PAGER and other Rapid Response engines............................................................ 62
3.2.18 QLARM, QUAKELOSS and QUAKELOSS2 via WAPMERR ............................ 63
3.2.19 RADIUS.................................................................................................................. 66
3.2.20 REDARS ................................................................................................................. 66
3.2.21 RiskScape................................................................................................................ 67
3.2.22 ROVER-SAT........................................................................................................... 67
3.2.23 SAFER..................................................................................................................... 68
3.2.24 SELENA and RISe.................................................................................................. 70
3.2.25 SES2002/ESCENARIS for SIGE, Spain................................................................. 71
3.2.26 SIGE, Italy............................................................................................................... 71
3.2.27 SP-BELA................................................................................................................. 72
3.2.28 StrucLoss/KOERILoss ............................................................................................ 72
3.3 Other Useful Possible Earthquake Integration Applications ...................................................73
3.3.1 VCH Tool for Disaster Updating ......................................................................... 73
3.3.2 Historic Loss Catalogues - NATHAN, EM-DAT, www.earthquake.it and USGS
based archives .......................................................................................................................... 73
Table of Contents
Daniell, April 2009 v
3.3.3 Hazard Modelling - NSHMP and GSHAP, OpenSHA........................................... 74
3.3.4 Complex Interseismic Hazard Modelling - QuakeSIM........................................... 75
3.3.5 TRANSFER............................................................................................................. 75
3.3.6 Proprietary Software for Insurance Companies....................................................... 76
4. SYNTHESIS OF ELE SOFTWARE PACKAGES FOR USE IN FUTURE OPEN SOURCE
PRODUCTION..................................................................................................................................... 77
4.1 Comparison of the ELE Software Packages ............................................................................77
4.2 Technical Aspects ....................................................................................................................77
4.3 Demand Module.......................................................................................................................83
4.3.1 Hazard types considered.......................................................................................... 83
4.3.2 Modes of Analysis Available .................................................................................. 84
4.3.3 Ground Motion Determination and Distributions ................................................... 86
4.3.4 Localised Site Effects.............................................................................................. 90
4.4 Vulnerability and Exposure Modules ......................................................................................91
4.4.1 Exposure Elements .................................................................................................. 91
4.4.2 Synopsis of Occupancy, Structural and Quality Criteria for Vulnerability............. 93
4.4.3 Vulnerability Methods and Damage Classes........................................................... 95
4.5 Specific Cost Module...............................................................................................................97
4.5.1 Social Losses ........................................................................................................... 97
4.5.2 Economic Losses..................................................................................................... 98
4.6 Rapid Response Use and Technological Intricacies ..............................................................100
5. ELE SOFTWARE PACKAGE PARAMETRIC STUDY..................................................... 102
5.1 Choice of platform to apply...................................................................................................102
5.2 Zeytinburnu Case Study Details ............................................................................................104
5.3 Results....................................................................................................................................110
5.3.1 Damage Distributions............................................................................................ 110
5.3.2 Economic Losses................................................................................................... 110
5.3.3 Social Losses ......................................................................................................... 112
5.4 Discussion and Conclusions for the Case Study....................................................................114
5.5 Important Optimisations for MDBELA and applications to other software..........................115
5.5.1 Sampling Size via Monte Carlo Simulation and other such methods ................... 115
5.5.2 Sampling Size for the material properties needed for DBELA............................. 120
5.5.3 Calculation Speed.................................................................................................. 120
5.6 Future Developments for this software type..........................................................................122
5.6.1 MHAZUS, MDBELA and SELENA changes ...................................................... 122
5.6.2 Expanding software into a global basis ................................................................. 123
Table of Contents
Daniell, April 2009 vi
6. OPEN SOURCE PROCEDURE FOR ASSESSMENT OF LOSS USING GLOBAL
EARTHQUAKE MODELLING........................................................................................................ 124
6.1 Introduction into Multicriteria Analysis ................................................................................124
6.2 Multicriteria Analysis with a Logic Tree Approach ..............................................................125
6.3 Future Developments - Framework for future production of OPAL-GEM2.........................127
6.3.1 Programming Language to be used....................................................................... 127
6.3.2 Choice of GIS........................................................................................................ 127
6.3.3 Collaboration with other partners.......................................................................... 128
6.3.4 Multi-level risk programming ............................................................................... 128
6.3.5 Forecasting, Post-Earthquake Use and Useable Output ........................................ 130
6.4 Conclusion of the Multicriteria Selection for OPAL-GEM2.................................................131
7. CONCLUSION...................................................................................................................... 132
REFERENCES ................................................................................................................................... 134
WEBOGRAPHY................................................................................................................................ 157
APPENDIX A: OVERVIEW OF CURRENT LITERATURE........................................................... A1
APPENDIX B: ADDITIONAL INFORMATION FOR ELE SOFTWARE PACKAGES................. B1
APPENDIX C: LOSS ANALYSIS AND SOFTWARE PRODUCTION:- ZEYTINBURNU CASE
STUDY FOR SELENA, MHAZUS AND MDBELA......................................................................... C1
APPENDIX D: MULTICRITERIA ANALYSIS AND SELECTION FROM OPAL-GEM1............ D1
APPENDIX E: CORRESPONDENCE WITH OTHER ELE SOFTWARE PACKAGE PRODUCERS
............................................................................................................................................................. E1

List of Figures
Daniell, April 2009 vii

LIST OF FIGURES

Page
Figure 1-1: Total Percentage of losses from 1950-2008 (MunichRE, 2009)........................................... 3
Figure 2-1: Identified Components of an Earthquake Loss Assessment (Rapid-Response, Post- or Pre-
Earthquake) ..................................................................................................................................... 7
Figure 2-2: Vision 2000 Structural Performance Objectives (SEAOC, 1995) (Porter, 2000)................. 8
Figure 2-3: Details of Exposure from the USGS PAGER Report (Jaiswal et al., 2008b)....................... 9
Figure 2-4: Earthquake Damage Assessment by Remote Sensing algorithms (UCAM Project, K. Saito,
WillisRe, 2008) ............................................................................................................................. 10
Figure 2-5: Example of a DPM by Whitman (1973) for a single building class ................................... 14
Figure 2-6: The parameter, p, for the binomial distribution DPM for a structural typology in Catalonia,
Spain (Roca et al., 2006)............................................................................................................... 15
Figure 2-7: Vulnerability function relationship between pga, damage factor and vulnerability index, I
v

(Guagenti and Petrini (1989), adapted by Calvi et al (2006), Crowley (2009)) ........................... 16
Figure 2-8: The difference between PGA vs. spectral displacement (as a fn of building period
vibration) validated for the 1995 Aegeon Earthquake in Greece (Rossetto and Elshinai, 2003).. 18
Figure 2-9: Macroseismic Method for EC8 DCM V=0.36, Q=2.5 and med-rise pre-code RC MRF
V=0.62, Q=2.3 from left, vulnerability curves, damage probabilities, fragility curves for pre-
code I=9, ductility=2.25. ............................................................................................................... 18
Figure 2-10: Production of Analytically derived Vulnerability Curves (Dumova-Jovanoska, 2004) ... 21
Figure 2-11: The Typical HAZUS Diagram which forms a basis for many ELE Engines, adapted from
Kircher et al., 2006 ....................................................................................................................... 23
Figure 2-12: The Capacity Spectrum Method (FEMA-440, ATC, 2005) ............................................. 24
Figure 2-13: Cumulative P(DS|Sd or Sa) for weak, medium and strong shaking and thus the
corresponding probability of being in a particular damage band (Kircher et al., 2006) ............... 24
Figure 2-14: Simplified Model for an equivalent SDOF system from Calvi et al. (2006). ................... 27
List of Figures
Daniell, April 2009 viii
Figure 2-15: Possible deformed shapes for the different limit states and in-plane failure modes (from
Calvi et al. (2006). ........................................................................................................................ 29
Figure 2-16: Comparing ADRS for various limit states in order to determine the associated damage
(Borzi et al., 2008) ........................................................................................................................ 32
Figure 2-17: Plan view of a prototype RC MRF Building as representative of a certain region (Borzi et
al. (2008)....................................................................................................................................... 33
Figure 2-18: Macrospatial Correlation of Ground Motion for 3 different models (Crowley et al., 2009)
....................................................................................................................................................... 40
Figure 2-19: Interperiod correlation (Temporal correlation) of ground motions from Baker and Cornell
(2006). ........................................................................................................................................... 40
Figure 2-20: Loss curves using the sum of three sites for PSHA and Stochastic Modelling (Crowley
and Bommer, 2006)....................................................................................................................... 42
Figure 2-21:- The processes contributing to earthquake loss from Khater et al. (2003) ....................... 43
Figure 2-22: Developed vs. Developing Economies (UNDP, 2008)..................................................... 44
Figure 2-23: Population density throughout the world (dark areas):- this can define locations of
greatest exposure (NASA visible earth, 2008).............................................................................. 45
Figure 2-24: Population growth in terms of % increase per year (-0 indicates a decrease in population)
(UNDP, 2008) ............................................................................................................................... 45
Figure 2-25: Resulting deaths from earthquakes from 1950-1999 (Coburn and Spence, 2002) ........... 46
Figure 2-26: Updated method for injury distributions and hence social losses (Spence, 2007) ............ 47
Figure 2-27: Comparison of Mean Damage Ratios for HAZUS and those suggested by Bal et al.
(2007). ........................................................................................................................................... 48
Figure 3-1: Methodology for EPEDAT (Eguchi et al., 1994). .............................................................. 56
Figure 3-2: LNECLoss Flowchart (Typical of many ELEs) Sousa et al. (2004) ............................... 60
Figure 3-3: PAGER System (PAGER Website, 2009).......................................................................... 63
Figure 3-4: Recent Italian Earthquake of 06/04/09 overlaying WAPMERR damage distribution and
USGS PAGER Population exposure via MMI (similar to PGA, PGV and MMI plots)............... 64
Figure 3-5: Flowchart of SAFER for various objectives and components (adapted from Zschau et al.,
2007). ............................................................................................................................................ 69
Figure 3-6: The difference between the static and dynamic components of SELENA for rapid response
modelling identification (Lindholm et al., Vilnius Conference, 2007)......................................... 70
Figure 3-7: The full SIGE setup which is transparent (providing insight into the Italian Government
methodology) Soddu et al. (2005) ............................................................................................. 72
Figure 5-1: The method for real-time estimation from the SELENA v3.5 manual. (Molina et al.,
2008a) ......................................................................................................................................... 103
List of Figures
Daniell, April 2009 ix
Figure 5-2: Image of the seismic temporal gap around Istanbul showing the fault segments S7 and S8,
the proximity of Zeytinburnu to the fault segments and the problem area, adapted from the
GONAF website using the work of Armijo et al., 2005. This shows the location of the close fault
monitoring system of GONAF as well as the Zeytinburnu district circle in Istanbul, with a
straight line to the point where the Mw 7.2 earthquake scenario was placed. ............................ 105
Figure 5-3: Evolution of the Zeytinburnu District from left to right Aerial Photos from 1946, 1966,
1982 and Present Day (Turkish Government Website, 2009) .................................................... 106
Figure 5-4: Flowchart detailing the approach of all three produced ELE software packages: SELENA
(all text files and some coding), MHAZUS (total), MDBELA (total), with the relevant input and
output (dashed) files.................................................................................................................... 108
Figure 5-5: Regular storey height (left panel) and the ground floor pier height (right panel) of the
URM buildings - Comparison between the statistical truncated normal at 2.0 and 2.85m (left) and
statistical F-noncentral (right) distributions (curves) and the histograms generated by Monte
Carlo simulations ........................................................................................................................ 109
Figure 5-6: Comparison for median GM field between the MDR distribution derived with the
MDBELA (left panel) and MHAZUS (right panel). The SELENA method is approx. the same as
the MDBELA and therefore has not been shown. ...................................................................... 111
Figure 5-7: Empirical Performance Point Limit State using optimally sampled buildings (adapting
diagram of Borzi et al. (2008)..................................................................................................... 118
Figure 5-8: Comparison between the fragility curves derived from MHAZUS (circles) and the curves
obtained from DBELA for C1M (circles) vs. RC5b buildings showing the limit states the
smooth curves are lognormal distributions (black rectangles) and the uniform distribution (no
symbol) shows the difference of the beam sway vs. column sway fragility curves.................... 119
Figure 6-1: Example of the GIS program, uDIG with possible economic loss geocells (uDIG, 2009)
..................................................................................................................................................... 128

List of Tables
Daniell, April 2009 x

LIST OF TABLES

Page
Table 1-1: The 10 most costly earthquakes in overall losses from 1980-2009 adapted from data from
MunichRE and USGS website data. ............................................................................................... 3
Table 2-1: The 11 Parameters used and the given weightings for the Boukri et al (2007) study of
Algiers, Algeria. ............................................................................................................................ 17
Table 2-2: Correlations between various ground motion intensity measures and building demand
parameters and percent loss estimates for various building classes modified from King et al.
(2005) and Stafford et al. (2007)................................................................................................... 19
Table 3-1: Overview of the ELE software packages selected for analysis............................................ 52
Table 3-2: Current and Future Tools that Global Earthquake Model will use for their system (GEM
Website, 2009) .............................................................................................................................. 75
Table 4-1: Availability of ELE Software Packages (as of April 2009) ................................................. 78
Table 4-2: Update and Development Status of ELE Software Packages selected................................. 79
Table 4-3: Hardware and Software Requirements for the ELE Software Packages.............................. 80
Table 4-4: Regional Applicability of reviewed ELE packages.............................................................. 82
Table 4-5: Earthquake-related hazards considered in ELE Software Packages reviewed..................... 84
Table 4-6: Analysis Models possible in the ELE Software Packages ................................................... 85
Table 4-7: Ground Motion Parameters for ELE Software Packages reviewed ..................................... 87
Table 4-8: Ground Motion type used for spatial distribution for the researched ELE software packages
....................................................................................................................................................... 89
Table 4-9: Local site effects modelling within the selected ELE Software Packages........................... 90
Table 4-10: Inventory elements considered for physical damage (normal) & information purposes
(italic) for the selected ELE Software Packages........................................................................... 92
Table 4-11: Building Criteria considered for the reviewed ELE software packages............................. 94
Table 4-12: Vulnerability Method and Damage Classes for the considered ELE Software Packages.. 96
Table 4-13: Social Losses for ELE Software Packages Reviewed........................................................ 97
List of Tables
Daniell, April 2009 xi
Table 4-14: Economic Losses for ELE Software Packages Reviewed.................................................. 99
Table 4-15: Rapid Response Capabilities computed in ELE Software Packages Reviewed............... 100
Table 5-1: Characterization of the building stock according HAZUS Code ....................................... 107
Table 5-2: Relative Building percentages in HAZUS-based damage classes...................................... 110
Table 5-3: Summary table of Economic Loss Estimates for Zeytinburnu District.............................. 111
Table 5-4: Uninjured, Injured and Death (I5) data for the collapse damage state (Spence, 2007)...... 112
Table 5-5: Summary table of Social Loss Estimates for Zeytinburnu District for Night-time
Earthquake Scenario ................................................................................................................... 113
Table 5-6: Summary table of Social Loss Estimates for Zeytinburnu District for Day-time Earthquake
Scenario....................................................................................................................................... 113
Table 5-7: Monte Carlo Simulation of 3000 buildings reduced to the damage state of 158 buildings in
the geocell and using 5 buildings expanded to the damage state of 158 buildings. .................... 119
Table 5-8: Fictitious case of using ANN buildings which are optimally chosen to portray the
characteristics using DBELA vulnerability curves ..................................................................... 120
Table 5-9: Speedup Using Matlab for the interpolation optimisation for a random test ..................... 121
Table 6-1: Multicriteria Analysis for 33 ELE Software Packages, some still under production, using
the extensive criteria detailed in Appendix D............................................................................. 126

List of Symbols and Abbreviations
Daniell, April 2009 xii

LIST OF SYMBOLS AND ABBREVIATIONS

ADRS = Acceleration-Displacement Response Spectrum
AGORA = Alliance for Global Open Risk Analysis
API = Application Programming Interface
ARE = Annual Rate of Exceedance
ATC = American Technical Council
BCF = Benefit Cost Function
BCR = Benefit/Cost Ratio
CAD = Computer-Aided Design
California OES = California Office of Emergency and Security
CAPRA = Central American Probabilistic Risk Assessment
CATS = Consequences Assessment Tool Set
CATS-JACE = CATS with Joint Assessment of Catastrophic Events
CAV
5
= Cumulative Absolute Velocity over 5 cm/s
CISN = California Integrated Seismic Network
COSMOS = Consortium of Organization for Strong Motion Observation Systems
CSM = Capacity Spectrum Method
CUBE = Caltech USGS Broadcast of Earthquakes System
CUREE-Caltech = Consortium of Universities for Research in Earthquake Engineering
CUS = Central United States
DAP = Displacement-based Adaptive Procedure
DBELA = Displacement-Based Earthquake Loss Assessment
DC = Damage Class
DCM = Ductility Class Medium
DGPC = Spanish General Direction of Civil Protection
DM = Damage Measure
DMT = Disaster Management Tool
DPM = Damage Probability Matrix
DSHA = Deterministic Seismic Hazard Assessment
DTRA = Defense Threat Reduction Agency
Daniell, April 2009 xiii
DV = Decision Variable
EAL = Expected Annual Loss
EC8 = Euro Code 8
ECLAC = Economic Commission for Latin America and the Caribbean
EDP = Engineering Demand Parameter
EIRD
= Estrategia Internacional para la Reduccin de Desastres (International Strategy
for Disaster Reduction)
ELA = Earthquake Loss Assessment
ElarmS = Earthquake Alarm Systems
ELE = Earthquake Loss Estimation
ELER = Earthquake Loss Estimation Routine
EM = Electro-magnetic
EM-DAT = Emergency Events Database
EMME = Earthquake Model for the Middle East Region
EMS = European Macroseismic Scale
EPC = Emergency Preparedness Canada
EPEDAT = Early Post Earthquake Damage Assessment Tool
EQ = EarthQuake
EQRM = EarthQuake Risk Management
EQSIM = EarthQuake damage SIMulation tool
ESCENARIS = Scenario-based tool for Catalonia and Pyrenees
ESPAS = Earthquake Scenario Probabilistic ASsessment
ESRI = Environmental Systems Research Institute
ETABS = Extended 3D Analysis of Building Systems (Software)
etc. = et cetera
ETHZ = Eidgenssische Technische Hochschule Zrich
FaMIVE = FAilure Mechanism Identification and Vulnerability Evaluation
FatalityVFs = Fatality Vulnerability Functions
FEMA = (U.S.) Federal Emergency Management Agency
FHWA = Federal HighWay Administration
FORM = First Order Reliability Method
Freq. Dep. = Frequency Dependent
Freq. Ind. = Frequency Independent
F- = Force-Displacement
GA = Geoscience Australia
GB = Gigabyte
Gebze IT = Gebze Institute of Technology
GEM = Global Earthquake Model
GeoFEST = Geophysical Finite Element Simulation Tool
GFDRR = Global Facility for Disaster Reduction and Recovery
GFZ = GeoForschungsZentrum (German Research Centre for Geosciences)
Daniell, April 2009 xiv
GHz = Giga Hertz
GIS = GeoInformation System
GM = Ground Motion
GMPE = Ground Motion Prediction Equation
GNDT
= Gruppo Nazionale per la Difesa dai Terremoti, Universita degli Studi di
Genova
GNP = Gross National Product
GNS = Geological and Nuclear Sciences Limited
GPRS = General Packet Radio Service
GPS = Global Positioning Systems
GSHAP = Global Seismic Hazard Assessment Program
GUI = Graphical User Interface
HAZ-Taiwan = HAZUS-Taiwan or Hazards-Taiwan
HAZUS-MH = Hazards U.S.-Multi-Hazard
HAZUS = Hazards U.S.
HAZUS-MH MR3 = Hazards U.S. Multi-Hazard Maintenance Release 3
HLA = High Level Assembly
i.e. = id est
IBC = International Building Code
IDNDR = International Decade for Natural Disaster Reduction
ILWIS = Integrated Land and Water Information System
IM = Intensity Measure
INGV = Istituto Nazionale di Geofisica e Vulcanologia (Italy)
InLET = Internet-based Loss Estimation Tool
INSAR = Interferometric synthetic aperture RADAR
INSN = Iranian National Broad-band Seismometer Network
ISET = Indian Society of Earthquake Technology
ISNet = Irpinia Seismic Network
ISTAT = Istituto Nazionale di Statistica (Italy)
JBDPA = Japan Building Disaster Prevention Association
JMA = Japan Meteorological Agency
JRA-3 = Joint Research Agency-3
KIT = Karlsruhe Institute of Technology
KOERILOSS = Kandilli Observatory Earthquake Research Institute Loss Estimation Software
LE = Loss Estimation
LiDAR = Light Detection and Ranging (Remote Sensing Method)
LNEC = Laboratorio Nacional de Engenharia Civil, Portugal
LNECLOSS = Laboratorio Nacional de Engenharia Civil LOSS model
LossCurveApp = Loss Curve Application
MADRS = Modified Acceleration Displacement Response Spectrum Method
MAEviz = Mid America Earthquake center VIZualisation
Daniell, April 2009 xv
MCE = Maximum Credible Earthquake
MCEER = Multidisciplinary Center for Earthquake Engineering Research
MCS = Mercalli-Cancani-Sieberg Intensity Scale
MDOF = Multi-Degree Of Freedom
MDR = Mean Damage Ratio
M-D- = Magnitude-Displacement-Variability triplet
MeBaSe = Mechanical Based Procedure for Seismic Risk Estimation of URM buildings
MIHEA = Mono-Image Height Extraction Algorithm
min = Minute
MIRISK = Mitigation Information and Risk Identification System Kyoto
Misc. = Miscellaneous
Ml = Local Magnitude
MM = Modified Mercalli
MMI = Modified Mercalli Intensity
MMSK86 = Modified Medvedev-Sponheuer-Karnik Intensity Scale of 1986 update
MSK = Medvedev-Sponheuer-Karnik Intensity Scale
MunichRE = Munich Reinsurance Company
Mw = MegaWatt
n/a = not available
n/f = not found
NASA = (U.S.) National Aeronautics and Space Administration
NATHAN = NATural Hazards Assessment Network
NEAREST = NEAR shore sourcES of Tsunamis
NEHRP = (U.S.) National Earthquake Hazards Reduction Program
NERIES = Network of Research Infrastructures for European Seismology
NGA = Next Generation Attenuation (Relations)
NHEMATIS = Natural Hazards Electronic Map and Assessment Tools Information System
NIBS = (U.S.) National Institute of Building Science
NIED = National Research Institute for Earth Science and Disaster Prevention
NIWA = (NZ) National Institute of Water & Atmospheric Research
NLA = Non-Linear Analysis
NOAA = (U.S.) National Oceanic and Atmospheric Administration
NORSAR = Research Council of Norway
NSHMP = National Seismic Hazard Mapping Program
NSP = Non-linear Static Procedure
NTHA = Non-linear Time History Analysis
NW = North-West
NZ = New Zealand
O.C. = Orange County
OPAL procedure = Overview, Preliminary research, Assessment and Loss analysis procedure
OPAL-GEM1 = 1
st
stage Open source Procedure for Assessment of Loss using Global
Daniell, April 2009 xvi
Earthquake Modelling
OPAL-GEM2
= 2
nd
stage Open source Program for Assessment of Loss for Global
Earthquake Modelling (to be produced)
OpenSees = Open System for Earthquake Engineering Simulation
OpenSHA = Open Seismic Hazard Analyses
OSRE = Open Source Risk Engine
OSSN = Italian Department of Civil Protection
P(DS|Sd or Sa) = Probability of Damage State exceeding Sd or Sa
P2 = PAGER version 2
P25 = 25-parameter
PAGER = Prompt Assessment of Global Earthquakes for Response
PAGER-CAT = PAGER - Consequence Assessment Tool
PARK = Boundary Element problem derived from PARKfield, California
PDA = Personal Data Assistant
pdf = Probability Density Function
PEER = Pacific Earthquake Engineering Research Center
PEER NGA = PEER Next Generation Attenuation of Ground Motions Project
pers. comm. = personal communication
PESH = Potential Earth Science Hazard
PGA = Peak Ground Acceleration
PGD = Peak Ground Displacement
PGV = Peak Ground Velocity
pop. = population
PP = Power Plant
PRC = Peoples Republic of China
PSHA = Probabilistic Seismic Hazard Assessment
PSI = Parameterless Scale of Intensity
QL2 = QUAKELOSS 2
QuakePy = Earthquake Python-based program
QuakeSim = Earthquake Simulation software
RADAR = RAdio Detection And Ranging
RADIUS = Risk Assessment tools for DIagnosis of Urban areas against Seismic disasters
RAFT = Rapid Aftershock Forecasting Toolbox
RAM = Random Access Memory
RC = Reinforced Concrete
RC MRF = Reinforced Concrete Moment Resisting Frame
REDARS = Risks from Earthquake DAmage to Roadway Systems
REDI = Rapid Earthquake Data Integration
RepairCostVFs = Repair Cost Vulnerability Functions
Res. = Research or Residential depending on context
RISe = Risk Illustrator for Selena
Daniell, April 2009 xvii
RISKMAN = RISK MANager software
Risk-UE = Risk Union Europeenne (EU RISK Project)
ROVER = Rapid Observation of Vulnerability and Estimation of Risk
ROVER-SAT = ROVER, Shakecast and ATc-20i
RTLo = Real-Time Location of hypocentre
RVT = Random Vibration Theory
s = second
Sa = Spectral Acceleration
SAFER = Seismic eArly warning system For EuRope
SAM = Scientific Annotation Middleware
Sd = Spectral displacement
SDOF = Single-Degree Of Freedom
SE = South-East
SEAHELLARC
= SEismic and tsunami risk Assessment and mitigation scenarios in the western
HELLenic ARC
SeisImpactTHES = Seismic Impact Software for THESsaloniki, Greece
SELENA = Seismic Loss EstimatioN using a logic tree Approach
SES2002 = Simulacion de Escenarios Sismicos 2002
SHA = Seismic Hazard Assessment
ShakeCast = ShakeMap SimulCasting by USGS
SIGE = Emergency Management Information System (Spain or Italy)
SPBELA = Simplified Pushover-Based Earthquake Loss Assessment
SPT/CPT = Standard Penetration Test/Cone Penetration Test
SRSS = Square Root of the Sum of Squares method
SS = Slope Stability
Std. = Standard
STEP project = Strategies and Tools for Early Post-Earthquake Assessment project
TRANSFER = Tsunami Risk ANd Strategies For the European Region
u/c = under construction
UAE = United Arab Emirates
UCAM = University of California AM
UML = Unified Modelling Language (open, object-oriented programming)
UNDP = United Nations Development Program
UN-HABITAT = United Nations Habitat Project
URM = Unreinforced Masonry
U.S. = United States
USC = University of Southern California
USGS = U.S. Geological Survey
VCH = Virtual Clearing House
Vs,30 = Shear Wave Velocity in the first 30 metres depth
vs. = versus
Daniell, April 2009 xviii
WAPMERR = World Agency for Planetary Monitoring and Earthquake Risk Reduction
WUS = Western United States
xmf = eXtensible Model and Format
ZEUS-NL = ZEUS-Non-Linear 3D static and dynamic analysis program
= shear-wave velocity in the top layer
= epsilon (applied variability)
= spectrum reduction factor
= sigma (Aleatory variability)
= Pi
= equivalent viscous damping

OPAL-GEM: with the advent of the Global Earthquake
Model (GEM-www.globalquakemodel.org), this thesis
provides a framework to produce a truly global open source
ELE software package Open source Program for
Assessment of Loss for Global Earthquake Modelling
(OPAL-GEM). An OPAL is Australias national GEM-stone
but is also found everywhere in the world, and even on Mars.
It shines with all colours of the spectrum and was believed in
the Middle Ages to be the all-encompassing and ruling GEM.
Thus, an OPAL is the perfect descriptor for an all-
encompassing open source Global Earthquake Model.

Chapter 1. Introduction
Daniell, April 2009 1

1. INTRODUCTION
Considerable research has been done to provide adequate earthquake loss estimation (ELE)
models for region specific scenarios and other studies. Many different software packages have
been produced around the world in order to provide accurate loss estimates. With the wealth
of software packages available for these risk assessment studies and economic, social and
infrastructure loss estimations, a synopsis of many available packages (and some not
available) has been undertaken. The authors of each of these packages have been contacted in
order to get the full version for analysis. Unfortunately, not all software packages are freely
available and many are proprietary, as the reinsurance and insurance industry requires
accurate estimations of loss due to natural hazards.

This thesis aims to provide the tools to produce a state-of-the-art global open source
ELE software package such as that produced through a GEM (Global Earthquake
Model).

This will be done by an in-depth view of current ELE and other earthquake software packages
using an Open source Procedure for Assessment of Loss using Global Earthquake Modelling
software (OPAL-GEM1) with the view of creating a truly Open source Program for
Assessment of Loss for Global Earthquake Modelling (OPAL-GEM2).
components.

Within the Open source Procedure for Assessment of Loss using Global Earthquake
Modelling (OPAL-GEM1), the following steps are proposed (OPAL):
1. Overview of all components of Earthquake Loss Assessment (2) Literature
Review;
2. Preliminary research and use or methodology of all current ELE software
packages (3) Data Collection and Familiarisation;
3. Analysis of the components of these ELE software packages (4) In-depth view
of ELE software package methods used; and
4. Loss analysis using the researched ELE software packages and produced ELE
software packages for familiarity with ELE systems and to identify avenues for
optimisation (5) Application and Innovation for a Project.

The literature review component of this thesis (Overview) consists of a section explaining the
theory of earthquake loss assessment through the exposure, vulnerability, hazard and specific
cost stages. Each of these stages is usually region-specific and hence adjustments need to be
made. Therefore, an ELE Software package should be able to accommodate these changes.

The various components of such assessments will be detailed, as well as the various available
approaches empirical, theoretical, analytical, force and displacement-based. A more detailed
view of displacement-based models will be examined, as these types of models have been
seen to provide a significant reduction in error in terms of calculating structural and non-
structural damage. The review outlines the knowledge behind each of the ELE software
components and the various advantages and disadvantages of available ELE software
packages, attempting to provide an overview of the information that is available on the subject
of Earthquake Loss Estimation. This information has resulted from the large economic and
social losses occurring in recent years.

It is shown in a study from 1988-2003 that 88% of damage in recent earthquakes has been
caused by ground shaking, rather than other causes such as landslides, tsunamis, liquefactions,
seiches and other secondary effects (Bird and Bommer, 2004). However, in the past 5 years,
these results would be significantly different because of the large tsunami of 2004, killing
over 227,000 people, and the Pakistan (2005) and Sichuan (2008) earthquakes where
landslides as well as ground shaking caused many deaths and injuries. In terms of economic
loss, there is an increased impact from earthquakes due to increased population growth in
areas of high risk causing higher vulnerability, as can be seen in recent earthquakes. However,
in these cases, ground shaking seems to be the major cause of damage and thus the study by
Bird and Bommer (2004) seems to still be valid for economic loss. For social losses (human-
related losses), future ELEs should possibly place more emphasis on other types of
earthquake effects and not just ground shaking. Nevertheless, for economic losses, reasonable
estimates could be obtained by only taking into account ground shaking.
Table 1-1: The 10 most costly earthquakes in overall losses from 1980-2009 adapted from data from
MunichRE and USGS website data.
Rank, Age Region
(Country, City)
Overall losses
(U.S.$Billion)
Insured losses
(U.S.$Billion)
Fatalities
(no. of people)
=1, 1995 Japan, Kobe 95-102.5 3 6,434
=1, 2008 China, Sichuan 85-146.5 1-2 69,000-88,000
=3, 1994 USA, Northridge 20-44 15.3 61-72
=3, 2007 Japan, Niigata 26 0.76 40-46
=5, 2009 Italy, LAquila 5(AIR) to 22 (Gov.) Minor Approx. 300
=5, 1988 Armenia, Spitak 14 Minor >25,000
=5, 1999 Taiwan, Chi-chi 14 0.75 2,370-2,416
=8 Japan 12.5 0.335 11
=8, 1999 Turkey, Izmit 12 0.6 17,000
=10, 1980 Italy, Irpinia 10-11.8 0.04 2,914
=10, 1989 USA, Loma Prieta 10 0.96 68
=10, 2004 Indonesia, Sumatra 7-14 Minor 170,000-300,000

Figure 1-1: Total Percentage of losses from 1950-2008 (MunichRE, 2009)
It can be seen from these figures, that fatalities from earthquakes make up more than half of
the total fatalities from natural disasters from 1950-2008. However, insured losses are
dominated by meteorological events. There is a bias due to the large hurricanes occurring in
recent years in the Gulf of Mexico but it can be seen that geophysical events are the second
highest in economic losses for natural disasters. Because of the temporal and spatial
randomness of earthquakes, they can cripple economies when they strike, such as within
Central America where a large percentage of the GNP can be seen to be lost (Nicaragua 1972
(40%), El Salvador (31%), Guatemala (18%)). They also cause extremely large economic
losses such as $146 billion U.S. (PRC Government, NDRC, 2009) or $85 million U.S.
(NATHAN, 2009) for the Sichuan earthquake in 2008 and $100 billion U.S. for the 1995
Kobe event, but they were a small percentage of the GNP (2-3%).

The need to quantify this damage has led to many ELE Software packages being developed.
ELE software packages are both closed (not freely available but documented) and open-
source (freely available or by contacting the developers), and the study first requires a
preliminary research, familiarisation and use of these ELE software packages.
An analysis will then be carried out to identify the components of all the relevant ELE
software packages and to identify the differences and similarities and possible criteria to test
these packages.

At least two software packages should then be applied to a loss assessment to look at the
differences in application. A test case using in the Zeytinburnu district in Istanbul has been
examined. Two software packages have been produced at this time to provide a good basis for
comparison coding of the DBELA (Displacement-Based Earthquke Loss Assessment) process
produced by Pinho et al. (2002) and the CSM-based method, HAZUS. These have both been
coded within Matlab. SELENA has also been modified and compared for the calculation of
economic, social and damage estimates.

Using the OPAL procedure, enough knowledge has been gained to produce a multi-criteria
analysis to select an avenue for further production and future developments of an open-source
program for global earthquake modelling loss assessment based on some of the coding
produced in this dissertation.

Further research is required in conjunction with GEM to complete this study; however, this
dissertation provides a good synopsis of the available packages for use and will help to
provide a basis for a possible GEM-type open source program to be produced (OPAL-
GEM2). It should be created under the constraints of the procedure used within this
dissertation (OPAL-GEM1). It is hoped that this dissertation will also aid other groups to
make their own informed decisions as to what the best method for production of an ELE
software package is, and to keep track of the state of different ELE software packages around
the world (i.e. dynamic updating).

This dissertation also yields some applications to the recent Italian earthquake in order to give
a recent perspective and some future possibilities for improvements to pre- and post-
earthquake studies, such as employing ROVER-SAT to the STEP project.

Chapter 2. Literature Review Earthquake Loss Assessment and Estimation

2. LITERATURE REVIEW EARTHQUAKE LOSS
ASSESSMENT AND ESTIMATION
2.1 Introduction
Earthquake Loss Assessments are produced in order to detect possible economic,
infrastructure and social losses due to an earthquake. In order to produce an effective ELE,
four components must be taken into account (Crowley et al., 2006), in that:-
Seismic Loss = Exposure * Vulnerability * Hazard * Damage Loss Conversion
Where:-
Exposure is defined as the amount of human activity located in the zones of seismic hazard as
defined by the stock of infrastructure in that location (usually defined by geocell);
Vulnerability is defined as the susceptibility of the infrastructure stock;
Hazard is defined by risk of a certain ground motion occurring at a location, which can be
defined by scenario modelling via stochastic catalogues, PSHA or other such methods, and
can include different types of earthquake effects;
Damage Loss Conversion, can be defined as the mean damage ratio (ratio of replacement &
demolition to repair & restoration cost (economically-speaking), or the social cost (i.e.
number of injuries, homeless and deaths).

Because of the myriad of ways that each of these components that make up seismic loss can
be determined, there is a large range of ELE methods available. For some regions one
particular method may be more applicable and because data collection is not the same all
around the world and regional effects do occur, it is impossible to have a 100% accurate
seismic loss estimate.

Thus, the ways in which each of these four components have been previously calculated must
be examined in order to define which methods are being used within all of the ELE software
packages.
2.2 What constitutes an ELE?
The general definition of an ELE has been given above. Nevertheless, following the tree
diagram portrayed below allows for a better insight into the components. It is necessary to
define an area of interest in which the seismic hazard should be located at every location. The
vulnerability of the infrastructure stock exposed to this hazard should be convolved with this
hazard and therefore a damage distribution can be established based on various classes of
infrastructure damage. From this damage distribution, economic and social losses can be
derived. All of these components constitute an ELE. Calculation of the losses can either be
done in a proactive way (pre-earthquake scenario modelling) or a reactive way (post-
earthquake fixed scenario modelling). In many ways, the methods are interchangeable and can
be used for production of both requirements with sufficient computer coding planning.

Figure 2-1: Identified Components of an Earthquake Loss Assessment (Rapid-Response, Post- or Pre-
Earthquake)
Scenario Earthquakes or
Given Earthquake
Given Exposure location or
Area for analysis set
Characterisation of infrastructure stock (material and
mechanical), population density, variability of infrastructure
within type, lifelines use of infrastructure.
Either set, or probabilistically defined by pre-existing
location, magnitude, fault type, source characterisation,
stochastic catalogues, historical earthquake activity, PSHA.
Hazard at the exposure
location or area
Path and site effects, tectonic regimes, distance from fault,
GMPEs/attenuation relations leading to ground motion
characterisation, hazard type (ground shaking, liquefaction,
tsunami, landslide, surface fault rupture, seiche or other),
NEHRP or other site class characterisation.
Set a damage scale to calculate vulnerability Damage Scale for exposure
Vulnerability Assessment
Procedure
Either Empirical (damage probability matrices or
vulnerability functions based on field surveys, typology or
expert judgement), Analytical (using capacity spectrum or
other NSPs, collapse mechanism-based or displacement-
based methods) or Hybrid (to fit exposed stock into damage
scale)
Economic, Social and
Infrastructure Losses
Loss assessment via economic means (direct and indirect
losses using the vulnerability results i.e. MDR ratio
between cost of repair and replacement for the entire
infrastructure stock), social losses using empirical tables or
previous data, or social vulnerability functions
Performance-based engineering is also becoming a large part of ELA and is best described in
terms of the PEER Framework which is much the same as the seismic risk definition given in
2 but is defined based on the structural performance for a given design level which is the
basic HAZUS- and DBELA-type framework. Location and Design via the PEER analysis
methodology are impacted upon by the probabilistic conditional relationships between IM
(hazard, intensity measure) and EDP (structural analysis, engineering demand parameter) and
DM (damage measure) and DV (loss, decision variable) (PEER Website, 2009).

Figure 2-2: Vision 2000 Structural Performance Objectives (SEAOC, 1995) (Porter, 2000)
For each earthquake design level, there is a given performance level that needs to be
maintained, depending on the exposure of the infrastructure to be designed or analysed.
2.3 Exposure
This component is increasing due to the increasing population around the world, and therefore
accurate characterisation is required. Unfortunately, although the simplest to characterise, it is
the most difficult component to collect. Data quality can vary from country to country and
region to region. Due to the sheer number of people in the world (6.78 billion as of mid-May
2009 and growing) living in various types of buildings, it is extremely difficult to derive an
accurate infrastructure stock for any country. However, this has been attempted by the USGS
for their PAGER (Prompt Assessment of Global Earthquakes for Response) exposure-type
program, in order to characterise possible losses in near real-time from any earthquake which
occurs around the world. This is based on an intensity-type system (Modified Mercalli) and
has come from a variety of publicly available data sources from the UN-Housing database,
UN-HABITAT, UN Statistical database on Global Housing (1993) housing censuses, PAGER
surveys, published literature, and the World Housing Encyclopaedia. In the end, due to a lack
of data, educated judgement was used in many cases.

Figure 2-3: Details of Exposure from the USGS PAGER Report (Jaiswal et al., 2008b)
Every ELE Software requires different inputs. However, EPEDAT inventory data include
building location, age, use, height, and structural type of buildings developed from data provided
by county assessors for five counties of southern California (Eguchi et al., 1997). Some other data
can include structural inconsistencies, construction age, number of storeys and population
information.

Of course, in order to do a detailed ELE, exposure needs to be undertaken on a smaller area
level and therefore other methods such as remote sensing techniques can be employed, but the
country-based generalisation does have merit, as government policy has a huge impact on
the effects of earthquakes and such social exposure should be taken into account (Birkmann,
pers. comm. 2008).
On a city or regional level, urban characteristics are required in order to produce a suitable
inventory, including location of lifelines and also building stock details. This is generally a
large amount of work requiring much money and time for a certain location, but, as stated
above, the level of detail that is able to be extracted during this part of the process determines
to a great extent the accuracy of the result at the end of the earthquake loss assessment. The
choice of how to go about creating an inventory is generally dependent on the loss estimation
technique/software used and therefore this must first be determined.

Remote sensing techniques have been employed with increasing success in recent years
(DellAcqua, 2009) in order to extract, verify and collate the asset and building type, footprint
and height using various optical methods such as LiDAR (light detection and ranging which
allows for canopy penetration and better vertical accuracy, but an aircraft is required),
hyperspectral and multispectral methods, other aerial remote sensing options such as
QuickBird imagery, LandSAT imagery (low quality) and aerial photography or RADAR
methods (InSAR (interferometric synthetic aperture RADAR) which collects backscattered
EM waves in a side-looking way in nature estimating distances and extracting geometric
features exactly). Manual digitising and image height extraction methods and algorithms can
also be employed in order to add height to building footprints. Segmentation of urban areas
can also be applied according to its morphology using various filters as well as different
frequency changes. Willis Research details a pre-earthquake and post-earthquake damage use
of remote sensing by using classification methods through the work at UCAM in going
through the process of image extraction to look at which sections of buildings are collapsed.

Figure 2-4: Earthquake Damage Assessment by Remote Sensing algorithms (UCAM Project, K. Saito,
WillisRe, 2008)
RiskScape is investigating different remote sensing techniques for use within their ELE
software as well as looking at Google Street View-type techniques which could be used for
sub-sampling type in regions or cities which are particularly at risk (RiskScape website,
2009). LiDAR, InSAR, and street view-type techniques could be used complementarily in
order to create an exposure asset inventory for use in an E.L.E.

More and more exposure and damage assessment has included remote sensing data in recent
years as it can provide an overview of damage, post-earthquake, independent of time of day or
weather. It can also supplement existing maps over a global scale, and is extremely resilient,
fast, and low risk (low chance of failure) (Eguchi and Mansouri, 2005). More information on
remote sensing techniques will not be included as they are not within the scope of this report,
however further reading for such techniques can come from Saraf et al. (2005), Sarabandi et
al. (2005) discussing the MIHEA (a height extraction method), Miura and Midorikawa
(2006), Sarabandi et al. (2006) and DellAcqua (2009). These techniques will become of
increasing importance as the technology continues to improve.

Further GIS system updates are being proposed in order to characterise where the damage
distribution of previous earthquakes has been electronically converted using survey and
inspection forms, collated post-earthquake. This type of data recovery aids software systems
to accurately check their software and methodology. Two of these projects are the
SeisImpactTHES (Savvaidis et al., 2004) and the Kobe earthquake project (Umemura et al.,
2002).

One of the most important aspects to be found in an exposure inventory production is that of
the lifelines within the region being analysed. These lifelines are intrinsically linked into the
specific cost section: however, they must be identified for rapid loss estimation procedures in
order to ensure that utility (potable water, waste water, oil (crude and refined), natural gas,
electric power and communication) and transportation systems (highway, railway, light rail,
bus, port, ferry and airport) can remain in place during a disaster. Critical systems (essential
and high potential loss facilities) must also be identified for rapid response and calculation
during the loss estimation module such as medical care facilities (hospitals and medical
clinics), schools (primary and secondary, colleges and universities) and emergency services
(fire stations, ambulance, police stations, army and civil service). In addition, industrial,
commercial and residential areas should also be identified. The HAZUS-MH MR3
methodology details these lifeline exposure inventories well (FEMA, 2003). Distances,
topography, land use and human demographics should also be modelled in the exposure
section (Schmidt et al., 2007).
2.4 Vulnerability
The real aim of a vulnerability assessment is to derive the probability of a certain level of
damage occurring to a certain infrastructure stock when subjected a certain scenario
earthquake.
2.4.1 Why is it so important to define large-scale vulnerability methods?
For analysis of existing buildings, it is desirable to use a non-linear analysis approach.
If it were possible to run a non-linear time history analysis for every single infrastructure item
within the infrastructure stock of the geo-cell and the information was in a freely available
form, then this would be done in order to calculate the vulnerability via production of a
capacity curve and then looking at the corresponding demand spectrum. However, this
information is not freely available for every infrastructure item and, in addition, it can take
days to produce a single NTHA of a building when given the construction plans as a new
model needs to be produced for every building. The computation time when running such an
analysis is immense, as the predicted ground motion to produce the capacity curve needs to be
changed in order to produce the entire curve within a NTHA, as each ground motion
corresponds to a single point on the F- curve. It simply uses the ground motion input and an
integration using the motion equation in order to produce this curve. There are numerous
software packages around, both proprietary and non-proprietary in order to undertake these
dynamic analyses for single buildings. ETABS (http://www.csiberkeley.com/
products_ETABS.html), SAP2000 (http://www.csiberkeley.com/products_SAP.html),
Ruaumoko (http://www.civil.canterbury.ac.nz/ruaumoko/), OpenSees (http://
opensees.berkeley.edu/index.php), ZEUS-NL (http://mae.ce.uiuc.edu/software_and_tools
/zeus_nl_registration.html) and Seismostruct (www.seismosoft.com) are some such programs
capable of NTHA to varying extents.

Non-linear static pushover can be used in order to reduce the computation time for a single
infrastructure item giving reasonable accuracy (it will never be as accurate as a direct NTHA
due to simplifications used) from a single run of a monotonic static pushover to produce the
capacity curve. There are numerous methods but all require definition of a model within the
software, and then a representative loading pattern is applied to the model which is either
conventional or changes with every loading step (adaptive). These latter methods can take
into account the stiffness degradation that occurs with increasing non-linearity and thus the
period lengthening and therefore then the change in mode shape as this occurs. These methods
are not considered within the scope of this report but note should be taken of the
Displacement-based Adaptive Procedure (DAP) method by Antoniou and Pinho (2004), as
this method could be employed for a future quick version of SPBELA. The DAP is
undertaken by applying a non-linear static pushover with changing displacement profile i.e.
first defining the nominal load vector and inertial mass, then calculating the computation of
the load factor, normalising and then updating the displacement vector. This is shown in
Seismostruct (www.seismosoft.com) which, being open-source, could be used for the
production of future ELEs when constructing the capacity curve. In order to define the actual
performance point with the DAP, to define the damage state, the adaptive capacity spectrum
method (Casarotti and Pinho, 2007) should be used which creates an equivalent SDOF
structure and uses damping and ductility reduction. The other methods that can be employed
as NSPs are the Capacity Spectrum Method (as applied in HAZUS as seen below), N2
Method (Fajfar and Fischinger, 1988), Modal Pushover Analysis (Chopra and Goel, 2002)
and Adaptive Modal Combination Procedure (Kalkan and Kunnath, 2006), the latter of which
attempts to better take into account higher mode effects. However, as defined previously, a
faster method of analysis is required when looking at many buildings within a district or
geocell.
2.4.2 Empirical Methods of Vulnerability Assessment
Empirical methods are vulnerability assessment methods based on observed damage data.
These methods have been employed to define the vulnerability of infrastructure stock from
the 1970s. Many of these methods only used macroseismic intensity or PGA rather than
spectral ordinates which created a large scatter of results; however, these were initially (with
the computer power available) the only possible method for large scale seismic risk analysis.

Damage probability matrices (DPM) are methods to determine damage due to strong motion
which are simply the conditional probability of obtaining a certain damage level (j) due to a
certain ground motion intensity (i), formed in the way (P|D=j/i) and were first proposed by
Whitman (1973) using data from the 1971 San Fernando Earthquake in which 1600 buildings
were surveyed over a variety of building classes. For a given building class, subjected to a
given intensity, there will be a certain percentage of buildings associated with the
combination which corresponds directly to a given damage ratio, as classed within the
damage state index used. Thus, for this percentage of buildings, the ratio of repair to
replacement cost corresponds to the values given in the DPM.

Figure 2-5: Example of a DPM by Whitman (1973) for a single building class
DPMs have also been used in the ATC-13 approach of 1985 which uses expert opinion in
order to derive DPMs. Many different earthquake engineering experts, civil engineers,
architects, building department officials and other system operators were asked to make
judgements as to possible values of the damage factor for certain building classes per intensity
(MMI range VI to XII) (Panel on Earthquake Loss Estimation Methodology by the ATC-13,
1985). These values were then weighted according to confidence in the expert as shown in
ATC-13 (ATC, 1985). These type of expert opinion approaches have been undertaken for
Bogota (Cardona and Yamin, 1997) as well as Basel (Faeh et al., 2001).
There have also been many instances of DPMs being used in Italy and other parts of Europe,
based on a binomial distribution which is simply a parameter defining the mean and standard
deviation for a given vulnerability class and intensity degree. For the following table,
buildings have been separated into four levels of vulnerability based on the EMS-98 scale,
where class D accounts for constructed buildings after 1980, which generally are better
seismically built. Thus for each damage level based on the MSK-76 damage level, i, from 0 to
5 (where 0 is classed as no damage, 1 = slight, 2 = moderate, 3 = heavy, 4 = destruction
(partial), 5 = total collapse) the following formula can be used in order to calculate the
probability (P
i
) of a vulnerability class having that type of damage level (i) for a particular
intensity, using the value p seen in Figure 2-6.
i i
i
p p
i i
P

=
5
) 1 (
)! 5 ( !
! 5
(2-1)
In some ways, this simplifies previous methods by only having one parameter but the mean
and standard deviation is also based on this single parameter.

Figure 2-6: The parameter, p, for the binomial distribution DPM for a structural typology in Catalonia,
Spain (Roca et al., 2006).
This method was first employed in Italy by Braga et al. (1982) using 1980 Irpinia earthquake
data and was also based on the MSK scale and has subsequently been used by Bramerini et al.
(1995). Dolce et al. (2003) also employed such DPMs for Potenza, Italy and Di Pasquale et al
(2005) used such DPMs as well but adapted the damage scale from MSK to MCS to fit better
the MCS-based Italian seismic catalogue. Giovinazzi and Lagomarsino (2004) also used a
macroseismic method based on a beta damage distribution and used Fuzzy Set Theory (i.e. the
gradual assessment of membership of elements in a set). It uses a vulnerability index based on
the region, building stock and classes to produce DPMs related to the building stock. This
method has been used extensively throughout risk assessments in Spain (Faro, Barcelona,
Catalonia) and Portugal (Lisbon) by Oliveira et al. (2004), Lantada et al. (2004), Roca et al.
(2006) and Oliveira et al. (2005) respectively. It is incorporated into the SES 2002 and
ESCENARIS earthquake loss assessment software for Spain as will be explained further
below in 3.2.25. This method is described in more detail within Oliveira et al., (2006) pp.
115-129.

A disadvantage of using macro-seismic methods for observed damage of building stock is that
the vulnerability and ground motion input are both based on observed damage due to
earthquakes which is not correct. If such observed damage values are going to be used, there
are also not many recordings of earthquakes with large intensities which occur close to cities.
Therefore there is a lack of data in the high damage and ground motion section of the
vulnerability matrix and so the statistical certainty is less towards the higher end of the
spectrum. Another issue results from the fact that PGA and spectral ordinates are generally
used for seismic hazard maps and these are not directly related to intensity scales which are
slightly subjective in nature. PGA when derived for empirical vulnerability does not take into
account the relationship of vibration frequency content of buildings versus that of the ground
motions and this is why spectral ordinates are more desirable.

The method described by Giovinazzi and Lagomarsino (2004) is one form of vulnerability
index. Empirical Vulnerability Index Methods are usually based on much survey data after
an earthquake in order to gain information as to relationships between damage and intensity
based on parameters influencing vulnerability. These methods have been used extensively
throughout Italy (based on GNDT Level I and GNDT Level II (Benedetti and Petrini, 1984;
GNDT, 1993)) previously using these parameters, and thus expert judgement can be used in
order to calculate the vulnerability index and then produce an indirect relationship with a
damage factor for a given PGA/macroseismic intensity. In the case below, the factors have
been derived for masonry buildings in Algiers, Algeria by Boukri et al., 2007 and contrary to
the usual vulnerability classes, D = the most vulnerable buildings (i.e. non-seismically built)
whereas A= the least vulnerable.

Figure 2-7: Vulnerability function relationship between pga, damage factor and vulnerability index, I
v

(Guagenti and Petrini (1989), adapted by Calvi et al (2006), Crowley (2009))

p
ppg
gga
aa (
((g
gg)
))
Table 2-1: The 11 Parameters used and the given weightings for the Boukri et al (2007) study of Algiers,
Algeria.
i Parameters KiA KiB KiC KiD Wi
1 Total Structural Resistance 0 5 20 45 1.5
2 Plan Configuration 0 5 25 45 0.5
3 Elevation Configuration 0 5 25 45 0.5
4 Wall connections 0 5 25 45 1
5 Wall type 0 5 25 45 0.25
6 Floor and Diaphragm 0 5 25 45 0.25
7 Roof type 0 15 25 45 0.25
8 Soil and Foundation conditions 0 5 25 45 0.75
9 Detailing 0 0 25 45 0.25
10 Maintenance 0 5 25 45 1
11 Modifications 0 5 25 45 0.5

In this study, with a value of 0-35, the buildings are classed as green and require no
intervention, 35-250 indicates orange, signifying possible future retrofitting, and 250-450
indicates red which require replacement and policy to be put in place to demolish these
buildings.

These methods have been also used also as part of the Risk-UE project which was undertaken
for seven European cities but an ATC-21 screening procedure (ATC, 1988) was undertaken to
define some parameters in conjunction with use of historic and construction data (www.risk-
ue.net). This methodology is easily adaptable to large-scale assessment of groups of buildings
but still requires expert opinion. It also requires the use of extensive field surveying which is
not available in many regions/countries. It is somewhat subjective and therefore is not exact,
and thus surveyors may have different ideas as to the definition of the building characteristics
without strict guidelines; therefore this gives discrepancies.

Continuous Vulnerability Curves are another such empirical method which has been used for
vulnerability assessment by directly utilising the probability of the damage of buildings to
earthquakes. These have been produced using ISTAT data (survey data from 1991 census
which includes building identification, damage, typology and damage) for the production of
seismic risk by taking DPMs to produce vulnerability curves in terms of spectral displacement
at the period of vibration (Crowley et al., 2009 via Rossetto and Elnashai, 2003).

Figure 2-8: The difference between PGA vs. spectral displacement (as a fn of building period vibration)
validated for the 1995 Aegeon Earthquake in Greece (Rossetto and Elshinai, 2003)
Similar vulnerability curves as those viewed above have also been derived using the Italian
ISTAT database by Colombi et al., 2008.
This method was undertaken in order to overcome the inaccuracies or continuous curves
based on PGA or macroseismic intensity converted from PSI (parameterless scale of
intensity) (Spence et al., 1992, Orsini, 1999). Vulnerability curves have also been derived by
Giovanizzi et al. (2006) corresponding to fragility curves after damage probabilities are
derived as shown below for a pre-code RC MRF and medium ductility class DCM for EC8.

Figure 2-9: Macroseismic Method for EC8 DCM V=0.36, Q=2.5 and med-rise pre-code RC MRF V=0.62,
Q=2.3 from left, vulnerability curves, damage probabilities, fragility curves for pre-code I=9,
ductility=2.25.
Similarly, fragility curves have been derived to translate ground motions into direct damage
levels for structure. King et al. (2005) have produced fragility functions for 22 different
ground motion intensity measures for various building types; the top 6 measures of which are
ranked for 5 building classes (W1 wooden residential, light frame, W2 wooden industrial
and commercial, S1 steel moment frames, C1 concrete moment resisting frame, C2
concrete frame with shear wall), mainly derived from data within the Northridge and Chi-Chi
Earthquakes. The variation of best intensity measures for use with building types is shown
with 20 different intensity measures represented in the 30 values presented. These earthquakes
were highly instrumented and thus data was available for buildings within a very close
distance to the strong motion recording stations. Thus, the data at these stations could be
presumed to be the same as that of the buildings and thus derived (undertaken within the
ATC-38 project). This was similarly done for the Chi-Chi Earthquake. Some different damage
states were used.
Table 2-2: Correlations between various ground motion intensity measures and building demand
parameters and percent loss estimates for various building classes modified from King et al. (2005) and
Stafford et al. (2007)
W1 W2 S1 C1 C2
Rank
IM IM IM IM IM
1 EPV 0.173 Tb -0.317 EPV 0.963 ASI 0.750 Tb -0.267
2 T90 -0.156 Sd 0.302 IDRmax 0.823 EPA 0.732 T90 -0.209
3 HI 0.143 Sv 0.297 PGV 0.799 AI 0.709 IMM 0.173
4 PGV 0.141 Td -0.282 R 0.766 RMS90 0.673 Sa -0.150
5 IMM 0.134 PGD -0.277 RMS 0.765 RMS 0.664 PGV 0.119
6 AI 0.134 T90 -0.249 Sv 0.738 RMSb 0.629 AI -0.113
Used Intensity Measures:- PGV Peak Ground Velocity, PGD Peak Ground Displacement, Td Total Record
Duration, T90 90% Cumulative Duration, Tb Bracketed Duration, RMS Root Mean Acceleration for Total
Duration, RMS90 Root Mean Acceleration for 90% Duration, RMSb Root Mean Acceleration for Bracketed
Duration, AI Arias Intensity, ASI Acceleration Spectral Intensity, EPA Effective Peak Acceleration, HI
Response Spectrum or Housner Intensity, MMI Modified Mercalli Intensity, IMM ShakeMap Instrumental
Intensity, R Roof Drift Ratio, IDRmax Maximum Interstory Drift Ratio, Sd Spectral Displacement at
Predominant Period, Sv Spectral Velocity at Predominant Period, Sa Spectral Acceleration at Predominant
Period

Screening Methods after earthquakes can also provide a method to determine which damage
state will occur based on correlated results. They can also be used pre-earthquake to
determine where retrofitting should occur. However, screening methods are generally a slow
process as they require individual building definition and expert judgement or training and
thus cause problems over a large number of buildings. Japanese vulnerability assessments
have particularly employed this method with the Japanese Seismic Index Method (JBDPA,
1990) which compared a seismic performance index (I
S
) with the seismic judgement index
(I
SO
); where I
S
is a function of basic structural performance (ultimate strength and ductility),
structural design (irregularity, stiffness and mass concentration) and time deterioration
(cracking, field investigation); and I
SO
is a function of the level of screenings (1-3) that have
been applied to the building, zone index (intensity at the building site), ground index
(amplification or topographic effects) and usage index (importance of the building).
Hence, if I
S
>I
SO
this indicates low vulnerability whereas where I
S
<I
SO
, depending on the
difference, further non-linear dynamic analysis is required, or if a large difference indicates
high vulnerability, demolition or retrofitting of the structure is required.
Rapid screening methods have also been utilised throughout Turkey, with a few methods
requiring the use of the dimensions of the walls, columns and floor area (Hassan and Sozen,
1997) as well as inconsistencies in the design and quality of workmanship (Yakut, 2004), yet
one of the most promising is the P25 approach by Bal et al. (2005). This method details 25
different structural procedures in order to assess the vulnerability of buildings. It is based on a
value-based system, and is to be used in conjunction with a large scale building assessment
method on the buildings that are deemed most likely to fail, thus providing a useful way of
undertaking a screening method. It is calibrated by detailed non-linear analyses but does
require entry to the building unlike the walk-down methods of FEMA 154-155 and the Simple
Screening Procedure of Sucuoglu et al. (2007). Some data were obtained from the 1989
Newcastle earthquake for over 6500 buildings from Geoscience Australia for possible use in
this thesis. Unfortunately, a walk-down method was used and it did not have the internal
parameters needed for displacement based analytical methods, and so was not used.
2.4.3 Analytical and Hybrid Methods of Vulnerability Assessment
Analytical methods are based on structural mechanics principles and are fast becoming the
preferred method of large-scale vulnerability assessment due to their proactive capacity, direct
correlation with damage and non-reliance on observed damage data. Analytical methods are
mainly based on non-linear analysis as this allows for stiffness degradation of existing
buildings to be taken into account. One of the analytical methods is analytically-derived
vulnerability and fragility curves and DPMs which can be produced by computational
intensive analyses rather than observed damage data to obtain the structural performance via a
given intensity measure. For European buildings, this has been undertaken in a variety of
studies such as Rossetto and Elnashai (2005), Dumova-Jovanovska (2004) and Masi (2004).

Figure 2-10: Production of Analytically derived Vulnerability Curves (Dumova-Jovanoska, 2004)
Singhal and Kiremidjian (1996) derived fragility curves and DPMs from Monte Carlo
simulation (random variation of material properties) for reinforced concrete frame structures
using a variety of ground motions with non-linear dynamic analysis (NTHA) in order to
produce the structural damage probabilities. The results for each of the nonlinear analyses was
done based on the Park and Ang (1985) damage index and then statistical analysis used in
order to produce the DPMs and fragility curves. The vulnerability curves were then updated
using Northridge data, essentially in the same way as the King et al. (2005) method, but also
using a weighting system (Singhal and Kiremidjian, 1998). Dumova-Jovanovska (2004) did
this for RC buildings in the Skopje region using a normal distribution for the damage
occurrence probability. A good description of the method to derive analytical fragility
functions is also detailed by them.
The Park and Ang (1985) damage index is the most used damage scale for analytical fragility
functions throughout these studies because it is the best known. However, there are many
other methods that are worth noting which are discussed within the OpenSees software.
Normalised (Absolute) Peak, the Mehanny-Deierlein (Mehanny and Deierlein 2001), the
Hysteretic Energy (Rahnama and Krawinkler, 1993) and the Kratzig (Kratzig et al., 1989) are the
methods discussed.
Masi (2004) derived his vulnerability curves using the EMS-98 scale, and NTHA with
synthetic and real accelerograms, utilising design code and handbook derived structural
models from Italian RC buildings from the 1970s onwards.
Rosetto and Elshinai (2005) used the damage scale derived from their 2003 paper and
produced adaptive pushover curves and thus via the capacity spectrum method the
performance point was defined to a damage state. ADRS of many ground motion records and
structural variability of buildings was modelled in order to create analytical displacement-
based vulnerability curves by the response surface method (Crowley et al. (2009)).
Choi et al. (2004) also derived vulnerability curves via combination of fragility curves for
various components of bridges and rather than using Monte-Carlo simulations, employed
FORM (First Order Reliability Method) to create bounds in order to increase the
computational efficiency of analytical vulnerability curves.
As described above, the computational time needed for analytical methods impacts upon their
usefulness for countries where there are many different construction types and characteristics,
and thus although not useful on their own, can be used to support empirical DPMs, fragility
and vulnerability curves via their use in hybrid DPM and vulnerability curve methods, which
use analytical methods to fill in the gaps in data within the damage band for certain intensity
levels where there is no empirical data in that location. This is therefore much faster than
analytical methods. Kappos et al. (1995, 1998) using the vulnerability index procedure in
combination with NLTHA and also Barbat et al. (1996) using weighted Monte Carlo
simulation combined with a vulnerability index method are two such methods employing
hybrid principles. However, as noted by Bommer and Crowley (2006), it is important both to
quantify variability accurately in these cases, as there are two sources of variability and to
ensure that the variability is not double-counted. In order to produce an accurate vulnerability
method, it is also important to be able to quantify this variability to each of the analytical and
empirical versions, calibrate both methods to the same value i.e. using the median (50
th

percentile) and hence define both sources of uncertainty.
Collapse based methods are discussed within the Calvi et al. (2006) ISET paper and have
recently been utilised by many different authors in order to derive a collapse multiplier from
which a damage distribution can be produced. According to Calvi et al. these methods have
mainly been done for masonry buildings by the use of mechanical concepts. The VULNUS
procedure by Bernardini et al. (1990), the FaMIVE procedure (DAyala and Sperenza, 2002)
and an approach by Cosenza et al. (2005) giving the uncertainties of the building material and
geometrical properties are detailed. For these collapse-based methods, no clear indication is
given as to how to derive the probability of exceeding a given limit state (with the exception
of VULNUS for the collapse limit state). Therefore, until this is quantified, loss assessment
models will most likely not incorporate these collapse-based methods.
The Capacity Spectrum Method is a method which is widely used within loss assessment
models due to its ability to relate the crossover point of capacity via a pushover curve and
demand via an ADRS to a given damage state. It was derived by Freeman et al. (1998) and is
implemented within HAZUS which has filtered through to many of the open source loss
assessment programs implemented HAZ-Taiwan (Yeh et al., 2000, 2006), Risk-UE
(Mouroux et al. 2004, 2006), EQRM (Robinson et al., 2005), SELENA (Molina and
Lindholm, 2005) and others. HAZUS (Hazard U.S.) was a project from the National Institute
of Building Science (NIBS), in cooperation with the Federal Emergency Management Agency
(FEMA), to develop a nationally applicable methodology for estimating the potential losses
from earthquakes on a regional basis (FEMA, 1999). The main modules of this methodology
are summarised in the diagram below from Kircher et al. (2006).

Figure 2-11: The Typical HAZUS Diagram which forms a basis for many ELE Engines, adapted from
Kircher et al., 2006
Essentially, the capacity spectrum method relies on an iteration method from the initial ADRS
(usually set at 5%) in order to relate it directly to the pushover curve to achieve the
performance point which defines the damage state taking into account both the equivalent
Indirect Economic Loss
Long-Term Effects and
Disruptions from Direct
non-linear damping and ductility (representing the horizontal displacement of the structure
under increasing horizontal loading). The iteration from FEMA-440 (ATC, 2005) shows the
ratio of area beneath the capacity curve (i.e. maximum strain energy) from the performance
point, to the total hysteresis loop area which is the energy dissipated by damping.

Figure 2-12: The Capacity Spectrum Method (FEMA-440, ATC, 2005)
As the ground motion increases (i.e. higher ground shaking), the amount of inelastic
deformation increases (i.e. larger displacements for a certain acceleration), and period
lengthening occurs for the structure. More ductile structures will have larger displacement
ductility associated with their capacity and stronger structures will be able to resist greater
accelerations for a certain displacement.
In the HAZUS methodology, from the displacement of the performance point, fragility curves
can be used in order to derive the probability of being in a particular damage band.

Figure 2-13: Cumulative P(DS|Sd or Sa) for weak, medium and strong shaking and thus the
corresponding probability of being in a particular damage band (Kircher et al., 2006)
The vulnerability assessment component of the procedure is contained within the direct
physical damage module and is based on the Capacity Spectrum Method. The capacity
response has been developed for many different building classes using model buildings
designed for different levels of design within the U.S.. The performance point obtained from
this average building provides the displacement input into the limit state vulnerability curves
to give the probability of being within a given damage band.

The vulnerability curves are calculated for Slight, Moderate, Extensive and Complete
structural damage states and Slight, Moderate, Extensive and Complete non-structural damage
states. Each vulnerability curve is characterised by mean and lognormal standard deviation
() values of PESH demand.

The standard deviation (
Sds
) combines the uncertainty of the damage threshold (
M(Sds)
), the
variability in the capacity (response) properties of the model building type (
C
) and the
uncertainty in the response due to the spatial variability of the ground motion demand (
D
).
[ ]

|
|
\
|
=
Sds d
d
Sds
d
S
S
S ds P
,
ln
1
(2-2)
where: Sds d S , is the median value of the spectral displacement at which the building reaches the threshold of the
damage state as created through a combination of test and field data, expert judgement and opinion, ds;
Sds
is
the SRSS combination of (
C
*
D
* Sds d S , ),(
M(Sds)
); is the standard normal cumulative distribution
function.

The median values of structural component fragility are based on drift ratios that describe the
threshold of damage states. These drift ratios are converted in spectral displacements through
the equation below:
h S
Sds R
Sds d =
2 ,
, (2-3)
where: Sds d S , is the median value of the spectral displacement, in inches, of structural components for the
damage state;
Sds R,
is the drift ratio at the threshold of structural damage state;
2
is the fraction of the
building (roof) height at the location of the pushover mode displacement; h- is the typical roof height, in inches,
of the building type of interest.

This second ratio can be calculated or it can be taken from the predefined tables in Chapter 5
of the HAZUS-MH document. Care must be taken if the HAZUS method is being used for a
full loss assessment, as the PSHA already takes into account the demand variability; and
therefore this must be removed in the calculation of the SRSS combination to calculate
Sds

by removing the
D
and Sds d S , values.

HAZUS is a very simple and hence useful and adaptive procedure. It has many simplifications
as it assumes the same capacity curve for a certain location. It is also difficult to adapt the
capacity curves to other locations in the world as the building classes have been derived for
limited height buildings in the U.S.. These buildings are also put into binned height classes
(Low, Moderate and High in most cases) and therefore the building capacity curves and
vulnerability functions are approximate. Thus, in order to adapt these buildings to other
locations in the world, a large amount of building information is required in order to carry out
a reliable non-linear static procedure such as those detailed above (DAP Antoniou and
Pinho (2004)).

Giovinazzi (2005) presents a displacement-based mechanical procedure to assess masonry
and RC frames by using a capacity curve which has been converted to a Sa-Sd plot. After
capacity spectra for the building classes have been created, the Capacity Spectrum method as
seen above, has been used; however, displacement thresholds derived from expert judgement
and prototype building analysis are used as well as a different method to model variability
from that of HAZUS. A binomial distribution has been used to model the damage distribution
in agreement with Braga et al. (1982), which is not necessarily correct due to the
simplification it assumes. This has been discussed within Calvi et al. (2006) in greater detail,
and Giovinazzi (2005) agrees that this has a fixed variability around the mean damage band.

Molina and Lindholm (2005) as part of their SELENA ELE software incorporate a logic tree
approach within their capacity spectrum methodology in order to reduce epistemic
uncertainty. This is further discussed within the SELENA methodology to be adapted.

Displacement-based methods have been produced recently due to their ability to relate to
damage states better than original force-based methods (as seen in Calvi (1999) in addition to
a spectral representation of earthquake demand) and proposed through Priestley (2003) as
damage is strain dependent. Thus, through strains, curvatures can be derived and subsequently
rotations and displacements.
The Displacement-Based Earthquake Loss Assessment method (DBELA) and also the
MeBaSe theory is based on a procedure proposed by Calvi (1999), which utilises the
principles of Direct Displacement-Based Design method (e.g. Priestley, 2003). It is also
presented in the work of Pinho et al. (2002) and is made into a fully probabilistic
methodology by Crowley et al. (2004). Within the method, Multi-Degree Of Freedom
(MDOF) buildings are modelled as equivalent linear Single-Degree Of Freedom (SDOF)
systems.

Figure 2-14: Simplified Model for an equivalent SDOF system from Calvi et al. (2006).
The displacement capacities of these SDOF systems are then compared with the demand from
a displacement response spectrum at effective response periods of vibration. Consequently,
the calculation of the period of vibration at different levels of damage is required and has been
derived for European buildings with and without infill panels by Crowley and Pinho (2004,
2006) deriving a yield period-height relation and subsequently by Bal et al. (2008a) for
masonry buildings. The damping of the structure is taken into account in the demand by
reducing the displacement response method with a damping reduction factor corresponding to
each of these damage levels. This reduction factor is a function of the predicted equivalent
viscous damping of the system at a given damage level, which in turn is a function of the
system ductility.

In this methodology, the different building classes are defined as a function of the assumed
response mechanism: beam-sway, more typically for reinforced concrete buildings that have
been designed to modern capacity design principles, and either beam or column-sway for
those which have not been seismically designed. A Sway Potential Index, which relates the
strength or stiffness of the beams and strength or stiffness of the columns, is used to help
identify the mechanism of a given building within a building class. The mechanism may be a
column-sway mechanism even if the building was seismically design and vice versa. Masonry
buildings are assumed to have a storey-sway response mechanism at the ground floor.

The sway potential index used to identify the sway mechanism of the reinforced concrete
buildings in the case of DBELA is the Stiffness Based Index given below.
c c
b b
L h
L h
/
/
(2-4)
-where h
b
and h
c
are the beam and column section depths and L
c
and L
b
are the column and beam lengths,
respectively.

The main advantage of the stiffness-based index is its ease of application and it has been
shown to be sufficiently reliable (Crowley et al. 2009). The stiffness-based index is assumed
to indicate a column-sway mechanism below 1.0 and a beam-sway mechanism above 1.2.
Normally, considering the fact that the beams carry negligible axial force, the cut-off for
column sway mechanism is shifted to a little higher than 1.0, representing the increase in
column strength and stiffness due to axial forces on the columns. However, for this
assessment, a somewhat artificially precise limit of 1.0 has been assumed to define the border
between the two mechanisms. This causes some issues if a distribution (lognormal) is to be
applied to produce a fragility curve for DBELA, as will be explained later.

After identifying the response mechanism of each building in the building class, the
calculations of the displacement capacities and periods of vibration for each damage state are
undertaken. The yield displacement capacity (
LS1
or
y
) is used to define the first limit state
while the post yield displacement capacities (
LS2
and
LS3
), which are obtained by adding the
plastic displacement components to the yield displacement, are used to define the second and
third limit states of damage. The first limit state of damage corresponds to the transition from
None/Slight to Moderate Damage, the second limit state corresponds to the transition from
Moderate to Extensive Damage and the third limit state corresponds to the transition from
Extensive Damage to Complete Damage. The formulae to compute the three displacement
capacities, as well as the corresponding yield and post-yield periods of vibration, are
presented in the following section.

The effective height for column sway cases is generally given as a function of storey height
but can also be used as a deterministic value of 0.5 depending on the study and is also a
function of storey height for beam sway of 0.64 ranging to 0.44 for a structure of greater than
20 storeys.

Non-structural components are calculated using inter-storey drifts. A shape factor, S, (a ratio
of effective height to roof displacement) and the limit state drift capacity of the buildings
drift-sensitive components (
i
) is used to calculate this non-structural damage.
NSLSi
= S
i
H
T
(2-5)
For further details, Crowley et al. (2004) can be consulted.

The following equations are from a simplified method for masonry buildings using rotation
capacities calculated from the limit state curvatures derived from Priestley (2003). Much work
has been done with respect to masonry buildings by Restrepo-Velez and Magenes (2004) and
Modena et al. (2005) that produced the Mechanical Based Procedure for the Seismic Risk
Estimation of URM buildings (MeBaSe). The MeBaSe uses the same limit state procedure
with adapted equations which take into account the deformed shapes seen below for masonry
structures.

Figure 2-15: Possible deformed shapes for the different limit states and in-plane failure modes (from Calvi
et al. (2006).
Out-of-plane failure mechanisms were also taken into account with the subsequent Australian
work of Doherty et al. (2002) and Griffith et al. (2003). For greater detail, the paper of Calvi
et al. (2006) should be referred to.
For Masonry buildings:-
The Storey-Sway Mechanism for the first limit state (yield) displacement capacity (
LS1
):
H k
y LS
=
1 1
(2-6)
The Limit state displacement capacity (
LSi
) Storey-sway mechanism is given as:-
( )
s y LS y LSi
h k H k + =
2 1
(2-7)
where
y
is the yield rotation capacity;
1
k is the effective height coefficient (to obtain the equivalent
height of the deformed MDOF system); H is the total height of the building;
LS
is the second or
third limit state rotation capacity;
s
h is the pier height;
2
k is effective coefficient of the masonry
piers.
Yield period (T
Y
):
87 . 0
062 . 0
T y
H T = (2-8)
Post yield period (T
LSi
):
LS y LSi
T T = (2-9)
where
T
H is the total height of the building;
LS
is the ductility at the limit state to be considered.
For Reinforced concrete buildings:-
The Storey-Sway Mechanism for the first limit state (yield) displacement capacity (
LS1
):
Beam sway mechanism
b
b
y T h
sway bm
Sy
h
l
H ef =

5 . 0
(2-10)
Column sway mechanism
c
s
y T h
sway cl
Sy
h
h
H ef =

43 . 0
(2-11)
The Limit state displacement capacity (
LSi
) Storey-sway mechanism is given as:-
Beam sway mechanism
( )
T h y LSi S LSi C
b
b
y T h
sway bm
SLSi
H ef
h
l
H ef + + =

7 . 1 5 . 0 5 . 0
) ( ) (
(2-12)
Column sway mechanism
( )
s y LSi S LSi C
c
s
y T h
sway cl
Sy
h
h
h
H ef + + =

7 . 1 5 . 0 43 . 0
) ( ) (
(2-13)
where
b
h and
b
l are the beam height and length;
h
ef is the effective height coefficient;
T
H is the
total height of the building;
c
h is the column length;
s
h is the storey height;
y
is the yield strain of the
reinforcement concrete;
) (SLi C
limit state concrete strains
) (SLi S
limit state steel strains.
Yield period (T
LS1
):
T y
H T 1 . 0 = (2-14)
(Crowley et al., 2004)
Post yield period (T
LSi
):
LS y LSi
T T = (2-15)
where the postyield stiffness has been neglected and where, as before,
T
H is the total height of the
building and
LS
is the ductility at the limit state to be considered.

Simplified pushover curves can also be generated using the DBELA methodology. The ratio
of the base shear force divided by the seismic height of the building can be calculated directly
from the yield displacement capacity (
y
) using the following formula:
2
2
4
y
y
gT
=

(2-16)

For both the DBELA and MeBaSe methodologies, a large amount of information is required
about the structure which cannot be gathered from a walk-down type exposure survey. Beam
and column dimensions are required for the RC buildings, and various wall shear strength
parameters for the MeBaSe methodology, and thus to produce the probabilistic density
function of each of the geometrical parameters a sample of buildings is required. Statistical
data on the material properties is derived from regional laboratory tests on reinforcing bars
and concrete cubes used in the building types sampled. Then, a Monte Carlo simulation can
be produced in order to create buildings based on a mean and standard deviation of the
sampled parameters. From this equations can be applied to determine the sway mechanism
and calculate the yield and limit-state periods of each building resulting finally in a relation to
damage by comparison of the equivalent SDOF bilinear capacity curve with the demand to
see which limit state results for each of the buildings via the performance point. The system
demand needs to be reduced to take into account the energy dissipation that occurs through
the cyclic response via hysteretic damping. Thus, this can be done by iterating through the
equivalent viscous damping of the MDOF structure at a certain level of ductility using the
Priestley et al. (2007) formula.
|
|
\
|
+ =
1
565 . 0 05 . 0
eq
(2-17)
Modification factors are then applied to the spectra via the following CEN (2005) formula.
+
=
5
10
LS
(2-18)

Figure 2-16: Comparing ADRS for various limit states in order to determine the associated damage (Borzi
et al., 2008)
Therefore, for the case above, the limit state would be the LS2 (Extensive damage) as the
third criterion is not satisfied.

If increasing spectral accelerations are applied to the structures after this process, total
vulnerability curves can be produced to show the uncertainty in the displacement capacity of
the building type examined.

The SP-BELA approach of Borzi et al. (2008) is based on the DBELA approach, and it uses a
prototype structure which is a usual Mediterranean RC MRF structure in that region to define
simple bilinear capacity curves.

Figure 2-17: Plan view of a prototype RC MRF Building as representative of a certain region (Borzi et al.
(2008)
Span lengths, storey heights and loads are based on typical Mediterranean values and the
prototype structure is created using code-based principles from which the flexural and shear
capacity of the sections can be calculated. The method then tests the column-sway mechanism
based on the shear capacity, noting that for each column of the frame the smallest shear
capacity of the following governs: 1)The shear capacity of the column; 2) The shear
corresponding to the flexural capacity of the column; and 3) The shear corresponding to the
flexural capacity of the beams supported by the column. The collapse multiplier for the
capacity curve is hence calculated using the following formula.
(2-19)
where:- W
T
is the global building weight, and W
i
is the weight associated with floor i located at height
z
i
, and V
C
i
is the lowest shear capacity on that floor.

The sway potential index and displacement capacities can therefore be calculated in much the
same way as shown above in the DBELA approach or by other approaches as shown in the
Borzi paper. The SPBELA approach, however, is not yet directly applicable to full scale loss
assessment yet.

In addition to the difficulty of data retrieval, there are many uncertainties in the DBELA
procedure relating to the difference in displacement of the MDOF structure as the period-
height relationship is fixed regardless of lateral strength and therefore this causes some errors
for different seismically designed buildings with other lateral strengths, therefore this must be
changed in future versions of DBELA. Of course, the problem of the limit state response
period is also a problem within SPBELA.

In DBELA and SPBELA it is necessary to define a beam or column sway mechanism and the
cutoff when used for implementation is very rigid, thus producing very different displacement
capacities for definition with fragility curves between the two different sway mechanisms.
Some other such uncertainties in both versions is the way that the equivalent viscous damping
is calculated, higher mode effects for taller structures, the way that the spectrum reduction
factor is estimated and the expected failure mechanism.
The uncertainties of the structural characteristics of the buildings are taken into account in the
pdfs, however by comparing the predictions with NSPs on the structures, the uncertainty can
be calculated via the following equation.
n LS LS
+ = ) ln( ) ' ln( (2-20)
2.5 Hazard
2.5.1 Potential Hazards analysed in Loss Models
Identification of possible hazards in addition to ground shaking has been undertaken by Bird
and Bommer (2004) as stated in the introduction and the impact to building damage of these
potential hazards has been found to be much less than the ground shaking due to earthquakes
(ground failure effects such as liquefaction, fault rupture, landslide, slope stability and bearing
capacity; tsunamis and seiche). However, these secondary effects cause a lot of damage to
lifelines (Bommer et al., 2006). There are various methods that have been employed to
analyse these secondary effects and these will be briefly summarised referencing the work of
Stafford et al. (2007), Foray and Bard (2008) and Stewart (2008).
(a) Ground Failure
Liquefaction involves the changing of soils from solid to liquid state and is usually caused by
induced cyclic shear. It is thus energy related and so is extremely complicated to incorporate
into loss models. There are many uncertainties over the area and extent of liquefaction but by
using simplified methods such as relationships between PGA and susceptibility such as that in
HAZUS by Liao et al. (1988), the strain potential index of Seed et al. (2001), the
displacement-based model where ground deformation is related to settlements (Ishihara,
1993) or horizontal displacement (Rauch and Martin (2000), Faris et al. (2006)), liquefaction
can be applied into loss models. Cetin et al. (2002) and Moss et al. (2006) allow for
determination of a factor of safety against liquefaction which can be applied to performance-
based approaches (Kramer et al., 2006, Foray, pers. comm., 2008) and probabilistic
approaches have been applied to loss models (Bird et al., 2006, Crowley et al., 2006).

Various intensity measures have been used for liquefaction potential with varying success.
PGA (Youd et al., 2001) and Arias Intensity (Kayen and Mitchell, 1997) are not as relevant
for performance-based engineering as they do not correspond directly to building damage.
Thus, Kramer and Mitchell (2006) looked at CAV
5
(cumulative absolute velocity above
0.005g) and this corresponded better with liquefaction potential and could be used for ELE
models.

Bearing Capacity failures using the methods of Richards et al. (1993) or Kumar et al. (2003)
can be used in order to determine the damage of structures due to loss of bearing capacity, but
currently there are no loss estimation models that take this into account (in the public sector).

Fault Rupture causes very localised effect (usually within a 200m width of a fault trace
Cotton, pers. comm., 2008) and thus is usually not incorporated into the ELE assessment. If
such an analysis needs to be carried out, that of Todorovska and Trifunac (2006) can be
applied. HAZUS simply sets up the possibility of application by predicting the maximum
displacement location and lowest displacement locations.

Landslides and Slope Stability are more difficult parameters to constrain because of the
need to determine the rainfall that has occurred in the area before the earthquake to have an
idea of possible landslides as they can also be rainfall-induced. Most of the analysis methods
include a simple ratio between PGA and the factor of safety (FS) is based on a critical
acceleration for the slide mass (FEMA, 2003). Many methods have been established including
a probabilistic framework by Del Gaudio et al. (2003) producing damage functions for
structures based on the movement and probability of slope failure. GIS elevation models can
simply identify susceptible areas to landslides, and this combined with an intensity measure
approach (Wilson, 1993) may be the best method for application.

Tsunamis (earthquake-induced waves) and Seiche (standing wave induced phenomena) have
increased in importance since the Boxing Day Tsunami of 2004 inducing many methods to be
introduced (eg. Geist and Parsons, 2006, Masson et al., 2006, NOAA Website, 2009).
Although important, this type of secondary effect should be generally applied at a rapid-
response level, due to the unknown nature of sea-floor bathymetry with undersea quakes and
the relative lack of knowledge in the area worldwide, given the depth and uncertainty of such
phenomena.
2.5.2 Methods of Seismic Hazard Assessment
There are two main methods of seismic hazard assessment: ones which are deterministic
(DSHA) and that include a single scenario earthquake (historical, MCE or user-defined); or a
probabilistic combination of earthquake scenarios in order to determine the hazard for the
given area (PSHA).
For a seismic hazard assessment, three components are required: recurrence relations
(magnitude function); source zones (background, fault and area); and earthquake catalogues
(historic and stochastic). The recurrence relationship comes about as a probabilistic result of
the minimum and maximum earthquake possible from the earthquake catalogue for the given
source to produce a probability density function giving the ARE of different magnitude
earthquakes. Earthquake catalogues are extremely important in hazard assessment and detail
the magnitude and spatial position of previous recorded earthquakes (from approx. 1875 via
accelerometers/seismometers) see 3.3. Source zones are the spatial regions where future
earthquakes are expected to occur, defined by tectonics, geology and observed seismicity.
Source, path and site effects calculated via GMPEs and local site conditions define the ground
motion field away from these sources.
A deterministic seismic hazard assessment (DSHA) consists of 3 main steps and has been
carried out for many locations where a complete worst case scenario or historical repetitive
earthquake is waiting to be modelled:
1) Define all the possible sources to cause significant hazard at a site from historic tectonic,
geologic or geotechnical data.
2) Choose a fixed distance, fixed magnitude earthquake and place it on the closest position to
the site on each source (this could be MCE (USGS, 1996), defined via empirical equations
(magnitude to fault dimensions) on the basis of geological evidence using Wells and
Coppersmith (1994) or Manighetti et al. (2007) or by just adding 0.5 magnitude units to the
largest historical earthquake (recently shown by geothermal gradients Kudo et al., 2009).
3) Estimate ground motions via GMPEs to determine the ground motions at the site in terms
of spectral ordinates. Variability can be modelled for the ground motions within a DSHA,
however, a common way is to use motions which are one logarithmic standard deviation
above the logarithmic mean (84
th
percentile motions) (Strasser et al., 2008). Each of these
ground motions corresponding to each source is considered for design.
DSHA is very useful for lifeline and critical facility locations and is increasingly being used
to supplement a PSHA.

A probabilistic seismic hazard assessment (PSHA) considers all M-D- combinations by
taking into account all probabilities and scenarios possible for magnitude and distance to
calculate the hazard. The steps involved are adapted from Akkar and Boore (2008) and
Crowley et al. (2009):
1) Define a probability of potential rupture locations for each source.
2) Determine of the temporal distribution via recurrence relationships. The Gutenberg-Richter
(Gutenberg and Richter, 1944) relationship is commonly used where N
m
is the mean annual
rate of exceedance of magnitude M, b is the activity parameter expressing likelihood of large
and small earthquakes, and, a, describes the yearly rate in logarithmic space of earthquakes
(M>0) i.e. the y-intercept,
bM a N
m
= ) log( (2-21)
,but other relationships adapt this to calculate a characteristic magnitude and thus describe
truncated normal and lognormal, exponential, uniform and Youngs and Coppersmith
Characteristic Equation and Delta magnitude recurrence relations (Akkar and Boore, 2008).
3) GMPEs are used for the range of distances for each magnitude to produce spectral
ordinates dependent on the tectonic regime with aleatory variability, (interevent and
intraevent) of each relationship taken into account as well as the applied variability,
(interevent and intraevent).
4) The hazard must then be integrated by combining the effects of different size, location,
source zone and occurrence probability earthquakes in order to calculate the expected number
of exceedance of ground motions due to the PDF of magnitude, distance between source and
site and also the probability calculating for spectral ordinate values away from the mean
value. From this, annual rates of occurrence are derived giving a hazard curve.
A PSHA assumption commonly made is the Poissonian model that takes the annual frequency
of exceedance from this analysis and assumes that each earthquake is independent of other
earthquakes, where q(z) is the probability of exceedance of a user-defined ground motion
level for a given time, t in years, where (z) is the annual rate of exceedance is thus:-

t
e z q

= 1 ) ( (2-22)
From this a 10% probability in 50 years gives the 1 in 475 year return period for this
earthquake (Campillo, 2007).
2.5.3 Modelling of Uncertainty
The PSHA results from the end of an analysis can next be disaggregated for a given
variability, distance, and magnitude triplet to perceive the hazard that contributes most for a
given ground motion parameter at the site for a given ARE. Thus, it can be seen which
earthquake scenarios contribute most to hazard (McGuire, 1995) and this can be repeated for
probabilistic risk estimates as done by Hong et al. (2006). Reduction in loss as a result of
retrofitting can then be calculated or uncertainty can be quantified via the epistemic
uncertainty and aleatory variability.
Epistemic uncertainty is the uncertainty from incomplete knowledge of earthquake process
(lack of information) and can be reduced through use of logic tree approaches (Molina and
Lindholm, 2005, Crowley et al., 2005). Bommer et al. (2005) also investigates logic trees but
in a ground motion sense.
For hazard analysis there are numerous decisions affecting the assessment, which then need to
be expertly weighted through the logic tree in order to produce a hazard curve with less
epistemic uncertainty. These have been summarised within Crowley et al. (2009) and by
Scherbaum (2008) and are presented below.
GMPEs - Even when the same dataset is used for GMPEs in the NGA project (2008),
there is variation due to different assumptions made. Thus weighting is very
important.
The maximum magnitude for every seismic source zone and the magnitude-recurrence
relationship
The earthquake catalogue completeness or stochastic nature and therefore the b-value
within the Gutenberg-Richter relation
The recurrence interval of characteristic earthquakes (Poissonian/time-dependent)
Boundaries and determination of seismic zones

Aleatory variability is the random variability as previously explained through intra and inter-
event variability which is simply measured by distribution to integrate into hazard
calculations. Inter-event variability is generally due to ground motion differences caused by
style-of faulting, directivity (Somerville, 2003), finite vs. point source theory, azimuthal
radiation variation, hanging-wall effects and stress drop/slip characteristics for a particular
magnitude earthquake. Intra-event variability is caused by the fact that path and site
conditions are different for every site with the same site classification and distance within an
earthquake. It is generally taken into account through the GMPEs and is the square root of the
sum of the squares. In terms of site and geotechnical conditions, the Vs30 brackets defined by
NEHRP are extremely broad and even the assumptions within a 30 metre shear wave velocity
cause aleatory variability when explicitly calculated. This is the same for path effects in terms
of variation of Q value with complex Earth effects. Thus amplification and deamplification
due to site effects can never be fully taken into account. For loss estimation, total variability is
often not appropriate (Crowley et al., 2009) and should be separated. Monte-Carlo simulation
can be applied in order to produce combinations of these variabilities to create an entire
ground motion field with different properties. Care must be taken that spatial and temporal
correlations are taken into account when ground motion fields are produced.
Macrospatial correlation occurs between ground motion pairs of sites as a result of their
proximity. This has been modelled by Boore et al. (2003), McGuire (1998) and Wang and
Takeda (2005) in various ways but is measured as a covariance matrix that can therefore be
applied to a given ground motion field between pairs of sites using the generalised formula of
Boore et al. (2003):-
( )
0
/ ) , ( r y x r
e

= (2-23)
where r is the distance between the site centres (x and y) and r
0
is the correlation length which
can be assumed to be a certain distance (5km for example).
A comparison has been undertaken in Crowley et al. (2009) showing similar results for the
three papers.

Figure 2-18: Macrospatial Correlation of Ground Motion for 3 different models (Crowley et al., 2009)
Inter-period correlation should also be taken into account as it has been shown for a certain
record that the epsilon changes from period-to period with a certain relationship. This was
undertaken by Baker and Cornell (2005, 2006) who present a correlation between pairs of
periods.

Figure 2-19: Interperiod correlation (Temporal correlation) of ground motions from Baker and Cornell
(2006).
Both of these methods (spatial and temporal) have been combined, moving on from their
work on probabilistic disaggregation to look at correlation coefficients for spatio-temporal
correlation coefficients by Goda and Hong (2008).
2.5.4 Other Ground Motion Issues for ELEs
In all cases, it is extremely important to apply the correct GMPEs to a particular location of
interest and that the desired magnitudes, locations and tectonic regime are accurate to the site.
Many ELE softwares give in-depth details as to the hazard but inexperienced users are able to
make errors. Thus, applicability sections should be presented for any new ELE software.
Spectral ordinates have been shown to relate better to building damage and therefore GMPEs
that allow a full acceleration response spectrum or displacement response spectrum should be
used in order to model possible ground motions at various frequencies.
The ability of ShakeMaps to take rapid ground motions via instrumental recordings and apply
fast empirical functions and kriging techniques can provide a first view of the shaking
intensity after an earthquake and allows for quick continuous representations to be produced
(Wald et al., 2005). It should be noted that these ShakeMaps have a large amount of error
associated with them.
The type of ground motion that is examined at the site can impact on the conversion to loss
estimates, as PGV (Akkar and Bommer (2007)) and Sa(1s) (Wu et al., 2004) has been shown
to correlate very well with deformation (damage). Other authors have investigated CAV and
Arias as energy based intensity measures (eg. Cabanas et al., 1997). As has been shown in the
vulnerability section, spectral ordinates are required in order to relate the frequency of ground
motion to that of the infrastructure, and no other intensity measures predict this as well for
ground shaking (MMI, JMA, MSK, EMS, PGA, PGV, PGD, Arias, CAV etc.).
Stochastic Catalogues are simply produced to fill in the temporal and spatial gaps within
historic earthquake records due to the lack of information available. This is done by simply
looking at minimum and maximum of certain magnitude earthquakes at a certain area, and
applying random generation techniques to give random numbers for a very long period
(100,000 or 1,000,000 years), and thus the ARE can be returned. The ground shaking can then
be applied to each earthquake in the stochastic catalogue in order to add intra-event and inter-
event variability and a GMPE predicts the logarithmic mean value. The predicted GM as a
result can then be produced. The result is produced by simply applying a historical
earthquake for a certain annual rate of occurrence and applying it along strike-slip sequential
faults in order to remove spatial incompleteness (Bommer, 2002). They have also been
produced by Monte Carlo methods (Musson, 1999) and this has been applied for areas of the
world where there is little data (Windeler et al., 2004).
The use of stochastic catalogues can therefore eliminate the shortcoming of PSHA in that
PSHAs treat hazard at all sites as fully correlated and at extremely high AREs the ground
motions are much higher (more than 3 standard deviations) greater than the median. This is
not the case in reality as for these very high magnitude earthquakes, as the ground motion
field will still be lognormally distributed. Thus PSHA produces higher losses for a low ARE
and lower losses for high AREs. In order to overcome this, PSHA ground motions must be
produced for all locations, due to every earthquake event and the associated losses which is
computationally expensive (Crowley et al., 2009).
This effect has been explored by Crowley and Bommer (2006) by applying random ground
motion fields for multiple earthquake scenarios in the northern Marmara sea for a given loss
model where Mean Damage Ratio is the ratio between repair and replacement cost.

Figure 2-20: Loss curves using the sum of three sites for PSHA and Stochastic Modelling (Crowley and
Bommer, 2006)
Thus, it can be seen that the theory matches the result and for reinsurance purposes (and also
for accuracy in ELE software), stochastic catalogues should be examined.
2.6 Damage Loss Conversion, Economic and Social Costs
By convolving the impacts of hazard, vulnerability and exposure, the conversion into a
damage loss and specific cost in terms of economic and social cost can be applied. These are
generally applied directly from the building damage classes: however, there are some broader
implications that should be first discussed.

Figure 2-21:- The processes contributing to earthquake loss from Khater et al. (2003)
2.6.1 Social and Economic Vulnerability
Increased social vulnerability can be looked at in advance, once danger has been identified
such as currently in Adelaide, Australia for a fictitious case. By using an in-depth assessment
technique such as DBELA and then applying it to Adelaide, to find out which types of houses
are most susceptible, people can be warned, prepared and educated. In some cases, the
government will undertake screening methods and then retrofitting methods (Daniell and
Luey, 2006, Oehlers, 2003). This has been found to work in advanced nations (Japan,
Australia and those below), but in developing nations this can be a problem and so the results
of ELE software need to work out some sorts of social vulnerability function based on the
nation in terms of building practice and not directly to damage. PAGER and QLARM are
currently applying formulas and some authors have attempted to take such demographic
issues into account such as Tsai et al. (2001) who look at age and fatality rate during the Chi-
Chi earthquake and find a clear correlation that with increasing age, an increasing fatality rate
from earthquakes occurs. This could be because elderly people may stay in the same house for
40-50 years and thus, the house is not as seismically protected as modern housing. By
including demographics data such as the current RiskSCAPE earthquake software, a more
accurate social loss model can be applied for each country in a holistic approach. Night and
day data (Yamazaki et al., 1996) is also extremely important for location-dependent social
losses. If business, industrial and residential data is inputted as part of the inventory, a quick
calculation can be made as businesses will have more people during the day versus residential
having more people during the night.

Advanced economies (dark-grey), Emerging and developing economies not least developed (light-grey)
Emerging and developing economies least developed (medium grey)
Figure 2-22: Developed vs. Developing Economies (UNDP, 2008)
It can generally be assumed that within developed countries, the seismic codes employed will
be more up to date than those of developing countries. Therefore, it can be assumed that less
damage will occur in these countries for an earthquake striking a high population area. It can
also be seen through the various earthquakes from 1950 onwards, that the total economic
impact per person killed is greatest in the advanced economies and there are a greater number
of people killed per dollar value lost in the least advanced economy countries (Masi, pers.
comm., 2009).
In most cases, if an earthquake strikes a location of high population density within these
countries, such as in the city of Kobe in Japan in 1995, it can be expected that the economic
loss will be high, with a large number of deaths and with this value of deaths increasing with
decreasing development (NATHAN, 2008). The GNP considerations have been considered in
the introduction.

Figure 2-23: Population density throughout the world (dark areas):- this can define locations of greatest
exposure (NASA visible earth, 2008)

Figure 2-24: Population growth in terms of % increase per year (-0 indicates a decrease in population)
(UNDP, 2008)
It can be seen that by correlating the population growth to the locations of population density
and also those of the development index, the increasing exposure of a nation to earthquakes
can be observed. Assuming the current population level increase, people will have to build in
locations of higher risk or with increasing speed and this will cause problems in the future.
This is the reason why China, India, Iran and Turkey have previously had high social losses
for the earthquakes that have occurred whereas locations like the U.S., NZ and Japan have
had lesser social losses for a similar amount of exposure.
2.6.2 Social Costs
Coburn and Spence (2002) is one of the most referenced social loss methodologies. Their
methodologies are based on much data over the past 100 years and show the predominance of
masonry buildings (as this is the most common building technique in the world) but, in
addition, also the influence of RC building in earthquake prone areas, as the number of deaths
from 1900-1949 was much less. The corresponding improvement in fire practices can also be
seen.

Figure 2-25: Resulting deaths from earthquakes from 1950-1999 (Coburn and Spence, 2002)
If this plot were to include the figures from earthquakes from 2000 to 2008, the other causes
would be much larger due to the number of deaths from the Boxing Day Tsunami. Coburn
and Spence (2002) also plot the predominant building type for the number of people killed vs.
the number of buildings heavily damaged and find a good correlation, except for the larger
number of deaths per collapsed RC building in which there is a much higher figure. This
shows that where collapse occurs, RC buildings kill more people than other types of
buildings.
Many current social loss models only take into account injury and death as a result of the total
collapse and heavily damaged structures and do not take into account the fact that even a light
or partially damaged house can cause injury (Murakami et al., 2004). For example, there are
infill wall failures etc.
The human casualty estimation of Coburn and Spence (2002) does not take into account this
fact and thus is based only on the percentage of collapsed buildings in the completely
damaged state (D
5
) where the number of deaths (K
S
) is a convolution of D
5
, with average
people in each collapsed building (M1), * percentage of occupants indoors at time of shaking
(M2), expected trapped occupants (M3), mortality at collapse (M4) and mortality post-
collapse (M5).
4 5 4 3 2 1 5
* ) * ) 1 (( * * * * M M M M M M D K
S
= (2-24)
The updated model of Spence (2007) as part of the LESSLOSS project is being currently
developed for masonry, RC, adobe and timber buildings in order to calculate the number of
deaths and 5 injury levels for all damage states. Only the complete damage state values have
been currently presented.

Figure 2-26: Updated method for injury distributions and hence social losses (Spence, 2007)
These social losses are all as a result of direct damage i.e. as a result of the hazard on the
building. Losses due to secondary effects are generally not included in ELEs due to the
complexities involved with the original vulnerability and hazard calculations. MAEviz (MAE,
2009) contains many details as to more complex social functions which have been collated
from the literature on the subject in the last 20 years.
2.6.3 Economic Costs
Economic costs are generally classified into two different sectors as a result of earthquakes.
Direct economic costs come from the direct damage (i.e. the impact of the hazard on the
infrastructure (ECLAC, 2003)) and are generally involved with the Mean Damage Ratio, i.e.
that is repair and reconstruction of the building stock to its original pre-event value, including
non-structural elements. This can be calculated via claims rates following earthquakes, cost of
retrofitting or post-earthquake loans given by governments, among other methods. Indirect
economic losses are those resulting from business downtime, disruption and the costs
involved in rescue and humanitarian efforts.
These direct economic losses have been estimated using damage ratios and a mean ground
floor area by Bal et al. (2008b) and FEMA (2003). The building plan area is generally
assumed to remain constant throughout the height of the building in order to assist with ease
of economic calculations. Approximate unit construction costs for new buildings in Turkey
found in Bal et al. (2007), for the U.S. in FEMA (2003) and Australia (Robinson et al., 2005)
can be seen and similar calculations are done for each ELE software, from which an estimate
of the direct economic loss resulting from damage to the building stock can be obtained.
Bal et al. (2007) have shown the dramatic difference between the mean damage ratios of
HAZUS, produced for American building stocks, and those which are proposed for Turkish
building stocks (Figure 2-27). Accordingly, they highlight the need to use damage ratios that
are country specific.
0
20
40
60
80
100
120
Sl i ght Moderate Severe Col lapse
D
a
m
a
g
e

r
a
t
i
o
Bal et al .
HAZUS

Figure 2-27: Comparison of Mean Damage Ratios for HAZUS and those suggested by Bal et al. (2007).
As mentioned, the cost of repairing (
rep
k j i
C
, ,
) is a proportion of the cost of replacement
according to Bal et al. (2007).
This cost is impacted upon by the Turkish Government that requires that severe and collapsed
buildings to be demolished and rebuilt. In all cases, the cost of demolition as discussed in Bal
et al. (2007), means that the MDR is greater than 100%. Further information on transportation
of rubble and demolition in Turkish conditions is discussed in this paper. This is not
transparently taken into account in the HAZUS values.
For each damage state (k), for each site or geocell (j) and for each building type (i), the cost of
replacement of the damaged buildings (
repl
k j i
K
, ,
) has been calculated using the equation below.
k u
repl
k j i
N n c A K =
, ,
(2-25)
where A is the ground floor area assumed independently on the building type, c
u
is the approximate
unit construction cost for new buildings, n is the number of storeys, and N
k
the number of buildings
that fall in the damage state k for each geocell and building type.
The Mean Damage Ratio (MDR) is an indicator of the total average loss divided by the value
of the building stock, i.e. represents the ratio between the cost of repairing and the
construction cost.
=
k
repl
k j i
k
rep
k j i
j i
K
C
MDR
, ,
, ,
,
(2-26)
The MDR can be disaggregated and thus the combination of hazard, vulnerability and
exposure which causes the highest economic loss can be calculated. Further recommendations
are discussed in the following chapter.

MAEviz (MAE, 2009) contains many details as to more complex social functions which have
been collated from the literature on the subject in the last 20 years.
2.7 Conclusion
An attempt has been made to search for all possible earthquake software programs that
correspond to the various methods that have been discussed above. Now the components of
risk have been identified, each of the modules; exposure; vulnerability; hazard and specific
cost can be discussed and examined for each of the ELE software packages selected. From
this, a state-of-the-art procedure to produce an open source ELE software package will be able
to be presented.

Chapter 3. Current ELE Software Packages Available

3. CURRENT ELE SOFTWARE PACKAGES AVAILABLE
3.1 Open Source vs. Closed Source
Earthquake Loss Estimation software packages are classed into two different types open
source and proprietary (closed source). The distinction is that open source software packages
are available to any user, and allow for freedom of information, whereas closed source are not
available to the public.

Open source is universal and allows for free exchange of data with benefits for the entire
world. Earthquakes are a problem that every point on the worlds surface must contend with
and thus the information to be able to protect against or to ascertain what effects could occur
at a certain point should be freely available to every person. Thus, many people have set up
earthquake loss estimation procedures that are open source. During this project, however, it
has been found that many are not freely available and that no one model has been tested for
the entire earth, with perhaps the exceptions of the PAGER and QLARM rapid response ELE
software. For a model to be truly open source, the internet should really be utilised to provide
source codes, documentation and transparency.

A lot of these open source-type packages are in programming languages such as Matlab
which is not freely available. However, there are open-source alternatives such as Octave
which can open Matlab files and is considered open source. Therefore, by avoiding many in-
built functions and by inlining coding, Octave can be used (gnu.octave.org).

GEM (Global Earthquake Model www.globalquakemodel.org) is one such worldwide
initiative which is attempting to produce such a risk analysis model that is adaptable to
changes in technology. It will most likely contain a combination of existing empirical-based
approaches and performance-based design principles to provide users with the option of a
faster, less-accurate analysis technique or a slightly slower, more-accurate analysis technique.
Adapted and added to by Crowley (2009), a state-of-the-art ELE modelling software is
required to have the following aspects:-
Recent developments in the field of SHA should be incorporated into the ELE
A probabilistic and deterministic model explicitly looking at all sources of uncertainty
Achieve the computational purpose desired
1) for rapid response speed vs. accuracy
2) for pre-earthquake computational intensity vs. detail of data vs.
confidence in results
3) for post-earthquake good comparison with results vs. amount of data
required vs. computational speed.
Easily adaptable worldwide to the different social, economic, environmental and
infrastructure construction practices encountered.
Dynamic modelling able to be changed with future information (new construction
types, retrofitting practices, further earthquakes, further developments in technology)
Allow for multiple hybrid modelling testing different vulnerability methodologies
(displacement vs. force based, empirical vs. analytical), spatio-temporal correlations
vs. uncorrelated ground motions, various platforms.
Open source to the largest extent possible.

Human Knowledge belongs to the world - Anonymous

Most closed source ELE packages without information were not reviewed within this study
and were discarded. However, some information is on the attached DVD for closed source
systems.
3.2 Overview of Worldwide Earthquake Loss Estimation Packages

With the speed with which ELE Software Packages are progressing, during this year-long
project of data collection (April 2008-April 2009) a dynamic model is required in order to
record changes for use within GEM. The data and up-to-date versions of software collected
and much documentation are available on the attached DVD.

Table 3-1: Overview of the ELE software packages selected for analysis
ELE Software ELE Past Influences Region Owner
CAPRA n/f Central America EIRD
CATS n/f Worldwide*/U.S. DTRI, FEMA
DBELA n/f Worldwide EUCENTRE
ELER HAZUS, SELENA, DBELA Europe JRA-3, NERIES
EmerGeo NHEMATIS, HAZUS Worldwide EmerGeo
EPEDAT n/f North America
EQE International, California
OES
EQRM HAZUS Australasia Geoscience Australia
EQSIM HAZUS Europe Karlsruhe University (KIT)
Extremum n/f Worldwide
Extreme Situations Res. Ctr.
Ltd.
HAZ-Taiwan HAZUS Asia National Science Council
HAZUS-MH n/f North America FEMA, NIBS
InLET HAZUS North America ImageCat Inc.
LNECLOSS n/f Europe LNEC
MAEViz HAZUS North America Uni. Illinois
OPENRISK OpenSHA-aided Worldwide AGORA, USGS, OpenSHA
OSRE MIRISK Worldwide Kyoto University, AGORA
PAGER USGS ShakeMap Worldwide USGS, FEMA
QLARM QUAKELOSS(2), Extremum Worldwide WAPMERR
RADIUS n/f Worldwide
Geohazards International,
IDNDR
REDARS n/f North America MCEER, FHWA
RiskScape n/f Australasia NIWA, GNS
ROVER-SAT
Includes ROVER,
ShakeCAST and ATC-20i
U.S. University of Boulder
SAFER
SELENA, EQSIM,
QUAKELOSS
Worldwide 23 worldwide institutions
SELENA HAZUS Europe/Worldwide NORSAR
SES2002 &
ESCENARIS
n/f Europe DGPC, Spain
SIGE n/f Europe OSSN, Italy
SP-BELA DBELA Europe EUCENTRE
StrucLoss KOERILoss, HAZUS Europe Gebze IT, Turkey

A short overview summarising the key points and interesting differences and similarities of
each of the software packages will now be detailed.
3.2.1 CAPRA
CAPRA (Central American Probabilistic Risk Assessment) is an initiative of the EIRD
between seven Central American countries (Nicaragua, Costa Rica, Honduras, Guatemala, El
Salvador, Belize and Panama) under a GFDRR Grant, in order to produce a region specific
Earthquake Loss Estimation model using a common methodology and giving tools via a
disaster risk information platform. It is based in Web 2.0 format and is located at
www.ecapra.org. It takes into account different natural hazards by probabilistic risk
evaluation using the usual platform combining GIS data through a Google Map Viewer-type
platform. It is envisaged to be running currently, and to be entirely open source. However, it
is under construction and currently not in use , but the source code will be modular, extensible
and open and will have an API with a first release date for beta version testing to take place in
the European summer of 2009 (Anderson, pers. comm., 2009). CAPRA also is employing
remote sensing data via ground truth calibration for their exposure data. It has Wiki, tools and
Map Viewer applications under headings of Hazard, Exposure, Countries and Risk, and
provides some current examples via the Examples heading. It is not just set up for
earthquakes and therefore the topographic information for floods and other hazards may be
the delay associated with this project. ILWIS 3.4 Open, other state-of-the-art GIS and other
software will attempt to be integrated into this software (Anderson, 2008).
3.2.2 CATS
CATS (Consequence Assessment Tool Set) predicts hazards and the consequences from these
hazards for both natural and man-made sources. It takes into account earthquakes, as well as
hurricanes and explosives. It uses ESRI ArcView as the GIS platform, as well as demographic
and infrastructure data. It was created out of cold war technology and is Windows-based. It
provides the facility to create realistic scenarios and assess the effects on the infrastructure
and population to allow for emergency management, resource deployment and to assess the
requirements for a sustained disaster response. CATS is owned by FEMA and also DTRA.
CATS has been tested for the earthquakes of Northridge, U.S., Kobe, Japan, and Izmit and
Duezce, Turkey.
CATS takes into account ground failure, tsunami, fire and ground shaking. It is extremely
detailed and even takes roadblock information into account for the U.S. version. It is likely
that these options are not as readily available for users outside the U.S.. It is available as part
of the ESRI CATS Bundle and work is continuing on various versions (CATS 6 is currently
used).
3.2.3 DBELA
DBELA has been well explained above within the vulnerability methods. DBELA has been
developed at the ROSE School/EUCENTRE in Pavia, Italy, and uses mechanical-based
formulae in order to determine the displacement capacity for reinforced concrete and masonry
buildings grouped by failure mechanism and also the building class. It is a fully
probabilistically-based method and uses statistical exposure data to formulate a probability
density function for each parameter, and then uses Monte Carlo simulation to produce the
building database on which the vulnerability methodology (displacement demand and
capacity produced for all periods) is applied for a given hazard or group of hazards. The
damage distribution for three limit states is then directly applied to the original building set
(the exposure data). It takes into account the uncertainties associated through the process for
demand and capacity. It is easily applicable to most ELEs: however, it does require accurate
sampling of a small group of buildings in the location of interest in order to formulate the
pdfs. It has currently been written in Matlab as well as Fortran. It is not directly open source;
nevertheless, the methodology has been made transparent through papers (Crowley et al.,
2004, Calvi et al., 2006, Bal et al., 2008a). It has been found to be more accurate than
HAZUS, but is more time consuming for some Istanbul datasets (Daniell et al., 2009).
3.2.4 ELER/NERIES
Through NERIES (NEtwork of Research Infrastructures for European Seismology), a study
has been done in order to identify some possible methodologies for real time estimation of
losses (Stafford et al., 2007). Through the work that has been done in the JRA-3 component
of the NERIES project, a multi-level methodology has been developed in conjunction with
Imperial College, NORSAR and ETHZ called ELER. ELER has been looked at in terms of
the 1999 Kocaeli Earthquake, in order to predict the losses by Erdik (2007). The five step
methodology includes a detailed ground motion prediction uncertainty and regional variability
analysis and source parameters, estimation of ground motion using geological and
geotechnical information in the region of interest, incorporation of strong motion data to
improve ground motion distribution in the form of a ShakeMap, thus through the exposure
and vulnerability methodology being able to produce a Loss Map with the regional losses
(Erdik et al., 2008). EMME (Earthquake Model for the Middle East Region) also are looking
to implement ELER as well as GEM architecture into their methodology (Tuzun, Tblisi
Conference, 2009).
3.2.5 EmerGeo (previously NHEMATIS)
NHEMATIS (Natural Hazards Electronic Map and Assessment Tools Information System) was
originally developed for the EPC (Emergency Preparedness Canada) but has since been updated
by a completely privatised company, EmerGeo (http://emergeo.net/hazard_models.aspx ), which
uses up-to-date ESRI ArcGIS software and an obviously updated version of the original routines
of NHEMATIS but still with the same principles. It is a multi-hazard tool for Canada, similar to
HAZUS and it includes many national databases. The model produces maps of MMI and PGA,
and needs at least one soil map to operate (EmerGeo website, 2009). It also takes into account
secondary effects of earthquakes, as well as damage and injury maps based on lifeline, building
and facility types. It includes a GPS-based setup which can be used to locate an expert on the
damage map and transfer damage information straight to the expert, which allows for on-the-
ground emergency taskforcing to occur quickly and with real-time information. It is also starting
to be used in Australia, UAE and other locations around the world (EmerGeo News, 2009) but is a
completely closed source.
3.2.6 EPEDAT
EPEDAT (Early Post Earthquake Damage Assessment Tool) was produced for California OES by
EQE International Inc (Eguchi et al., 1994) to give a real-time information system in Southern
California to assist the local and state governments to produce not only response plans and
organise resources by simulating, but also by real-time infrastructure damage and casualty
estimates. The method uses observed post-earthquake information from satellite and aerial survey
to update model-based predictions of damage via loss estimation after an earthquake. It is
particularly useful with respect to significant ground deformation, as seen via liquefaction. The
methodology consists of 5 models which include an earthquake scenario generator, building
inventory models, building and lifeline damage models, casualty estimation models and displaced
individuals modelling. The real time information works in conjunction with data through CUBE
(Caltech USGS Broadcast of Earthquakes System) and REDI (Rapid Earthquake Data
Integration). It uses Modified Mercalli Intensity, and links can be seen between it and the current
PAGER system. It is Windows-based; however, not much data can be found from 1997 onwards
on it except that it is being used for building data in a current NASA QuakeSim project.

Figure 3-1: Methodology for EPEDAT (Eguchi et al., 1994).
3.2.7 EQRM
EarthQuake Risk Management (EQRM) is a model for regional earthquake risk assessment
that has been developed by Geoscience Australia (GA) for application to Australian cities.
The model is utilised in the form of a Python or Matlab-based program founded on the
HAZUS (Hazards United States) model that is widely used for risk assessment purposes
around the world. It has been adapted to Australian conditions with the building types and
other changes, especially the geological conditions within the Hazard section. It is a
reasonably straightforward program to use and the current version is that of February, 2009. It
does not require any GIS software and is based on the convolution of the four key areas that
make up seismic risk, i.e.Hazard (including a regional seismicity model, attenuation model
and regolith site response model), Elements at Risk (Social demographics, building
inventory), Vulnerability of those elements at risk (Building vulnerability model (capacity),
casualty and injury model and economic loss model) and Risk (the final earthquake loss
assessment). Significant studies have been undertaken in order to look at uncertainties of the
EQRM model (Patchett et al., 2005), (Robinson et al., 2005, 2006) as EQRM can be thought
to be a proxy of the HAZUS procedure.
3.2.8 EQSIM
Another methodology that is heavily based on HAZUS is EQSIM, an EarthQuake damage
SIMulation tool (including the integrated Disaster Management Tool, DMT) which has been
developed by the University of Karlsruhe. However, this is not open source. The tool has been
used for a test location in Bucharest, Romania, with some adaptations to European conditions
(14 HAZUS classes were added). It uses up-to-date reconnaissance techniques (damage
detection using airborne laserscanning data and response tools for coordination,
communication and information after an earthquake as part of the DMT). In the same way as
EPEDAT, there is a detection support system that analyses data after the earthquake to
combine with pre-earthquake data. This includes an augmented reality system which enables
individual buildings to be viewed in terms of their structural weaknesses post-earthquake.
This is an extremely detailed method proposed by Markus et al. (2004) and unfortunately is
closed to the public, despite an attempt to contact the authors. Many papers, nevertheless,
have been sourced to provide an insight into the system tools used. An open source version
will be available soon and is named eEQSIM, www.eeqsim.com.
3.2.9 Extremum
The Extremum software tool is a combination of many different tools developed at Extreme
Situations Research Center Ltd., including Emercom and SIGE, Russian Academy of
Sciences (Frolova et al., 2006). The system is extremely closed source. However, a
partnership with ETH Zurich has spawned QUAKELOSS (Wyss, 2004a) which is a version
of Extremum. Extremum uses an updateable model of settlements throughout the world based
on various scales, population distribution via mathematical models combined with hazard and
exposure data, and lifeline and hazardous system information, in order to produce damage
distributions for infrastructure and human loss, as well as rapid assessment methodologies.
Past event impact is required for calibration of the tool, and uncertainties need to be taken into
account. By combining it with the knowledge of QUAKELOSS, an extremely accurate tool
has been produced. It is Windows-based and incorporates GIS data. The integrated risk tool
also takes into account tsunami and other such secondary effects of earthquakes (more details
can be seen in QLARM and in the comparison in 4).
3.2.10 HAZ-Taiwan
HAZ-Taiwan is built based on HAZUS-based methodologies which have been tested for
Taipei City and other cities (Chien et al., 2002, Shaw et al, 2007, Yeh et al., 2006). It uses the
same type of building size discrimination as in HAZUS, which forms the database. The
economic loss estimation model assumes damage based on spectral displacement (Sd) for the
structural systems and drift-sensitive non-structural components; yet spectral acceleration is
used for the acceleration-sensitive non-structural components. It also incorporates a
probabilistic risk analysis methodology to produce exceedance probability curves based on
the mean and standard deviation for regional losses on multiple event philosophies.
3.2.11 HAZUS
HAZUS (Hazards-U.S.) as described above in the vulnerability section is an all encompassing
multi-hazard tool which was originally developed in the 1990s by FEMA to account for
earthquakes (ground shaking, ground failure, liquefaction, rupture and landslide (Kircher et
al., 1997). Later work has included flood and tropical storms, which is not within the scope of
this earthquake-based report. It is closed source and is developed entirely for U.S. scenarios
(48 conterminous states plus 2 and Puerto Rico), which are its two primary drawbacks. The
current version is HAZUS-MH MR3 as of April 2009, and allows for changeable population
and changes to the fire modules after earthquakes (FEMA website, 2009).

Federally-collected data has also been integrated into the program and the database contains
information on every building in hazard prone areas. The inventory is classed based on 36
different types of building based on construction standards and material as well as size and
building use. The software uses C++ and Visual Basic routines to implement loss models and
Microsoft SQL as a relational database, interfacing also with ArcGIS and many other GIS
programs in order to express the damage states for the building stock and lifelines (as well as
essential and large-potential loss facilities) (Schneider et al., 2006, Kircher et al., 2006). The
U.S. census tract is used to calculate losses, including social (death, injury, homelessness and
disruption) and economic losses. From this, pre-planning for major earthquake scenarios
(M
w
<8.5) has been carried out. Because of the large amount of data that the U.S. has gathered
on assets in digital form, the specialised nature of the program and the U.S. typology of
buildings, HAZUS can be applied successfully. However, most countries do not have this
level of data available and may have different combinations of hazard (landslides and
volcanoes) or vulnerability (building types and construction codes), which therefore render
the HAZUS methodology not as effective. An example of one such adapted code is for an
area in Andhra Pradesh for India, by Siddiqui et al. (2007).

HAZUS is closed source as previously stated, but the mathematics of the methods as well as
most of the functions have been reproduced and shown by Porter (2008c), EQRM, SELENA,
MAEviz, HAZ-TAIWAN and numerous other open-source software and documents on the
internet.
3.2.12 InLET
InLET (Internet-based Loss Estimation Tool) has been developed by the University of California
and ImageCat Inc and is a complete web-based real-time earthquake loss estimation tool. It is part
of the RESCUE project funded by the NSF, encompassing several California universities. Data,
model updates and results are completely stored online in order to provide availability to users at
all times. USGS ShakeCast notifications provide loss estimates within a minute of building
damage and casualties using simplified HAZUS damage functions and GIS databases. This
method also provides traffic information and is easily changeable for federal and local scenarios
(ImageCat Inc., 2008). As stated, just the internet is required (Web 2.0), which is a problem if the
internet in the area of interest has been cut off by the earthquake (but they will possibly be able to
get help from outside), and also uses high resolution imagery, including street view from
Microsoft Virtual Earth. The internet-based methodology is also a large advantage due to the fact
that updates to the software can be simultaneously shown to customers rather than having to send
out DVD updates. The program claims differences with HAZUS of less than 5%, which for
emergency purposes is extremely useful. Unfortunately, it is rather specialised in location at
present, with only California data. It is essentially EPEDAT in internet form.
3.2.13 LNECLOSS
LNECLOSS was produced under a European Integrated Project focusing on Risk Mitigation
for Earthquakes and Landslides (www.lessloss.org) at the Laboratorio Nacional de
Engenharia Civil, Lisbon, Portugal. It uses the following methodology tested for the Lisbon
metropolitan area in order to produce earthquake disaster scenario outputs of economic loss,
social loss and economic damage. This method uses a spectrum-based approach in order to
calculate the damage state probabilities. This is not open source; however, Sousa et al. (2004)
provides an in-depth profile of the method and also the various LESSLOSS deliverables.

Figure 3-2: LNECLoss Flowchart (Typical of many ELEs) Sousa et al. (2004)
3.2.14 MAEviz
Another HAZUS-based application but applied to the middle states of the U.S. is MAEViz
(Mid America Earthquakes visualization). At first glance, it seems specialised; however, the
huge potential is shown by the flowchart of analysis procedures (48 and counting) and its
complete HAZUS system, including more detailed algorithms. It has an extremely good
format for the software (Windows-based) and has been expanded to take into account datasets
from Turkey and worldwide in recent versions. It is completely open source and features in-
built GIS, although there is some reliance on the Internet connection. It was made by the
University of Illinois as part of the Mid-America Earthquake Center for looking at earthquake
loss estimation particularly for the New Madrid Seismic Zone where there was a series of
damaging earthquakes from 1811-1812. The visual driven system of MAEViz uses a
combination of Sakai (an open source web portal), NEESgrid (a framework of tools to allow
researchers to collaborate) and SAM (Scientific Annotation Middleware) in order to allow for
users to add their own hazard data. Damage estimates can be established with options for
multiple mitigation strategies, testing of scientific and engineering principles and also
estimating the earthquake hazard impact on lifelines, social or economic systems. This is fully
user oriented and should be considered within GEM, with special note to the economic and
social algorithms which can be applied or changed.
3.2.15 OPENRisk
OPENRisk is a suite of programs which has been produced by Porter et al. (2007) and Porter
(2008c) in building an open source ELE software program that combines the vulnerability of
CUREE-Caltech vulnerability functions and HAZUS fragility functions with OpenSHA
Hazard and user-defined exposure data with up-to-date HAZUS social and economic loss
functions. It has been produced as part of the AGORA (Alliance for Global Open Risk
Analysis) project and is entirely in Java, UML format; it also uses the USGS ELE Software
ResRisk. It consists of a Hazard Loss Exceedance Frequency Curve program, Fragility
Function Calculator and a Benefit/Cost Ratio application, allowing users to analyse whether it
is better to retrofit and have less losses or not retrofit and have higher losses. These decision-
making machines take into account HAZUS mathematical functions for deaths and repair
costs and allow for open-source viewing of the financial loss models (EAL, LE, BCF) of
types of commercial software (3.3.6) in a single-site risk calculation or portfolio risk
calculation algorithm.
3.2.16 OSRE via MIRISK
MIRISK (Mitigation Information and Risk Identification System Kyoto) is one of the
software packages to be produced by Kyoto University and incorporated into AGORA
(Scawthorn, 2007, Mina et al., 2004). The project is mainly based on infrastructure to be built
(not existing infrastructure) and is not just earthquake-based, but takes floods and volcanoes
into account (but no secondary effects). It helps development managers to consider natural
hazards risk and ways to reduce that risk (Scawthorn, 2007). It was developed into OSRE.
OSRE (Open Source Risk Engine) has been developed through the AGORA workshop and is
designed to produce a freely available multihazard risk analysis software code which is Web-
and GUI-enabled and object-oriented. The 3
rd
version, OSREIII or OSRE3, is the current
version (April, 2009). It works on the same principle as the other fully open source systems
(EQRM and SELENA) available on the internet, but uses Java coding in this version, as
opposed to Visual Basic/C++ and Fortran coding in their first two versions, and this allows
for a fully open framework. A user can apply his or her own hazard, vulnerability function
and loss data into the methodology as well as the asset data to supplement the data already
obtained from NIED for hazard (PGA, MMI, PGV and JMA) and the ATC-13 matrix for
catalogue vulnerability. OSRE applications can be created to stand alone or be published on
the internet. The source code is available and could be adapted to other vulnerability
methodologies.
3.2.17 PAGER and other Rapid Response engines
PAGER (http://earthquake.usgs.gov/eqcenter/pager/), as stated previously, is an attempt by
the USGS to produce a rapid response Modified Mercalli intensity-based system to look at the
population exposure to any significant earthquake through the USGS ShakeCast throughout
the world. It is currently employed for any earthquake greater than approximately Mw4.3 in
the world, and Mw3.5 in the USA, but is really useful if a population of greater than 1000
people feel an earthquake. It also uses a function, Did you feel it?, which collects intensity
data via a series of online questions to supplement the intensity-distance relationship. It is an
extension of ShakeMaps, converting hazard into population risk. Although only intensity-
based, it makes very quick calculations as it employs the online data and other sensor data. It
is currently only used for ground shaking, and does not consider secondary effects. It is one of
two readily available near real-time population exposure systems for ground shaking (the
other is QLARM, to be discussed next). For tsunamis, NOAA Pacific Warning System
(http://www.prh.noaa.gov/ptwc/?region=2&id=hawaii.2009.03.19.200910) and also Japan
Meteorological Survey relay warnings for all oceans worldwide in response to the 2004
Sumatra tsunami of 26/12/04.

Figure 3-3: PAGER System (PAGER Website, 2009)
A new PAGER system lossPAGER (or PAGER Version 2) is set for production (it was
supposed to be late 2008) and this will be similar to the QLARM system. It is proposed to be
open source and open methodology. This will be invaluable as a worldwide model, as the
exposure and vulnerability data is worldwide and is sourced from many locations. It will then
use intensity-based or CSM-based methods in order to estimate fatalities (Porter, 2008a).
It uses a three-tier system including empirical, semi-empirical and comparison with real data
in order to formulate this.
3.2.18 QLARM, QUAKELOSS and QUAKELOSS2 via WAPMERR
QLARM, QUAKELOSS and QUAKELOSS2 (in production) are the rapid response program,
and versions of their ELE, respectively, from WAPMERR (World Agency for Planetary
Monitoring and Earthquake Risk Reduction). QUAKELOSS and QUAKELOSS2 were
spawned from EXTREMUM in 2000, and have since been relabelled as QLARM online for
the rapid determination version of social losses to help rescue teams. It claims to have
averaged earthquake loss estimates 28 minutes after the earthquake generally dependent on
the USGS source parameters (average 26 min after an earthquake). For the recent Italian
earthquake after 30 minutes a minimum of 50 dead and a maximum of 500 dead, and injuries
between 100 and 1300 were given, which was within the bounds of the final value of 300
fatalities and about 1179 injured. The moment tensor solution of USGS is used for further
updates if required.

Figure 3-4: Recent Italian Earthquake of 06/04/09 overlaying WAPMERR damage distribution and USGS
PAGER Population exposure via MMI (similar to PGA, PGV and MMI plots).
The difference can be seen in Figure 3-4 in that the damage distribution and hypocentre is
based further north in the WAPMERR rapid response, versus that of PAGER which assumes
a different focal mechanism (not extending the intensity as far NW and SE as WAPMERR),
thus reducing estimates of intensities for these towns. The WAPMERR estimates of the 5
damage states of buildings have only been released for the Sichuan earthquake, indicating a
cut-off in fatalities. The two methods can be seen to be reasonably close when relating
intensity directly to damage. In a real-time test for the aftershock of the 6/4/09 Italian EQ
measuring Mw5.6, PAGER was 16 minutes faster than QLARM (18 min vs. 34 min) with
QLARM not being able to give accurate estimates, because it was an aftershock and building
characteristics were significantly changed.

QLARM claims an 85% detection of major events vs. non-major events and some errors can
be seen such as those of Honshu, Japan, in June, 2008 where over 1000 dead were predicted
as a minimum, and only 13 died. However, with speed increases comes a reduction in
accuracy in locations without good seismological networks. This is conversely the case for
countries with detailed seismological networks having greater accuracy and quicker response
times (U.S., Taiwan, Japan) than those countries without such detailed networks (mostly
developing countries). With the Sichuan, China, earthquake, a oneday delay was shown with
approx. fatalities between 40,000 and 100,000 people, which was accurate. Wyss (2004b)
explains that the results are poorest for earthquakes that are small to moderate.

For the loss estimation module QLARM uses population data similar to PAGER in order to
create the population exposure for any particular event and also uses PAGER and World
Housing Encyclopaedia collapse rates or social loss matrices due to building class and EMS-
98 vulnerability, depending on location. The assumptions for some settlements would have
come from the Extremum system (Larionov, 1999), and fragility curve calibration for 5
fragility classes has come from approximately 1000 earthquakes worldwide (Shakhramanjyan
et al., 2001).

QLARM models elements-at-risk in each city to consider the available data possible at each
location (age, height and seismic protection level of buildings, structural classes) and uses soil
amplification factors that are dependent on topography, Vs30 estimations or local and national
soil classifications. The hazard component is in terms of PGA and macroseismic intensity
using GMPEs dependent on the tectonic regime and source, path and site effects. Many
different test cases have been undertaken, and much dependence is on hypocentre location
(Wyss, 2004b). Over 2 million settlements worldwide have been added to the database which
also takes into account developing countries. The final result is a range of fatalities and
injured with 3 classes of severity, as well as damage distribution (5 classes None, Slight,
Moderate, Extreme and Collapsed) for each settlement. It also incorporates QuakePy and
XMLQuake in order to create and update the historical earthquake catalogue and to produce
the warnings. There should be more study of these as they are freeware. A new version,
named QUAKELOSS2, has been released recently for the SAFER Project and includes more
up-to-date settlement information through satellite imagery; however, it is not yet finished
(Wyss, pers. comm., 2009). These changes may be transmitted to QLARM, depending on the
speed of calculation.

Max Wyss has been contacted for a password to the entire structure, including QUAKELOSS.
However, the public user password for simply viewing results is U:wapmerr, P:quakewap.
The method is transparent through papers and should be further explored as part of my
doctoral work with the University of Karlsruhe.
3.2.19 RADIUS
As part of the IDNDR initiative in the late 1990s, an excel-based tool for earthquake loss
assessment was created, named RADIUS (Risk Assessment tools for DIagnosis of Urban
areas against Seismic disasters). Nine case studies were undertaken worldwide comparing for
magnitude and seismic risk through the usual convolution of hazard, vulnerability, exposure
and specific cost. It does not include a GIS data program and is a simple program, more suited
to providing awareness of major earthquakes and a very quick assessment of a city and is thus
extremely user-friendly. It gives MMI, building damage, lifeline and casualty data when users
apply historic earthquakes or theoretical data. No authors have been able to be contacted and
it would appear that this tool is outdated. More information is available through the company
that supported the project, GeoHazards International, http://www.geohaz.org/projects/radius.
RADIUS is therefore deemed as a closed source, as Fumei Kaneko of OYO Corp. had to be
contacted in order to retrieve the relevant data.
3.2.20 REDARS
REDARS (Risks from Earthquake DAmage to Roadway Systems) is currently in its second
version, REDARS 2, produced by Geodesy, Seismic System Engineering Consultants and
ImageCatInc. for MCEER and FHWA and is a completely closed source. Documentation has
been gained as to how the system works, however. It uses the Los Angeles area as a testing
area, as well as locations in Tennessee. It looks specifically at the effects of earthquakes on
traffic, time delays, road closures and repair costs. It allows adaptation to other regions of the
original data and models within the program via updating with new user-defined values. In the
new version, updated HAZUS fragility functions for bridges and roads are examined. (Eguchi
et al., 2003, Werner et al., 2006).
3.2.21 RiskScape
The RiskScape model comes from New Zealand and is a fully integrated multi-hazard system
with a full economic, social and damage estimation for regional scenarios in NZ (Schmidt et
al., 2007). It is developed by a joint partnership between NIWA and GNS. It has the
advantage of being a stand-alone Windows-based system without any need for a GIS system.
H, it is fully integratable with GIS. It currently takes into account earthquakes and tsunamis in
the first phase of its production. Moreover, it has secured funding for the next 8 years, from
2009-2016, and will include landslides among other hazards. The front-end looks to be
extremely sophisticated and a detailed infrastructure inventory will continue to be improved
through remote-sensing techniques with existing data, and will provide social exposure data
(age, demographic, cultural and seasonal variation), as well as lifeline and important
infrastructure (transport and utilities) details. For tsunamis, the use of LiDAR is being
employed and there are detailed soil and geology maps for earthquake use. A hybrid database
has been produced, as no NZ survey had accurate enough data. For the vulnerability
component, existing fragility functions will be used and then updated with future post-
earthquake survey data. The authors have not been contacted but Riskscape should be
available due to the open nature of funding. The front-end of the program may be of use even
though the earthquake methodology is reasonably basic.
3.2.22 ROVER-SAT
ROVER (Rapid Observation of Vulnerability and Estimation of Risk) is a screening method
which has been applied into an open source Python script program available for download. It
was created as part of the work at University of Boulder, Colorado by Porter et al. (2007). It
is essentially FEMA154 in open source form which allows for a rapid inventory of buildings
and vulnerability via the risk calculation to be recorded. It includes soil and hazard lookup as
well as database, GIS and other features. It is the only screening method that has been found
in this study, which is open source and will allow automatic data handling and inventory for
emergency planning. It is a U.S.-based screening method which is a walk-down method, and
could be applied to the rest of the world to supplement data in any area.
ShakeCast (available from USGS) allows a user to watch for relevant earthquakes and
therefore this can be installed to allow for checking of the vulnerable buildings identified in
ROVER.
ATC-20i is another free program that is ATC-20 in pocket form, which can share data with
ROVER using GPS for mapping. ATC-20 is a standard post-earthquake safety inspection,
which can be used in order to give a red/yellow/green tag method and thus an exact database
without paper forms.
For instance, for the recent LAquila Earthquake, if ROVER had been used before the
earthquake on possibly vulnerable buildings, after one of the previous ELE softwares had
identified the most vulnerable buildings, the disaster management team could have responded
to the ShakeCast earthquake for LAquila, and shown the relevant buildings. Management
teams could then move quickly to the most vulnerable buildings and begin the ATC-20 style
post-earthquake safety inspection.
These programs are available with relevant documentation on the attached DVD.
This combination of programs, ROVER-SAT, as described by ROVER (2009) should be
looked at for possible application to the GEM project, given its open source nature, ROVER-
(http://code.google.com/p/emcode/wiki/RedROVER), Shakecast-
https://sslearthquake.usgs.gov/resources/software/shakecast/downloads/, ATc-20i-
http://www.atcouncil.org/ATC20i.shtml and could also be applied to the EUCENTRE
STEP project after modification to European conditions, in conjunction with the work
on aftershock sequencing or any other management system.
3.2.23 SAFER
SAFER is a real-time Seismic eArly warning system for Europe which has been developed in
a large collaboration between 23 worldwide universities and institutions coordinated by GFZ
Potsdam, in order to undertake a wide range of actions for critical systems in the first seconds
to minutes after an earthquake (Zschau et al., 2007).


Figure 3-5: Flowchart of SAFER for various objectives and components (adapted from Zschau et al.,
2007).
The process also looks at lifelines, transportation and industries and the supply of
information, including aftershock hazard, to undertake a holistic real-time earthquake loss
estimation methodology. Real-time estimates of source parameters, damage assessment and
reduction strategies (control mechanisms for critical systems), shake-maps and aftershock
hazard assessment were culminated into an End User Interface.
The Aftershock Hazard Assessment program (including the Rapid Aftershock Forecasting
Toolbox (RAFT)) attempts to take into account the effects of aftershocks on structural losses.
This would be particularly relevant for the recent Italian aftershock (Mw5.6, 7/4/09) where
many houses were already collapsed before the aftershock. Aftershocks generally have
different source mechanisms and effects from the main shock, as has been described in
Daniell (2008). A full 30 month loss assessment occurred between many different users with
5 test cases (Athens, Bucharest, Cairo, Istanbul and Naples). Due to the number of members,
very different seismic risk software such as PreSEIS (a neural network-based approach to
earthquake early warning systems Koehler et al. (2007)), RTLoc (Real-Time Location of
hypocentre Zschau et al., 2007) SELENA (used as a real-time damage computing system in
this case Lang et al., 2008a), ISNet use (Weber et al., 2007) and ElarmS (INSN network use
in real-time due to P-wave offsets Olivieri et al., 2008) has been tested.
Prediction
O.1
O.1
O.1
O.2
O.2
O.3
O.4

3.2.24 SELENA and RISe
SELENA (Seismic Loss EstimatioN using a logic tree Approach) has been produced by
NORSAR with support from the International Centre for Geohazards, Norway, and is
essentially the HAZUS damage probability methodology in a stand-alone Matlab format. It is
fully open source as long as the user has Matlab. The method uses the Capacity Spectrum
Method, as has been explained previously in HAZUS, EQRM and 0. The difference with
SELENA is that it uses a logic tree approach based on the weighting of the input parameters,
in order to consider epistemic uncertainty (Molina and Lindholm, 2005, 2007). Some
calibration has been undertaken to the test case of Oslo, Norway. Further modifications are
being carried out to speed up the process in terms of logic tree branches and other real-time
information for the SAFER Project (Lang et al., 2008a). Many versions have been released,
with versions 3.5 and 4.0 being the current versions (Molina et al, 2008a, 2008b). GIS
viewers such as ArcView can be implemented to display losses. However, a program named
RISe (Risk Illustrator for SElena) has been produced which is the associated GIS viewer that
is part of the SELENA package, which allows for easy viewing of the results from the
SELENA analysis (Lang et al., 2008b).

Figure 3-6: The difference between the static and dynamic components of SELENA for rapid response
modelling identification (Lindholm et al., Vilnius Conference, 2007).
3.2.25 SES2002/ESCENARIS for SIGE, Spain
SES2002 (Simulacion de Escenarios Sismicos started in 2002 Barranco et al., 2002) and
ESCENARIS (RSE, 2003) are two software tools that have been developed for Spain and
Catalonia respectively using a methodology that has been detailed in Roca et al. (2006). The
systems are scenario-based for hazards and use epicentral intensity and also GIS systems. The
vulnerability matrix method is as shown above in Figure 2-6. The losses are just based on the
building stock and all classes of social impact. It links in with the General Direction of Civil
Protection (DGPC) system SIGE for Spain for emergency management which is a GIS-based
system developed under MapObjects 2.1. The methodology again is reasonably basic with the
use of empirical vulnerability functions. It has large-scale use in Spain, but the software is
closed source. It uses a 2 level system using EMS98 intensities looking at empirical DPMs as
well as vulnerability functions and indices developed through the RISK-UE project
(Giovinazzi, 2005). ESCENARIS is used for the Pyrenees ISARD (Information System for
Automatic Regional Damage) project which is actually based in Catalonia, where the software
was developed for (Dominique et al., 2007).
3.2.26 SIGE, Italy
SIGE is an all-encompassing earthquake response and scenario system for Italy which has been in
production since 1992 by the Italian Department of Civil Protection (OSSN). The loss estimation
is intensity-based, but does involve lifeline, facility and population information, including
building damage, casualty information and many other statistical maps integrating GIS, GPS,
GPRS, PDA and WEB data into a user-friendly format (Windows-based system named Quater)
(Soddu et al., 2005). The real-time earthquake results of INGV are incorporated directly into the
SIGE system through which a variety of mediums are used to transmit information. It also uses
Bayesian updating techniques to correct intensity values determined. The structure of the
vulnerability modules, as seen from Di Pasquale et al. (2004), appears to be exactly the same of
that of SES2002 with some scenario updating modifications for intensity with extra information.
Updates may have been made since 2005, but little documentation has been seen. It has a
reasonable advanced loss system for housing loss calculating collapsed, uninhabitable and
damaged dwellings.

Figure 3-7: The full SIGE setup which is transparent (providing insight into the Italian Government
methodology) Soddu et al. (2005)
3.2.27 SP-BELA
SP-BELA is a displacement-based method developed at the ROSE School/EUCENTRE in
Pavia, Italy. As explained previously in 2.4.3, it follows DBELA in many ways but has the
difference of a code built structure representing the structure and thus shear capacity, and
other assumptions can be used. It does not have the adaptability to different building types of
DBELA for structures which are outside the European-Mediterranean region. SP-BELA has
not been applied to a large dataset and is currently not in software form. Such a methodology
may be able to be applied to SELENA, EQRM or OSRE with ease or be used to update other
HAZUS methodologies. The method is open source.
3.2.28 StrucLoss/KOERILoss
StrucLoss 1.4 (Earthquake and Structural Department of Gebze Institute of Technology,
Istanbul, Turkey) is the updated version of KOERILoss 1.0 (Kandilli Observatory and
Earthquake Research Institute Loss Estimation Software). It was produced as part of the
LESSLOSS project for the test case of Istanbul and allows many user options, including
damage estimation via both macroseismic intensity and HAZUS-like spectral displacement
vulnerability methodologies and the ability to consider both deterministic and probabilistic
approaches for earthquakes, in a 2 option type system. The source code is MapBasic, Excel
and Matlab and runs under MapInfo, which is useful for GIS display with a Windows-based
user interface. Four building damage classes are used. Direct economic and social (4 levels of
injury severity) losses are derived from the damage classes as well as vulnerability curves for
various building types. Unfortunately, an email response by the authors was not obtained but
there is enough transparency in the documentation.
3.3 Other Useful Possible Earthquake Integration Applications
3.3.1 VCH Tool for Disaster Updating
VCH (Virtual Clearing House) is an online Google Maps-style application which requires
work (http://quake.kuciv.kyoto-u.ac.jp/vch/). Everything is set up to be interactive but there is
limited information and nowhere near real time, as it seems to have finished in 2007.
The VCH website is to provide ease of disaster investigation, so that there is no doubling up
of information between different entities for ground response (even if this is useful). A
Wikipedia-style idea is very useful for this type of on-the-spot disaster response collaboration
as it allows for search and pictures, but it has been unused since the Niigata earthquake of
2007. It is currently in Japanese and English with Google Maps and provides the location of
data, what is missing, and various disasters (tropical cyclones, floods and earthquakes as well
as tsunamis). The software uses Apache, MySQL and Xoops with connection through to the
internet open-source content management systems. This is all open source software and
could be integrated into disaster response software.
3.3.2 Historic Loss Catalogues - NATHAN, EM-DAT, www.earthquake.it and USGS
based archives

NATHAN (http://mrnathan.munichre.com/) is an open-source GIS-based historic catalogue
for previous earthquakes from MunichRe. It includes most known earthquakes and social and
economic losses within the previous 2000 years. It also includes volcanoes and many other
insurance hazards. It is very useful and is country based. It is completely available online and
therefore it could be used to incorporate the historical catalogue at any point on the Earth.
EM-DAT (http://www.emdat.be/) is another open-source historical catalogue project from
Belgium, which also could be supplemented to include loss.
USGS (http://earthquake.usgs.gov) and www.earthquake.it also attempt to provide such
historical catalogues. These are mentioned as they provide the stepping stones for use in
earthquake loss assessment models.
If a historic loss catalogue is simply required for the hazard component, the PEER NGA
database (http://peer.berkeley.edu/nga), the CISN database (www.cisn.org), the European
strong motion database (http://www.isesd.cv.ic.ac.uk/ESD/), Italian accelerometric archive
(http://itaca.mi.ingv.it/) or COSMOS (http://www.cosmos-eq.org) can be used to formulate
these catalogues. The PAGER-CAT archive is attached on the DVD.
In order to complete earthquake catalogues, the Python and Matlab coding by USC and INGV
is available on the web at http://completeness.usc.edu.
3.3.3 Hazard Modelling - NSHMP and GSHAP, OpenSHA
To produce ELE software, many different hazard modelling software can be integrated
depending on the location in the world. For any U.S. application NSHMP (National Seismic
Hazards Mapping Program) is available online and uses the new NGA dataset. The NGA
(Next Generation Attenuation relationships) database has been used by some of the worlds
foremost experts in GMPEs in order to create extended worldwide datasets and updated
equations for use within the world on the set of equations that was produced in the late 1990s.
These equations were produced for various tectonic regimes and each was given the same
datasets to use, although the different assumptions used by each of the authors (Campbell and
Bozorgnia, 2008, Boore and Atkinson, 2008, Abrahamson and Silva, 2008, Idriss, 2008,
Chiou and Youngs, 2007) gave different GMPEs. These updated versions are implemented
within OpenSHA, an Open source Seismic Hazard Analysis program available on the internet
(www.opensha.org) and include updates such as directivity and hanging wall effects in some
of the new GMPEs, and the ability to produce ShakeMaps easily. This is a very useful tool to
adapt into any future open source program for ELE such as will be produced within the GEM
project. SEISRISK, FRISK and CRISIS can also be used for assessment where available.
GSHAP (Global Seismic Hazard Assessment Program - http://www.seismo.ethz.ch/gshap/)
was the precursor to the GEM project in that it provided hazard levels for the entire world for
a 1 in 475 year earthquake with MSK intensity based on various relationships and
assumptions. It was completed during the 1990s, but could still provide some insights for
hazard in locations which have not been looked at in depth through current ELE software. The
GEM project is attempting to produce a fully open source earthquake loss assessment
procedure for the entire world, integrating various software packages, and could also include
rapid response capabilities and provide the possibility of producing full assessments at any
location, rather than just for macroseismic intensity at a 1 in 475 year level.

Table 3-2: Current and Future Tools that Global Earthquake Model will use for their system (GEM
Website, 2009)
Hazard Risk Socio-Economic Impacts
Seismic hazard software
Global active fault database
Global earthquake history & catalogue
Global instrumental earthquake catalogue
Geodetic Data Standards Development
Global NGA Models
Global Soil Database
Adapted Global Inventory
Adapted Global
Vulnerability Fragility
Population Assessments
Consequence Functions
(deaths, injuries, homeless)
Risk Indicator Development
Socio-Economic
Indicator Development
Loss Functions (direct
and indirect, social
losses, long-term
economic losses)
Decision Tools
Development
3.3.4 Complex Interseismic Hazard Modelling - QuakeSIM
QuakeSIM has been produced through the Jet Propulsion Laboratory at California Institute of
Technology through NASA using complex modelling techniques (2D and 3D Finite and
Boundary Element modelling) in order to capture the interseismic tectonic simulation and
progression of earthquakes and hence provide accurate analysis at a site through accurate
source and path characterisation. Three open source programs have been produced which they
believe will help earthquake risk estimation GeoFEST, PARK and Virtual California are the
three source codes available (http://quakesim.jpl.nasa.gov/download.html). They use
palaeoseismic fault data, GPS deformation, InSAR and seismicity data to predict earthquakes
and site ground motions via forecasting and also to undertake simulations (with attenuation
and site modelling) for accurate ground motion hazard calculations in California. This could
also be updated to other locations in the world for use in a fully adaptive ELE software.
3.3.5 TRANSFER
The TRANSFER project is the Tsunami-based version of the ELER project from the FP6
European Commission and brings together the knowledge of 29 European institutions. The
purpose is to produce Tsunami Risk ANd Strategies For the European Region
(www.transferproject.eu) by looking at tsunami catalogues, the seismic and non-seismic
sources of tsunamis, in addition, and also development of instrumental signals and networks
for the development of early warning systems. They are looking to do this by improving
numerical modelling and adapting remote sensing techniques, inundation maps and risk
assessment strategies in a probabilistic and statistical way. This knowledge can then be used
to reduce the potential impacts of tsunamis through collaboration with NOAA and other
Tsunami authorities such as JMA.
3.3.6 Proprietary Software for Insurance Companies
There are many other proprietary ELE software for reinsurance companies which have not
been reviewed due to lack of knowledge. Reinsurance companies (the main four being
Munich Re, Swiss Re, Berkshire Hathaway and Hannover Re) require up-to-date ELE
software in order to set premiums for the insurance companies that they cover. Some have
their own resources and there are also many private all-encompassing ELE software
(developed by ABS Consulting (RISKMAN), Risk Engineering and Degenkorb (FRISK),
AIR Worldwide (CATrader), PBS&J (HAZUS)) which are proprietary and unavailable to
non-paying users.
These programs are not all encompassing for countries in most cases (AIR website, 2009) but
try to provide benefit/cost and assessment procedures with up-to-date technologies. Many
insurance companies are bringing out country-specific models based on stochastic catalogues
(historic data based) and these can be found for nearly every country (eg. Bulgaria
AONBenfield (GapQUAKE Bulgaria) and various world bank projects (WillisRE). Thus,
reinsurers and insurers require new technology as well as relying on proprietary methods.
Thus funding for open source projects such as GEM is essential. Proprietary types of ELE
software have not been applied, as not enough documentation is provided.

A model is only as good as the information that is fed into it. Prof. J. Stewart (2008)

Chapter 4. Synthesis of ELE Software Packages for use in Future Open Source Production

4. SYNTHESIS OF ELE SOFTWARE PACKAGES FOR USE IN
FUTURE OPEN SOURCE PRODUCTION
4.1 Comparison of the ELE Software Packages
This comparison builds upon the preliminary work of Stafford et al. (2007) and provides up-
to-date and corrected information, a significantly increased study (from January 2007 to April
2009) and new insights not covered within the NERIES study into ELE software packages
sourced worldwide for the closed source ELER project. Much coding and programs have been
sourced and reviewed from March 2008 April 2009 and this study acts as a current view of
the open source knowledge available.
4.2 Technical Aspects
The development status of the selected programs discussed in Chapter 3 is shown below. The
open source and closed source status is in order to define how readily available the software
is. Some software was sourced or will be sourced once available (CAPRA is not online as yet
and still under development). Many software systems claim open source status; however, to
be truly open source, code and the program should be available on the internet. The contact
that has occurred in order to gain code, information etc. is detailed below. The asterisk (*)
corresponds to an uncertainty in the result, or that the status of that result may change very
soon in the future. The only truly open source ELE software packages as of April 2009 are
EQRM, SELENA, OpenRisk (using ResRisk and OpenSHA), OSRE, MAEViz and ROVER-
SAT (a suite of programs; ROVER, ShakeCAST and ATC-20i). However, various
collaborations can be made and therefore software sourced. CAPRA and RiskScape will both
be finished and available online. QUAKELOSS2 is also moving towards being open source,
but is still under development from WAPMERR. Versions of DBELA are also in production
for distribution through this thesis. A truly open-source platform with GIS is desired before
release, unlike OSRE where several versions have been released. Where data was unavailable,
much documentation was sourced and authors contacted in order to make the methodology
transparent. From this documentation, it is possible to code these programs, or at least to
critically view the components, in order to decide what the best combination of programs is.
This has all been supplied on the attached DVD. For CAPRA, RiskScape and components of
ELER, there is no source code or exact documentation due to the development status. In
addition, the proprietary and governmental nature of CATS, EmerGeo, EPEDAT, Extremum,
HAZ-Taiwan, InLET, PAGER, REDARS, SES2002 and SIGE has made it extremely difficult
to retrieve information; nevertheless, the best effort has been made in all cases.
Table 4-1: Availability of ELE Software Packages (as of April 2009)
ELE Software
Closed or
Open
Source Availability
Method
Documentation Contacted
CAPRA Open N/A, online reg.
YES, no source
code
Yes, Francis Ghesquiere, Ed
Anderson
CATS Closed N/A YES Yes, no response
DBELA Closed* N/A YES
Yes and Matlab version made (this
thesis)
ELER Closed* N/A YES Yes, no response
EmerGeo Closed N/A YES Online Preview available
EPEDAT Closed N/A YES Yes, Ron Eguchi sent details
EQRM Open Available YES Yes, numerous discussions
EQSIM Closed* N/A YES Yes, papers sent
Extremum Closed N/A* YES
Yes, no response but QUAKELOSS
uses it
HAZ-Taiwan Closed N/A YES Yes, no response
HAZUS-MH Closed N/A YES Not needed
InLET Closed N/A YES Yes, no online access granted
LNECLOSS Closed* N/A YES Yes, no response
MAEViz Open Available YES n/a
OPENRISK
Open
Available YES n/a
OSRE Open Available YES n/a
PAGER Half* Usable
YES, no source
code Catalogue available online.
QLARM Half* Usable*
YES, no source
code
Yes, Max Wyss contacted QL2
updates are not yet applied*
RADIUS Half* N/A YES
Yes, should be available but outdated
tech.
REDARS Closed N/A YES No
RiskScape Open* N/A YES Yes
ROVER-SAT Open Available YES n/a
SAFER Closed N/A YES Yes, available once in Karlsruhe
SELENA Open Available YES n/a
SES2002 &
ESCENARIS Closed* N/A YES No
SIGE Closed N/A YES No
SP-BELA Closed* N/A YES Yes, details available
StrucLoss Closed N/A YES Yes, no response

Further details are available in Appendix E.
The use of such programs is usually determined by the update status. In the field of
earthquake loss assessment, there have been many updates, new research ideas, technological
changes and collaborations which have yielded updated software packages (QLARM is one
such example, which is a real time loss estimation software from WAPMERR which has
continued refining their methodologies. DBELA also has had many updates and changes
since its original inception). In addition, there have been various software packages which
have not been updated since 2007 and before, and therefore can be assumed to be final
versions. RADIUS, Extremum and EPEDAT are such examples. These methods have been
integrated in some cases to new software such as QUAKELOSS (like QLARM). By looking
at the availability, update, development status, hardware, software and licensing, the financial,
training and implementation cost of using such a package can be gleaned. It is unknown if
LNECLOSS is still under production.
Table 4-2: Update and Development Status of ELE Software Packages selected
ELE Software
Update
since 2007? Development Status as of April 2009
CAPRA YES
Web portal with Linux source code under construction,
beta versions for testing should be available in August,
2009 (Anderson, pers. Comm, 2009)
CATS YES CATS-JACE with ESRI CATS Bundle
DBELA YES
Fortran version (Bal, Crowley 2009), Matlab (this thesis),
Python in progress (2009)
ELER YES Level 0 and Level 1 in testing phase
EmerGeo YES Version 4.0 (This Quarter)
EPEDAT NO EPEDAT unchanged since 1997
EQRM YES
Python Version 1.0, svn873 (Feb 2009), Matlab version
1.3 ended production but has also been sourced
EQSIM YES Version unknown since 2004 (Wenzel, pers. comm, 2008)
Extremum NO Russia possibly but integrated in 2001 into QUAKELOSS
HAZ-Taiwan YES In use since late 2000
HAZUS-MH YES HAZUS-MH MR3 Patch 3 April 2009, MR4 June 2009
InLET YES InLET version online (data from 2008)
LNECLOSS YES Version not updated since Istanbul test
MAEViz YES Version 3.1.1 from Vers. 2.3
OPENRISK YES
LossCurveApp, BCR Application, OpenSHA,, Frag 1.0.3,
FatalityVFs, RepairCostVFs
OSRE YES OSRE3.0
PAGER YES 2004 onwards
QLARM YES QUAKELOSS2: still under production, QLARM: online
RADIUS NO Not updated since 2000
REDARS YES REDARS 2.0 since 2006
RiskScape YES 1st version still in production
ROVER-SAT YES ROVER, ShakeCAST 2.0.4. and ATC-20i all working
SAFER YES Yet to be released
SELENA YES Version 4.0 released
SES2002 &
ESCENARIS YES* In production since 2003 and 2002 respectively
SIGE YES* In production since 1992, used since 1995
SP-BELA YES No version tested
StrucLoss YES Version 1.4 (March 2008)

The hardware and software requirements are extremely important in the decision-making
process. GIS licensing can be extremely costly, and not all places in the world have the same
computing power. It is desirable that a truly open source global model can be run without the
need for a separate GIS licence and platform and that all programs are freeware. In addition,
the coding must be so, that there is as greater accuracy possible for a certain non-exorbitant
computational method (i.e. not running a NLTHA for every building within the building stock
for a range of ground motion values). Hence, the ELE software packages have been reviewed
and the various hardware and software needed detailed. InLET is a web-based application like
MAEviz (if the data is not downloaded), QLARM and PAGER. These can be run on a
standard PC (1.6GHz with 2GB RAM, 200GB Hard Drive), as with all the other software,
although DBELA, EmerGeo and other computationally expensive algorithms and GIS-based
systems may require more computing power. The actual physical computation is generally not
computationally expensive and it is usually the large amount of data (exposure) that causes a
problem with memory. For deterministic use in post-earthquake studies, all of the codes can
be used in reasonable time (depending on if it is for an entire region and the data is available).
Where the software has not been personally tested, Untested has been written; however,
there has been an attempt to find details in the documentation. For the case of ROVER-SAT,
a handheld PC is useful for such a screening method. SELENA has built its own GIS system,
RISe, in order to make the package self-contained. MAEviz, InLET, RiskScape, SIGE,
Extremum, PAGER, QLARM and REDARS have similar packages. DBELA, LNECLOSS,
OSRE, OPENRISK, SPBELA and EQRM are all applied without a final GIS application, and
therefore the user must apply the loss estimation into such a program. Nonetheless, the open
methodology source is such that it should be easy to apply a GIS to such data, either within
the program or externally.

Table 4-3: Hardware and Software Requirements for the ELE Software Packages
ELE Software Hardware Source code software Licensed Software needed
CAPRA Standard PC
Linux, Windows, Web 2.0 with
Wiki-style updating.
None
CATS Untested, Std. Windows, ESRI ArcGIS
ESRI ArcView, StreetMap &
Analyst
DBELA Extended PC Fortran and Matlab Matlab and GIS
ELER Untested, Std. Google, Windows None*
EmerGeo Untested, Slow Windows, GIS, Google
OpenGIS (ESRI ArcGIS),
WebEOC
EPEDAT Untested, Std. Windows, MapInfo MapInfo
EQRM Standard PC Matlab, Python None
EQSIM Untested, Std. Windows, HLA, Access, Oracle9i ArcView, MapInfo
Extremum Untested, Std. Windows None
HAZ-Taiwan Untested, Std. not found not found
HAZUS-MH Untested, Std. C++, Visual Basic, Microsoft SQL ESRI ArcGIS 9.2, HAZUS
InLET Untested, Std. Windows, Java None
LNECLOSS Untested, Std. Windows, language unknown GIS platform
MAEViz Untested, Std. Java, Windows None
OPENRISK Standard PC Windows, Java None
OSRE Standard PC Windows, Java None
PAGER Standard PC Web-based None
QLARM Standard PC Web-based, C++, xmf, Windows None
RADIUS Standard PC Excel optional GIS via ArcView
REDARS Untested, Std. Windows, Unix None
RiskScape Untested, Std. Windows, in-built GIS None
ROVER-SAT Std., Handheld PC Windows, Python None
SAFER Untested, Std. Matlab, other coding Not described as yet
SELENA Standard PC Matlab, RISe Matlab
SES2002 &
ESCENARIS
Untested, Std. Windows-based MapObjects
SIGE Untested, Std. Windows, Unix GIS Platform
SP-BELA Untested, Std. Fortran Needs GIS or visualisation
StrucLoss Untested, Std. Windows, Unix GIS Platform

The regional applicability of the ELE packages is important, as it determines which software
tools could be used for the production of a global level ELE software. In some cases, the test
regions determine how the software can be applied to other regions as fragility and
vulnerability functions are region-specific. Where the exposure module is not hard-wired into
the software, then it is possible to easily change the regional data, and so it can be applied
worldwide once relevant checks have been made to the applicability of the vulnerability,
hazard and specific loss modules. RADIUS is a different application in that it was actually
applied to 9 cities in order to bring awareness of natural disasters. A simplified approach was
used in its application (Shaw, 2000). Truly global tested models are PAGER, Extremum,
QLARM (and QUAKELOSS), ROVER-SAT, OSRE and OPENRISK. These software
packages work on algorithms to estimate the loss from settlements, or universal loss functions
or are screening methods. In addition, there are also user-defined region software packages
which allow direct application without modification of the vulnerability and hazard
methodology. These are CAPRA, DBELA, EQRM, EQSIM, MAEviz, SAFER, SELENA and
StrucLoss. 16 of the 28 identified full ELE packages can have a worldwide application, as
seen below in the table. Those with an open source code method can also be modified after
changing hazard and vulnerability data.

Table 4-4: Regional Applicability of reviewed ELE packages
ELE Software Applicable Regions Test regions
CAPRA
Central America, User-
defined
Central America
CATS U.S. Northridge, Turkey, Japan
DBELA User-defined Marmara
ELER Europe Istanbul
EmerGeo Worldwide Canada, multiple regions
EPEDAT U.S. California
EQRM User-defined*, Australia Newcastle, Perth
EQSIM User-defined Bucharest
Extremum Russia, Worldwide Russia
HAZ-Taiwan Taiwan Taipei
HAZUS-MH U.S. Northridge
InLET California Los Angeles, O.C.
LNECLOSS Europe Lisbon
MAEViz
Central U.S., Turkey, User-
defined
New Madrid (U.S.), Zeytinburnu
(Turkey)
OPENRISK Worldwide U.S.
OSRE Worldwide Japan
PAGER Worldwide U.S.
QLARM Worldwide Numerous (2000+)
RADIUS Worldwide 9 cities worldwide
REDARS U.S. California
RiskScape New Zealand Hawkes Bay, N.Z.
ROVER-SAT Worldwide/U.S. U.S.
SAFER User-defined
Athens, Bucharest, Cairo, Istanbul,
Naples
SELENA User-defined Oslo, Norway
SES2002 & ESCENARIS Spain Spain & Catalonia
SIGE Italy 2002 Molise EQ
SP-BELA Mediterranean/Europe Italy
StrucLoss User-defined Istanbul
4.3 Demand Module
The demand module is intrinsically linked into the applicability of the system since the
ground motion parameters and response is different in every location of the world and
depending on tectonic regime, GMPEs are applied differently.
4.3.1 Hazard types considered
Different hazards are calculated in each of the packages, with some being part of a larger
group of hazards, including climatological (fire, drought etc.), hydrological (floods, mass
water movements etc.) and meteorological (hurricane, cyclone and others) types (RiskScape,
HAZUS-MH, CAPRA etc.). However, for the scope of this report a focus on earthquake-
related hazards that the ELE software packages have calculations for has been included.
Ground shaking, as shown in the introduction, contributes most to the social and economic
losses in earthquakes and therefore only ELE software packages which consider ground
shaking have been tabulated. As demonstrated by Bird and Bommer (2004) in 50 earthquakes
reviewed from 1980-2003, secondary effects such as liquefaction, fault rupture, landslides and
slope stability, tsunami and standing waves can cause much damage. This is particularly the
case in terms of transportation, utility and critical systems where secondary effects (ground
failure via liquefaction, fault rupture or a landslide) may impede emergency services and/or
cause earthquake-related social problems (disease, traffic problems, water/power shortages,
and lack of the internet (this was the case for the undersea cable near Taiwan)). Tsunamis are
also being considered exclusively by a few different projects (TRANSFER, NEAREST and
SEAHELLARC).

The aftershock sequences are very difficult to model due to the fact that the earthquake has
already damaged buildings, and that the location of the aftershocks is difficult to constrain in
terms of intensity. Both ELER and SAFER are two software packages which take the
aftershocks into account. However, they have not been released yet and are primarily for
deterministic post-earthquake modelling. The STEP project is also identifying aftershock
hazards. In addition, earthquakes can be seen to be related to volcanoes in some cases and
therefore these have also been accounted for, albeit briefly, by CAPRA, HAZUS-MH and
RiskScape. A screening method such as ROVER-SAT can be used for any of these hazards.
Fire after earthquakes is considered with HAZUS and is more an indirect effect. The Sichuan
disaster in May 2008 also formed many so-called Quake Lakes (formed by landslides
blocking rivers) and is equivalent to a dam burst when the water pressure builds up behind the
weakly consolidated rock structure. These have, as yet, not been considered by such packages
but could be in the future.
Table 4-5: Earthquake-related hazards considered in ELE Software Packages reviewed
ELE Software
Ground
Shaking
Liquefaction Fault Rupture Landslide/SS
Tsunami/
Seiche
Other
CAPRA YES YES YES* YES YES Volcanoes
CATS YES YES* YES* YES YES NO
DBELA YES YES NO NO NO NO
ELER YES YES* YES* YES* YES* Aftershocks
EmerGeo YES
YES,
potential
NO NO NO NO
EPEDAT YES YES YES NO NO NO
EQRM YES NO NO NO NO NO
EQSIM YES NO NO NO NO NO
Extremum YES NO NO NO NO NO
HAZ-Taiwan YES YES YES YES YES NO
HAZUS-MH YES YES YES YES YES Volcanoes
InLET YES NO YES* NO NO NO
LNECLOSS YES NO NO NO NO NO
MAEViz YES YES YES YES NO NO
OPENRISK YES NO NO NO NO NO
OSRE YES NO NO NO NO NO
PAGER YES NO NO NO NO NO
QLARM YES NO NO NO NO NO
RADIUS YES NO NO NO NO NO
REDARS YES YES YES Future Future NO
RiskScape YES NO NO Future YES Volcanoes
ROVER-SAT YES YES* YES* YES* YES*
Applied to
any situation
SAFER YES NO NO YES NO Aftershocks
SELENA YES NO NO NO NO NO
SES2002 &
ESCENARIS
YES NO NO NO NO NO
SIGE YES NO NO NO NO NO
SP-BELA YES NO NO NO NO NO
StrucLoss YES NO NO NO NO NO
4.3.2 Modes of Analysis Available
Table 4-6 considers the various possibilities between the modes of analysis that can be
undertaken in order to gain the loss due to certain situations. The difference between
probabilistic and deterministic SHA has been shown above and thus a desirable software
package should allow for both methods, including using real-time, historical and user-
specified data in order to provide a pre- and post- earthquake analysis tool.
The temporal distribution of earthquakes in probabilistic methods is generally looked at in
two ways: a poissonian distribution process in which earthquake probability is independent of
time from the last earthquake (earthquakes are a random process as shown by the Parkfield
prediction exercise) or; Time-dependent methods which assume that earthquake events are
linked temporally. Considering the difficulty of interseismic Coulomb stress modelling, a
poissonian distribution process is a reasonable assumption.
For the single scenario (deterministic) predicted methods; these methods are the ability of the
software to be used for a certain chosen earthquake by the user. PAGER and QLARM are the
only methods which do not allow this, due to their real-time nature. A user-defined event for
the ground motion can sometimes be applied, allowing the user to apply a complex theoretical
model or any model desired.
In contrast, observed values are also used in various packages, utilising either historical
ground motions or corresponding to ShakeMap ground motions from an automated near real-
time network (i.e. strong-motion networks). This can usually only be applied for a few
locations in the world, but the new methodologies of PAGER and QLARM make it possible
to apply ground-motion maps.
Table 4-6: Analysis Models possible in the ELE Software Packages
Deterministic-Predicted Deterministic-Observed Probabilistic
ELE Software
User-
specified EQ
User event
for GM
Historical
GMs
Automated GMs
(Real-Time) Poissonian
Time-
Dependent
CAPRA YES* YES* YES* YES* YES NO
CATS YES NO NO YES NO NO
DBELA YES user-built user-built NO YES
User-
defined
ELER YES YES YES YES YES*
YES and
stress
transfer
EmerGeo YES NO NO NO NO* NO
EPEDAT YES NO NO YES NO NO
EQRM YES NO NO NO YES NO
EQSIM YES NO YES YES NO NO
Extremum YES NO YES YES NO NO
HAZ-Taiwan YES YES YES YES YES NO
HAZUS-MH YES YES YES YES YES NO
InLET YES NO YES YES NO NO
LNECLOSS YES YES NO NO NO NO
MAEViz YES YES YES NO YES NO
OPENRISK YES YES YES NO YES NO
OSRE YES NO YES NO YES NO
PAGER NO NO YES YES NO NO
QLARM* YES/NO NO YES YES NO NO
RADIUS YES NO NO NO NO NO
REDARS YES YES YES YES YES NO
RiskScape YES* NO* YES* NO* NO* NO*
ROVER-SAT YES NO YES* YES NO NO
SAFER YES* YES* YES* YES* YES* NO*
SELENA YES YES NO YES YES NO
SES2002 &
ESCENARIS YES* NO NO NO YES NO
SIGE YES NO NO YES via updates NO NO
SP-BELA YES user-built user-built NO YES
User-
defined
StrucLoss YES NO NO NO YES NO

4.3.3 Ground Motion Determination and Distributions
There are generally two methods of ground motion determination; intensity-based and
spectrum-based methods. Intensity-based models are generally easier to apply than spectrum-
based methods as they inherently have information of the structural damage within the
intensity measure and therefore a direct probability of damage is calculated from the intensity
measure that is given to the location. It is more a traditional method, as can be seen by the use
in EPEDAT, EmerGeo (NHEMATIS based), SIGE etc. and also in the rapid determination
models of PAGER and QLARM, as they are also faster with respect to coding. There are
many different intensity-based methods, including many different intensity scales, being used
(as detailed in the Appendix A):- Conversion from peak ground parameters such as MMI
conversion to PGA, PGV, JMA or vice versa, regional GMPEs such as in SIGE, SES2002
(Ml to MCS), and EQRM. It must be noted that the conversions seen in LNECLOSS and
OSRE, where intensity is converted to a response spectrum (via GMPEs), are a bad
assumption, as intensity is a non-linear and discrete blocked method. Only in cases where
observations are made with macroseismic intensity (with no other method used) should this
be used. Nevertheless, most ELE software packages reviewed choose a more accurate way,
converting intensity from peak ground parameters.
Spectrum-based methods are used in the majority of new methods, as response spectral
ordinates can be used to directly compare to the vulnerability through structural capacity of
the building stock, and then a level of damage is derived. The ground motion information
required at each step of the process is generally only calculated based on 3 points, in order to
reduce computation time (and is a reasonable assumption considering the complexity of
ground motion). These are usually PGA, 0.3 seconds and 1.0 second in order to take short and
long period ordinates into account. HAZUS-MH also includes a 3.0 second period ordinate.
The best ELE software packages should allow the user to choose between intensity-based and
response-spectrum-based methods for ground motion calculation. ELER, CAPRA
(hopefully), EQRM, QUAKELOSS2, RiskScape and StrucLoss all take into account both
methods.
Table 4-7: Ground Motion Parameters for ELE Software Packages reviewed
ELE Software Intensity
Response
Spectrum
Details
CAPRA unknown unknown Yet to be online
CATS NO YES
Response spectrum based on PGA, Sa=0.3s and 1.0s; or
inferred PGA spectrum
DBELA NO YES Full Response Spectrum estimated
ELER YES YES
Level 0 and 1:MMI, PGA and PGV from MMI, Level
2:HAZUS-MH style
EmerGeo YES NO Site response corrected MMI from PGA
EPEDAT YES NO MMI from bedrock PGA (Eguchi et al., 1994)
EQRM YES YES
Full Response Spectrum, and allows for conversion to MMI
via Gaull GMPE also, Can use UHS as an option.
EQSIM NO YES
Response spectrum based on PGA, Sa=0.3s and 1.0s with
guidelines given to create the shape from PGA*
Extremum YES NO MMSK86 intensities
HAZ-Taiwan NO YES
guidelines given to create the shape from PGA*
HAZUS-MH NO YES
Response spectrum based on PGA, Sa=0.3s, 1.0s, 3.0s; or
InLET NO YES
LNECLOSS NO YES
Full Response Spectrum estimated, or intensity converted
spectrum
MAEViz NO YES Full response spectrum obtained
OPENRISK YES NO Spectral Acceleration OpenSHA derived
OSRE YES NO MMI and then PGA, PGV, JMA converted from MMI
PAGER YES NO MMI near-real time
QLARM YES YES, future
EMS98 intensities and PGA near real time, QL2: PGV and
Spectral Accelerations
RADIUS YES NO Site response corrected MMI from PGA
REDARS NO YES
guidelines given to create the shape
RiskScape YES YES, future MMI and Spectral Accelerations to be confirmed
ROVER-SAT YES NO Screening method (Related to direct damage - ShakeCast)
SAFER NO YES
PGV, Response spectrum based on PGA, Sa=0.3s, 1.0s,
3.0s or inferred PGA spectrum
SELENA NO YES
SES2002 &
ESCENARIS
YES NO EMS98 intensities
SIGE YES NO MCS intensities via Housner intensities (Local)
SP-BELA NO YES Full Response Spectrum estimated
StrucLoss YES YES
EMS98 intensities OR PGV, Response spectrum based on
PGA, Sa=0.2s,0.3s and 1.0s

Within the intensity-based and spectrum-based methods, there are three main ways in which
the spatial distributions for ground motions are derived. These are observed, empirical and
theoretical methods.
1) Observed spatial ground motion distributions generally use past earthquake catalogues
or real-time ground motions, in order to develop the ground motions. Generally, real-
time ground motions are not always available due to the lack of seismometer
networks. These past earthquake catalogues can also be used in regions of high
seismicity (and locations where the seismic sources are known) to create ShakeMaps
for a typical scenario. These scenarios may then provide a realistic ground-motion
distribution (even with locations without a real time network). It can also be classified
that this is using observed ground motions in a theoretical manner. Most of the ELE
software packages reviewed allow the option of applying observed spatial
distributions (mainly in the form of ShakeCast motions, or historical earthquakes
relevant to the region that they were tested in as seen in Table 4-4).
Where only earthquake source characteristics are provided, then theoretical and empirical
methods are the only methods available.
2) Empirical ground motions are the most common spatial distribution ground motion
method of the three, due to the use of GMPEs applied to the region using empirical
data regression. There are also cases where GMPEs are simulated and then calibrated
by local data for low seismicity regions (eg. EQRM where Australian data not always
available). The regionality of these GMPEs is generally a problem in implementation
as they are usually built into the ELE software package and therefore cannot be easily
changed (this is the same with regional vulnerability functions as commented on
later). Where this is a problem, user-defined empirical GMPEs should be allowed by
adding a quick user-defined spectrum loop. The new NGA equations work towards
solving the problem as they contain worldwide information; however, this is no
substitute for user-defined possibilities.
3) Theoretical ground motions derived from seismological models for various earthquake
scenarios have also been allowed through this user-defined setting in a few different
ELE software packages (DBELA, EQSIM, OPENRISK, REDARS and SP-BELA and
most likely in CAPRA, QUAKELOSS2 and SAFER). Theoretical ground motions are
time-consuming due to the computational inefficiency of such methods for large areas.
The input parameters usually cannot also be constrained, which makes these methods
uncertain.
Where it is possible to update the ground motions based on shakemap data (preliminary
magnitude, distance and depth data), this has been included in the table. Some methods also
allow updating of spatially distributed ground motions via aftershock data for use in loss
assessment. This is especially useful in rapid post-earthquake situations.
Table 4-8: Ground Motion type used for spatial distribution for the researched ELE software packages
ELE
Software
Observed Empirical Theoretical Details Updated GMs?
CAPRA YES* YES* u/c O:Past earthquake catalogue ShakeMap based
CATS Prop. YES Prop. GMPE based. ShakeMap based
DBELA YES YES YES user-defined displacement spectrum NO
ELER YES YES n/f
O: real time GMs E: a)Regional intensity
attenuation relationships (Wald et al.,
1999), b)Intensity correlations with
attenuation relationship based PGV,
PGA, and Spectral Amplitudes and,
c)Intensity correlations with synthetic
Fourier Amplitude Spectrum.
ShakeMap based,
Aftershock STEP
EmerGeo YES* YES NO
O:USGS ShakeMap feed, E:Latest
GMPEs
ShakeMap based
EPEDAT YES YES NO
O: near real-time GMs, E: 1 WUS
(Campbell & Bozorgnia)
ShakeMap based
EQRM NO YES NO
5 GMPEs:Australian adapted, can user
adapt to an observed or other emp.
GMPE, all Sa based
NO
EQSIM YES YES YES
O:recorded time history response
spectra,E: regional attenuation to EC8 ,
type, T, O, E: all can also be user-defined
ShakeMap based
Extremum NO YES NO Standard attenuation functions (M,D) ShakeMap based
HAZ-Taiwan YES YES YES Built-in, historical, user-defined ShakeMap based
HAZUS-MH YES YES YES Built-in, historical, user-defined ShakeMap based
InLET YES YES NO O:ShakeCast, T:1997 or 2000 Campbell ShakeMap based
LNECLOSS YES YES YES
O:user-defined, E:European spectral
ordinate GMPE, T: RVT
NO
MAEViz YES YES YES
E: 3 NGA, 6 WUS, 6 CUS
O and T: user-defined.
NO, could be
implemented
OPENRISK YES YES YES
E:12 GMPEs + USGS Combined,
O:Scenario desired (historical) or User-
defined
NO
OSRE YES NO NO Given Hazard Curve using NIED data NO
PAGER YES YES NO
O:real-time GMs and attenuation
equations
ShakeMap based
QLARM YES YES NO
O:real-time GMs, E:Standard attenuation
functions (M,D)
ShakeMap based
RADIUS NO YES NO 3 pre-defined GMPEs NO
REDARS YES YES YES user-defined ShakeMap based
RiskScape YES* YES* NO*
Still in production but historical
earthquake and GMPEs
NO*
ROVER-
SAT
NO NO NO Screening Method YES
SAFER YES* YES* NO*
Still in production but ShakeMaps to be
produced
ShakeMap based*,
Aftershock RAFT
SELENA NO YES NO
user-defined logic based weighting of
GMPEs
ShakeMap based
SES2002 &
ESCENARIS
NO YES NO Spanish intensity GMPE NO
SIGE NO YES NO
Regional intensity GMPEs, weighted
GMPE from local Intensity to intensity
Updated on new
earthquake info
(ESPAS)
SP-BELA YES YES YES user-defined displacement spectrum NO
StrucLoss n/f YES n/f Empirical GMPEs NO
O: Observed, E:Empirical, T:Theoretical, M:Magnitude, D:Distance
4.3.4 Localised Site Effects
The observed, empirical and theoretical methods take into account the source and path effects.
However, local site effects can be modelled in many different ways. Site amplification can
cause a large change in ground motions. The site classification scheme is generally either
geotechnical (based on material type, V
s,30
, site class or SPT/CPT), geological (i.e. derived
from the rock structure or surface geology of the site), or borehole methods using a soil
profile (only used in LNECLOSS by calculating a transfer function from the soil profile for
each grid cell and is extremely time consuming).
The correction factor is defined by modification of bedrock ground motions in most cases and
therefore can be defined as frequency dependent or independent. Frequency dependent
methods include that of HAZUS which is detailed by spectral modification of the bedrock
spectrum via the NEHRP method of 1997 (code design spectra) (ATC, 1997). Soft soil
deposits have been found to have longer periods and may contribute more to damage of taller
structures. This can be seen by the general formula (Kramer, 1996):-
min , 0
4 / h V f
surf s
= (4-1)
Frequency independent methods include multiplying by specific factors (PGA or MMI) or by
geotechnical scales etc. This is not technically correct due to the nature of intensities being
discrete quantities; however, it is a fast method.
Table 4-9: Local site effects modelling within the selected ELE Software Packages
ELE
Software Site Classification Scheme
Correction
Factor Details
CAPRA u/c u/c Not yet given
CATS Geological Freq. Ind.
Geologically based correlation between surface
geology and relative intensity MMI
DBELA Geotechnical and user-defined n/a Considered within the GM application
ELER Geotechnical Freq. Dep. Vs30 expected or within GM
EmerGeo Geotechnical Freq. Ind. Site class factors multiplied by PGA - MMI
EPEDAT Geological Freq. Ind. Intensity geologically based
EQRM user-defined Freq. Dep.
user-defined amplification of bedrock spectra which
is site class, bedrock PGA and Mw binned.
EQSIM *Geotechnical n/a
Considered within the GM application site class B
for Bucharest test case
Extremum Geological*, n/a n/a
Most likely background model supplemented by soil
data where given
HAZ-Taiwan Geotech Freq. Dep. NEHRP (1997) changing of bedrock spectrum
HAZUS-MH Geotech Freq. Dep. NEHRP (1997) changing of bedrock spectrum
InLET Geotech Freq. Dep. NEHRP (1997) changing of bedrock spectrum
LNECLOSS Borehole Freq. Dep.
Transfer function from 1D NLA (using 37 Lisbon soil
profiles)
MAEViz Geotech Freq. Dep. NEHRP or user-defined
OPENRISK Geotech Freq. Ind.
Changing of the bedrock from user-input therefore
can be applied how wanted many source options
also using fault modelling
OSRE NO NO Within the GMPE used to produce NIED data
PAGER Geotech Freq. Ind. Conversion from Vs30
QLARM
Geological,
QL2:Geotechnical=national soil
classification or inferred from
radar for Vs30
QL2: Freq.
Dep. QL2: Radar or simple f-dependent
RADIUS Geotech Freq. Ind. Geotech scale used to modify PGA or user-defined
REDARS Geotech Freq. Dep. NEHRP (1997) changing of bedrock spectrum
RiskScape u/c u/c u/c
ROVER-SAT NO NO
SAFER u/c u/c u/c but could be QL2 or SELENA/HAZUS based
SELENA Geotech Freq. Dep. NEHRP (1997) changing of bedrock spectrum
SES2002 &
ESCENARIS NO NO Not considered
SIGE Geological Freq. Ind.
Local Intensity corrected geologically before
converting to MCS
SP-BELA n/a n/a Considered within the GM application
StrucLoss Geological Freq. Ind. Intensity geologically based
4.4 Vulnerability and Exposure Modules
The vulnerability and exposure modules of the various software packages are very different as
most have been developed for different purposes (post-earthquake, regional calculations,
awareness, forecasting or historical modelling). For the decision making process associated
with models for a global integrated model, these factors should be integrated.
4.4.1 Exposure Elements
The general building stock is looked at in all cases apart from REDARS which is a
transportation-based model. The general building stock detail differs depending on the region
where the models have been produced. For emergency relief, such methods are extremely
important.
Large loss potential facilities are those which could result in many deaths if breached. This is
identified but not covered in HAZUS in terms of nuclear powerplants, military installations
and dam breaches. Critical elements on the other hand are those which are needed for
emergency relief and/or the most important links in the infrastructure examined. Identification
of such elements has been undertaken by only a few models (such as MAEviz). The test area
for MAEviz was reasonably small and therefore could go into more detail; however, their
system is extremely good. Transportation and utility systems are taken into account in more
software packages and should be observed in terms of disruption to communities as a result of
earthquakes and the economic and social loss implications (RADIUS, ELER, CATS, HAZUS
based and StrucLoss looked at these). HAZUS and MAEviz consider loss of functionality as
well as direct damage in their application of utility and transport systems. For secondary
effects, such information is invaluable as ground failure plays a large role in the physical
damage of transportation and utility systems. These systems could be quickly added through
the use of remote sensing identification in a region.
Table 4-10: Inventory elements considered for physical damage (normal) & information purposes (italic)
for the selected ELE Software Packages
ELE Software General Building Stock
Large Loss
Potential
Critical Transportation Utility
CAPRA Central American stock u/c u/c u/c u/c
CATS
85 structures, government,
industrial, many databases
YES, complex Emergency services YES, complex
Oil and Gas, Electric
Power,
Communications,
Water supply etc.
DBELA
Classified by type, fail mode,
design level - Residential,
Commercial/Industrial
User-defined,
future
User-defined, future
User-defined,
future
User-defined, future
ELER Detailed but u/c YES*
PP, high precision,
manufacturing, high
energy, refineries,
oil pipelines, gas,
water pipelines,
airport ops,
computers*
YES*, Trains,
Tunnels, Bridges
Power stations, oil,
gas, computer
centres,
telecommunications
EmerGeo
Industrial, Commercial,
Residential, Education,
Medical
n/f n/f YES n/f
EPEDAT
Residential, Agricultural,
Mining Charity/Religion,
Governmental, Education,
Misc.
NO NO
Highways and
major roadways
NO
EQRM
Australian versions of
HAZUS
NO NO NO NO
EQSIM
HAZUS + 14 European
types
NO NO NO NO
Extremum No discrimination NO NO NO NO
HAZ-Taiwan
Charity/Religion,
Governmental, Education
n/f n/f
Highways,
Railways, Bus,
Ports, Harbours,
Ferry, Airports,
Bridges, Tunnels
Potable & waste
water, oil, natural gas,
electrical power,
communication,
buried pipelines
HAZUS-MH
Charity/Religion,
Governmental, Education
Nuclear PP,
Military, Dams
Police, fire stations
Highways,
Railways, Bus,
Ports, Harbours,
Ferry, Airports
Potable & waste
water, oil, natural gas,
electrical power,
communication
InLET
Residential
NO NO
Highways and
major roadways
NO
LNECLOSS Residential NO NO NO NO
MAEViz
Residential (12 types),
Commercial (10 types),
Industrial, Agricultural,
Religion/Non-profit,
Government, Education.
Not found (but
could be in a
function
somewhere)
Fire Stations, S, M,
L Hospital, Medical
offices/clinics, Police
Stations, Grade
Schools, Colleges
and Universities
Bridges
Buried Pipe, Water
Tank, Electrical
Substation or Power
Plant, Potable Water,
Electric Power,
Pipeline, Water
Network, Utility
Builder
OPENRISK CUREE-Caltech types NO NO NO NO
OSRE ATC-13 and Japanese types NO NO NO NO
PAGER NO NO NO NO NO
QLARM NO, QL2: incorporated
NO, QL2:
incorporated
NO, QL2:
incorporated
NO, QL2:
incorporated
NO, QL2:
incorporated
RADIUS
Residential, Education,
Medical
NO
Medical - need to
increase grid cell
importance
Roads, Bridges,
Tunnels
Water, Reservoirs,
Gas, Electric Power
REDARS NO NO NO
Highways and
Bridges
NO
RiskScape u/c u/c u/c u/c u/c
ROVER-SAT
Dwelling, Other Residential,
Public Assembly,
Emergency Services,
Commercial, Offices,
Industrial, Government,
Historic, School
NO, could be
user-coded
NO, could be user-
coded
NO NO
SAFER u/c u/c u/c u/c u/c
SELENA
Norwegian versions of
HAZUS
User-coded User-coded User-coded User-coded
SES2002 &
ESCENARIS
Residential, Medical,
Commercial and Industrial,
Dams, Rivers,
Nuclear PP,
Chemical
Fire Stations,
Hospitals
Airports,
Highways,
Railways
Identified
SIGE Residential Identified Identified Identified Identified
SP-BELA
Classified by type, fail mode,
design level - Res,
Commercial/Industrial
NO NO NO NO
StrucLoss YES NO NO YES
Potable & waste
water, natural gas,
electric power,
communications,
Koeripipe

4.4.2 Synopsis of Occupancy, Structural and Quality Criteria for Vulnerability
In order to identify the building criteria within the inventory classes, the same criteria has
been used as in the ELER study report (Stafford et al., 2007), however with updated
definitions as required for a diverse model.
The occupancy criteria refer to the building use within the modelling. The occupancy rate is
usually calculated by this use, but is in some cases defined through the relative percentage of
people using that structure at three times during the day. Detailed social loss systems
(MAEviz, EPEDAT, EmerGeo) include all types of buildings and possible combinations
including dynamic movement of people.
The structural criteria are first based on the basic structural features. Most building stocks
are all wood, steel (different levels), RC including precast concrete, concrete shear walls,
URM, and reinforced masonry and provide information for calculation for all types. DBELA
and SP-BELA have as yet only been developed for RC and masonry buildings. As most of the
worlds damage is within these two classes of buildings, this is not a big problem. The second
criteria are behaviour-influencing structural features. This is usually based on building height
but can also include features such as irregularities in plan and elevation or soft-storey
recording. Where a number is provided, this refers to the number of height bins. This
information should be used if provided in a region. A screening method or remote sensing can
be employed for future detection of such features. A factor-based system could be employed
for a percentage of buildings based on a sampling system. If the software package calculates
complex failure mechanisms such as column-sway and beam-sway for the DBELA system,
shear capacity for the SP-BELA system, or soft-storey for EQSIM this can be used to group
the buildings. EQSIM also uses DMT, which is a before and after screening mechanism,
which can then be used in disaster recovery, but the data is also used for calibration of future
models.
The quality criteria take into account the relative quality of the construction. In locations
where building practices have changed over the past 100 years with the advent of seismic
building codes this affects the level of design and therefore this can be used to determine the
construction confidence for vulnerability calculations. Generally four age classes are used, but
this depends on the location in the world, so therefore this has not been included for all
software packages, as it is variable. It is only included for those fixed location studies
(LNECLOSS, SIGE, SES2002, and EQSIM-Bucharest), whereas in some cases such as
DBELA, SP-BELA, QLARM and EQSIM, there is effort made to look at the variability in
construction materials. This variability can be modelled through the application of a
probability density function and applying a Monte Carlo simulation to calculate the
differences between buildings in a particular class. This is both a quality and structural
criterion and is also associated with more complex models.
Table 4-11: Building Criteria considered for the reviewed ELE software packages
Occupancy Criteria Structural Criteria Quality Criteria
ELE
Software
Use
Occupancy
Rate
Basic
Structural
Behav.
struc
features
Complex
Failure
Age,
Classes
Variability
CAPRA YES u/c YES u/c YES u/c
u/c but
desired
CATS YES Proprietary YES Prop. Prop. Prop. Prop.
DBELA if provided if provided YES YES
Disp. Based
(column or
beam)
YES YES, pdf
ELER YES n/f YES YES NO YES NO
EmerGeo YES YES NO NO NO YES n/f
EPEDAT YES
YES, day
and night
for all
classes
YES
YES, 3
levels
NO
YES, 7
levels
NO
EQRM YES NO YES YES NO YES NO
EQSIM YES NO YES, in- YES, 4 Soft storey, YES, 5 YES occ.
depth Levels DMT class
Extremum Pop-based NO YES YES NO YES
YES -
inherent
HAZ-Taiwan YES NO YES YES NO YES NO
HAZUS-MH YES NO YES YES NO YES NO
InLET YES NO YES YES NO YES NO
LNECLOSS NO NO YES YES NO
YES, 3
levels
n/f
MAEViz YES YES YES YES NO YES
allows
input
OPENRISK YES NO YES YES NO
Design
life
NO
OSRE YES NO YES YES NO YES NO
PAGER Pop-based NO NO NO NO NO NO
QLARM
QLARM:-
Pop-based
NO
QL2:YES
YES
QL2:YES
YES,
QL2:YES
NO YES
YES,
inherent
RADIUS YES NO YES YES NO YES NO
REDARS NO NO
YES,
bridges
YES,
bridges
NO YES NO
RiskScape u/c u/c u/c u/c u/c u/c u/c
ROVER-
SAT
YES NO YES YES NO YES NO, single
SAFER YES u/c YES YES u/c YES u/c
SELENA YES
YES: user-
defined
YES YES NO YES NO
SES2002 &
ESCENARIS
YES NO YES
YES, 3
levels
NO
YES, 3
levels
NO
SIGE YES NO YES
YES, 3
levels
NO
YES, 6
levels
NO
SP-BELA if provided if provided YES YES
Shear based
(column or
beam)
YES YES, pdf
StrucLoss NO NO YES YES NO YES NO

4.4.3 Vulnerability Methods and Damage Classes
Both analytical and empirical vulnerability methods have been presented above in 2.4. These
were identified for each of the selected ELE software packages. The capacity is defined by the
vulnerability in that the screening method, vulnerability index and damage probability matrix
methods define the conditional probability in a given damage state using each vulnerability
class based on a certain intensity measure. Extremum and QLARM use a mathematical type
of system in the same way to calculate the damage. For HAZUS-type capacity spectrum
method software, simple empirical pushover curves are used, as described previously and
lognormal probability is calculated from the fragility curves. This is similar for DBELA and
SP-BELA but mechanical principles and probability physical demand vs. capacity for a
certain ratio of buildings are used (they do not use binning). StrucLoss has implemented, and
ELER is to implement both analytical and empirical methods in a three level process. The
damage classes are mainly crack-size-based, as defined by HAZUS (i.e. direct damage), and 5
damage classes are used in most of these methods. There are many variations to the number of
damage classes with DBELA using 4 damage states and some systems using gradation-based
systems. In some cases damage distributions can be updated as data comes available and can
be seen in SELENA, EPEDAT, REDARS, SAFER, EQSIM and will be implemented in
ELER. The truly open source methods can be modified in order to add this feature. An
analysis of capacity spectrum-based vs. displacement-based methods will be undertaken in
order to identify the critical differences between the two and to look at the application for a
global system.
Table 4-12: Vulnerability Method and Damage Classes for the considered ELE Software Packages
ELE
Software
Vuln. Method
No. of DCs and
basis
Damage Classes
CAPRA Analytical or Empirical
HAZUS or
displacement-based
Limit States or Damage States
CATS Vuln. Index/DPM 4, intensity based None, Light, Moderate, Severe
DBELA
Displacement-based or
Vuln. Fns. based on
DBELA properties
4, mechanism based Strain based LS1, LS2, LS3
ELER
L0 and L1: EMS98
Vuln.Index, DPM,
L2: HAZUS or other Sd
method
L0 and L1: 6, L2: 5
D0-D5 for levels 0 and 1, HAZUS for
level 2
EmerGeo Vuln. Index/DPM 6, st. dev. Based sd=-1 to 0, Mean, 0-1, 1-2, 2-3, 3+
EPEDAT Vuln. Index/DPM 6, repair based
None, Slight, Light, Mod., Heavy,
Major
EQRM
HAZUS CSM OR MMI
damage curve basis
5, HAZUS None, Slight, Mod, Ext, Complete
EQSIM
HAZUS derived capacity
curves for use in fragility
functions
Extremum
Mathematical vulnerability
functions
6, damage based d=0-5, same as QLARM
HAZ-Taiwan HAZUS CSM 5, HAZUS None, Slight, Mod, Ext, Complete
HAZUS-MH CSM 5, Crack width, disp. None, Slight, Mod, Ext, Complete
InLET
HAZUS MH damage
functions
LNECLOSS HAZUS CSM 5, HAZUS None, Slight, Mod, Ext, Complete
MAEViz
HAZUS CSM OR
USGS/NEHRP
4, physical view Insignificant, Mod, Heavy, Complete
OPENRISK User-made Vuln. Curves, Damage curves Not specified
OSRE
Vulnerability Curves,
DPMs
Damage Curves Damage, 7 in DPM to form
PAGER
P2: Vulnerability Fns, just
pop exposure at moment
5 vulnerability codes,
3 PAGER Ratings
No, not currently P2
QLARM
Mathematical Vulnerability
Fns, QL2: Vuln. Index
QL2:6, probabilistic
(damage)
QL2: D0-D5
RADIUS DPM-based gradation based Zigong-No, Slight, S-L, L-M, M-H, C
REDARS HAZUS Fragility curves 5, HAZUS None, Slight, Mod, Ext, Complete
RiskScape
Empirical then future
may be analytical
unknown Unknown
ROVER-SAT Screening Method 3, Rating System Red, Orange, Green
SAFER HAZUS CSM or QL2 type 5, HAZUS None, Slight, Mod, Ext, Complete
SELENA
HAZUS CSM or MADRS
(Modified CSM method)
SES2002 &
ESCENARIS
Binomial DPMs or Vuln.
functions
6, crack based D0-D5
SIGE
Empirical fragility curves
(binom. DPM based)
6, crack based D0-D5
SP-BELA
Simplified Pushover (D-
based)
4, mechanism based Strain based - LS1, LS2, LS3
StrucLoss Vuln. curve OR HAZUS 5, HAZUS None, Slight, Mod, Ext, Complete
4.5 Specific Cost Module
The specific cost module is the way that the damage data is converted to social and economic
loss estimates as well as how these methods can be integrated.
4.5.1 Social Losses
As was commented upon during the literature review, there is increasing social vulnerability
and this can be correlated to development and other issues. Including social vulnerability,
such as has been done in MAEviz, allows for a greater level of information to be obtained
from the vulnerability modelling results for use in disaster response (homeless shelter needs
etc. as detailed below in Table 4-13. Nevertheless, social losses are included to some degree
in most of the ELE software packages reviewed. In some cases for the post-earthquake
methods, disruption details are also included. DBELA has been updated to include social
losses for Turkish systems. The occupancy criteria above affect the complexity of the
calculations of social losses. For the injury data, the level indicates the severity of injuries
(minor, moderate, major etc.). Many of the software packages calculate a level of homeless
based on the top three levels of damage (Moderate, Severe and Collapsed) for HAZUS-based
models, and similarly for other systems.
Table 4-13: Social Losses for ELE Software Packages Reviewed
ELE Software Deaths Injured Homeless Other
CAPRA YES YES YES unconfirmed
CATS YES
YES,
Complex
YES, Complex Detailed system & road disruptions
DBELA YES
YES, 4
levels
YES via Bal et al. (2008a)
ELER YES YES YES
L0:Regionally adjusted fatality vs. EMS98
relationships AND L1, L2:damage-based,
L0=Samardzhieva et al. (2002), L1=D4+D5
=deaths, 4xdeaths = injuries (ATC13)
EmerGeo YES YES YES Privatised System
EPEDAT YES YES YES Day and night etc.
EQRM YES
YES, 1
level
NO
EQSIM YES
YES,
Hazus
YES Advanced Recovery System
Extremum YES
YES, 1
level
YES
HAZ-Taiwan YES
YES, 3
levels
YES & shelter
needs, long-
term housing
Night and Day, Indoor/Outdoor/Commuting,
Disruptions
HAZUS-MH YES
YES, 3
levels
YES & shelter
needs, long-
term housing
Night and Day, Indoor/Outdoor/Commuting,
Disruptions
InLET YES
YES, 3
levels
YES
LNECLOSS YES YES YES
MAEViz YES
YES,
Complex
YES, Complex
Social Vulnerability Analysis, Business
Interruption Algorithms, Dislocation Analysis,
Short Term Shelter Needs, Shelter Supply,
Temporary Housing Data Analysis and
Optimisation
OPENRISK YES* NO NO Damage-based
OSRE NO NO NO
PAGER
NO,
P2:YES
NO,
P2:YES
NO, P2:YES
Population exposure, P2: integrates vulnerability
and uses a 3 tier approach based on the
PAGER-CAT. Prone areas = derive social loss
from previous EQs, Developed = analytical
models due to improved standards, Semi-
empirical = function of collapse rates, occupancy
and vulnerability for every country.
QLARM YES
YES, 1
level
NO, not shown
Values are rapid estimations, QL2: improved
estimates.
RADIUS YES YES NO
REDARS NO NO NO Travel time delays, Disruptions
RiskScape YES YES
YES (shelter
needs)
Is looking to be HAZUS based.
ROVER-SAT NO NO NO
SAFER YES YES YES
SELENA YES
YES, 3
levels
NO Coburn and Spence adapted
SES2002 &
ESCENARIS
YES YES YES Coburn and Spence based
SIGE YES YES YES
SP-BELA YES
YES, 4
levels
YES via Bal et al. (2008b)
StrucLoss YES
YES, 3
levels
NO D4 and D5 combined = deaths

4.5.2 Economic Losses
The economic losses can either be calculated as a result of direct or indirect loss (as explained
previously). Most methods simply use the damage data and the MDR calculated from it, in
order to derive an economic loss. No economic loss estimate will ever be exactly accurate;
however MAEviz and HAZUS-MH should be identified due to the complexity of their
calculations. OPENRISK investigates economic loss in depth using benefit-cost ratios, EAL
and LE damage exceedance matrices. This is a proxy for the proprietary software packages. In
addition, disaggregation is also possible within many economic loss softwares. This is where
the economic information can be either defined for different ground motions or for particular
magnitude-distance combinations, or as a full analysis (EQRM Patchett et al., 2005). As
many of these ELE software packages have been produced for a certain test region,
retrofitting studies have also been undertaken in order to compare whether it is better to
retrofit before a disaster to reduce losses (but have the cost of retrofitting), or to not retrofit
the building stock and simply have increased costs in the disaster. This is being currently
taken into account for economics but could also include the social consequences (StrucLoss,
LNECLOSS and OPENRISK).
Table 4-14: Economic Losses for ELE Software Packages Reviewed
ELE
Software
Direct Indirect Methods Disaggregation
CAPRA YES Expect
Cost-benefit analysis, Damage-based, Indirect
Module expected, HAZUS type etc.
YES
CATS YES* YES* Now proprietary - Damage-based 10 yrs ago YES*
DBELA YES YES Damage-based user-defined methods YES
ELER NO* NO*
Not seen in documentation but can be produced
via damage calculations.
NO
EmerGeo YES* YES* Proprietary YES*
EPEDAT YES NO Direct intensity based - insurance NO
EQRM YES NO
Australian adapted, damage-based, 3 struc.
Components, aggregated loss
YES
EQSIM NO NO Future possible from Damage-based NO
Extremum YES NO Damage-based NO
HAZ-Taiwan YES YES HAZUS based YES
HAZUS-MH YES YES
Direct: damage-based, structural, non-structural
(Acc and Disp. Sensitive), contents, Indirect: Loss
Module based on historic, time, disruption etc.
YES
InLET Future Future Damage-based unknown
LNECLOSS YES YES Damage-based, Retrofitting YES
MAEViz YES YES
Full Direct as HAZUS and Indirect including Fiscal
impact, shelter, temp. housing, pop. Dislocation
for buildings and utilities. Also utility functions.
YES
OPENRISK YES NO
Portfolio EAL, Single BCR, LE, EAL via Damage
Exceedance Matrices
Possible
OSRE YES NO Loss curve NO
PAGER NO NO NO
QLARM
YES,
QL2: YES
QL2:YES Direct damage cost, disaster management QL2:Possible
RADIUS NO NO NO
REDARS YES YES Network Analysis, component performance, YES
RiskScape YES YES
Unknown methods as yet. Community disruption,
downtime for indirect costs etc.
unknown
ROVER-SAT NO NO NO
SAFER YES unknown u/c unknown
SELENA YES NO Structural damage - direct NO
SES2002 &
ESCENARIS
YES NO %GNP and value, damage-based NO
SIGE YES NO Loss curve - direct NO
SP-BELA YES YES Damage-based User-defined methods YES
StrucLoss YES NO Geocell Damage-based, Retrofitting also YES
4.6 Rapid Response Use and Technological Intricacies
The rapid-response possibility for an ELE software package is a function of the data that is
provided and the number of lifeline, critical and general building stock details it has. This will
be further described in the ensuing matrix and flowchart analysis. Many of these software
packages are only produced for rapid response (eg. PAGER, QLARM, ELER and SAFER)
and therefore there is a trade-off between accuracy and speed of results. They use fast
mathematical models and assumptions. However, the new SAFER model using a fast-based
SELENA algorithm is the most advanced as it is the first to use an analytical model in near
real-time.

Technology has a major role in the improvement of ELE software packages. Modern GMPEs
have slightly less epistemic uncertainty than previous models and increased modelling of
complex features such as directivity and hanging wall effects. However, this reduction is
minor compared to the gains being made in characterising exposure via remote sensing data
and the increasing number of layers of GIS data for use in these models. GIS data for soil
layering and increased knowledge of topography and fault positions will improve modelling.
Table 4-15: Rapid Response Capabilities computed in ELE Software Packages Reviewed
ELE Software Present?
Onsite
Details
Methods Used GIS/Remote
CAPRA YES YES* As user defined - can tie into a fast response GIS/Remote
CATS YES YES Consequence Assessment Tool Set GIS/Remote
DBELA NO NO Not currently - but proposed in Chapter 4.2 GIS
ELER YES YES*
USGS-style Shake Maps and Loss Maps
produced
GIS/Remote
EmerGeo YES YES
Quick outputs of PGA and MMI, Crisis
Management System
GIS/Remote
EPEDAT YES NO Shakemaps GIS
EQRM NO NO GIS
EQSIM YES YES
Advanced Disaster Management Tool, CAD
Damage Output
GIS/Remote
Extremum YES NO
Quick MMSK86 intensity calculations on
mathematical formulas
Settlement-based
HAZ-Taiwan YES YES
Management Systems, Shake Maps, Loss
Maps
GIS/Remote
HAZUS-MH YES YES
Management Systems, Shake Maps, Loss
Maps
GIS/Remote
InLET YES YES User-defined shakemaps GIS/Remote
LNECLOSS NO NO None defined GIS/Remote
MAEViz YES YES
Social Solutions, GIS Mapping, ability to
produce ShakeMaps
GIS
OPENRISK NO NO NO
OSRE NO NO NO
PAGER YES NO Shakemaps Settlement-based
QLARM YES NO Lossmaps
GIS
QL2:GIS/Remote
RADIUS NO NO Not in coding, but strategies GIS/Remote
REDARS YES YES
Post-earthquake decisions for transport
systems
GIS/Remote
RiskScape YES NO Shakemaps GIS/Remote
ROVER-SAT YES YES
Post-earthquake Screening Method, ATC-
20, ShakeCast
GIS/Remote
SAFER YES YES
Rapid versions of SELENA, QUAKELOSS,
GIS, ElarmS, Aftershocks
GIS/Remote
SELENA YES NO Rapid CSM via real-time data, less branches GIS/Remote
SES2002 &
ESCENARIS
YES YES Management System GIS/Remote
SIGE YES YES
ESPAS is a probabilistic module which can
be used to do rapid response modelling of
the affected area - Management System
GIS/Remote
SP-BELA NO NO GIS
StrucLoss NO NO GIS/Remote
Chapter 5. ELE Software Package Parametric Study

5. ELE SOFTWARE PACKAGE PARAMETRIC STUDY
It was the original plan of this thesis to input displacement-based methods into models such as
SELENA. However, the identification of the diversity within ELE software packages
identified the need for a synthesis of the current state of knowledge and to produce a
framework for a truly open source ELE software package. One of the main differences that
directly impacts on the calculation is intensity-based versus spectrum-based methods.
Nevertheless, for the use of a forecasting ELE software package, it is desirable to use a
spectrum-based method. Therefore, displacement-based and CSM-based methods were
chosen to test. Both DBELA and HAZUS are such packages. Thus, a comparison between
ELE software packages has been undertaken to create extra information as to which platforms
should be applied on a more global scale, and these will be identified through the decision
analysis in 0. Data was gained as part of a project for Zeytinburnu, Turkey and thus this was
used as the testing location.
5.1 Choice of platform to apply
A few software packages were identified through preliminary analysis of software packages
for being possible to apply to a case study.

The MAEviz data was found to be slightly different and contained 17000 buildings for
Zeytinburnu. As not enough data was recovered for Zeytinburnu for the DBELA study (11500
buildings), it was removed from the comparison. MAEviz allows easy viewing as a software
package and has Zeytinburnu as one of its tutorials, which allows for an in-depth view into
the social and economic loss functions that it presents. For future analysis in other systems,
these algorithms will be invaluable; however, the level of data required is not always available
in other regions. Therefore, a combination of PAGER2 data and future-derived equations will
need to be used for future versions.

EQRM will be a useful tool to adapt for a Python-based version of DBELA, as it has useful
disaggregation tools and is very HAZUS based. However, it is not a straight forward
modelling software, with many errors to do with the python libraries. Much correspondence
with the authors (Duncan Gray, David Robinson and Trevor Dhu, pers. comm., Geoscience
Australia, 2008/09) was made and some errors with the production were found. Python is
better than Matlab for Object Oriented Programming and therefore will be used for a future
version of open source DBELA.

SELENA requires much data input and has a nice GUI for data input, despite the coding
errors found which required a lot of hard-wired details in the coding to be changed. The new
updated version 4.0 which is less hardwired has a slightly better method of formulating
economic and social losses.

SAFER uses a real time version of SELENA which is included in the current versions 3.5 and
4.0; however, this is also being upgraded currently. It works on the principle of checking
geocells for incoming ground motion data and drawing circles in order to check whether a
certain number of centres have data. Otherwise, a larger radius circle is drawn. It then applies
the spectral ordinates to each of those centroids, as the ground motion will be reasonably
similar and also takes soil amplification into effect.

Figure 5-1: The method for real-time estimation from the SELENA v3.5 manual. (Molina et al., 2008a)
Another faster calculation that is done involves the use of logic trees for SELENA. Because
each of the logic tree decisions takes time, for faster use, they recommend the use of only 3
sources, 3 soil types, 3 attenuation relations and 1 vulnerability function at a time, in order to
reduce the time taken (Lindholm, 2007).

DBELA is essentially an algorithm for displacement-based vulnerability assessment with
some loose guidelines as to how to apply the exposure, hazard and specific cost modules.
Thus, it was decided to code DBELA using the Monte Carlo methodology of Borzi et al.
(2008) applying social functions of Spence (2007), economic and exposure data of Bal et al.
(2008a) and other generic functions for GM determination. Since HAZUS is not open source,
users are either required to buy HAZUS-MH or to code HAZUS such as has been done for
EQRM and SELENA.
The equations for DBELA and HAZUS are as shown in 2.4.3 and only the key points will be
summarised below. A DBELA and HAZUS version was produced and compared with a
modified SELENA software in order to gain a comparison between the two produced codes,
as it is strongly HAZUS-based and provides extra checking for the software produced. The
hardwiring was bypassed by changing the code. There were also a few errors which were
corrected throughout the code, especially in att_sub.m file. The social calculations were
modified to be consistent with Spence (2007) for SELENA.
5.2 Zeytinburnu Case Study Details
The source code for DBELA was derived using the material data of Bal et al. (2008a) for the
Zeytinburnu area in Istanbul. Armijo et al. (2005) have shown that there exists a seismic gap
on fault segments S6, S7 and S8 of the North Anatolian Fault and thus this may correlate to a
direct problem for cities around this fault, namely, Istanbul (Erdik, 2004). The Duezce
earthquake in 1999 was of Mw7.2 and thus a similar magnitude earthquake was set as the
scenario earthquake used for this study. The details of the location and fault segments can be
seen below in Figure 5-2.
The DBELA-based and a HAZUS-based code has been produced using Matlab which allows
damage estimates, economic and social losses to be calculated given a scenario Mw7.2
earthquake for the worst situation on the fault segment which is the closest point to the site
(Further details are available in Appendix C). The coding also allows for GIS visualisation of
the blocks affected. For the purposes of this report, DBELA will be called MDBELA
(Matlab-coded DBELA) and MHAZUS (Matlab-coded HAZUS).


Figure 5-2: Image of the seismic temporal gap around Istanbul showing the fault segments S7 and S8, the
proximity of Zeytinburnu to the fault segments and the problem area, adapted from the GONAF website
using the work of Armijo et al., 2005. This shows the location of the close fault monitoring system of
GONAF as well as the Zeytinburnu district circle in Istanbul, with a straight line to the point where the
Mw 7.2 earthquake scenario was placed.

For SELENA and the HAZUS-based version produced in Matlab, the following system was
used to define the 37 pre-assigned building classes given. Unlike the 37 pre-assigned building
classes for the DBELA method, as the HAZUS method is extremely dependent on confidence
level in the building stock to define the building class, a thorough analysis was undertaken in
order to define what type of building classes would be used.

In order to gain an understanding of the age of the buildings in the Zeytinburnu district, and
hence an understanding of which earthquake code was used to design the building, aerial
photos from 1946, 1966, 1982 and present day were analysed (Figure 5-3).


Figure 5-3: Evolution of the Zeytinburnu District from left to right Aerial Photos from 1946, 1966,
1982 and Present Day (Turkish Government Website, 2009)
It can be seen that a large proportion of the building stock was built between 1946 and 1966.
Design codes that were produced in Turkey before 1975 were the only provided advice for the
construction of the buildings and hence only building stock after 1975 can be considered to be
designed to a seismic design code (Bal et al., 2007). It can be seen that most of the rest of the
building stock was designed between 1966 and 1982. 65% of the RC stock in Istanbul was
designed under the 1975 seismic code (Bal et al., 2008a). Around the late 1980s, a lot of the
northern area in Zeytinburnu (away from the coast) was changed from industrial areas into
rapidly built housing to deal with the increase of people in Istanbul (Bal, pers. comm., 2009).
A lot of the soil classes in the Zeytinburnu area are man-made and thus higher ground
motions can be expected due to the lower shear wave velocity (less consolidated).

Within the HAZUS methodology, it is necessary to define the level design code to which the
buildings have been constructed. The HAZUS methodology is defined using high, moderate,
low and pre-code levels for capacity and fragility curves. In California, high code seismic
design level is defined by post-1973 construction, where a high level of seismic code existed.
The moderate and low code seismic design levels are designed for areas where there has been
seismic design in practice during construction. Unfortunately, although a proportion of the
building stock in Zeytinburnu was built between 1966 and the present, the seismic design
code which has been in place since 1975 has not always been adhered to. There are problems
with construction malpractice, illegal building practices and non-adherence to the codes in
place (Sucuoglu, pers. comm., 2009). Thus, a pre-code seismic design level is appropriate.
There is only a small difference between low and pre-code in the HAZUS methodology;
however, it is important to assume the buildings as pre-code as this is most likely the level to
which they have been constructed in Zeytinburnu district. This is a very subjective part of the
earthquake loss estimation and because the yield and ultimate displacement and accelerations
are calculated based on the tables within the HAZUS-MH Chapter 5 document for damage
estimation, the capacity curves are created by this assumption.

Under the HAZUS definition of building classes, this translates to only five equivalent classes
(Chapter 5 Table 5.1 - Model Building Types). The RC buildings are generally moment
resisting frames, and therefore are defined under the C1 class of buildings. In the building
stock within Zeytinburnu District, the buildings range from 1 to 9 storeys. Thus, there are 3
classes in the C1 subsection of buildings which can classify these types. Differences in steel
type and storey height are not accounted for in this methodology. C1L defines all the classes
of reinforced concrete buildings in the district that are 1-3 storeys high, C1M defines those
from 4-7 storeys, and C1H defines those between 8-9 storeys. The masonry buildings in
Turkey are unreinforced and can be defined under unreinforced masonry bearing walls
URML and URMM; URML for buildings from 1-2 storeys and URMM for the 3-4 storey
buildings within the district.
Table 5-1: Characterization of the building stock according HAZUS Code
Turkish Building Stock Code HAZUS Code
RC-1-a,RC-2-a,RC-3-a,SRC-3-a,RC-1-b,RC-2-b,RC-3-b, SRC-2-b,SRC-3-b C1L
RC-4-a,RC-5-a,RC-6-a,RC-7-a,SRC-4-a,SRC-5-a,SRC-6-a, SRC-7-a,RC-4-
b,RC-5-b,RC-6-b,RC-7-b,SRC-4-b,SRC-5-b,SRC-6-b,SRC-7-b
C1M
RC-8-a,RC-9-a,SRC-8-a,SRC-9-a,RC-8-b,RC-9-b,SRC-8-b,SRC-9-b C1H
M-1, M-2 URML
M-3, M-4 URMM


Figure 5-4: Flowchart detailing the approach of all three produced ELE software packages: SELENA (all
text files and some coding), MHAZUS (total), MDBELA (total), with the relevant input and output
(dashed) files.
MHAZUS
Define Earthquake
GMPEs, and
calculate GM with
site effects
MDBELA SELENA
Deterministic
Analysis chosen
GM.m fault characteristics, GC distance via
coords. and site conditions
BJF1997horiz.m corrected 1997 GMPE for BJF
GM.m production of matrices for 100 GMs, 11
periods for spatially correlated and uncorrelated
versions using Boore (2003) formula, and lhsnorm
and normrnd functions outputs matrices as well
as median field. (GM with Latin Hypercube, rand.)
HEADER.txt - headers required
earthquake.txt fault modelling
attenuation.txt includes details to
use Boore (1997) for Sa=0, 0.3
and 1.0 sec.
att_sub.m modified to allow use
of this GMPE
soil_files.txt lat, long, site class
via NEHRP
Computation of
Building Damage
gmotionsceni.txt
.MAT files for 100 GM fields - spat. corr.
and uncorr. (100x50x11) and median, GIS
MDBELAmain.m
random generation of
3000 buildings using
pdfs of Bal et al. data

1
,
2
,
3
, T etc.
calculated for each
building including
Using hazard .MAT files
in the Sd-T domain
using the , iteration
loop described in 2.4.3
MHAZUS.m
redefinition of 37 classes
into 5 classes.
Pre-code assumption,
fragility + capacity
curves made
Using hazard .MAT files
in the ADRS domain
using the , iteration
loop described in 2.4.3
for CSM
.MAT files of LS damage in cube form no.
of buildings per geocell+build type (H+D)
builtarea.txt area specs.
numbuild.txt no. of buildings
for each geocell and class
ocupmbt.txt occupancy %
vulnerfiles.txt fragility1.txt
fragility values from 5 classes of
HAZUS and capacity curves
cpfile.txt chosen MADRS
method or CSM method as
described in 2.4.3
computetool.m main run
program after startwin.m
douti.txt output damage
sqmctdout.txt m
2
damage
tree weights etc. output
Economic Loss
Calculations
MDBELAmain.m and MHAZUS.m economic
calculations done directly using the formulas
below in text for repair as in Bal et al. (2007)
Redefinition of DBELA from 4 to 5 damage states
ecfiles.txt economic loss defn
needed for 4 damage states based
on Turkish data eloss*d.txt
.txt files produced with MDR for each cell and
statistics over the 2x100+1 damage dists., GIS
eclossesi.txt for each logic
tree branch i, repair cost
given for each geocell
MDBELAmain.m and MHAZUS.m social losses
for day and night populations calculated via eqns
for night and day. Deaths, injuries and homeless by
Spence (2007) equations.
.txt files produced with social losses for each
cell and full statistics, mean, stdev and %
population.txt pop. data
poptime.txt population % at time
injuryi.txt injury % for ocupmbt
Social Loss
Calculations
hlbyinjuri.txt injuries,
deaths by building type
totalinjuri.txt total social
Comparison between individual ELE studies and total results of the ELE assessment rerun of analysis
After the typologies were set, the DBELA algorithm, in order to produce a Monte Carlo
Simulation, required a probability density function for each material property. This was
defined within the Matlab code for truncated normal, truncated exponential, delta, Youngs
characteristic, exponential, normal, logarithmic and F non-central distributions. Thus, this is
setup for nearly every geophysical distribution commonly used. Only 2 examples have been
shown below.
2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
0
20
40
60
80
100
120
140
160
180
200
Regular storey height, h
urm
[m]
p
d
f
1 1.5 2 2.5 3
0
100
200
300
400
500
600
700
800
Ground floor pier height [m]
p
d
f

Figure 5-5: Regular storey height (left panel) and the ground floor pier height (right panel) of the URM
buildings - Comparison between the statistical truncated normal at 2.0 and 2.85m (left) and statistical F-
noncentral (right) distributions (curves) and the histograms generated by Monte Carlo simulations

The differing magnitudes between the ground motion fields suggest a more realistic
representation on the intra-event variability by the smooth spatial correlated version in
contrast to that of the unsmoothed spatially uncorrelated version. Two patches of ground
motion can be seen in the spatially correlated version (right) vs. the randomness of the
spatially uncorrelated version.

This allowed 100 different economic loss scenarios to be produced for the correlated and
uncorrelated types and therefore a corresponding standard deviation of results to be produced
for both MHAZUS and MDBELA methods.
Mean=2.4m
Cov=15%
Mean=2.62m
Cov=8%
5.3 Results
From the damage distributions resulting from both the MDBELA, SELENA and MHAZUS
methods considering the median, the spatially uncorrelated and the spatially correlated ground
motions, the economic and social losses have been estimated.
5.3.1 Damage Distributions
The difference between HAZUS, SELENA and DBELA is that the HAZUS method
distinguishes between no damage and slight damage states. In contrast, for DBELA the no
damage and slight damage limit states are grouped together. Thus in calculating the direct
economic losses, DBELA overestimates the contribution from these 2 damage states. In an
attempt to prevent this, 50% of the buildings were applied in the no damage state for
MDBELA and thus remain undamaged whilst the remaining 50% incur slight damage.
MHAZUS is seen to overestimate each end of the damage classes due to the algorithm used to
define the performance point for the pre-code system. For the adjusted version of this code, a
different area-based function is being applied which should solve this problem and therefore it
should be close to the values of MDBELA and SELENA. SELENA and MDBELA are in
reasonable agreement. Within each of these damage states, the 37 building types have been
calculated separately for each of the 50 geocells for MDBELA and 5 building types for the
MHAZUS and SELENA scripts.
Table 5-2: Relative Building percentages in HAZUS-based damage classes
Method Type None Slight Moderate Extensive Complete
MDBELA Median 8 8 29 29 26
MDBELA Correlated 8 9 28 28 27
MHAZUS Median 16 13 17 4 50
MHAZUS Correlated 30 6 5 4 55
SELENA Median 4.9 8.7 29.6 29.7 27.1
5.3.2 Economic Losses
The MDR was calculated for every cell for every method used. There is much variation in the
MDR due to the hazard associated by the site classification and the reducing ground motion
for the locations further from the fault (due to attenuation).

Longitude []
L
a
t
i
t
u
d
e

[
]
HAZUS: MDR map for the median GM
28.89 28.9 28.91 28.92
40.98
40.985
40.99
40.995
41
41.005
41.01
41.015
41.02
41.025
41.03

Figure 5-6: Comparison for median GM field between the MDR distribution derived with the MDBELA
(left panel) and MHAZUS (right panel). The SELENA method is approx. the same as the MDBELA and
therefore has not been shown.
100 different economic loss models could be produced for the MDBELA and MHAZUS for
the spatially correlated and uncorrelated ground motion, in addition to the one median ground
motion. This is directly comparable with the SELENA values. The SELENA value then
includes variation as well for the 16
th
and 84
th
percentiles. The values are reasonably similar.
Both the MHAZUS and MDBELA methods produce the same spatial distribution of social
losses despite having considerably different estimates. This type of information can be very
useful for emergency response planning and it is encouraging that all methods show the same
patterns for both day-time and night-time events. Also of note are the different spatial
distributions for the day-time and night-time populations, indicating a largely residential area
towards the south of the Zeytinburnu district. Again, this information can prove valuable for
civil defence and emergency relief planning.
Table 5-3: Summary table of Economic Loss Estimates for Zeytinburnu District
Economic
Costs
Type
Mean
(Billion )
Standard Dev.
(Billion )
Min. for 100
runs (Billion )
Max. for 100 runs
(Billion )
MDBELA Correlated 1.78 0.44 0.67 2.48
MHAZUS Correlated 1.62 0.65 0.14 2.53
MDBELA Uncorrelated 1.79 0.13 1.50 2.20
MHAZUS Uncorrelated 1.65 0.23 1.09 2.26
Longitude []
L
a
t
i
t
u
d
e

[
]

28.89 28.9 28.91 28.92
40.98
40.985
40.99
40.995
41
41.005
41.01
41.015
41.02
41.025
41.03
0
0.
1
0.
2
0.
3
0.
4
0.
5
0.
6
0.
7
0.
8
0.
9
1
MDBELA: MDR map for the median GM
SELENA Median 1.63 0.58 n/f 2.53 (Max MDR)
MDBELA Median 1.86 n/a n/a n/a
MHAZUS Median 1.76 n/a n/a n/a
Maximum MDR possible = 1.05x2.407billion (full extensive damage state), n/a = not applicable
5.3.3 Social Losses
If generic social loss estimates were based on these classes of one person per building in half
of the Extensive and the full Complete damage bands (approximate to the D4 and D5 setting
used in ELER level 1 from KOERI (2002)), the number of buildings damaged is
approximately 40% for the MDBELA and SELENA cases and approximately 52-57% for the
MHAZUS case from Table 5-2. Approximately 4500 people killed in MDBELA and 6500
people killed in MHAZUS is much less than the estimates further on. This is also the case for
the ATC-13 injuries consisting of approximately 18,000 people and 26,000 people
respectively.

Social losses for this study have been only calculated from the collapse damage state and are
therefore set a bit higher than normal to account for possible losses in lower damage states
(Spence, 2007). This has been done for the day and night-time populations.

The proportion of the people living in each building type (j) and geocell (i) has been
evaluated, assuming the same number of people for each storey, within a particular cell.
Hence, such a proportion (p
i,j
) has been calculated as the ratio between of the number of
collapsed storeys (
storey
j i
n
,
) and the total number of storeys (
storey
j i
N
,
) multiplied by the total
population estimated to be in those building during the night and during the day for P
n
and P
d
,
respectively.
d n
storey
j i
storey
j i nightorday
j i
orP P
N
n
p =
,
,
,
(5-1)
Table 5-4: Uninjured, Injured and Death (I5) data for the collapse damage state (Spence, 2007)
Type Floors UI I1 I2 I3 I4 I5
Masonry (1F) 23.60% 50.00% 12.00% 8.00% 0.40% 6.00%
Masonry (2&3F) 16.50% 50.00% 15.00% 10.00% 0.50% 8.00%
Masonry (>4F) 9.40% 50.00% 18.00% 12.00% 0.60% 10.00%
RC (1F) 32.90% 30.00% 19.00% 3.00% 0.20% 15.00%
RC (2&3F) 20.80% 30.00% 23.00% 4.00% 0.20% 22.00%
RC (>4F) 9.70% 30.00% 27.00% 5.00% 0.30% 28.00%

In future additions to the equations, social losses will be calculated from all different damage
states. For SELENA, the same method as for the MDBELA and MHAZUS was applied. This
is based on the occupancy rating, number of storeys/building and the damage state. Due to the
increased collapse layer in MHAZUS, more deaths are recorded than for other damage states.
SELENA was calculated for the median and also one standard deviation variation. It must be
re-stated that this is not exactly the same as running 100 different ground motion fields and
taking the mean and standard deviation but gives some basis for comparison.
Table 5-5: Summary table of Social Loss Estimates for Zeytinburnu District for Night-time Earthquake
Scenario
NIGHT
Mean exc. SELENA Standard Deviation (log) (%)
Social
Losses
Ground
Motion
Type Uninj. Injuries Deaths Uninj. Injuries Deaths Uninj. Injuries Deaths
MDBELA Correlated 284138 48136 21383 n/p 34154 15053 80.3 13.6 6.0
MHAZUS Correlated 230532 85909 37216 n/p 67294 29579 65.2 24.3 10.5
SELENA Median 297460 39756 16441 n/p 46885 20157 84.1 11.2 4.6

Table 5-6: Summary table of Social Loss Estimates for Zeytinburnu District for Day-time Earthquake
Scenario
DAY
Mean exc. SELENA Standard Deviation (log) (%)
Social
Losses
Ground
MotionType Uninj. Injuries Deaths Uninj. Injuries Deaths Uninj. Injuries Deaths
MDBELA Correlated 265419 61197 27041 n/p 46808 20385 75.0 17.3 7.6
MHAZUS Correlated 190532 114947 48178 n/p 96942 41597 53.9 32.5 13.6
SELENA Median 271011 59784 22862 n/p 70647.08 29583 76.6 16.9 6.5

A tighter boundary was applied on the MHAZUS approach so this will be re-coded as part of
the redesign of the software and this should mean that there will not be as many extremes in
the calculation of the limit states. Therefore, there will be less buildings in the collapse limit
state and the deaths and injuries will be approximately the same as the MDBELA and
SELENA codes.

All in all, the results are extremely close between the MDBELA and SELENA coding. There
is more confidence that MDBELA is giving the correct results during this method as it takes
into account mechanical principles and can pick up better a soft storey collapse better. Indeed,
it also may pick up that the pre-code assumption used in MHAZUS is too strong, and, if so,
thus a low-code or moderate-code assumption building dependent should be used.
5.4 Discussion and Conclusions for the Case Study
HAZUS-MH is designed for U.S. conditions and therefore there were some limitations when
applying it to the Turkish building stock. SELENA is much the same where it is built for
Norwegian and U.S. conditions but not for Turkish conditions. Many assumptions need to be
made as to the material properties, pre-code, low-code, moderate-code or high-code definition
and hence the MDR from the MHAZUS results. In MHAZUS, the buildings were grouped
into the classes provided (i.e. C1L, C1M, C1H etc.); consequently this does not allow for
changes in the material properties in the Monte Carlo Simulation buildings. MDBELA on the
other hand is able to be defined based on these properties and therefore can be directly
adapted to Turkish conditions. This is the main advantage of the DBELA methodology.

MDBELA was only used for ground shaking in this case, and non-structural effects were not
accounted for. Many software packages include non-structural effects and this could be a
future improvement to this MDBELA code. DBELA could also have been used for
liquefaction and this may be applicable to the class E sites which are built on manmade soil
classes within Zeytinburnu. Both MHAZUS and SELENA were not modified to take
liquefaction effects into account. The HAZUS methodology provides damage estimates for
many associated earthquake effects such as tsunamis and landslides, which were also not
considered.

Both methods are unable to take into account higher mode effects. Within DBELA it is
necessary to define a beam or column sway mechanism and the cutoff when used for
implementation is very rigid, thus producing very different displacement capacities for
definition with fragility curves, as was seen in the fragility curve diagram produced.

MHAZUS can be seen to have a higher ratio of buildings within the collapsed limit states, and
a very rigid cut off results due to the low ductility in the pre-code assumption. Therefore, the
deaths and injuries due to MHAZUS are about two times larger than those of MDBELA. On
the other hand, because buildings need to be demolished in Turkey under the new seismic
regulations if they lie within the Severe or Complete Damage states, MDBELA and
MHAZUS have reasonably similar economic losses.
For Zeytinburnu District such an earthquake would be catastrophic and by looking at the
information provided as to the locations of the social effects such as deaths and injuries, as
well as the infrastructure and lifeline damage locations, disaster response planning can be put
in place in order to greatly reduce the amount of casualties. SAFER, ELER and most other
major software packages have attempted to model the Istanbul scenario earthquake. Policy is
currently in place within Zeytinburnu in order to retrofit buildings within the district to
seismic standards in order to reduce the approximately 25,000 deaths (mean as predicted by
MDBELA), 50,000 deaths (by MHAZUS) or 20,000 deaths (by SELENA) depending on the
time of day. This can be done on a district level or on a cell by cell level. The Zeytinburnu
district with an approximate value of 2.4 billion Euros using a general assumption will have
repair costs for a mean disaster of approx. 1.6-1.9 billion Euros, which is substantial. This is
because repair costs are higher in Turkey than in the HAZUS manual. Standard deviations
over the 100 ground motions also provide a good prediction of the uncertainty of these figures
for insurance and reinsurance. By undertaking earthquake loss estimation projects such as this
and using logic tree approaches such as that shown within this thesis, the Turkish government
can further implement policy in order to save as many lives as possible when the scenario
earthquake hits in Istanbul. A fully probabilistic approach could have also been performed or
further logic tree branches added to SELENA in a probabilistic way.
5.5 Important Optimisations for MDBELA and applications to other software
5.5.1 Sampling Size via Monte Carlo Simulation and other such methods
Statistical distributions such as Monte-Carlo simulations are very computationally expensive
in some cases and therefore the number of buildings used needs to be carefully looked at
where data is provided for a region. If the building stock and hence building types are
unknown, then this number of buildings cannot be reduced, as using 100 buildings to
represent the 2000 buildings in a geocell and in a particular building type is unrealistic.
However, once the maximum number of buildings in a geocell is found, statistically this could
be classed as a minimum sample size desired. For a Monte Carlo simulation, by the law of
large numbers it is desirable to have as large a number of buildings as possible to sample
from, in order to get the statistical distribution for the material properties. The law of large
number states that for a random distribution, the error will reduce by a factor of 1/n where n
is the sample number. Thus, for the Zeytinburnu District, two criteria were set in order to
determine the minimum number of buildings to be used in the analysis (also consequent
checking of these assumptions proved the theory).

1) The number of buildings used in the Monte Carlo simulation was greater than the
maximum number of buildings per geocell and building type (158 buildings in this
district of Zeytinburnu).
2) By the law of large numbers and considering the amount of variability within the
analysis procedures a limit of 10% error total from the distribution was wanted
(corresponding to 100 buildings sampled) of course the mean and standard deviation
may be closer (the 10% error is an upper limit).

Using both criteria, the suggested value of 3000 buildings per geocell per type give an error
limit of approx. 2%, and a value of 200 buildings, as suggested by the study of Daniell et al.
(2009) to satisfy both criteria gives an error limit of approx. 7%. Thus, there will not be a
significant difference in estimates from a Monte Carlo run earthquake loss estimation for
3000 buildings versus the same analysis for 200 buildings, district dependent. This rule also
depends on many properties such as the variance of every component of the material
properties being used in the analysis, and the confidence of the distributions fitted (i.e. the
data given for analysis) (Kreyszig, 2002).

Quasi-Monte Carlo methods may be able to be used in the future in order to improve this
component and speed up the process. The Quasi-Monte Carlo method uses a more efficient
low discrepancy sequence (quasi-random numbers) in order to apply to a given sample space
and thus better samples the distribution, and is faster via this pseudo-random sampling
(Calfisch, 1998). Random walk methods, neural networking or genetic algorithms may also
be able to be used (Daniell and Parken, 2005, Anderson et al., 2006). Artificial neural
networks are an efficient way of applying a non-linear statistical model. They have been used
in early warning systems for ground motion detection (Boese et al., 2003) and are being used
as part of the SAFER project (Fleming, 2008). For DBELA to be used, the most accurate
representation of random samples is wanted for the smallest computational time. Therefore, if
20 representative houses using a genetic algorithm or neural networking could approximately
represent the damage distribution that 3000 Monte Carlo simulation houses would give, then
this is desired. This will need to be looked into further for computational efficiency but also
for the possibility of applying displacement-based methods to a fast post-earthquake
calculation. For the purposes of this Masters thesis, this point has only been discussed. This is
also a problem of limit states. A set limit state imposes a need for more than 20 buildings.
However, if a percentage-based system were employed, although not exactly accurate in
saying that the house was 31.4% destroyed, this accuracy is available via the DBELA
algorithm by using a gradation system within each limit state (i.e. using vulnerability curves).
An attempt will be made to model this for the Zeytinburnu study, in order to allow for a quick
analysis to be made using random properties. A quick example is shown below.

In one geocell, there are 158 buildings, and 2 different amounts of Monte Carlo Simulations
have been undertaken, one with 3000 buildings with random properties, and one using a
genetic algorithm, bootstrapping or neural network method producing 20 buildings with
random properties, using a new gradation based system.

Two ideas were established for this method, the latter of which is preferred due its direct
application. For the purposes of this study, a gradation-based system on the chosen buildings
could be used to calculate a fragility curve type value based on the ratio of reduction factor
and ductility between certain limit states. This would need further calibration and would not
be exactly accurate: i.e. the 20 optimised buildings return values of 2.6, 1.4, 0.3, 3.8 etc. and
then they are distributed into the various classes based on a distribution system (binomial or
otherwise), thus representing not just one limit state for each. Hence, an accurate
representation of the 158 buildings can be gleaned.

Figure 5-7: Empirical Performance Point Limit State using optimally sampled buildings (adapting
diagram of Borzi et al. (2008)
This is very empirical as it reduces the method to something like a discrete intensity-scale
based system. However, it does take into account the variability within a limit state and
therefore can capture the variation of the building group thus meaning that 20 optimised
buildings could represent 158 buildings accurately.

The gradation based system could also be produced as a function of ductility, in which case
there is no need for application to limit states until the final step. This solves the problem of
needing to run the analysis for 3000 buildings and uses vulnerability curves made from the
3000 buildings (3 points - LS1, LS2 and LS3) based on the original DBELA principles.

It then uses the 20 selected buildings via the DBELA vulnerability curves, using the
performance point crossing over with the demand to define the displacement of each building.
This would be optimized by choosing buildings which represent the entire range of possible
buildings and both column and beam sway mechanisms (bootstrapping or the neural network
process). This is a way to reduce the time caught up within data storing and also that taken up
by the iteration to work out if a certain building is within LS1, LS2 or LS3.
2
.
5
5

2
.
6
8


Figure 5-8: Comparison between the fragility curves derived from MHAZUS (circles) and the curves
obtained from DBELA for C1M (circles) vs. RC5b buildings showing the limit states the smooth curves
are lognormal distributions (black rectangles) and the uniform distribution (no symbol) shows the
difference of the beam sway vs. column sway fragility curves
This idea is best applied in the Sa-Sd domain where a number of pushover curves are
produced and then iterated for in a HAZUS-type method (capacity spectrum but using
DBELA capacity curves and reduction factors). Thus, this is essentially applying DBELA
over a smaller number of optimized buildings chosen to represent the sample to a HAZUS-
type method using fragility curves.
Table 5-7: Monte Carlo Simulation of 3000 buildings reduced to the damage state of 158 buildings in the
geocell and using 5 buildings expanded to the damage state of 158 buildings.
Monte Carlo
(3000) None LS1 LS2 LS3
Using 3000 build. 307 777 789 1127
Probability 0.102 0.259 0.263 0.376
In 158 buildings 16 41 42 59
Monte Carlo (5) None LS1 LS2 LS3
In 5 buildings 0 1 0 4
In 158 buildings 0 32 0 126

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
S
d
[m]
P
(
D
S
>
=
S
d
)
Fragility Curve, DBELA RC5b LS1,2 and 3, vs. HAZUS C1M Sl,Mo,Ex,Co
Table 5-8: Fictitious case of using ANN buildings which are optimally chosen to portray the
characteristics using DBELA vulnerability curves
ANN Buildings None LS1 LS2 LS3
Building 1 0.31 0.49 0.15 0.05
Building 48 0.015 0.08 0.352 0.553
Building 293 0.028 0.14 0.319 0.513
Building 781 0.143 0.23 0.23 0.397
Building 2008 0.034 0.15 0.301 0.515
Probability 0.106 0.218 0.270 0.406
In 158 buildings 17 34 43 64

This will be tested in a version of DBELA being applied to SELENA over the coming
months. It is hopefully a faster way of retrieving approximately the same estimate of damage.
Eventually, the vulnerability curves could also be optimized using other methods.
5.5.2 Sampling Size for the material properties needed for DBELA
Once an efficient method has been programmed to model the Monte Carlo Simulation, a more
efficient way needs to be found in order to sample the data for each country if a DBELA-type
algorithm is going to be used worldwide, as data of material properties is required.
A future possibility involves a country development-based approach to random building
properties. This could be based on any data that is gained from a UNDP study, as defined by
social vulnerability (or the new PAGER vulnerability code system, as found on the DVD and
explained in Appendix B).
It has been found through the literature review that the adherence to seismic codes applied and
level of development are linked, and cause a large difference in building practices (for
example are completely different in Turkey, China and the U.S.). Thus the DBELA system
could use this to characterize material properties or uncertainty in material properties.
This is the main advantage of a DBELA type system. It calculates this variability and
translates it into more accurate results, taking into account real uncertainty and more accurate
analysis procedures. Thus, this part of the algorithm could be improved for faster results by
producing probability density functions for building materials on small amounts of local data
with calibration to this PAGER2 value.
5.5.3 Calculation Speed
A large amount of time (up to 90 percent) was found to be involved with the searching of the
limit state which the performance point corresponds to as checked above in the equations
shown in Figure 5-7. There are a number of ways that this has been reduced within the
researched software packages. Not all of these have been coded in Matlab; however, the
optimisation technique involved with reducing the number of periods at which the GMPE is
calculated for can save time in the storage and also the searching process. The shape of the
displacement response spectra is reasonably consistent and only requires at the most four
ordinates. Current HAZUS-based applications mostly use PGA, a short period spectral
ordinate (0.2 or 0.3s) and a moderate/long period ordinate at 1.0s. The current HAZUS model
also recommends a 4
th
spectral ordinate at a long period of 3.0s. This allows for accurate
conversion between the acceleration and displacement spectra using the formula:-
T
where
g Sa
Sd

2
,
.
2
= =
For future applications in DBELA, this can also be applied in order to speed up processing
time as it is a reasonable assumption. In most cases an EC8 or IBC code-based spectra is built
on these points as part of the coding, thus providing equations rather than a random hazard
spectra which does not require an interpolation to calculate the value.

The total number of buildings in each damage state per cell using both HAZUS and DBELA
was calculated for 100 ground motion fields for the spatially uncorrelated and correlated
cases, as well as the median for every cell and every building type wherever there were
buildings within the Zeytinburnu district within that cell/building-type combination. An
optimisation loop was undertaken in the Matlab script to fully optimise this procedure to
decrease computation time.
Table 5-9: Speedup Using Matlab for the interpolation optimisation for a random test
Calls Seconds Total Calls/second Speedup
interp1 51079 58.31 876 1
interp1q 60345 41.28 1462 1.67
Inlined* 142385 3.82 37274 42.6

A lot of time is taken within the interpolation in DBELA and hence this can be optimised in
many ways from the original 1D interpolation function written in Matlab. By removing the
error checking, an interp1q call is 1.67 times faster than interp1. However, by inlining a small
part of this and using vectorisation, this can be made up to 42.6 times faster. A screening
method can also be used which just takes into account where the interpolation point was and
then compares it to the spectral demand vector for each DBELA run. This was attempted, but
the average speedup was only 10-15 times over interp1. Other speedups include using a ceil
function and equi-spaced periods at which the function is calculated in order to apply faster
the interpolation function.

However, as the data is extremely dense i.e. a 4D 201 (ground motion) x4 (damage states)
x37(building classes) x50(geocells) matrix for DBELA and a 201x5x37x50 matrix for the
HAZUS-based method many other speedups such as zeroing matrices and vectorisation were
used to take full advantage of this programming language (Matlab). When coding DBELA in
another language, other advantages may be gained. A Fortran version has already been
produced by members of EUCENTRE (Bal et al., 2007), which is one step to providing a
faster code.
5.6 Future Developments for this software type
Many improvements are planned for these packages which are detailed in the following short
subsections. These concern immediate changes to the current code, modifying the coding to a
global sense, moving to open source faster format, and including a MAEviz type interface
(user-interactive).
5.6.1 MHAZUS, MDBELA and SELENA changes
The MHAZUS code will be rewired to take into account MADRS and also CSM and
use a different algorithm to determine the performance point to align results
approximately with SELENA and MAEviz.
The initial improvements will be the modification of the MDBELA and MHAZUS
codes to include easy file insertion and a GUI-style interface better than that of
SELENA. All of the functions will be un-hardwired (as was programmed for the
purposes of this study) and the GIS system will be further developed to require no
reinput of data for plotting.
A SELENA-style approach involving probabilistic, real time and deterministic
approaches will be added to MDBELA and MHAZUS.
A further update will include a Google Earth visualisation overlay for the boxes in
order to show social, economic and building damage on a global sense.
There will be improved social functions via the new Spence logic tree social functions.
MDBELA will be improved to take into account non-structural elements as well as to
calculate other secondary effects such as liquefaction.
The fastest method to make MDBELA and MHAZUS truly open source will be
explored using Octave (an open source form of Matlab). Little changes are required
only the great circle calculation, as everything else is inlined. All Matlab functions
will need to be checked for compatibility.
FastDBELA will also be explored to add MDBELA into a HAZUS type vulnerability
version with improved definition of fragility curves to be calculated based on the
region. i.e. adding MDBELA into SELENA.
5.6.2 Expanding software into a global basis
The PAGER-CAT database will be applied for a global dataset to supplement any
information (using those functions from PAGER2, to be applied, to attempt to use
worldwide data on vulnerability, Jaiswal (2008b), Jaiswal et al. (2008c), Porter
(2008b), Allen and Wald (2007)).
Development ratings will also be explored to define the distribution for variation in
materials for production of the Monte Carlo simulation buildings. This will require
calibration.
The integration of remote sensing techniques is also required in order to produce this
system which will allow for fast determination of the building height and other such
values. The PAGER-CAT database with 89 representative types (Masonry, RC etc.)
could be supplemented with regional beam and column data.
Period-height relationships are required for timber, steel and adobe buildings. In this
case, however, the coding can be put in place to have dynamic updating as new
information like this becomes available.
Further collaboration with other software developers to integrate worldwide datasets.

All of the above ideas in 5.5 can then be attempted. However, the need to identify which are
the software packages that should be used to convert to this global sense, and which current
packages should the new software be modelled on still needs to be undertaken in a
quantitative sense. Therefore this will be discussed in 0.
Chapter 6. Open Source Procedure for Assessment of Loss using Global Earthquake Modelling

6. OPEN SOURCE PROCEDURE FOR ASSESSMENT OF
LOSS USING GLOBAL EARTHQUAKE MODELLING
6.1 Introduction into Multicriteria Analysis
Many current ELE and other software packages have been identified, reviewed and explained
within Section 3 and 4.1. Gathering the information and synthesising it into key areas allows
for the development of a future system to be produced. The ELER project has recently
attempted a similar identification and produced a 3 level system, Level 0 and 1 being
empirically-based with Level 2 being analytically-based. This has an advantage of allowing
variability and also the difference in speed and accuracy to be accounted for. There are
currently many systems with each of these characteristics, but not quite as accurate.

CAPRA (Anderson, pers. comm., 2009) is attempting to be completely open source and will
allow collaborators through the internet to add code in a Wikipedia-style format. This is
similar to the idea from the Risk-AGORA group led in coding by Porter (2007a, 2008a) who
brings ideas from HAZUS to open source. PAGER2 data is available on the PAGER website
in order to be open source. This is a large project which has included worldwide social and
economic loss models, as well as hazard, vulnerability and exposure data (examples of which
are explained within the appendices). Again, they have used multiple systems to account for
different locations, but only one estimate. QUAKELOSS2 is also developing such a system
(Wyss, 2009, Trendafiloski, 2009) as well as eEQSIM, and the SAFER projects are ongoing.

Within the Open source Procedure for Assessment of Loss using Global Earthquake
Modelling (OPAL-GEM1), the following steps were proposed (OPAL):
1. Overview of all components of Earthquake Loss Assessment (2);
2. Preliminary research and use or methodology of all current ELE software
packages (3);
3. Analysis of the components of these ELE software packages (4); and
4. Loss analysis using the researched ELE software packages and produced ELE
software packages for familiarity with ELE systems and to identify avenues for
optimisation (5)
By following this procedure, a non-user of ELE software is able to have the knowledge to
make an informed decision about the available techniques for use in the myriad of ELE
literature, choose an avenue for future production of a software package, or use existing
software packages as a knowledgeable user. Depending on the desired end-use, each of the
software packages presented has merit. For the purposes of further developing a useable
conclusion from this OPAL-GEM1 (Open source Procedure for Assessment of Loss using
Global Earthquake Models) report for use in OPAL-GEM2 (Open source Program for
Assessment of Loss for Global Earthquake Modelling), a multicriteria analysis will now be
undertaken to determine the software packages and components of these packages.
6.2 Multicriteria Analysis with a Logic Tree Approach
This can be seen in the full version for the criteria applied within each of the generalised
criteria. The generalised criteria are evaluated for OPAL-GEM2. Criteria used in the analysis
of the 28 different ELE software packages and 5 in-production ELE software packages are
presented below in Appendix D. This list of criteria is by no means exhaustive but these are
the components determined through this thesis to be the most important. It is required that
users make their own informed opinions once they have gone through the OPAL procedure.
The format for OPAL-GEM2 has been applied by taking into account the optimal criteria and
will attempt to be implemented with the help and collaboration of those software packages
identified within this report. With respect to the many ELE software packages reviewed and
assessed, a combination of qualitative assessment (Green = best, Yellow, Orange, Red=worst)
and quantitative (numerical) assessment has been undertaken (10=best, 1=worst) for the
criteria with a relative 1-10 ranking extracted from the combination of these, and this has been
weighted for its impact to the final proposed product.

Due to their complexity, hazard, vulnerability and exposure will be equally weighted. The
true open source model availability, global user ability and all-encompassing software
package are the key components of this process and have been included at the same level as
hazard, vulnerability and exposure.
Table 6-1: Multicriteria Analysis for 33 ELE Software Packages, some still under production, using the
extensive criteria detailed in Appendix D
Setup and updating of
the software Methodology
Results and
Usefulness
ELE Software
O
p
e
n
-
S
o
u
r
c
e

S
o
f
t
w
a
r
e

D
e
t
a
i
l
s

U
s
e
r

K
n
o
w
l
e
d
g
e

T
e
c
h
n
o
l
o
g
y

E
x
p
o
s
u
r
e

H
a
z
a
r
d

/

D
e
m
a
n
d

V
u
l
n
e
r
a
b
i
l
i
t
y

D
a
m
a
g
e
-
L
o
s
s

C
o
n
v
e
r
s
i
o
n

F
o
r
e
c
a
s
t
i
n
g

P
o
s
t
-
E
a
r
t
h
q
u
a
k
e

O
u
t
p
u
t
TOTAL
(/100)
Weighting 1.5 1 0.5 0.5 1.5 1 1 1 0.5 0.5 1 10
OPAL-GEM2 9 9 10 9 9 9 9.5 10 10 8 10 93
CAPRA 9 8 8 8 7 9 7 8 9 8 7 79.5
MAEViz 7 7 8 6 7 8 6 9 8 7.5 6 71.75
SAFER 5 7 6 8 7 8 8 7 8 9.5 8 71.75
ELER 5 5 6 7 6 7 8 8 6 9 8 66.5
PAGER2 3 4 7 7 8 7 5 7 7 9 8 62.5
MDBELA 5 6 6 4 6 7 6 7 7 2 7 59
ROVER-SAT 8 6 7 7 7 3 4 6 1 6 7 59
MHAZUS 5 5 6 4 6 7 5 7 7 5 7 58.5
HAZUS-MH 2 4 7 7 6 8 5 7 9 7 7 58
SELENA 5 6.5 6 4 5 6 5 6 7 6 8 58
EmerGeo 1 4 7 8 7 7 6 7 7 7.5 7 57.75
QL2 4 3 7 7 8 5 4 5 8 9 7 57.5
EQRM 7 7 4 5 6 7 5 4 7 3 5 57
CATS 1 5 6 7 6 7 5 8 7 8.5 7 56.75
OPENRISK 6 6 7 4 6 5 6 7 7 2 4 56
EQSIM 1 5 7 8 6 6 6 5 8 9 7 55.5
DBELA 4 6.5 3 4 6 7 6 6 7 2 6 54.5
HAZ-Taiwan 1 4 6 6 5 7 5 7 9 8 7 53.5
SES2002 &
ESCENARIS 3 3 6 7 5 4 7 7 7 7.5 6 52.75
StrucLoss 3 5 6 6 6 5 7 6 6 1 6 52
PAGER 6 5 7 7 8 6 1 1 1 8 6 51.5
SIGE 3 3 6 7 5 5 4 7 7 7.5 6 50.75
RiskScape 6 5 6 8 4 3 3 4 8 7 6 50.5
InLET 1 5 7 7 3 6 5 7 7 7.5 7 50.25
LNECLOSS 4 4 6 5 6 4 5 7 6 1 6 50
REDARS 1 5 8 7 2 7 4 7 7 8 7 49.5
Extremum 0 6 6 4 7 4 4 5 8 5 6 47
QLARM 4 5 7 7 7.5 5 1 1 1 8 6 46.75
EPEDAT 3 4 6 3 4 6 4 5 7 5 5 45
OSRE 6 6 7 3 6 3 4 2 6 1 3 44.5
SP-BELA* 4 2 2 4 6 5 6 6 7 1 2 43
RADIUS 1 4 8 2 5 2 3 2 4 4 6 35
*A full version of SP-BELA has not been sourced as yet
Once more users have read this document and have the necessary experience, a combination
of peoples opinions can be used via an expert opinion in the same way that the old empirical
Damage Probability Matrices were made, in order to create the package. It is difficult to apply
a rating on the generalised criteria; however, the best software packages reviewed are seen to
be towards the top of the ranking, as expected. The lack of knowledge about different
software packages could be supplemented by uncertainty as part of a logic tree approach.

It can be seen that the proposed OPAL-GEM2 will hopefully alleviate a lot of the problems
associated with open source, software licensing and other important setup issues. It will also
look to provide multiple methods to allow user selection in terms of utilising an empirical and
at least one analytical model for vulnerability. MAEviz is by far the best current production
ELE software package that is available for download and that would fit the criteria best to
progress to the next stage. By combining components of other ELE software packages (i.e. the
interface of MAEviz, the technology and post-earthquake production via SAFER, the disaster
management system of EQSIM and the open-source and software detailing of CAPRA and
EQRM), an extremely good holistic package can be produced.
6.3 Future Developments - Framework for future production of OPAL-GEM2
6.3.1 Programming Language to be used
A Python program will subsequently be produced using this technology, as Python allows
increased coding speed through its superior object-oriented programming and combining it
with C++ will improve the time, including a slightly different iteration method which should
be faster. EQRM has already been reviewed and the coding examined, in order to attempt to
apply MDBELA into a form of EQRM over the coming months this will be high on the
priority list to produce a beta version. Then other methods will be added as options (QL2 and
MHAZUS/SELENA types).
6.3.2 Choice of GIS
This Python programming will be incorporated with a GIS package. Although the GIS
method developed as part of this project in Matlab works well, it lacks many of the options of
a modern GIS package, such as ESRI ArcGIS 9.1. However, licensing is costly for such
applications. In order to gain the useability of MAEviz, the interface that is utilised contains
the software interface and GIS system of the program, uDIG. The GUI is actually open
source, and has been downloaded and used in readiness to provide a complete GIS solution
for data plotting of results and the addition of exposure, hazard and vulnerability layers. It is
Java-based and will be interfaced into the Python programming. Another option for the GIS is
another open source program which is called ILWIS 3.5 Open and is a simple GIS machine,
but is very effective.

Figure 6-1: Example of the GIS program, uDIG with possible economic loss geocells (uDIG, 2009)
6.3.3 Collaboration with other partners
There is the need for collaboration with other partners to improve the ideas behind the coding
(from proprietary software packages and other contacted partners). During this project, a
number of different partners have been contacted in order to gain information and exchange
comments about their software. Looking at the results of the multicriteria analysis, Dr. Ed
Anderson, through CAPRA, and the open source architecture manager of CAPRA, Dr. Stuart
Gill, will be very useful to further adapt the ideas of open source extensible architecture,
Wikipedia-style layout, a virtual community and development and global accessibility. In
addition, the team of Duncan Gray, David Robinson and Trevor Dhu producing EQRM will
be of use for Python programming. Prof. Max Wyss who is working with the WAPMERR
QLARM and QUAKELOSS2 projects is also keen for discussion and collaboration and the
architecture which they are using to make their software open source will also be of use. The
social functions should give a new insight (Wyss, 2008) as well. ELER should also be
contacted, as the work of Stafford et al. (2007) and Strasser et al. (2008) provides insight into
more research in this field for decision-making.
6.3.4 Multi-level risk programming
Several projects (CAPRA, ELER and SAFER) have expressed the need for multi-level
programming within the seismic risk module. Through the criteria shown and the OPAL
procedure, it has been seen that no one vulnerability, hazard and damage-loss conversion will
characterise the seismic risk for the exposed inventory. The exposed inventory will first be
defined for 5 different classes (building stock, critical loss facilities, large loss facilities, and
utility and transportation systems). It will be able to be added to eventually in a Wiki style
way, so that users can add data (checked by a moderator) for download by anyone. In order to
gain a global cover, technology such as remote sensing will be employed, as well as further
consultation with the PAGER-CAT detailing. Temporal and spatial changes should also be
accounted for.
For the vulnerability component, a 3 stage process is proposed initially, in order to provide
the user with options as to which system is desired:
Level 1 should consist of an empirical model (vulnerability function-based) which is
set up along the lines of the PAGER2 system or QUAKELOSS2 system which use
globally derived loss functions as well as vulnerability codes and development
indexes;
Level 2 should be a HAZUS-styled CSM-based method in order to take into account
the previous MHAZUS coding and has the advantage in that fragility curves can be
globally adapted; and
Level 3 should be a displacement-based method, adapted from DBELA in order to
correlate better with damage than other analytical methods.
In terms of the hazard components applied with these models, both spatially and temporally
correlated ground motions should be able to be applied by user-defined and user-selection of
observed, theoretical and empirical ground equations, as defined in Appendix D.
The technology behind SAFER and EQSIM should be looked at carefully, as well as
integrating the tsunami results from TRANSFER and a screening process such as ROVER-
SAT for online use, if desired, in a particular location around the world. OPAL-GEM2 will be
able to set up a user-defined architecture with empirical GMPE hazard data that is pre-
defined, selected based on the region, for lower level users.
From the hazard and vulnerability, the calculation of building damage can be done and will be
presented on the GIS in multiple ways.
Up-to-date equations will be sourced for social and economic losses, but as they are desired
for a global context, they should be development-based. Thus, the recent USGS PAGER2
work and QUAKELOSS2 will be consulted, as well as social and economic vulnerability
researchers in other fields. The algorithms of Porter (2008c) used in OPENRISK and those of
MAEviz (MAE, 2009) should be sourced and applied in addition to the future generation of
Spence (2007) equations. The MAEviz data, in terms of social vulnerability, is very useful for
governments and also to relate to the general public, and the algorithms used are a direct
function of the HAZUS system with greater improvements.
The simplicity of the GUI in SELENA to implement all of these components should be taken
into account but a Windows style interface on top of the Python script should be coded.
6.3.5 Forecasting, Post-Earthquake Use and Useable Output
It is envisaged that the three step process could be used for forecasting of earthquakes, or in
post-earthquake use, utilising the criteria detailed in Appendix D. Integration with seismic
networks is therefore essential and the work of QLARM, ELER, SAFER or PAGER could be
implemented in conjunction to gain a range of uncertainties. Instead of re-implementing these
techniques, it may be better to apply such systems globally and to look at the output and
management issues associated. The DMT of EQSIM (soon to be eEQSIM) should also be
closely examined. According to Tiedemann (1992) over 100 helpers in Mexico City were
killed, but the DMT can help to see block movement, and so adding this to a proactive
ROVER-SAT method, with the addition of the OPAL-GEM2 program, would provide
estimates and visualisation data for a real-time project and reduce these losses of emergency
workers and helpers.
For forecasting, the three-level process will be optimised for speed and accuracy. The first
steps have been detailed in 5.6 to improve accuracy and speed, and the change of format to
Octave (or immediately to Python) and cleaning up of the code using these optimisation
techniques will allow for improved forecasting and more emphasis to be put on the output.
It is desirable for the output to have a user-friendly GIS production, which has already been
discussed. In addition, a one page summary comprising loss estimates, the hazard,
vulnerability and exposure data and models used, as well as some graphs, will be presented.
One output that will be improved will be the output file, in order to reduce the time lost
between running the analysis and getting useable results. In the future, as the global database
is improved, recommendations for the communities and region-specific recommendations for
codes can be added to the OPAL-GEM program when a loss analysis in a particular location
is undertaken.
6.4 Conclusion of the Multicriteria Selection for OPAL-GEM2
Within this 1
st
stage, preliminary tools have been applied with a lot of setup and procedural
framework in order to choose the best possible avenue for improvement. Much information
and software has been sourced and much experience obtained. However, with the huge
amount of information and ever-changing methods, there is no one correct opinion and thus
continued collaboration and freedom of information between similar projects will lead to
further adaptation of the methodology. Moreover, through combining the OPAL procedure
with the multicriteria analysis, it is believed that a good avenue to produce the OPAL-GEM2
program has been established.

Chapter 7. Conclusion

7. CONCLUSION
This project (OPAL-GEM1) is an ongoing open source process to produce a truly global open
source ELE software package.
components:
1) Overview of current and new components of earthquake loss assessment (vulnerability,
hazard, exposure, specific cost and technology) identified the disadvantages and advantages
of methods used;
2) Preliminary research, acquisition and familiarisation with all available ELE software
packages was gained;
3) Assessment of these software packages was undertaken in order to identify the ELE
methods used; and
4) Loss analysis was undertaken using 3 ELE Software packages for a test case.
This test case was undertaken for a deterministic earthquake (Mw7.2) for the Zeytinburnu
district, Istanbul, Turkey, by production and use of 2 software packages: displacement-based
MDBELA (Matlab-based DBELA); and CSM-based MHAZUS (Matlab-based HAZUS), as
well as by adapting SELENA for use. This was done to gain an understanding of the process
differences, in order to produce damage, economic and social loss estimates. It was found
that the Zeytinburnu district, in addition to having a repair cost of between 1.6-1.9 billion,
would also have between 16,000-21,500 deaths, and approx 40,000-50,000 injuries if during
the night, and approx. 5000 more deaths and 15,000 more injuries if during the day.

MDBELA was found to be more computationally expensive than the other coding systems
due to the level of complexity, taking into account variability. However, other mediums and
optimisation techniques have been presented for future application in OPAL-GEM2.

Optimisation of the software and ELE components needed for OPAL-GEM2 were identified
through a multi-criteria analysis applied to all ELE software packages, using the knowledge
Chapter 7. Conclusion
gained through the OPAL-GEM1 process. Future improvements to the step 4 in the OPAL
procedure have been recognised and will be undertaken in future work, including conversion
to Octave and Python. From the experience gained through the in-depth view of the state of
earthquake loss assessment, a view forward has been gained through this Masters Thesis for
integrating new technological techniques such as remote sensing data and GIS data into the
coding. OPAL-GEM2 will be a dynamic, open-source, multiple level, Python-based (Java-
compatible) ELE Software Package integrated with an open source GIS package, produced to
to allow for pre-planning for earthquakes in places such as Zeytinburnu and to provide a
solution for both forecasting and rapid loss estimation.

References

REFERENCES
Abrahamson, N.A., Silva, W.J. [2008] Abrahamson & Silva NGA Ground Motion Relations For The
Geometric Mean Horizontal Component Of Peak And Spectral Ground Motion Parameters, Final
Report prepared for the Pacific Earthquake Engineering Research Center, February 2008.
Akkar, S., Bommer, J.J. [2007] New Empirical Prediction Equations for Peak Ground Velocity
Derived from Strong-Motion Records from Europe and the Middle East, Bulletin of the Seismological
Society of America, vol. 97, No. 6; p. 2152-2170.
Akkar, S., Boore, D. [2008] Fundamentals of Seismic Hazard Assessment Course Notes, Rose School,
University of Pavia, Pavia, Italy.
Allen, T.I., Wald, D.J., Hotovec, A.J., Lin, K., Earle, P.S., Marano, K.D. [2008] An Atlas of
ShakeMaps for selected global earthquakes, U.S. Geological Survey Open-File Report 2008-1236.
Allen, T.I., Wald, D.J. [2007] Topographic slope as a proxy for global seismic site conditions (V
S
30
)
and amplification around the globe, U.S. Geological Survey Open-File Report 2007-1357.
Anderson, E. [2008] Central American Probabilistic Risk Assessment, World Bank, Presentation at
EIRD Barbados Conference 8/7/08.
Anderson, N.L., Daniell, J.E., Luey, L.W., Quince, M.J. [2006] Rotation of Reinforced Concrete
Members, Honours Thesis, <published book>, University of Adelaide Press, Adelaide.
Antoniou, S., Pinho, R [2004] Development and verification of a displacement-based adaptive
pushover procedure, Journal of Earthquake Engineering, Vol. 8, No. 5, pp. 643-661.
Applied Technology Council [1985] Earthquake Damage Evaluation Data for California, Report
ATC-13, Applied Technology Council, Redwood City, California, U.S.A.
Applied Technology Council [1988] Rapid Visual Screening of Buildings for Potential Seismic
Hazards: A Handbook, Report ATC-21, Applied Technology Council, Redwood City, California,
U.S.A.
References
Applied Technology Council [1997] NHERP Guidelines for the Seismic Rehabilitation of
Buildings, FEMA Report No. 273, Federal Emergency Management Agency, Applied Technology
Council, Washington, D.C., U.S.A.
Applied Technology Council [2005] Improvement of Nonlinear Static Seismic Analysis Procedures,
FEMA 440 Report, Redwood City, California, U.S.A.
Arias, A. [1970] A Measure of Earthquake Intensity- Seismic Design for Nuclear Power Plants, edited
by R. J. Hansen. The MIT Press, Cambridge, U.S.A., pp. 438-483.
Armijo, R., Pondard, N., Meyer, B., Mercier de Lepinay, B., Ucarkus, G., Malavieille, J., Dominguez,
S., Gustcher, M-A., agatay, N., Cakir, Z., Imren, C., Kadir, E., Natalin, E., [2005] Submarine fault
scarps in the Sea of Marmara pull-apart (North Anatolian Fault): implications for seismic hazard in
Istanbul, Geochemistry Geophysics Geosystems, Vol. 6, No. 6, ; pp. 1-29.
Baker, J.W., Cornell, C.A. [2005] A vector-valued ground motion intensity measure consisting of
spectral acceleration and epsilon, Earthquake Engineering & Structural Dynamics, Vol. 34, No. 10,
pp. 1193-1217.
Baker, J.W., Cornell, C.A. [2006] Which spectral acceleration are you using?, Earthquake Spectra,
Vol. 22, pp. 293-312.
Baker, J.W., Cornell, C.A. [2006b] Correlation of response spectral values for multicomponent
ground motions, Bulletin of the Seismological Society of America, Vol. 96, No. 1, pp. 215-227.
Bal, .E. [2005] Rapid assessment techniques for collapse vulnerability of reinforced concrete
buildings, MSc Thesis, stanbul Technical University, Civil Engineering Department, Istanbul,
Trukey (in Turkish).
Bal, I.E., Crowley, H., Pinho, R. [2008a] Displacement-Based Earthquake Loss Assessment for an
Earthquake Scenario in Istanbul, Journal of Earthquake Engineering, Vol. 11 No. 2, pp. 12-22.
Bal, I.E., Crowley, H., Pinho, R., Glay, G. [2007] Structural characteristics of Turkish building
stock in Northern Marmara Region for Loss Assessment Applications, ROSE Research Report
2007/03, IUSS Press, Pavia, Italy.
Bal, I.E., Crowley, H., Pinho, R., Glten Glay, F. [2008b] Detailed assessment of structural
characteristics of Turkish RC building stock for loss assessment models, Soil Dynamics and
Earthquake Engineering, Vol. 28, pp. 914-932.
References
Barbat, A.H., Ypez Moya, F., Canas, J.A. [1996] Damage scenarios simulation for seismic risk
assessment in urban zones, Earthquake Spectra, Vol. 12, No. 3, pp. 371-394.
Bard, P-Y. [2007] Engineering Seismology Course Notes, Universit Joseph Fourier, Grenoble,
France.
Barranco, L., Izquierdo, A. [2002] Estimacion rapida preliminary de danos potenciales en Espana por
terremotos: simulacion de scenarios sismicos (SES 2002), Direccion General de Proteccion Civil y
Instituto Geografico Nacional, Spain.
Benedetti, D., Petrini, V. [1984] Sulla Vulnerabilit Di Edifici in Muratura: Proposta Di Un Metodo
Di Valutazione, Lindustria delle Costruzioni, Vol. 149, No. 1, pp. 66-74.
Bernardini ,A., Gori, R., Modena, C. [1990] An Application of Coupled Analysis Models and
Experimental Knowledge for Seismic Vulnerability Analysis of Masonry Buildings, in Engineering
Aspects of Earthquake Phenomena. A. Koridze (ed); vol. 3, pp.161-180.
Bird, J.F., Bommer, J.J. [2004] Earthquake Losses due to Ground Failure, Engineering Geology,
Vol. 75, No. 2, pp. 147-179.
Bird, J.F., Bommer, J.J., Crowley, H., Pinho, R. [2006] Modelling liquefaction-induced building
damage in earthquake loss estimation, Soil Dynamics and Earthquake Engineering, Vol. 26, pp. 15-
30.
Boese, M., Erdik, M., Wenzel, F. [2003] Artificial Neural Networks for Earthquake Early Warning,
Fall Meeting, American Geophysical Union.
Bommer, J.J., Bird, J.F., Crowley, H [2006] Earthquake Losses due to Ground Failure (Liquefaction
and Landslides) Proceedings of the 1st European Conference on Earthquake Engineering and
Seismology, Geneva.
Bommer, J., Scherbaum, F., Bungum, H., Cotton, F., Sabetta, F., Abrahamson, N. [2005] On the use
of logic-trees for ground-motion prediction equations in seismic-hazard analysis, Bulletin of the
Seismological Society of America, Vol. 95, No. 2, pp. 377389.
Bommer, J., Spence, R., Erdik, M., Tabuchi, S., Aydinoglu, N., Booth, E., del Re, D., Peterken, O.
[2002] Development of an earthquake loss model for Turkish catastrophe insurance, Journal of
Seismology, Vol. 6, pp. 431-446.
References
Bommer, J.J., Crowley, H. [2006] The Influence of Ground-Motion Variability in Earthquake Loss
Modelling, Bulletin of Earthquake Engineering, Vol. 4, No. 3, pp. 231-248.
Boore, D. M., Joyner, W. B., Fumal, T. E. [1997] Equations for estimating horizontal response
spectra and peak acceleration from Western North American earthquakes: a summary of recent work,
Seismological Research Letters, Vol. 68, No. 1, pp. 128153.
Boore, D.M., Atkinson, G.M. [2008] Boore-Atkinson NGA Ground Motion Relations for the
Geometric Mean Horizontal Component of Peak and Spectral Ground Motion Parameters, PEER
Research Report 2007/01, Pacific Earthquake Engineering Research Center, Berkeley, California,
U.S.A..
Boore, D.M., Gibbs, J.F., Joyner, W.B., Tinsley, J.C., Ponti, D.J. [2003] Estimated Ground Motion
From the 1994 Northridge, California, Earthquake at the Site of the Interstate 10 and La Cienega
Boulevard Bridge Collapse, West Los Angeles, California, Bulletin of Seismological Society of
America, Vol. 93, No.6, pp. 2737-2751.
Borzi, B., Crowley, H., Pinho, R. [2008] Simplified pushover-based earthquake loss assessment (SP-
BELA) method for masonry buildings, International Journal of ArchitecturalHeritage, Vol. 2, No. 4,
pp. 353-376.
Borzi, B., Pinho, R., Crowley, H. [2008a] Simplified pushover-based vulnerability analysis for large
scale assessment of RC buildings, Engineering Structures, Vol. 30, No. 3, pp. 804-820.
Boukri, M., Bensaibi, M. [2007] Indice de Vulnrabilit des Btiments en Maonnerie de la Ville
dAlger, au 7me colloque National (AFPS), 4 au 6 juillet 2007 lEcole Centrale de Paris, France.
Braga, F., Dolce, M., Liberatore, D. [1982] A Statistical Study on Damaged Buildings and an
Ensuing Review of the MSK-76 Scale, Proceedings of the Seventh European Conference on
Earthquake Engineering, Athens, Greece, pp. 431-450.
Bramerini, F, Di Pasquale, G. Orsini, G. Pugliese, A., Romeo, R., Sabetta, F. [1995] Risichio Sismico
del Territorio Italiano. Proposta di una Metodologia e Resultati Preliminari, Rapporto Tecnico
SSN/RT/951.
Cabanas, L., B. Benito, M. Herraiz [1997] An approach to the measurement of the potential structural
damage of earthquake ground motions, Earthquake Engineering & Structural Dynamics, vol. 26, No.
1, pp. 79-92.
References
Calfisch, R.E. [1998] Monte Carlo and quasi-Monte Carlo methods, Acta Numerica. 7, Cambridge
University Press, Cambridge, U.K., pp. 149.
Calvi, G.M., Pinho, R., Magenes, G., Bommer, J.J., Restrepo-Vlez, L.F., Crowley, H. [2006] The
development of seismic vulnerabilty assessment methodologies over the past 30 years, ISET Journal
of Earthquake Technology, Vol. 43, No. 4, pp. 75-104.
Calvi, G.M. [1999] A Displacement-Based Approach for Vulnerability Evaluation of Classes of
Buildings, Journal of Earthquake Engineering, Vol. 3, No. 3, pp. 411-438.
Campbell, K. W. [1997] Empirical near-source attenuation of horizontal and vertical components of
peak ground acceleration, peak ground velocity, and pseudo-absolute acceleration response spectra,
Seismological Research Letters, Vol. 68, No. 1, pp. 154179.
Campbell, K. W., [2000] Erratum to Empirical near-source attenuation relationships for horizontal
and vertical components of peak ground acceleration, peak ground velocity, and pseudo-absolute
acceleration response spectra by Kenneth W. Campbell, Seismological Research Letters, Vol. 71, pp.
353-355.
Campbell, K.W., Bozorgnia, Y. [2008] Campbell-Bozorgnia NGA Ground Motion Relations for the
Geometric Mean Horizontal Component of Peak and Spectral Ground Motion Parameters, Pacific
Earthquake Engineering Research Center, Berkeley, California, U.S.A.
Campillo, M. [2007] Advanced Theoretical Seismology Course Notes, Universit Joseph Fourier,
Grenoble, France.
Cardona, O.D., Yamin, L.E. [1997] Seismic Microzonation and Estimation of Earthquake Loss
Scenarios: Integrated Risk Mitigation Project of Bogot, Colombia, Earthquake Spectra, Vol. 13, No.
4, pp. 795-814.
Casarotti, C., Pinho, R. [2007] An Adaptive Capacity Spectrum Method for assessment of bridges
subjected to earthquake action, Bulletin of Earthquake Engineering, Vol. 5, No. 3, pp. 377-390.
Cetin, K.O., Kiureghian, A.D., Seed, R.B. [2002] Probabilistic models for the initiation of seismic
soil liquefaction, Structural Safety, Vol. 24, No. 1, pp. 67-82.
Chen, W-F., Scawthorn, C. [2003] Earthquake Engineering Handbook, CRC Press, Japan
References
Chien, S-W., Chen, L-C., Chang, S-Y., Chiu, G-H., Chu, C-L. [2002] Development of an After
Earthquake Disaster Shelter Evaluation Model, Journal of the Chinese Institute of Engineers, Vol.
25, No. 5, pp. 591-596.
Chiou, B.S.-J., Youngs, R.R. [2007] Chiou and Youngs PEER-NGA Empirical Ground Motion Model
for the Average Horizontal Component of Peak Acceleration andPseudo-Spectral Acceleration for
Spectral Periods of 0.01 to 10 Seconds, Pacific Earthquake Engineering Research Center, Berkeley,
California, U.S.A.
Choi, E. S., DesRoches, R., Nielson, B. [2004] Seismic fragility of typical bridges in moderate
seismic zones, Engineering Structures, Vol. 26, No. 2, pp. 187-199.
Chopra, A.K., Goel, R.K. [2002] A modal pushover analysis procedure for estimating seismic
demands for buildings, Earthquake Engineering & Structural Dynamics, Vol. 31, pp. 561-582.
Coburn, A., Spence, R. [2002] Earthquake Protection, John Wiley, Chichester, U.K.
Colombi, M., Borzi, B., Crowley, H., Onida, M., Meroni, F., Pinho, R. (2008), Deriving vulnerability
curves using Italian earthquake damage data, Bull. Earthquake Eng. 6:pp.485-504.
Comit Europeen de Normalization (CEN) [2005] Eurocode 8: Design of Structures for Earthquake
Resistance - Part 2: Bridges, EN 1998-2, Brussels, Belgium.
Cosenza, E., Manfredi, G., Polese, M.,Verderame, G.M. [2005] A Multi-Level Approach to the
Capacity Assessment of Existing RC Buildings, Journal of Earthquake Engineering, Vol. 9, No. 1,
pp. 1-22.
Cotton, F. [2008] Theoretical Seismology Course Notes, Universit Joseph Fourier, Grenoble, France.
Crowley, H., Pinho, R., Bal, I. [2009] Earthquake Loss Estimation Course Notes, Rose School,
Crowley, H. [2005] An investigative study on the modelling of earthquake hazard for loss
assessment, Individual Study, European School for Advanced Studies in Reduction of Seismic Risk
(ROSE School), University of Pavia, Italy.
Crowley, H. Pinho, R. [2006] Simplified Equations for Estimating the Period of Vibration of Existing
Buildings, Proceedings of the First European Conference on Earthquake Engineering and
Seismology, Geneva, Switzerland, Paper No. 1122.
References
Crowley, H., Bommer, J.J. [2006] Modelling Seismic Hazard in Earthquake Loss Assessment with
Spatially Distributed Exposure, Bulletin of Earthquake Engineering, Vol. 4, No. 3, pp. 249-273.
Crowley, H., Bommer, J.J., Pinho, R., Bird, J.F. [2005] The Impact of Epistemic Uncertainty on an
Earthquake Loss Model, Earthquake Engineering & Structural Dynamics, Vol. 34, No. 14, pp. 1653-
1685.
Crowley, H., Colombi, M., Borzi, B., Faravelli, M., Onida, M., Lopez, M., Polli, D., Meroni, F.,
Pinho, R. [2009] Comparison of seismic risk maps for Italy, Bulletin of Earthquake Engineering,
Volume 7, No. 1, pp. 149-180
Crowley, H., Pinho, R. [2004] Period-Height Relationship for Existing European Reinforced
Concrete Buildings, Journal of Earthquake Engineering, Vol. 8, No. 1, pp. 93-119.
Crowley, H., Pinho, R., Bommer, J.J. [2004] A Probabilistic Displacement-Based Vulnerability
Assessment Procedure for Earthquake Loss Estimation, Bulletin of Earthquake Engineering, Vol. 2,
No. 2, pp. 173-219.
Crowley, H., Pinho, R., Bommer, J.J., Bird, J.F. [2006] Development of a Displacement-Based
Method for Earthquake Loss Assessment, Report 2006/01, European School for Advanced Studies in
Reduction of Seismic Risk (ROSE School), IUSS Press, Pavia, Italy.
DAyala, D., Speranza, E. [2002] An Integrated Procedure for the Assessment of Seismic
Vulnerability of masonry structures, Proceedings of the 12th European Conference on Earthquake
Engineering, London, U.K., Paper No. 561.
Daniell, J. [2008] Aftershocks: Present Knowledge and their Implications, Active Faults, Universit
Joseph Fourier, Grenoble, France. (in submission)
Daniell, J., Luey, L. [2006] FRP Retrofitting Solutions for RC Buildings, FRP Retrofitting of
Structures, The University of Adelaide, Adelaide, Australia.
Daniell, J., Maley, T., Peres, R., Villani, M. [2009] Earthquake Loss Estimation Group Assignment
II Zeytinburnu District, Course of Earthquake Loss Estimation, European School for Advanced
Studies in Reduction of Seismic Risk (ROSE School), University of Pavia, Pavia, Italy.
Daniell, J., Parken, C. [2005] Microgenetic Algorithms and Artificial Neural Networks in
Geophysics, Internal Geophysics Conference, University of Adelaide, Adelaide.
References
Del Gaudio, V., Pierri, P. Wasowski, J. [2003]. "An approach to time-probabilistic evaluation of
seismically induced landslide hazard." Bulletin of the Seismological Society of America 93(2): 557-
569.
DellAcqua, F. [2009] First Steps Towards a Framework for EO-Based Seismic Vulnerability
Evaluation, ROSE School Seminar, ROSE School, Pavia, Italy.
Di Pasquale, G., Ferlito, R., Orsini, G., Papa, F., Pizza, G., Van Dyck, J., Veneziano, D. [2004]
Seismic scenario tools for emergency planning and management, XXIX General Assembly of the
European Seismological Commission, Potsdam, Germany.
Di Pasquale, G., Orsini, G., Romeo, R.W. [2005] New Developments in Seismic Risk Assessment in
Italy, Bulletin of Earthquake Engineering, Vol. 3, No. 1, pp. 101-128.
Doherty, K.T., Griffith, M.C., Lam, N., Wilson, J. [2002] Displacement-Based Seismic Analysis for
Out-of-Plane Bending of Unreinforced Masonry Walls, Earthquake Engineering & Structural
Dynamics, Vol. 31, No. 4, pp. 833-850.
Dolce, M., Masi, A., Marino, M., Vona, M. [2003] Earthquake Damage Scenarios of the Building
Stock of Potenza (Southern Italy) Including Site Effects, Bulletin of Earthquake Engineering, Vol. 1,
No. 1, pp. 115-140.
Dominique, P., Goula, X., Colas, B., Jara, J. A., Romeu, N., Susagna, T., Irizarry, J., Sedan, O.,
Figueras, S., Roulle, A., Olivera, C. [2007] Systme transfrontalier de rponse rapide en cas de
sisme dans les Pyrnes Orientales, 7me Colloque National de lAFPS, Paris, France, July 2007,
CD-ROM, Paper ID V-143. [in French]
Dumova-Jovanoska, E. [2004] Fragility Curves for RC Structures in Skopje Region, Proceedings of
the 13th World Conference on Earthquake Engineering, Vancouver, Canada, Paper No. 3.
Eguchi, R.T., Goltz, J.D., Seligson, H.A., Flores, P.J.N., Blais, C., Heaton, T.H., Bortugno, E. [1997]
Real-Time Loss Estimation as an Emergency Response Decision Support System: The Early Post-
Earthquake Damage Assessment Tool (EPEDAT), Earthquake Spectra, Vol.13, No. 4, p. 815-832.
Eguchi, R.T., Huyck, C.K., Cho, S., Ghosh, S., Basoz, N., [2003] Review of REDARS 1.0; Seismic
Risk Analysis Software (Task B1-4), ImageCat Inc., January.
Eguchi, R.T., Mansouri, B. [2005] Use of Remote Sensing Technologies for Building Damage
Assessment after the 2003 Bam, Iran, EarthquakePreface to Remote Sensing Papers, Earthquake
Spectra, Volume 21, Issue S1, pp. S207-S212.
References
Eguchi, R.T., Seligson, H.A. [1994] Status Report: Development of an early post-earthquake damage
assessment tool for Southern California, EQE International Report for USGS Award Number 1434-
93-G-2306, Irvine, California.
Erdik, M. [2007] Discussion of: Istanbul at the threshold: an evaluation of the seismic risk in
Istanbul by Pyper Griffiths, J.H., Irfanoglu, A, and Pujol, S., Earthquake Spectra, Vol. 23, No.3, pp.
727729.
Erdik, M., Demincioghu, M., Sesetyan, K., Durukal, E., Siyahi, B. [2004] Earthquake hazard in
Marmara Region, Turkey, Soil Dynamics and Earthquake Engineering, Vol. 24, No. 8, pp. 605-631.
Erdik, M., Safak, E. [2008] Earthquake Early Warning and Rapid Response System Istanbul,
Bogazici University Presentation ELER, Istanbul, Turkey.
Fh, D., Kind, F., Lang, K., Giardini, D. [2001] Earthquake Scenarios for the City of Basel, Soil
Dynamics and Earthquake Engineering, Vol. 21, No. 5, pp. 405-413.
Fajfar, P., Fischinger, M. [1988] N2 A method for non-linear seismic analysis of regular buildings,
Proceedings of the Ninth World Conference in Earthquake Engineering, Tokyo-Kyoto, Japan, 5, pp.
111-116.
Faris, A.T., Seed, R.B., Kayen, R.E., Wu, J. [2006] A semi-empirical model for the estimation of
maximum horizontal displacement due to liquefaction-induced lateral spreading, Proceedings of the
8th U.S. National Conference on Earthquake Engineering, San Francisco, California, U.S.A..
FEMA [1999] HAZUS99 Technical Manual, Federal Emergency Management Agency, Washington,
DC, U.S.A.
FEMA [2003] HAZUS-MH Technical Manual, Federal Emergency Management Agency, Washington,
DC, U.S.A.
Fleming, K. [2008] Alert Map generation and distribution within the context of the Self-Organising
Seismic Early Warning Information Network, European Seismological Commission (ESC), Crete.
Foray, P., Bard, P-Y. [2008] Engineering Seismology Course Notes, Universit Joseph Fourier,
Grenoble, France.
Freeman, S. A. [1998] Development and use of capacity spectrum method, Proceedings of the 6th
U.S. National Conference on Earthquake Engineering, Oakland, California, U.S.A.
References
Frolova, N., Larionov, V., Bonnin, J. [2006] Multi-Hazard Risk Assessment at Different Levels with
Extremum System Application. The Third International Conference on Early Warning, Bonn,
Germany.
Geist, E., Parsons, L.T. [2006] Probabilistic Analysis of Tsunami Hazards, Natural Hazards, Vol.
37, No. 37, pp. 277-314.
Giovanizzi, S., Lagomarsino, S., Pampanin, S. [2006] Vulnerability Methods and Damage Scenario
for Seismic Risk Analysis as Support to Retrofit, Proceedings of NZSEE Conference 2006, Napier,
New Zealand.
Giovinazzi, S. [2005] The Vulnerability Assessment and the Damage Scenario in Seismic Risk
Analysis, PhD Thesis, Technical University Carolo-Wilhelmina at Braunschweig, Braunschweig,
Germany and University of Florence, Florence, Italy.
Giovinazzi, S., Lagomarsino, S. [2004] A macroseismic method for the vulnerability assessment of
buildings, Proceedings of 13th World Conference. on Earthquake Engineering, Vancouver, Canada.
GNDT [1993] Rischio Sismico Di Edifici Pubblici, Parte I: Aspetti Metodologici, Proceedings of
CNR-Gruppo Nazionale per la Difesa dai Terremoti, Roma, Italy.
GNDT [2000] The Catania Project: earthquake damage scenarios for a high risk area in the
Mediterranean, CNR - Gruppo Nazionale per la Difesa dai Terremoti, Roma, Italy.
Goda, K., Hong, H. P. [2008] Spatial Correlation of Peak Ground Motions and Response Spectra,
Bulletin of Seismological Society of America, Vol. 98, No. 1, pp. 354365.
Griffith, M.C., Magenes, G., Melis, G., Picchi, L. [2003] Evaluation of Out-of-Plane Stability of
Unreinforced Masonry Walls Subjected to Seismic Excitation, Journal of Earthquake Engineering,
Vol. 7, No. 1, pp. 141-169.
Grnthal, G. [1998] European Macroseismic Scale 1998, Cahier du Centre Europen de
Godynamique et de Sismologie, Vol. 15, Luxembourg, Luxembourg.
Guagenti, E., Petrini, V. [1989] The Case of Old Buildings: Towards a Damage-Intensity
Relationship, Proceedings of the Fourth Italian National Conference on Earthquake Engineering,
Milan, Italy, pp. 145-153 (in Italian).
Gutenberg, B., Richter, C.F. [1944] Frequency of earthquakes in California., Bulletin of the
Seismological Society of America, Vol. 34, No. 4, pp. 1985-1988.
References
Hassan, A.F., Sozen, M.A. [1997] Seismic Vulnerability Assessment of Low-Rise Buildings in
Regions with Infrequent Earthquakes, ACI Structural Journal, Vol. 94, No. 1, pp. 31-39.
Hong, H. P. Goda, K. [2006] "A Comparison of Seismic-Hazard and Risk Deaggregation", Bulletin of
the Seismological Society of America 96(6): 2021-2039.
Idriss, I.M. [2008] An NGA Empirical Model for Estimating the Horizontal Spectral Values
Generated By Shallow Crustal Earthquakes, Earthquake Spectra, Vol. 24, No. 1, February 2008, pp.
217-242.
ImageCat, Inc. [2008] Loss Estimation Online Using Inlet: The Inter-Based Loss Estimation Tool,
ImageCat, Inc. Whitepaper, c/o Ron Eguchi.
International Code Council [2006] International Building Code (IBC-2006), ICC, United States,
January 2006, 664 pp.
Ishihara, K. [1993] "Liquefaction and Flow Failure During Earthquakes", Geotechnique 43(3): 351-
415.
Jaiswal, K.S., Wald, D.J., Hearne, M. (in prep) [2008c] Estimating casualties for large worldwide
earthquakes using an empirical approach, U.S. Geological Survey Open-File Report.
Jaiswal, K.S., Wald, D.J. [2008a] Developing a global building inventory for earthquake loss
assessment and risk management, Proceedings of 14th World Conference on Earthquake
Engineering, Beijing, China.
Jaiswal, K.S., Wald, D.J. [2008b] Creating a global building inventory for earthquake loss assessment
and risk management, U.S. Geological Survey Open-File Report 2008-1160, available from URL:
http://earthquake.usgs.gov/pager.
JBDPA [1990] Standard for Seismic Capacity Assessment of Existing Reinforced Concrete Buildings,
Japanese Building Disaster Prevention Association, Ministry of Construction, Tokyo, Japan.
Joyner, W.B., Boore, D.M. [1993] Methods for regression analysis of strong-motion data, Bulletin
of the Seismological Society of America, Vol. 83, No. 2, pp. 469-487.
Kalkan, E., Kunnath, S. K. [2006] Effects of fling step and forward directivity on seismic response of
buildings, Earthquake Spectra, Vol. 22, No. 2, pp. 367-390.
References
Kappos, A.J., Pitilakis, K., Stylianidis, K., Morfidis, K., Asimakopoulos, N. [1995] Cost- Benefit
Analysis for Seismic Rehabilitation of Buildings in Thessaloniki., Proceedings of the Fifth AFPS-
EERI Conference of Seismic Zonation, Nice, France.
Kappos, A.J., Stylianidis, K.C., Pitilakis, K. [1998] Development of Seismic Risk Scenarios Based
on a Hybrid Method of Vulnerability Assessment, Natural Hazards, Vol. 17, No. 2, pp. 177- 192.
Kayen, R.E., Mitchell, J.K. [1997] "Assessment of Liquefaction Potential During Earthquakes by
Arias Intensity." Journal of Geotechnical and Geoenvironmental Engineering, Vol. 123, No. 12, pp.
1162-1174.
Khater, M., Scawthorn, C., Johnson, J. [2003] Earthquake Engineering Handbook, Chapter 31: Loss
Estimation, eds, Chen, W-F., Scawthorn, C., CRC Press LLC, Florida.
King, S.A., Kiremidjian, A.S., Sarabandi, P., Pachakis, D. [2005] Correlation of Observed Building
Performance with Measured Ground Motion, California Strong Motion Instrumentation Program
(SMIP), SMIP03 Seminar on Utilization of Strong-Motion Data May 22, 2003, Oakland, California,
U.S.A.
Kircher, C.A., Whitman, R.V., Holmes, W.T. [2006] HAZUS Earthquake Loss Estimation Methods,
Natural Hazards Review, Vol. 7, No. 2, pp. 45-59.
Kircher, C.A., Nassar, A.A., Kustu, O., Holmes, W.T. [1997] Development of Building Damage
Functions for Earthquake Loss Estimation, Earthquake Spectra, Vol. 13, No. 4, pp. 663-682.
Koehler, N., Wenzel, F., Boese, M. [2007] PreSEIS eine Methodik zur Erdbebenfruehwarnung im
Rahmen des SAFER-Projekts, German Geophysical Society, DGG, Aachen, Germany.
KOERI [2002] Earthquake Risk Assessment for Istanbul Metropolitan Area. Kandilli Observatory and
Earthquake Research Institute, Istanbul, available from URL:
http://www.koeri.boun.edu.tr/depremmuh/EXEC_ENG.pdf.
Kramer, S.L., Mayfield, R.T., Anderson D.G. [2006] Performance-Based Liquefaction Hazard
Evaluation: Implications for Codes and Standards, Proceedings of the 8th U.S. National Conference
on Earthquake Engineering, San Francisco, California.
Kramer, S.L., Mitchell, R.A. [2006] Ground motion intensity measures for liquefaction hazard
evaluation, Earthquake Spectra, Vol. 22, No. 2, pp. 413-438.
Kramer, S.L. [1996] Geotechnical Earthquake Engineering, Prentice-Hall, New Jersey, U.S.A.
References
Kratzig, W. B., Meyer, I. F., Meskouris, K. [1989] Damage evolution in reinforced concrete members
under cyclic loading, Proceedings of the 5th International Conference on Structural Safety and
Reliability, San Francisco, CA, pp. 795-802.
Kreyszig, E. [2002] Advanced Engineering Mathematics: 8th Edition, John Wiley and Sons, Inc., New
York.
Kudo, T., Tanaka, T., Furumoto, M. [2009] Estimation of the Maximum Earthquake Magnitude from
the Geothermal Gradient, Bulletin of the Seismological Society of America; February 2009; v. 99; no.
1; p. 396-399
Kumar, J. Rao, V. [2003] "Seismic bearing capacity of foundations on slopes", Geotechnique 53(3):
347-361.
Lang, D.H., Molina, S., Lindholm, C. [2008] Towards near real-time damage estimation using a
Csm-based tool for seismic risk assessment, Journal of Earthquake Engineering, Vol. 12 No. S2, pp.
199210.
Lang, D.H., Gutierrez, V., Lindholm, C.D. [2008b] RISe v1.0 - User and Technical Manual v1.0,
NORSAR, December 2008, 22 pp.
Lantada, N., Pujades, L.G., Barbat, A.H. [2004] Risk Scenarios for Barcelona, Spain, Proceedings
of the 13th World Conference on Earthquake Engineering, Vancouver, Canada, Paper No. 423.
Larionov, V.I. [1999] Forecast of Emergencies, Mechanics of Destruction, in Theoretical basis of
response to emergency situations, pp. 1-276, in Russian, Military Engineering University, Moscow
Liao, S., Veneziano, S.D., Whitman, R.V. [1988] Regression Models for Evaluating Liquefaction
Probability, Journal of Geotechnical Engineering, Vol. 114, No. 4, pp. 389-411.
Lindholm, C.D., Molina-Palacios, S., Lang, D.H. [2007] SELENA-Seismic loss estimation using a
logic tree approach Vilnius Conference and Athens Conference, Greece and Lithuania.
Manighetti, I., Campillo, M., Bouley, S., Cotton, F. [2007] Earthquake scaling, fault segmentation,
and structural maturity, Earth and Planetary Sciences Letters, Vol. 253, pp. 429-438.
Markus, M., Fiedrich, F., Leebmann, J., Schweier, C., Steinle, E. [2004] Concept for an Integrated
Disaster Management Tool, Proceedings of the 13th World Conference on Earthquake Engineering,
Vancouver, B.C., Canada.
References
Masi, A. [2004] Seismic Vulnerability Assessment of Gravity Load Designed R/C Frames, Bulletin.
Earthquake Engineering, Vol. 1, No. 3, pp. 371-395.
Masson, D.G., Harbitz, C.B., Wynn, R.B., Pedersen, G., Lovholt, F. [2006] Submarine landslides:
processes, triggers and hazard prediction, Philosophical Transactions of the Royal Society a-
Mathematical Physical and Engineering Sciences, Vol. 364, No. 1845, pp. 2009-2039.
McGuire, R.K. [1995] Probabilistic Seismic Hazard Analysis and Design Earthquakes: Closing the
Loop, Bulletin of the Seismological Society of America, Vol. 85, No. 5, pp. 1275-1284.
McGuire, R.K. [1988] Seismic risk to lifeline systems: critical variables and sensitivities,
Proceedings of 9th World Conference on Earthquake Engineering Vol. VII, pp. 129-134.
Medvedev, S., Sponheuer, W. [1969] MSK scale of seismic intensity, Proceedings 4th World
Conference on Earthquake Engineering,: Chilean Association for Seismology and Earthquake
Engineering, Santiago, Chile.
Mehanny, S.F., Deierlein, G.G. [2001] Seismic damage and collapse assessment of composite
moment frames Journal of Structural Engineering, ASCE, 127(9): 1045-1053
Mina, C., Higuchi, M., Danno, K., How Tion, P., Scawthorn, C. [2004] Capstone Project OSRE,
Risk-Agora, Kyoto University, Japan.
Miura, H., S. Midorikawa [2006] Updating GIS building inventory data using high-resolution
satellite images for earthquake damage assessment: Application to Metro Manila, Philippines,
Earthquake Spectra, Vol. 22, No. 1, pp. 151-168.
Modena, C., Loureno, P.B., Roca, P. (editors) [2005] Structural Analysis of Historical Constructions
Possibilities of Numerical and Experimental Techniques, Taylor and Francis, London, U.K.
Molina S., Lindholm, C. [2007] Estimation the confidence of earthquake damage scenarios: examples
from a logic tree approach, Journal of Seismology 11(3): 399310.
Molina, S., Lang, D.H., Lindholm, C.D. [2008a] SELENA v3.5.1 - User and Technical Manual
v3.5.1, NORSAR, May 2008, 69 pp.
Molina, S., Lang, D.H., Lindholm, C.D. [2008b]. SELENA v4.0 - User and Technical Manual v4.0,
NORSAR, October 2008, 85 pp.
References
Molina, S., Lindholm, C. [2005] A Logic Tree Extension of the Capacity Spectrum Method
Developed to Estimate Seismic Risk in Oslo, Norway, Journal of Earthquake Engineering, Vol. 9,
No. 6, pp. 877-897.
Moss, R.E.S., Seed, R.B., Kayen, R.E., Stewart, J.P., Kiureghian, A.D., Cetin, K.O. [2006] CPT-
based probabilistic and deterministic assessment of in situ seismic soil liquefaction potential, Journal
of Geotechnical and Geoenvironmental Engineering, Vol. 132, No. 8, pp. 1032-1051.
Mouroux, P., Bertrand E., Bour, M., Le Brun, B., Depinois, S., Masure, P., RISK-UE team [2004]
The European Risk-UE Project: An Advanced Approach to Earthquake Risk Scenarios, Proceedings
of the 13th World Conference on Earthquake Engineering, Vancouver, B.C., Canada.
Mouroux, P., Le Brun, B. [2006] RISK-UE project: an advanced approach to earthquake risk
scenarios with application to different European towns, In: Assessing and Managing Earthquake
Risk, Oliveira, C. S. Roca, A., and Goula, X., Eds. Springer, Dordrecht, pp. 479508.
Murakami, H.O., Ohta, Y. [2004] Human Entrapment in the 195 Kobe Earthquake- Comparison of
Urban and Rural Environment, 3rd Taiwan-Japan Workshops on Lifeline Performance and Disaster
Mitigation Workshop, Taiwan.
Musson, R.M.W. [1999] Determination of design earthquakes in seismic hazard analysis through
Monte Carlo simulation Journal of Earthquake Engineering; Vol. 3, No.4, pp. 463-474.
Nazri, F. [2007] Development of Design Response Spectra For Penang Island, MSc Thesis,
Universiti Sains Malaysia.
NEHRP [1997] Recommended provisions for seismic regulations for new buildings and other
structures, Report FEMA-303, Building Seismic Safety Council, Federal Emergency Management
Agency, Washington DC., U.S.A.
Oehlers, D.J. (Ed.) [2003] Retrofitting reinforced concrete structures by steel and FRP plating, John
Wiley and Sons, Sydney, Australia.
Oliveira, C., Redondo, E., Lambert, J., Riera, A, Roca, A. (editors) [2006] Els terratremols dels segles
XIV i XV a Catalunya, Institut Cartografic de Catalunya, 407, pp.115-129.
Oliveira, C., Roca, A., Goula, X. [2006] Assessing and managing earthquake risk: geo-scientific and
engineering knowledge for earthquake risk mitigation: developments, tools, techniques, Springer,
Netherlands.
References
Oliveira, C.S., Ferreira, M.A., Mota de S, F. [2004] Seismic Vulnerability and Impact Analysis:
Elements for Mitigation Policies, Proceedings of the XI Congresso Nazionale on Lingegneria
Sismica in Italia, Genova, Italy.
Oliveira, C.S., Mota de S, F., Ferreira, M.A. [2005] Application of Two Different Vulnerability
Methodologies to Assess Seismic Scenarios in Lisbon, Proceedings of the International Conference:
250th Anniversary of the 1755 Lisbon Earthquake, Lisbon, Portugal, Paper No. 37.
Olivieri, M., Allen, R.M., Wurman, G. [2008] The potential for earthquake early warning in Italy
using ElarmS, Bull. Seism. Soc. Am., 98, pp. 495-503.
Orsini, G. [1999] A Model for Buildings Vulnerability Assessment Using the Parameterless Scale of
Seismic Intensity (PSI), Earthquake Spectra, Vol. 15, No. 3, pp. 463-483.
Park, Y.J., Ang, A.H.S. [1985] Mechanistic Seismic Damage Model for Reinforced Concrete,
Journal of Structural Engineering, Vol. 111, No. 4, pp. 722-739.
Patchett, A., Robinson, D., Dhu, T., Sanabria, A. [2005] Investigating Earthquake Risk Models and
Uncertainty in Probabilistic Seismic Risk Analyses, Geoscience Australia Record 2005/02, Canberra,
Geoscience Australia, p. 83.
Pinho, R., Bommer, J.J., Glaister, S. [2002] A Simplified Approach to Displacement-Based
Earthquake Loss Estimation Analysis, Proceedings of the 12th European Conference on Earthquake
Engineering, London, U.K., Paper No. 738.
Porter, K.A. [2000] Assembly-Based Vulnerability of Buildings and Its Uses in Seismic Performance
Evaluation and Risk-Management Decision-Making. PhD Thesis from Department of Civil and
Environmental Engineering. Palo Alto, California, Stanford University, 148.
Porter, K. A. [2008c] Cracking an open safe: HAZUS vulnerability functions in terms of structure
independent spectral acceleration, Earthq. Spectra. In submission.
Porter, K.A., Jaiswal, K.S., Wald, D.J., Greene, M., Comartin, C. [2008b]. WHE-PAGER Project: a
new initiative in estimating global building inventory and its seismic vulnerability, Proc. 14th World
Conf. Earthq. Eng., Beijing, China 8 pp.
Porter, K., Jaiswal, K., Wald, D., Earle, P., Hearne, M. [2008a]. Fatality models for the U.S.
Geological Survey's Prompt Assesment of Global Earthquake for Response (PAGER) system, Proc.
14th World Conf. Earthq. Eng., Beijing, China, 8 pp.
References
Porter, K.A., Scawthorn, C.R. [2007a] OpenRisk: Open Source Risk Estimation Software, SPA
Risk, available from http://risk-agora.org
Priestley, M.J.N. [2003] Myths and Fallacies in Earthquake Engineering, Revisited: The Mallet Milne
Lecture, 2003, IUSS Press, Pavia, Italy.
Priestley, M.J.N., Calvi, G.M., Kowalsky, M.J. [2007] Displacement-Based Seismic Design of
Structures, IUSS Press, Pavia, Italy.
Rahnama, M., Krawinkler, H. [1993] Effects of soft soil and hysteresis model on seismic demands,
John A. Blume Earthquake Engineering Center Report 108, Stanford University, Palo Alto, California,
U.S.A.
Rauch, A.F.,, Martin, J.R., II. [2000] EPOLLS model for predicting average displacements on lateral
spreads,J. Geotech. Geoenviron. Eng., 1264 , 360371.
Restrepo-Vlez, L.F., Magenes, G. [2004] Simplified Procedure for the Seismic Risk Assessment of
Unreinforced Masonry Buildings, Proceedings of the 13th World Conference on Earthquake
Engineering, Vancouver, Canada, Paper No. 2561.
Richards, R., Elms, D.G., Budhu, M. [1993]. Seismic Bearing Capacity and Settlements of
Foundations Journal of Geotechnical Engineering, ASCE, Vol. 119, No. 4, pp. 662-674.
Robinson, D., Dhu, T., Schneider, J. [2006] Practical probabilistic seismic risk analysis: a
demonstration of capability, Seismological Research Letters, Vol. 77, No. 4, pp. 452-458.
Robinson, D., Fulford, G., Dhu, T. [2005] EQRM: Geoscience Australia's Earthquake Risk Model
Geoscience Australia Record 2005/01, Geoscience Australia, Canberra, Australia.
Roca, A., Goula, X., Susagna, T., Chvez, J., Gonzlez, M., Reinoso, E. [2006] A simplified method
for vulnerability assessment of dwelling buildings and estimation of damage scenarios in Spain,
Bulletin of Earthquake Engineering, Vol. 4, No.2, pp. 141158.
Rossetto, T., Elnashai, A. [2003] Derivation of vulnerability functions for European-type RC
structures based on observational data, Engineering Structures, Vol. 25, pp. 1241-1263.
Rossetto, T., Elnashai, A. [2005] A New Analytical Procedure for the Derivation of Displacement-
Based Vulnerability Curves for Populations of RC Structures, Engineering Structures, Vol. 7, No. 3,
pp. 397-409.
References
RSE [2003] Manual dutilitzacio del programa de calcul i representacio descenaris de danys,
Internal Report, Institut Cartografic de Catalunya, Barcelona, 56p.
Saito, K. [2008] Streamlining the creation of building inventories using remote sensing and
geospatial data, Cambridge University Centre, Willis Research Network Presentation, University of
Cambridge, 08/07/08.
Samardjieva, E., Badal, J. [2002] Estimation of the Expected Number of Casualties Caused by Strong
Earthquakes, Bulletin of the Seismological Society of America, Vol. 92, No. 6, pp. 2310-2322.
Sarabandi, P., Adams, B.J., Kiremidjian, A.S., Eguchi, R. [2005] Infrastructure Inventory
Compilation using Single High Resolution Satellite Images, 3
rd
International Workshop on Remote
Sensing for Post-Disaster Response, Chiba, Japan.
Sarabandi, P., H. C. Chung, B. J. Adams [2006] Remote Sensing for Building Inventory Updates in
Disaster Management, SPIE 11th Annual Symposium, Nondestructive Evaluation for Health
Monitoring and Diagnostics, San Diego, California, U.S.A.
Saraf, A.K., Mishra, B.P., Choudhury, S. and Ghosh, P. [2005] Digital elevation model (DEM)
generation from NOAA-AVHRR night-time data and its comparison with USGA-
DEM, International Journal of Remote Sensing, 26, 3879-3887
Savvaidis, P., Doukas, I., Hatzigogos, Th., Tziavos, I.N., Kiratzi, A., Roumelioti, Z., Savvaidis, A.,
Theodulidis, N., Drakatos, G., Koutoupes, S., Karantonis, G., Sotiriadis, A. [2004] Database
development and evaluation of earthquake damage reports under the SEISIPACT-THES system,
presented in 10th Congress of the Greek Society of Geology, Thessaloniki, paper included in the
Bulletin of the Geol. Soc. of Greece, vol. XXXVI.
Scawthorn, C. [2007] MIRISK, Department of Urban Management, Kyoto University, Japan.
Scherbaum, F. [2008] Probabilistic Seismic Hazard Assessment Course Notes, Universit Joseph
Fourier, Grenoble, France.
Schmidt, J., Turek, G., Matcham, I., Reese, S., Bell, R., King, A. [2007] RiskScape an innovative
tool for multi-hazard risk modelling, Geophysical Research Abstracts, Vol. 9, 05782.
Schneider, P.J., Schauer, B.A. [2006] HAZUS Its Development and Future,Natural Hazards
Review, Vol. 7, No. 2, May 2006, pp. 40-44.
References
SEAOC [1995] Vision 2000: Performance Based Seismic Engineering of Buildings Part 2:
Conceptual Framework, Vision 2000 Committee, Structural Engineers Association of California,
Sacramento, California, U.S.A.
Seed, R. B., Cetin, K.O., Moss,R.E.S., Kammerer, A.M., Wu, J., Pestana, J.M., Riemer, M.F. [2001]
Recent Advances in Soil Liquefaction Engineering and Seismic Site Response Evaluation,
Proceedings of the 4th International Conference on Recent Advances in Geotechnical Earthquake
Engineering and Soil Dynamics, San Diego, California, U.S.A.
Shakhramanjyan, M.A., Nigmetov, G.M., Larionov, V.I., Nikolaev, A.V., Frolova, N.I., Sushchev,
S.P., Ugarov, A.N. [2001] Advanced procedures for risk assessment and management in Russia.,
International Journal of Risk Assessment and Management, 2 (3/4), 303-318.
Shaw R. K. [2000] RADIUS Evaluation for Asian Cities: Bandung, Tashkent and Zigong In United
Nations International Strategy for Disaster Reduction (UN ISDR) RADIUS Year-later Evaluation,
Geneva, Switzerland.
Shaw,D., Yeh, C-H., Jean, W-Y., Loh, C-H., Kuo, Y. [2007] A Probabilistic Seismic Risk Analysis
of Building Losses in Taipei: An Application of HAZ-Taiwan with its Pre-processor and Post-
processor, Journal of the Chinese Institute of Engineers 30(2 )
Siddiqui, Z., Kumar, R. [2007] Estimation of Risk due to earthquake hazard in AP, India an IT
Based Approach, Tech Report-IIIT Hyderabad, India.
Singhal, A., Kiremidjian, A.S. [1996] Method for probabilistic evaluation of seismic structural
damage, Journal of Structural Engineering., ASCE, Vol. 122, No. 12, pp. 1459-1467.
Singhal, A., Kiremidjian, A.S. [1998] Bayesian updating of fragilities with application to RC
frames, Journal of Structural Engineering, ASCE, Vol. 124, No. 8, pp. 922-929.
Soddu P., Martini M.G., [2005] Seismic emergency management: Technologies at work, The First
International Symposium on Geo-information for Disaster Management, Delf.
Somerville, P. G. [2003] Magnitude scaling of the near fault rupture directivity pulse, Physics of the
Earth and Planetary Interiors, Vol. 137, No. 1-4, pp. 201-212.
Sousa, M.L., Campos Costa, A., Carvalho, A., Coelho, E. [2004] An Automatic Seismic Scenario
Loss Methodology Integrated on a Geographic Information System, Proceedings of the 13th World
Conference on Earthquake Engineering, Vancouver, Canada, Paper No. 2526.
References
Spence, R. Ed [2007] Earthquake disaster scenario predictions and loss modelling for urban areas,
LESSLOSS Report 7, IUSS Press, Pavia, Italy.
Spence, R., Coburn, A.W., Pomonis, A. [1992] Correlation of Ground Motion with Building
Damage: The Definition of a New Damage-Based Seismic Intensity Scale, Proceedings of the Tenth
World Conference on Earthquake Engineering, Madrid, Spain, Vol. 1, pp. 551-556.
Stafford, P.J, Strasser, F.O., Bommer, J.J. [2007] Preliminary Report on the Evaluation of Existing
Loss Estimation Methodologies, Proceedings of Neries-JRA3 Meeting 22-23 January 2007, Istanbul,
Turkey.
Stewart, J. [2008] Earthquake Geotechnical Engineering II. Applications Course Notes, Rose School,
Strasser, F.O., Bommer, J.J., eetyan, K., Erdik, M., anan, Z., Irizarry, J., Goula, X., Lucantoni,
A., Sabetta, F., Bal, I.E., Crowley, H., Lindholm, C. [2008] A Comparative Study of European
Earthquake Loss Estimation Tools for a Scenario in Istanbul, Journal of Earthquake Engineering,
Vol. 12, No. 1, pp. 246-256.
Sucuolu, H., Yazgan, U., Yakut, A. [2007], A Screening Procedure for Seismic Risk Assessment in
Urban Building Stocks, Earthquake Spectra, Vol. 23, pp.441-458.
Tiedemann, H. [1992] Earthquakes and Volcanoes: a Handbook for Risk Assessment, Swiss
Reinsurance Company, Zurich, Switzerland.
Todorovska, M.I., Trifunac, M.D. [2006] A Note on Probabilistic Assessment of Fault Displacement
Hazard, Proceedings of the 8th U.S. National Conference on Earthquake Engineering, San Francisco,
California, U.S.A.
Trendafiloski, G., Wyss, M., Rosset, P., Marmureanu, G. [2009] Constructing city models to estimate
losses due to earthquakes worldwide: application to Bucharest Romania, Earthquake Spectra, in
press.
Tsai, Y.-B., Yu, T.U., Chao, H.-L., Lee, C.-P. [2001] Spatial Distribution and Age Dependence of
Human-Fatality Rates from the Chi-Chi, Taiwan, Earthquake of 21 September 1999, Bulletin of the
Seismological Society of America, Vol. 91, No. 5, pp. 1298-1309.
Tuzun, C. [2009], EMME Earthquake Model of the Middle East, International Conference on
Sustainable Development and Geohazards in the Southern Caucasus, Tbilisi.
References
Umemura, K., Murao O., Yamazaki, F. [2002] Development of GIS-based building damage database
for the 1995 Kobe earthquake, Proc. of the 21st Asian Conference on Remote Sensing, Taiwan, Vol.
I, pp. 389-394
United Nations [1993] Housing in the World- Graphical presentation of statistical data: United
Nations, United Nations, New York, 177 p.
Wald, D.J., Quitoriano, V., Heaton, T.H., Kanamori, H. [1999] Relationships between Peak Ground
Acceleration, Peak Ground Velocity, and Modified Mercalli Intensity in California, Earthquake
Spectra, Vol. 15, No.3, pp. 557-564.
Wald, D.J.,. Worden, B.C, Quitoriano, V., Pankow, K.L. [2005] ShakeMap manual: technical
manual, user's guide, and software guide, U.S. Geological Survey Techniques and Methods, book 12,
section A, chap. 1. Reston, Virginia, U.S. Geological Survey: 132.
Wald, D.J.,Earle, P.S., Allen, T.I., Jaiswal, K., Porter, K, Hearne, M. [2008] Development of the U.S.
Geological Survey's PAGER system (Prompt Assessment of Global Earthquakes for Response),
Proceedings of 14th World Conference on Earthquake Engineering, Beijing, China.
Wang, M., Takada, T. [2005] Macrospatial correlation model of seismic ground motions,
Earthquake Spectra, Vol. 21, No. 4, pp. 1137-1156.
Weber, E., Iannaccone, G., Zollo, A., Bobbio, A., Cantore, L., Corciulo, M., Convertito, M., Di
Crosta, M., Elia, L., Emolo, A., Martino, C., Romeo, A., Satriano, C. [2007] Development and testing
of an advanced monitoring infrastructure (ISNet) for seismic early-warning applications in the
Campania region of southern Italy, Earthquake early warning systems, Springer, 2007.
Wells, D.J., Coppersmith, K.J. [1994] New empirical relationships among magnitude, rupture length,
rupture width, rupture area, and surface displacement, Bulletin of the Seismological Society of
America; Vol. 84, pp. 974-1002.
Werner, S.D., Cho, S., Taylor, C.E., Lavoie, J-P, Huyck, C.K., Chung, H., Eguchi, R.T. [2006].
Technical Manual: REDARS 2 Methodology and Software for Seismic Risk Analysis of Highway
Systems, Report MCEER-06-SP08,: Multidisciplinary Center for Earthquake Engineering Research,
Buffalo NY.
Whitman, R.V., Reed, J.W., Hong, S.T. [1973] Earthquake Damage Probability Matrices,
Proceedings of the Fifth World Conference on Earthquake Engineering, Rome, Italy, Vol. 2, pp. 2531-
2540.
References
Wilson, R.C. [1993]. Relation of Arias Intensity to magnitude and distance in California, Open File
Report 93-556, Reston, Virginia, U.S. Geological Survey: 42.
Windeler, D., Morrow, G., Williams, C.R., Rahnama, M., Molas, G., Pea, A., Bryngelson, J. [2004]
Earthquake risk estimates for residential construction in the western United States, Proceedings of
13th World Conference on Earthquake Engineering, Vancouver, B.C., Canada.
Wu, Y.M., Hsiao, N.C., Teng, T.L. [2004] Relationships between strong ground motion peak values
and seismic loss during the 1999 Chi-Chi Taiwan Earthquake, Natural Hazards, 32 (3), pp. 357-373.
Wyss, M., Zibzibadze, M. [2009] Delay times of worldwide global earthquake alerts. Natural
Hazards, DOI 10.1007/s11069-009-9344-9.
Wyss, M. [2004a] Earthquake loss estimates in real-time begin to assist rescue teams,
worldwide, EOS, Vol. 85, p. 567.
Wyss, M. [2004b] Real-time prediction of earthquake casualties, paper presented at International
Conference on Disasters and Society From Hazard Assessment to Risk Reduction, edited by D.
Malzahn and T. Plapp, Logos Publishers, Universitt Karlsruhe, Karlsruhe, Germany, pp.165-173.
Wyss, M. [2008] Estimated human losses in future earthquakes in central Myanmar., Seismological
Research Letters,Vol. 79, pp. 520-525.
Yakut, A. [2004] Preliminary Seismic Performance Assessment Procedure for Existing RC
Buildings, Engineering Structures, Vol. 26, No. 10, pp. 1447-1461.
Yamazaki, F., Nishimura, A., Ueno, Y. [1996] Estimation of human casualties due to urban
earthquakes, Proceedings of the 11th World Conference on Earthquake Engineering, Acapulco,
Mexico.
Yao, B., Xie, L., Huo, E. [2004] Study Effect of Lifeline Interaction Under Seismic Conditions,
Proceedings of the 13th World Conference on Earthquake Engineering, Vancouver, B.C., Canada.
Yeh, C.H., Loh, C.H., Tsai, K.C. [2006] Overview of Taiwan Earthquake Loss Estimation System,
Natural Hazards, Vol. 37, No. 1-2, pp. 23-37.
Yeh, C.H., Jean, W.Y. and Loh, C.H. [2000] Damage Building Assessment for Earthquake Loss
Estimation in Taiwan, Proceedings of the 12th World Conference on Earthquake Engineering,
Auckland, New Zealand, Paper No. 1500
References
Youd, T.L., Idriss, I.M., Andrus, R.D., Arango, I., Castro, G., Christian, J.T., Dobry, R., Finn, W.L.,
Harder, L.F., Hynes, M.E., Ishihara, K., Koester, J.P., Liao, S.S.C., Marcuson, W.F., Martin, G.R.,
Mitchell, J.K., Moriwaki, Power, M.S., Robertson, P.K., Seed, R.B., Stokoe, K.H. [2001]
Liquefaction resistance of soils: Summary report from the 1996 NCEER and 1998 NCEER/NSF
Workshops on Evaluation of Liquefaction Resistance of Soils, Journal of Geotechnical and
Geoenvironmental Engineering, Vol. 127, No. 10, pp. 817-833.
Zschau, J., Gasparini, P., Papadopoulos, G., the SAFER Consortium [2007] Status of the SAFER
Project (Seismic eArly warning For EuRope), American Geophysical Union (AGU) Fall Meeting,
San Francisco, USA.

Webography

WEBOGRAPHY
AIR Worldwide Reinsurance [2009] AIR Worldwide Proprietary ELE Software, available from
URL: http://www.air-worldwide.com.
Applied Technology Council [2009] ATC-20i Mobile Postearthquake Building Safety Evaluation
Data Acquisition System (Version 1.0), available from URL:
http://www.atcouncil.org/ATC20i.shtml.
Bulgaria, Willis RE [2009] Willis Research Network Bulgarian Quake Model, available from
URL: http://natkat.insurance.bg/2008/bg/files/10.pdf
CAPRA [2009] CAPRA Central American Probabilistic Risk Assessment Portal available from
URL: http://www.ecapra.org.
CISN [2009] Welcome to CISN, available from URL: www.cisn.org.
Completeness [2009] Completeness of Earthquake Catalogues, available from URL:
http://completeness.usc.edu.
COSMOS [2009] Consortium of Organisations for Strong Motion Observation Systems, Forum
Services Advocacy, available from URL:http://www.cosmos-eq.org.
Earthquake.it [2009] Earthquake. Terremoti dallItalia e dal mondo, (Archive), available from URL:
http://www.earthquake.it.
ECLAC (2003), Handbook for estimating the socio-economic and environmental effects of
disasters, available from URL: http://www.cepal.org.mx
eEQSIM [2009] Earthquake Risk Simulation, available from URL: http://www.eeqsim.com.
EM-DAT [2009] EM-DAT: Emergency Events Database Welcome to EM-DAT, available from
URL: http://www.emdat.be.
Webography
EmerGeo Solutions Inc. [2009] Hazard Models, available from URL:
http://emergeo.net/hazard_models.aspx.
ETABS Computers & Structures, Inc. [2009] ETABS Features Integrated Analysis, Design and
Drafting of Building Systems, available from URL:
http://www.csiberkeley.com/products_ETABS.html.
European strong motion database, the [2009] Abstract, available from URL:
http://www.isesd.cv.ic.ac.uk/ESD.
FEMA [2009] FEMA Website, available from URL: http://www.fema.gov.
GAPQuake Bulgaria [2009] Benfield Group New Bulgaria Quake Model, available from URL:
http://www.benfieldgroup.com/MEDIA%20CENTRE/PRESS%20RELEASES/Pages/NewBulgarianE
arthquakeModelLaunchedatWorldBankWorkshop.aspx.http://natkat.insurance.bg/2008/bg/files/10.pdf
GEM [2009] GEM Global Earthquake Model, available from URL:
http://www.globalquakemodel.org.
Geohazards International [2009] RADIUS Risk Assessment Tools for Urban Areas Against
Seismic Disasters, available from URL: http://www.geohaz.org/projects/radius.
Global Seismic Hazard Assessment Program [2009] Global Seismic Map, available from URL:
http://www.seismo.ethz.ch/gshap.
GONAF [2009] GONAF Website, available from URL: http://www.gonaf.de.
ILWIS [2009] ILWIS 3.5 OPEN Website, available from URL:
http://52north.org/index.php?option=com_content&view=category&layout=blog&id=33&Itemid=67.
Istituto Nazionale di Geofisica e Vulcanologia [2009] Itaca - Italian Accelerometric Archive,
available from URL: http://itaca.mi.ingv.it.
Jet Propulsion Laboratory California Institute of Technology [2009] QUAKESIM Download
QuakeSIM codes, available from URL: http://quakesim.jpl.nasa.gov/download.html.
LESSLOSS [2009] LESSLOSS Home, available from URL: http://www.lessloss.org.
MAE [2009] ZEUS-NL Registration, available from URL:
http://mae.ce.uiuc.edu/software_and_tools/zeus_nl_registration.html.
Webography
MunichRE Group [2009] Mnchener Rck Munich Re Group, available from URL:
http://www.munichre.com/en/homepage/default.aspx.
NASA Visible Earth [2009] Population Density details, available from URL:
http://visibleearth.nasa.gov/view_rec.php?id=116.
NATHAN MunichRE [2009] NATHAN Natural Hazards Assesment Network, available from
URL: http://mrnathan.munichre.com.
National Development and Reform Commission [2009] Figures on Sichuan Earthquake, available
from URL: http://en.ndrc.gov.cn/
National Weather Service [2009] NOAA - Pacific Tsunami Warning System, available from URL:
http://www.prh.noaa.gov/ptwc/?region=2&id=hawaii.2009.03.19.200910.
NGA PEER [2008] Next Generation of Attenuation of Ground Motion (NGA) Project, available
from URL: http://peer.berkeley.edu/products/nga_project.html.
NOAA [2009] NOAA National Oceanic and Atmospheric Administration United States
Department of Commerce, available from URL: http://www.noaa.gov.
NORSAR [2009] Link to all NORSAR activities, available from URL: http://www.norsar.no/
NSHMP USGS [2009] Earthquake Hazards Program Products and References, available from
URL: http://earthquake.usgs.gov/eqcenter/pager/prodandref/index.php.
Octave [2009] Octave Current News, available from URL: http://www.gnu.org/software/octave.
OpenSees [2009] OpenSees Parallel Workshop, available from URL:
http://opensees.berkeley.edu/index.php.
OpenSHA [2009] Open Seismic Hazard Analysis, available from URL: http://www.opensha.org.
OSRE [2009] OSREIII, available from URL: http://quake.kuciv.kyoto-u.ac.jp/OSRE/.
PAGER archives USGS [2009] Earthquake Hazards Program PAGER Archives, available from
URL: http://earthquake.usgs.gov/eqcenter/pager/archives.php.
PAGER USGS [2009] Earthquake Hazards Program PAGER: Prompt Assessment of Global
Earthquakes for Response, available from URL: http://earthquake.usgs.gov/eqcenter/pager.
Webography
PEER [2009] Pacific Earthquake Engineering Research Center: NGA Database, available from
URL: http://peer.berkeley.edu/nga.
PEER [2009] PEER Pacific Earthquake Engineering Research Center, available from URL:
http://peer.berkeley.edu.
RiskScape [2009] RiskScape, available from URL: http://www.riskscape.org.nz.
RISK-UE [2009] An advanced approach to earthquake risk scenarios with application to different
European towns, available from URL: http://www.risk-ue.net.
ROVER Emcode [2009] RedROVER - A simple phyton Rover-InCast Web conversion Script,
available from URL: http://code.google.com/p/emcode/wiki/RedROVER.
Ruaumoko [2009] The Maori God of Earthquakes and Volcanoes, available from URL:
http://www.civil.canterbury.ac.nz/ruaumoko.
SAP2000 Computers & Structures, Inc. [2009] SAP 2000 - Integrated Software for Structural
Analysis and Design, available from http://www.csiberkeley.com/products_SAP.html.
SeismoSoft [2009] SeismoStruct: A computer program for static and dynamic nonlinear analysis of
framed structures, available from URL: http://www.seismosoft.com.
ShakeCast USGS [2009] ShakeCast Pre-Download Agreement, available from URL:
https://sslearthquake.usgs.gov/resources/software/shakecast/downloads.
TRANSFER [2009] Transfer Project Home Page Tsunami Risk ANd Strategies For the European
Region, available from URL: http://www.transferproject.eu.
Turkish Government Website [2009] Zeytinburnu District Aerial Photos, available from URL:
http://sehirrehberi.ibb.gov.tr/Default.aspx?&ap=istanbulresim&cx=84389&cy=94583&scl=3.
uDIG [2009] Open-Source GIS Software from Refractions Research, available from URL:
http://www.udig.com.
UNDP [2008] UNDP Website, available from URL: http://www.undp.org.
UN-HABITAT [2007] Housing in the world- Demographic and Health Survey, available from
URL: http://www.unhabitat.org.
USGS [2009] Earthquake Hazard Programme, available from URL: http://earthquake.usgs.gov.
Webography
VCH [2009] VCH @ Kyoto Univ. for Disaster Investigation Coordination, available from URL:
http://quake.kuciv.kyoto-u.ac.jp/vch.
Willis RE [2009] Willis Research Network Home Confronting the Normality of Extremes,
available from URL: http://www.willisresearchnetwork.com.
World Housing Encyclopaedia [2009] Worldwide Housing > Find Reports, available from URL:
http://www.world-housing.net

Personal Communication also conducted with E. Anderson, I. Bal, J. Birkmann, F. Cotton, H.
Crowley, R. Eguchi, P. Foray, Duncan Gray, David Robinson, Trevor Dhu (Geoscience Australia), A.
Masi, J.Stewart, H. Sucuoglu, F. Wenzel, M. Wyss in 2008-09.

Appendix A: Overview of Current Literature

Daniell, April 2009 A1

APPENDIX A: OVERVIEW OF CURRENT LITERATURE

Definition of Ground Motion Intensity Parameters
A number of main ground motion parameters have been used within this report (not including
the 30 identified by King et al. (2005)). These encompass many different methods, including
peak ground acceleration, velocity and displacement (PGA, PGV and PGD), Arias intensity,
and Fourier Amplitude Spectrum, as well as Pseudo-spectral acceleration, velocity and
displacement (Sa, Sv, Sd) with various levels of damping.

These are used in varying degrees for the transference to various intensity scales, most of
which are discrete. Arias Intensity is the integral of the square of the acceleration-time
history, which can be measured in m/s or otherwise (Arias, 1970). It is found to be good for
measurement of liquefaction and slope stability as stated in the text. Significant durations (i.e.
5%-95% and 5%-75%) of Arias Intensity are also sometimes used. Peak ground acceleration
is technically the acceleration of a particle attached to a ground with respect to time. PGA is
equivalent to the spectral acceleration at T=0sec. Similarly, the definition also extends to
PGV for velocity-time and PGD for displacement-time particle behaviour by being the peak
of the first integration of acceleration-time, and the second integration, respectively. PGV has
been found to correlate best with damage.

The definition of pseudo-spectral acceleration is classified by the geometric mean of the
components of the accelerometer or by a single horizontal parameter, depending on the
situation (Baker and Cornell, 2006). Spectral acceleration is the maximum acceleration that
ground motion causes for a linear oscillator, given a certain damping (0%, 2%, 5%, 10%, 20%
are common values) for a specified period. It is related to Spectral Displacement (the
maximum relative displacement response), as shown by the equation:-
T
where
g Sa
Sd

2
,
.
2
= = .

The maximum relative velocity response (Sv, spectral velocity) is simply the spectral
displacement multiplied by the natural frequency.
The Fourier Amplitude Spectrum can be defined using the Fourier spectrum. The Fourier
spectrum reproduces the system as a sum of cosine and sine functions (Fourier analysis) with
respect to frequency (not damping). The Fourier Amplitude Spectrum is simply the inverse
transform of the time domain function presented in the frequency domain (Bard, 2007).

Definition of Intensity Scales
In terms of the intensity scales presented in the various ELE software packages, there are
many different types. The most commonly used is the Modified Mercalli (most ShakeMap
based software packages) and the MSK (Medvedev et al., 1969) scale (Extremum). These two
methods are essentially the same, using 12 different discrete levels to define the intensity of
an earthquake. The MSK scale uses a slightly smaller
II value and larger III value than the MM scale. JMA
is also used in Japan and is different from the others
in that it utilises 7 discrete levels rather than 12.
Mercalli-Cancani-Sieberg (MCS) is also another scale
(SIGE) used and EMS98 (European Macroseismic
Scale) is also used and is the 1998 update of the MSK
scale. These are all defined by the damage that the
earthquake causes, and how the earthquake was felt
by people. It can be influenced by site effects and
other amplification and deamplification factors
(Cotton, 2007). These are generally related to PGA
and other spectral ordinates via empirical equations.

Comparison of MMI, JMA, MSK and EMS (Nazri et al.,
2007)


The EMS-98 Macroseismic Scale, Gruenthal et al. (1998) comparative to the MSK, MM and MCS.

Definition of Magnitude
There are also many definitions of magnitude, which are used within the different ELE
software packages. The general magnitude which is used is moment magnitude (Mw) which
is simply a measure of the seismic energy released. It is measured in dyn.cm or N.m and is the
moment of the earthquake, which is equal to the convolution of the rigidity of the Earth,
average fault slip, and fault area. Other magnitude scales include local magnitude (or Richter
scale) which is measured by the displacement on a Wood-Anderson Seismometer calibrated at
T=0.8sec. Body Wave Magnitude (Mb) measures the P-wave amplitude at first arrival and
corresponds approximately to T=1sec. Surface Wave Magnitude (M
S
) similarly measures the
surface wave amplitude and corresponds to a period of approximately 20 seconds. M
JMA
is a
magnitude scale derived from the JMA scale. All of the other magnitude scales, apart from

M
W
, suffer from saturation at higher magnitudes (M
b
at approximately 6.0, and M
S
at
approximately 8.2).

A comparison of magnitude scales (Chen et al., 2003)

Damage Scales used for Vulnerability
There are many different damage scales for the ELE software packages, which have been
summarised. However, for a comparison between damage scales, Rossetto and Elnashai
(2003) have published a comparison of the most common method, screening, HAZUS and
D0-D5, etc. This will give some comparison for the data provided in Table 4-12:
Vulnerability Method and Damage Classes for the considered ELE Software Packages.



Various Damage States ATC-21, MSK, EMS98, Vision 2000, HAZUS have been presented for different
ELE Software Packages (Rossetto and Elnashai, 2003).
DBELA is somewhat similar to HAZUS, with the first part being none and slight together
before LS1.

The NEHRP Site Classification Scale
This scale is used in many different ELE software packages and therefore should be
presented. It is classified by S-wave velocity in the top 30 metres.

NEHRP site classification as shown in ICC (2006), FEMA (1997)


The Modified Acceleration-Displacement Response Spectrum (MADRS)
This method is included as it is in the most recent version of SELENA. It can be added as an
option and is simply a different method of picking the performance point using different
periods, similar to the N2 method previously stated.

Performance Point determination using MADRS (Molina et al., 2008b) from FEMA 440.

Appendix B: Additional Information for ELE Software Packages

Daniell, April 2009 B1

APPENDIX B: ADDITIONAL INFORMATION FOR ELE
SOFTWARE PACKAGES

PAGER-2

A sample of the new PAGER-2 database produced for the new version of PAGER, USGS (2009)
The PAGER Database contains information which includes a vulnerability code set by
Development rating and location, as well as a PAGER rating, and splits up the databases into
Urban and Non-Urban, Residential and Non-Residential. There are then 89 different building
types for which a comparative percentage of buildings has been applied. From these building
types, capacity curves have been derived for use in PAGER2. The country and the source
from which the data was established are included. These can be modified in order to suit extra
information that has been gained or as a dynamic-type system for future applications.
PAGER_database.xls is on the attached DVD. PAGER2 will be out soon and will feature
global social and economic loss functions as well as improved global settings (Jaiswal et al.,
2008a, Jaiswal et al., 2008b).
Appendix C: Loss Analysis and Software Production: Zeytinburnu Case Study

Daniell, April 2009 C1

APPENDIX C: LOSS ANALYSIS AND SOFTWARE
PRODUCTION:- ZEYTINBURNU CASE STUDY FOR
SELENA, MHAZUS AND MDBELA
The Scenario Earthquake:-Mw7.2 Earthquake located at approximately 28.84E, 40.9N
between fault segments 7 and 8 on the Marmara Sea fault, as shown on the map in the main
body of the text.
The Site:- Zeytinburnu District, a mainly business district to the north, with primarily
residential in the south.
The Data used:- 37 building types, 50 geocells:- 4 masonry types, 33 RC types, 11250
buildings.
The GMPE used:- Boore et al., 1997 with the erratum, Joyner and Boore (1993) distance
measure
ELE Software Packages:- MHAZUS, MDBELA (coded and produced in MATLAB),
SELENA (modified and adapted in MATLAB).
The Ground Motions used:- 100 correlated GM fields, 100 uncorrelated GM fields for
MHAZUS and MDBELA and 1 median GM field and variations for SELENA, MHAZUS and
MDBELA.
The Vulnerability used:- Capacity Spectrum Method and MADRS for SELENA, Capacity
Spectrum Method (with modified iteration) for MHAZUS, Displacement-based design for
MDBELA.

The geocells are as shown below. For the conversion to HAZUS it is possible to include some
of the data to allow for reproduction of readers for use in the OPAL procedure.
%GEOUNIT LONGI LATIT SOILT C1L C1M C1H URML URMM
2434 28.892 40.993 5 21.0 121.0 12.0 13.0 3.0
2489 28.898 41.008 5 8.0 57.0 4.0 4.0 0.0
2490 28.898 41.002 5 36.0 411.0 24.0 23.0 5.0
2491 28.898 40.998 5 50.0 439.0 31.0 103.0 20.0
2492 28.898 40.993 4 28.0 427.0 38.0 45.0 9.0

2493 28.898 40.987 4 31.0 499.0 43.0 52.0 10.0
2494 28.898 40.983 4 2.0 14.0 0.0 23.0 5.0
2543 28.903 41.028 4 3.0 6.0 0.0 4.0 0.0
2544 28.903 41.023 4 4.0 6.0 0.0 4.0 0.0
2545 28.903 41.017 4 3.0 5.0 0.0 3.0 0.0
2546 28.903 41.013 4 4.0 6.0 0.0 4.0 0.0
2547 28.903 41.008 4 20.0 311.0 22.0 22.0 5.0
2548 28.903 41.002 4 31.0 499.0 32.0 52.0 10.0
2549 28.903 40.998 4 46.0 416.0 30.0 98.0 20.0
2550 28.903 40.993 3 23.0 744.0 47.0 105.0 22.0
2551 28.903 40.987 3 248.0 552.0 35.0 41.0 8.0
2552 28.903 40.983 4 2.0 9.0 0.0 17.0 3.0
2601 28.907 41.028 4 9.0 12.0 0.0 5.0 1.0
2602 28.907 41.023 4 8.0 12.0 0.0 5.0 1.0
2603 28.907 41.017 3 4.0 8.0 0.0 4.0 0.0
2604 28.907 41.013 3 3.0 5.0 0.0 3.0 0.0
2605 28.907 41.008 3 28.0 423.0 31.0 29.0 6.0
2606 28.907 41.002 3 40.0 601.0 38.0 62.0 12.0
2607 28.907 40.998 3 19.0 286.0 17.0 29.0 6.0
2608 28.907 40.993 4 43.0 620.0 45.0 30.0 6.0
2609 28.907 40.987 4 4.0 22.0 0.0 38.0 8.0
2657 28.913 41.032 4 10.0 15.0 0.0 5.0 1.0
2658 28.913 41.028 4 8.0 12.0 0.0 5.0 1.0
2659 28.913 41.023 4 4.0 7.0 0.0 4.0 0.0
2660 28.913 41.017 3 2.0 2.0 0.0 1.0 0.0
2661 28.913 41.013 3 22.0 147.0 8.0 45.0 9.0
2662 28.913 41.008 3 14.0 93.0 5.0 28.0 5.0
2663 28.913 41.002 4 42.0 542.0 35.0 71.0 14.0
2664 28.913 40.998 4 30.0 405.0 26.0 53.0 10.0
2665 28.913 40.993 4 2.0 11.0 0.0 20.0 4.0
2666 28.913 40.987 4 0.0 0.0 0.0 1.0 0.0
2716 28.918 41.028 3 4.0 7.0 0.0 4.0 0.0
2717 28.918 41.023 3 2.0 4.0 0.0 3.0 0.0
2718 28.918 41.017 3 8.0 58.0 4.0 18.0 3.0
2719 28.918 41.013 3 8.0 64.0 4.0 20.0 4.0
2720 28.918 41.008 4 2.0 25.0 1.0 8.0 2.0
2721 28.918 41.002 4 2.0 11.0 0.0 17.0 3.0
2722 28.918 40.998 4 2.0 11.0 0.0 17.0 3.0
2723 28.918 40.993 4 1.0 0.0 0.0 6.0 2.0
2773 28.922 41.028 4 1.0 0.0 0.0 1.0 0.0
2774 28.922 41.023 3 4.0 7.0 0.0 4.0 0.0
2775 28.922 41.017 3 17.0 109.0 5.0 33.0 7.0
2776 28.922 41.013 3 25.0 157.0 8.0 48.0 10.0
2777 28.922 41.008 4 9.0 69.0 4.0 20.0 4.0
7678 28.907 40.983 4 1.0 0.0 0.0 4.0 0.0
7679 28.913 40.983 5 0.0 0.0 0.0 0.0 0.0
7680 28.918 40.987 4 0.0 0.0 0.0 0.0 0.0


%MDBELAmain.m The main processor which runs through the MDBELA Script
close all;
warning off
profile on

%INPUTS:
%Random population values coming in from a specified amount of buildings, %Numbuild = specified no. of
buildings, %random selection of the total set based on those numbers, %User input via a Graphical User
Interface, %Any other user inputs could also be produced and an automated response, %also seen.
%
%Look up each of these numbers and extract the relevant values into, %arrays...lb=length of beam, %hb =
depth of beam, %hc = depth of column, %hs = height of bottom storey (poor classification buildings)
%hs = height of storey (good classification buildings), %epsy = yield strain (refers to limit state 1)
%epsc2 = yield strain concrete (refers to limit state 2), %epss2 = yield strain steel (LS2)
%epsc3 = yield strain concrete (refers to limit state 3), %epss3 = yield strain steel (LS3)

%calculate mean and standard deviation for each of these parameters.
%Also the distribution type - whether lognormal or normal etc.

%-----------------------------------------------------------%
%MASONRY INPUT
%-----------------------------------------------------------%
n=200;
numbuild=n;
nbins=20;
gm=100;

buildings = dlmread('buildinginfo.txt',' ',[0 0 51 38]);

buildvalue(1:50,1:37) = buildings(1:50,3:39);
for b=1:37
for z=1:50

if buildvalue(z,b)>0
SiteBuild(z,b)=z;
end
end
end
Mw=7.2;
[sdB,saB,SamatrixB,SdmatrixB,SamatrixBspat,SdmatrixBspat] = GM(Mw,numbuild,gm);
% THUS THE GROUND MOTIONS HAVE BEEN CALLED

Then the distributions were added example as shown below.

%Masonry ground floor pier height

Mhsmean=2.62;
Mhscov=0.08;
Mhsstdev=Mhsmean*Mhscov;
Mhslambda=15;
Mhsd1=16;
Mhsd2=20;
Mhsmin=1.35;

Mhsmax=2.85;
Mhs1st = randraw('fnoncentral', [Mhslambda Mhsd1 Mhsd2], n);
Mhs=Mhsmax-(Mhs1st*(Mhsmax-Mhsmin))/(max(Mhs1st)-min(Mhs1st));
% figure(2);
% hold on
% % hist(Mhs1st,nbins);
% % X_Mhs=[min(Mhs1st):(max(Mhs1st)-min(Mhs1st))/nbins:max(Mhs1st)];
% % Y_Mhs=ncfpdf(X_Mhs,Mhsd1,Mhsd2,Mhslambda);
% % % Y_Mhsmod=Y_Mhs(end:-1:1);
% % scale_Mhs =(max(Mhs1st)-min(Mhs1st))*numbuild/nbins;
% % plot(X_Mhs,Y_Mhs*scale_Mhs,'r','linewidth',2.0); hold on
% X_Mhs1st=[min(Mhs1st):(max(Mhs1st)-min(Mhs1st))/nbins:max(Mhs1st)];
% Y_Mhs1st=ncfpdf(X_Mhs1st,Mhsd1,Mhsd2,Mhslambda);
% scale_Mhs1st =(max(Mhs1st)-min(Mhs1st))*numbuild/nbins;
% Y_Mhs = Mhsmax-(Y_Mhs1st*(Mhsmax-Mhsmin))/(max(Y_Mhs1st)-min(Y_Mhs1st));
% X_Mhs=[min(Mhs):(max(Mhs)-min(Mhs))/nbins:max(Mhs)];
% Y_Mhsmod=Y_Mhs1st(end:-1:1);
% scale_Mhs =(max(Mhs)-min(Mhs))*numbuild/nbins;
% hist(Mhs,nbins); hold on
% plot(X_Mhs,Y_Mhsmod*scale_Mhs1st,'r','linewidth',2.0); hold on
% set(gca,'Fontsize',14)
% xlabel('Ground floor pier height [m]');
% xlim([1 3])
% ylabel('pdf');

% Limit States for Masonry - Rotational Capacity
masLS=[1 2 3];

for i=1:3

chLSmean=[0.0027 0.005 0.01];
chLScov=[0.25 0.25 0.25];
chLSstdev=chLSmean.*chLScov;
chLS = randraw('normaltrunc', [chLSmean(i)-
4*chLSstdev(i),chLSmean(i)+4*chLSstdev(i),chLSmean(i),chLSstdev(i)], n);
clLSmean=[0.005 0.0077 0.015];
clLScov=[0.25 0.25 0.25];
clLSstdev=clLSmean.*clLScov;
clLS = randraw('normaltrunc', [clLSmean(i)-4*clLSstdev(i),clLSmean(i)+4*clLSstdev(i),clLSmean(i),clLSstdev(i)],
n);

More distributions were then made for all RC and Masonry buildings eg. for RC buildings

1 2 3 4 5 6 7 8
0
50
100
150
200
250
300
350
400
length of beam, l
b
[m]
p
d
f
0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8
0
50
100
150
200
250
300
height of beam, h
b
[m]
p
d
f

Characteristics of the beam: length (left panel) and height (right panel) RC buildings - Comparison
between the statistical truncated lognormal distributions (red curves) and the histograms generated by
Monte Carlo simulations (with a sample of 3000 buildings) (Daniell et al., 2009)

GM.m-A function was called for the ground motions with an example below of how to
have a distribution of the geocells and the GIS code for plotting
function [sdB,saB,SamatrixB,SdmatrixB,SamatrixBspat,SdmatrixBspat] = GM(Mw,numbuild,gm)

Mw = 7.2 %SET DETERMINISTIC VALUE

Alatb4=[40 48.0062]; Alongb4=[028 09.0269];
Blatb4=[40 53.5918]; Blongb4=[028 50.3282];
Clatb4=[40 43.0733]; Clongb4=[029 14.1314];

Alat=dm2degrees(Alatb4); Along=dm2degrees(Alongb4);
Blat=dm2degrees(Blatb4); Blong=dm2degrees(Blongb4);
Clat=dm2degrees(Clatb4); Clong=dm2degrees(Clongb4);

%B is always the shortest distance, as checked by an
interp along the sites

coordinates = dlmread('coordinates.txt',' ',[1 0 52 3]);
siteclasses = dlmread('siteclasses1.txt',' ',[0 0 51 2]);

for i=1:52
if siteclasses(i,3)==5
Vs(i)=150;
elseif siteclasses(i,3)==4
Vs(i)=250;
Vs(i)=520;
end
end

Geocells = coordinates(:,1);
Mean=3.37
Cov=38%
Mean=0.60
Cov=16%

Latitudes = coordinates(:,4);
Longitudes = coordinates(:,3);

for i=1:52
JBDistB(i) = deg2km(DISTANCE(Latitudes(i),Longitudes(i),Blat,Blong));
end;

TRIAL TO SEE ALONG THE FAULT IN ORDER TO CHECK INTERPOLATION
Xa=[Alat Blat]; Ya=[Along Blong]; ABLat=Alat:0.001:Blat;
ABLong=interp1(Xa,Ya,ABLat,'linear');

Xc=[Clat Blat]; Yc=[Clong Blong]; CBLat=Clat:0.001:Blat;
CBLong=interp1(Xc,Yc,CBLat,'linear');

ABCBLong = horzcat(ABLong,CBLong);
ABCBLat = horzcat(ABLat,CBLat);

coordinates = dlmread('coordinates.txt',' ',[1 0 52 3]);
siteclasses = dlmread('siteclasses1.txt',' ',[0 0 51 2]);
% fid = fopen('siteclasses1.txt', 'r');
% SC_text = textscan(fid, '%s', 0, 'delimiter', '|');
% siteclasses = textscan(fid, '%f %f %c', 'CollectOutput', 0);
% fclose(fid);

for i=1:52
if siteclasses(i,3)==5
Vs(i)=150;
Vs(i)=250;
Vs(i)=520;
end
end

Geocells = coordinates(:,1);
Latitudes = coordinates(:,4);
Longitudes = coordinates(:,3);

JBDistABC=zeros(52,270);

for i=1:52
for l=1:270
JBDistABC(i,l) =
deg2km(DISTANCE(Latitudes(i),Longitudes(
i),ABCBLat(l),ABCBLong(l)));
end;
end

for i=1:52
JBDistFin(i)=min(JBDistABC(i,:));
end;

for i=1:52

for j=1:11
if j==11
sdB(i,j)=sdB(i,10);
saB(i,j)=(sdB(i,11))/9.81*(2*pi/15)^2;
seB(i,j)=log(saB(i,11));
sigmaB(i,j)=sigmaB(i,10); %assumption
else
T=[0.001 0.1 0.15 0.2 0.3 0.4 0.5 0.75 1 2 15];

[saB(i,j), seB(i,j), sigmaB(i,j)] = BJF_1997_horiz(Mw, JBDistB(i), T(j), 1, Vs(i), 0);
saB(i,j)=saB(i,j);
seB(i,j)=seB(i,j);
sdB(i,j)=saB(i,j)*9.81*(T(j)/(2*pi))^2;

end
end;
end;

epsilonrand = gm;
for i=1:52
for m=1:gm
epsilonmatrix(i,m)=normrnd(0,1);
end
end; %CHECKED

% SdmatrixA=zeros(52,47,101);
SdmatrixB=zeros(52,11,gm);
SamatrixB=zeros(52,11,gm);
SdmatrixBspat=zeros(52,11,gm);
SamatrixBspat=zeros(52,11,gm);
% SdmatrixC=zeros(52,47,101);

%First Loop to find values
%using correlation consistent
for i=1:52
for j=1:11
for m=1:gm
SamatrixB(i,j,m)=9.81*(exp(seB(i,j)+epsilonmatrix(i,m)*sigmaB(i,j)));
SdmatrixB(i,j,m)=exp(seB(i,j)+epsilonmatrix(i,m)*sigmaB(i,j))*9.81*(T(j)/(2*pi))^2;
end;
end;
end;

%Spatial correlation using BOORE (2003)
CorrDist = zeros(52,52);

for i=1:52
for j=1:52
CorrDist(i,j)=
deg2km(DISTANCE(Latitudes(i),Longitudes(i),L
atitudes(j),Longitudes(j)));
pxy(i,j)=exp(-CorrDist(i,j)/5);
end;

end;

mucol=zeros(1,52);
epsilonmatrixspat = lhsnorm(mucol,pxy,gm);
epsilonmatrixspat = epsilonmatrixspat';

for i=1:52
for j=1:11
for m=1:gm
SamatrixBspat(i,j,m)=9.81*(exp(seB(i,j)+epsilonmatrixspat(i,m)*sigmaB(i,j)));
SdmatrixBspat(i,j,m)=exp(seB(i,j)+epsilonmatrixspat(i,m)*sigmaB(i,j))*9.81*(T(j)/(2*pi))^2;
end
end
end

for i=1:52
for j=1:11
seB(i,j)=9.81*seB(i,j);
saB(i,j)=9.81*saB(i,j);
end
end

plot(T,SdmatrixBspat(1,:,1));
x=Longitudes;
y=Latitudes;
%T=0.2secs, Uncorrelated Ground Motions

% GIS CODE

random=[1 2 3];
for j=1:3
for i=1:52
input=SdmatrixB(:,4,random(j));
end;

%creating grid with 0.005 degrees resolution
cellsize=0.005; minx=min(x)-0.0025;
maxx=max(x)+0.0025; miny=min(y)-0.0025;
maxy=max(y)+0.0025;

xi=(minx+0.0025:cellsize:maxx+0.0025);
yi=(miny+0.0025:cellsize:maxy+0.0025);
[X,Y]=meshgrid(xi,yi); [m,n]=size(X);

%populate the grid
xind=floor((x-minx)./cellsize)+1; yind=floor((y-
miny)./cellsize)+1;

zind=unique([yind,xind],'rows');
Sd=ones(m,n).*NaN;

for k=1:length(zind)
Longitude []
L
a
t
i
t
u
d
e

[
]
28.89 28.9 28.91 28.92
40.98
40.985
40.99
40.995
41
41.005
41.01
41.015
41.02
41.025
41.03
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22

ind=find(xind==zind(k,2) &...
yind==zind(k,1)==1);
Sd(zind(k,1),zind(k,2))=mean(input(ind));
end

%plot the results
figure
imagesc(xi,yi,Sd);
set(gca,'ydir','normal')
axis equal

c1=colorbar;
caxis([0 max(input)])
set(get(c1,'ylabel'),'string','Sd (m)')
title(['Uncorrelated Ground Motions Trial No.', num2str(random(j)),' for T=0.2secs']);
end

%T=1.0secs, Uncorrelated Ground Motions
for j=1:3
for i=1:52
input=SdmatrixB(:,9,random(j));
end;

%creating grid with 0.005 degrees resolution
cellsize=0.005;
minx=min(x)-0.0025;
maxx=max(x)+0.0025;
miny=min(y)-0.0025;
maxy=max(y)+0.0025;

xi=(minx+0.0025:cellsize:maxx+0.0025);
yi=(miny+0.0025:cellsize:maxy+0.0025);
[X,Y]=meshgrid(xi,yi);

%populate the grid using gridcell averaging
xind=floor((x-minx)./cellsize)+1;
yind=floor((y-miny)./cellsize)+1;

zind=unique([yind,xind],'rows');
Sd=ones(m,n).*NaN;

for k=1:length(zind)

ind=find(xind==zind(k,2) &...
yind==zind(k,1)==1);

Sd(zind(k,1),zind(k,2))=mean(input(ind));

end

%plot the results
figure

imagesc(xi,yi,Sd);
set(gca,'ydir','normal')
axis equal

c1=colorbar;
set(get(c1,'ylabel'),'string','Sd (m)')
caxis([0 max(input)])
title(['Uncorrelated Ground Motions Trial No.', num2str(random(j)),' for T=1.0secs']);
xlabel('Longitude');
ylabel('Latitude');
end

BJF_1997_horiz.m Boore 1997 GMPE with erratum changed (used for MHAZUS and
MDBELA)
function [sa, se, sigma] = BJF_1997_horiz(M, R, T, Fault_Type, Vs, arb)
% NEED ALL COEFFICIENTS FROM BJF AND THEN CAN SELECT ANY DESIRED TO FORM THE DISTRIBUTIONS
period = [ 0.001 0.1 0.11 0.12 0.13 0.14 0.15 0.16 0.17 0.18 0.19 0.2 0.22 0.24 0.26 0.28 0.3 0.32 0.34 0.36 0.38
0.4 0.42 0.44 0.46 0.48 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 ];
B1ss = [ -0.313 1.006 1.072 1.109 1.128 1.135 1.128 1.112 1.09 1.063 1.032 0.999 0.925 0.847 0.764 0.681
0.598 0.518 0.439 0.361 0.286 0.212 0.14 0.073 0.005 -0.058 -0.122 -0.268 -0.401 -0.523 -0.634 -0.737 -0.829 -
0.915 -0.993 -1.066 -1.133 -1.249 -1.345 -1.428 -1.495 -1.552 -1.598 -1.634 -1.663 -1.685 -1.699 ];
B1rv = [ -0.117 1.087 1.164 1.215 1.246 1.261 1.264 1.257 1.242 1.222 1.198 1.17 1.104 1.033 0.958 0.881
0.803 0.725 0.648 0.57 0.495 0.423 0.352 0.282 0.217 0.151 0.087 -0.063 -0.203 -0.331 -0.452 -0.562 -0.666 -
0.761 -0.848 -0.932 -1.009 -1.145 -1.265 -1.37 -1.46 -1.538 -1.608 -1.668 -1.718 -1.763 -1.801 ];
B1all = [ -0.242 1.059 1.13 1.174 1.2 1.208 1.204 1.192 1.173 1.151 1.122 1.089 1.019 0.941 0.861 0.78 0.7
0.619 0.54 0.462 0.385 0.311 0.239 0.169 0.102 0.036 -0.025 -0.176 -0.314 -0.44 -0.555 -0.661 -0.76 -0.851 -
0.933 -1.01 -1.08 -1.208 -1.315 -1.407 -1.483 -1.55 -1.605 -1.652 -1.689 -1.72 -1.743 ];
B2 = [ 0.527 0.753 0.732 0.721 0.711 0.707 0.702 0.702 0.702 0.705 0.709 0.711 0.721 0.732 0.744 0.758 0.769
0.783 0.794 0.806 0.82 0.831 0.84 0.852 0.863 0.873 0.884 0.907 0.928 0.946 0.962 0.979 0.992 1.006 1.018
1.027 1.036 1.052 1.064 1.073 1.08 1.085 1.087 1.089 1.087 1.087 1.085 ];
B3 = [ 0 -0.226 -0.23 -0.233 -0.233 -0.23 -0.228 -0.226 -0.221 -0.216 -0.212 -0.207 -0.198 -0.189 -0.18 -0.168 -
0.161 -0.152 -0.143 -0.136 -0.127 -0.12 -0.113 -0.108 -0.101 -0.097 -0.09 -0.078 -0.069 -0.06 -0.053 -0.046 -
0.041 -0.037 -0.035 -0.032 -0.032 -0.03 -0.032 -0.035 -0.039 -0.044 -0.051 -0.058 -0.067 -0.074 -0.085 ];
B5 = [ -0.778 -0.934 -0.937 -0.939 -0.939 -0.938 -0.937 -0.935 -0.933 -0.93 -0.927 -0.924 -0.918 -0.912 -0.906 -
0.899 -0.893 -0.888 -0.882 -0.877 -0.872 -0.867 -0.862 -0.858 -0.854 -0.85 -0.846 -0.837 -0.83 -0.823 -0.818 -
0.813 -0.809 -0.805 -0.802 -0.8 -0.798 -0.795 -0.794 -0.793 -0.794 -0.796 -0.798 -0.801 -0.804 -0.808 -0.812 ];
Bv = [ -0.371 -0.212 -0.211 -0.215 -0.221 -0.228 -0.238 -0.248 -0.258 -0.27 -0.281 -0.292 -0.315 -0.338 -0.36 -
0.381 -0.401 -0.42 -0.438 -0.456 -0.472 -0.487 -0.502 -0.516 -0.529 -0.541 -0.553 -0.579 -0.602 -0.622 -0.639 -
0.653 -0.666 -0.676 -0.685 -0.692 -0.698 -0.706 -0.71 -0.711 -0.709 -0.704 -0.697 -0.689 -0.679 -0.667 -0.655 ];
Va = [ 1396 1112 1291 1452 1596 1718 1820 1910 1977 2037 2080 2118 2158 2178 2173 2158 2133 2104 2070
2032 1995 1954 1919 1884 1849 1816 1782 1710 1644 1592 1545 1507 1476 1452 1432 1416 1406 1396 1400
1416 1442 1479 1524 1581 1644 1714 1795 ];
h = [ 5.57 6.27 6.65 6.91 7.08 7.18 7.23 7.24 7.21 7.16 7.1 7.02 6.83 6.62 6.39 6.17 5.94 5.72 5.5 5.3 5.1 4.91
4.74 4.57 4.41 4.26 4.13 3.82 3.57 3.36 3.2 3.07 2.98 2.92 2.89 2.88 2.9 2.99 3.14 3.36 3.62 3.92 4.26 4.62 5.01
5.42 5.85 ];
sigma1 = [ 0.431 0.44 0.437 0.437 0.435 0.435 0.435 0.435 0.435 0.435 0.435 0.435 0.437 0.437 0.437 0.44 0.44
0.442 0.444 0.444 0.447 0.447 0.449 0.449 0.451 0.451 0.454 0.456 0.458 0.461 0.463 0.465 0.467 0.467 0.47
0.472 0.474 0.477 0.479 0.481 0.484 0.486 0.488 0.49 0.493 0.493 0.495 ];
sigmac = [0.160 0.134 0.141 0.148 0.153 0.158 0.163 0.166 0.169 0.173 0.176 0.177 0.182 0.185 0.189 0.192
0.195 0.197 0.199 0.200 0.202 0.204 0.205 0.206 0.209 0.210 0.211 0.214 0.216 0.218 0.220 0.221 0.223 0.226
0.228 0.230 0.230 0.233 0.236 0.239 0.241 0.244 0.246 0.249 0.251 0.254 0.256];
sigmar = [ 0.460 0.460 0.459 0.461 0.461 0.463 0.465 0.466 0.467 0.468 0.469 0.470 0.473 0.475 0.476 0.480
0.481 0.484 0.487 0.487 0.491 0.491 0.494 0.494 0.497 0.497 0.501 0.504 0.506 0.510 0.513 0.515 0.518 0.519
0.522 0.525 0.527 0.531 0.534 0.537 0.541 0.544 0.546 0.550 0.553 0.555 0.557];

sigmae = [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]; %inter
variability = 0
sigmalny=sigmar;

% interpolate between periods if necessary
if (length(find(period == T)) == 0)
index_low = sum(period<T);
T_low = period(index_low);
T_hi = period(index_low+1);

[sa_low, sigma_low] = BJF_1997_horiz(M, R, T_low, Fault_Type, Vs, arb);
[sa_hi, sigma_hi] = BJF_1997_horiz(M, R, T_hi, Fault_Type, Vs, arb);

x = [log(T_low) log(T_hi)];
Y_sa = [log(sa_low) log(sa_hi)];
Y_sigma = [sigma_low sigma_hi];
sa = exp(interp1(x,Y_sa,log(T)));
sigma = interp1(x,Y_sigma,log(T));

else
i = find(period == T);

% compute median and sigma
r = sqrt(R^2 + h(i)^2);

if(Fault_Type == 1)
b1 = B1ss(i);
elseif(Fault_Type == 2)
b1 = B1rv(i);
else
b1 = B1all(i);
end
lny= b1 + B2(i)*(M-6) + B3(i)*(M-6)^2 + B5(i)*log(r) + Bv(i)*log(Vs / Va(i));
sa = exp(lny);
se = lny;
if (arb) % arbitrary component sigma
sigma = sigmalny(i);
else % average component sigma
sigma = sqrt(sigmar(i)^2 + sigmae(i)^2);
end
end

DBELA Script contains the main calculations of the DBELA curve performance points
function
[delta1sd,delta2sd,delta3sd,RCLS1T,RCLS2T,RCLS3T,LS2eta,LS3eta,CollectLS,CollectLSspat,CollectLSmed]=RCScr
ipt(col,floor,sttype,totalmatrix,sdB,SdmatrixB,SdmatrixBspat,SiteBuild,numbuild,gm)

efhBS=zeros(numbuild,1); efhCS=zeros(numbuild,1); delta1sd=zeros(numbuild,1); delta2sd=zeros(numbuild,1);
delta3sd=zeros(numbuild,1);

numparams=22;
RCparams=zeros(numbuild,numparams); RCLS1T=zeros(numbuild,1); RCLS2T=zeros(numbuild,1);
RCLS3T=zeros(numbuild,1); LS2zeta=zeros(numbuild,1); LS3zeta=zeros(numbuild,1); LS2eta=zeros(numbuild,1);

LS3eta=zeros(numbuild,1);

for i=1:numbuild
RCparams(i,1)=totalmatrix(i,20); %random beam length
RCparams(i,2)=totalmatrix(i,21); %random beam depth
if col<=3; RCparams(i,3)=totalmatrix(i,16); %random column depth
elseif col==4; RCparams(i,3)=totalmatrix(i,17); %random column depth
elseif col==5; RCparams(i,3)=totalmatrix(i,18);
elseif col>=6; RCparams(i,3)=totalmatrix(i,19);
end

RCparams(i,4)=totalmatrix(i,15); %random storey height
if floor==1
RCparams(i,5)=col*RCparams(i,4); %buildingheight
a(i) = (RCparams(i,2)/RCparams(i,1))/(RCparams(i,3)/RCparams(i,4));
if a(i)<1; BCfailtype(i)=1; %beam sway
else ; BCfailtype(i)=2; %column sway
end
elseif floor==2
RCparams(i,5)=col*RCparams(i,4)+(totalmatrix(i,14)-1)*RCparams(i,4);
a(i) = (RCparams(i,2)/RCparams(i,1))/(RCparams(i,3)/(RCparams(i,4)*(totalmatrix(i,14))));
if a(i)<1; BCfailtype(i)=1; %beam sway
else; BCfailtype(i)=2; %column sway
end ; end;
if sttype==1
RCparams(i,6)=totalmatrix(i,22);
elseif sttype==2
RCparams(i,6)=totalmatrix(i,23);
end;
RCparams(i,7)=BCfailtype(i); %type of failure mode
RCparams(i,8)=totalmatrix(i,24); %LS2, concrete
RCparams(i,9)=totalmatrix(i,25); %LS2, steel
RCparams(i,10)=totalmatrix(i,26); %LS3, concrete
RCparams(i,11)=totalmatrix(i,27); %LS3, steel
RCparams(i,12)=RCparams(i,5)*0.1; %Ty, period
%column versus a beam section, see which governs
end

%STRUCTURAL DISPLACEMENT
sitevalue=nonzeros(SiteBuild);
CollectLS=zeros(length(sitevalue),gm,numbuild,'single');
CollectLSspat=zeros(length(sitevalue),gm,numbuild,'single');
CollectLSmed=zeros(length(sitevalue),1,numbuild,'single');

%calculation of the efh-beam sway based on storey height
for j=1:numbuild
if(BCfailtype(j)==1);
if (col<=4) ; efhBS(j) = 0.64;
elseif (col>4 && col<20) ; efhBS(j) = 0.64-0.0125*(col-4);
else ; efhBS(j) = 0.44;
end
% deltaRC_y(i,j)=0.5*efh_B(j)*Ht_RC(i,j)*eps_y(i,j)*RC(i,7)/RC(i,8); % yield displ
% deltaRC_LS2(i,j)=deltaRC_y(i,j)+0.5*(RC(i,3)+RC(i,5)-1.7*eps_y(i,j))*efh_B(j)*Ht_RC(i,j); % displ_LS2
% deltaRC_LS3(i,j)=deltaRC_y(i,j)+0.5*(RC(i,4)+RC(i,6)-1.7*eps_y(i,j))*efh_B(j)*Ht_RC(i,j);


Ts2(j)=0.1*RCparams(j,5)*sqrt(1+((RCparams(j,8)+RCparams(j,9)-
1.7*RCparams(j,6))*RCparams(j,2))/(RCparams(j,6)*RCparams(j,1)));
1.7*RCparams(j,6))*RCparams(j,2))/(RCparams(j,6)*RCparams(j,1)));

delta1sd(j)=0.5*efhBS(j)*RCparams(j,5)*RCparams(j,6)*RCparams(j,1)/RCparams(j,2);

delta2sd(j)=0.5*efhBS(j)*RCparams(j,5)*RCparams(j,6)*RCparams(j,1)/RCparams(j,2)+0.5*((RCparams(j,8)+RCp
arams(j,9)-1.7*RCparams(j,6))*efhBS(j)*RCparams(j,5));

delta3sd(j)=0.5*efhBS(j)*RCparams(j,5)*RCparams(j,6)*RCparams(j,1)/RCparams(j,2)+0.5*((RCparams(j,11)+RC
params(j,10)-1.7*RCparams(j,6))*efhBS(j)*RCparams(j,5));

else

efhCS(j) = 0.5;
%CALCULATION OF THE PERIODS FOR EACH LIMIT STATE OF THE 3000 RANDOM BUILDINGS
2.1*RCparams(j,6))*RCparams(j,3))/(0.86*efhCS(j)*RCparams(j,5)*RCparams(j,6)));
2.1*RCparams(j,6))*RCparams(j,3))/(0.86*efhCS(j)*RCparams(j,5)*RCparams(j,6)));
% CALCULATION OF THE DISPLACEMENT OF EACH LIMIT STATE
delta1sd(j)=0.43*efhCS(j)*RCparams(j,6)*RCparams(j,5)*RCparams(j,4)/RCparams(j,3);

delta2sd(j)=0.43*efhCS(j)*RCparams(j,6)*RCparams(j,5)*RCparams(j,4)/RCparams(j,3)+0.5*((RCparams(j,8)+RC
params(j,9)-2.14*RCparams(j,6)))*RCparams(j,4);

delta3sd(j)=0.43*efhCS(j)*RCparams(j,6)*RCparams(j,5)*RCparams(j,4)/RCparams(j,3)+0.5*((RCparams(j,10)+R
Cparams(j,11)-2.14*RCparams(j,6)))*RCparams(j,4);

end
% RECORDING VALUES IN A MATRIX
RCparams(j,13)=4*pi^2*delta1sd(j)/(9.81*(RCparams(j,12))^2); %collapse multiplier ;
RCparams(j,14)=Ts2(j); RCparams(j,15)=Ts3(j);
RCparams(j,16)=delta1sd(j);
RCparams(j,19)=(RCparams(j,17)/RCparams(j,16)); %LS2 ductility
RCparams(j,20)=(RCparams(j,18)/RCparams(j,16)); %LS3 ductility
RCLS2T(j)=RCparams(j,12)*sqrt(RCparams(j,19));
RCLS3T(j)=RCparams(j,12)*sqrt(RCparams(j,20));
RCLS1T(j)=RCparams(j,12);
LS2zeta(j)=0.05+0.565*((RCparams(j,19)-1)/(RCparams(j,19)*pi));
LS3zeta(j)=0.05+0.565*((RCparams(j,20)-1)/(RCparams(j,20)*pi));
LS2eta(j)=sqrt(7/(2+100*LS2zeta(j)));
LS3eta(j)=sqrt(7/(2+100*LS3zeta(j)));
%Periods used
x=[0.001 0.1 0.15 0.2 0.3 0.4 0.5 0.75 1 2 15]';
%Output matrix for Sd
SdLS1out=SdmatrixB; SdLS2out=LS2eta(j)*SdmatrixB; SdLS3out=LS3eta(j)*SdmatrixB;
for site=1:length(sitevalue)
for gmval=1:gm
% SdLS1out(site,:,gm)
y1=SdLS1out(sitevalue(site),:,gmval)';
xi1=RCLS1T(j);


% SPEED UP VECTORISED CAN BE INCLUDED WITH A SELENA OR OTHER TYPE TO MAKE EVEN FASTER.
% SPEED UP FOR SCREENING ALSO CODED BUT MORE COMPUTATIONALLY EFFICIENT LIKE THIS vs. interp1
% Special scalar xi case
%tic
r = max(find(x <= xi1));
r(xi1==x(end)) = length(x)-1;
u = (xi1-x(r))./(x(r+1)-x(r));
Sdvalue1=y1(r,:)+(y1(r+1,:)-y1(r,:)).*u(:,ones(1,size(y1,2)));
%toc

% Sdvalue1 = interp1q(T,SdLS1out(sitevalue(site),:,gm)',RCLS1T(j));
if Sdvalue1<RCparams(j,16)
CollectLS(site,gmval,j)=0;
else

xi2=RCLS2T(j);

u = (xi2-x(r))./(x(r+1)-x(r));

else
xi3=RCLS3T(j);

u = (xi3-x(r))./(x(r+1)-x(r));

else
end
end
end
% CHECK TO SEE THAT IT
IS WORKING
% if j==31 && site ==1 && gm==98
% figure
% SdLS1out(1,:,98)
% SdLS2out(1,:,98)
% SdLS3out(1,:,98)
%
Cl ass RC-5b, DBELA (Spati al l y correl ated)
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
No Damage Sl i ght Moderate Severe Col l apse
%

o
f

b
u
i
l
d
i
n
g
s
Geocel l 1
Geocel l 5

% Ability to plot any geocell and class plot(x,SdLS1out(1,:,98),'r','linewidth',2.0); hold on
% plot(x,SdLS2out(1,:,98),'r','linewidth',2.0); hold on
% plot(x,SdLS3out(1,:,98),'r','linewidth',2.0); hold on
% plot(RCLS1T(j),RCparams(j,16),'ro'); hold on
% xlabel('T [secs]');
% xlim([0 2.0])
% ylabel('S_d [m]');
% title('DBELA Capacity vs. Demand - RC');
% b=CollectLS(1,98,31);
% end% WORKING
end
end

PART OF THE DBELA MASONRY CODING capacities and iteration looping
(searching for which limit state in which it is)
function
[CollectLSspatMAS,CollectLSMAS,CollectLSMASmed]=MASScript(MAS,SiteBuild,sdB,SdmatrixB,SdmatrixBspat,n
umbuild,gm)

sitevalue=nonzeros(SiteBuild); CollectLSMAS=zeros(length(sitevalue),gm,numbuild,'single');
CollectLSspatMAS=zeros(length(sitevalue),gm,numbuild,'single');
CollectLSMASmed=zeros(length(sitevalue),1,numbuild,'single');

for j=1:numbuild
x=[0.001 0.1 0.15 0.2 0.3 0.4 0.5 0.75 1 2 15]';

%Adding in the different reduction factor for each of
SdLS1out=SdmatrixB; SdLS2out=sqrt(7/(2+10))*SdmatrixB; SdLS3out=sqrt(7/(2+15))*SdmatrixB;

for gmval=1:gm

y1=SdLS1out(sitevalue(site),:,gm)';
xi1=MAS(j,4);
u = (xi1-x(r))./(x(r+1)-x(r));

if Sdvalue1<MAS(j,11)
CollectLSMAS(site,gmval,j)=0;
else

xi2=MAS(j,12);


u = (xi2-x(r))./(x(r+1)-x(r));
%INTERPOLATION SEQUENCE (ALSO BY 0, 0.3, 1.0, 2.0)
CollectLSMAS(site,gmval,j)=1;
else
xi3=MAS(j,15);

u = (xi3-x(r))./(x(r+1)-x(r));

if Sdvalue3<MAS(j,13); CollectLSMAS(site,gmval,j)=2;
else ; CollectLSMAS(site,gmval,j)=3;
End; end ; end; end; end

%spatial correlation
SdLS1outspat=SdmatrixBspat; SdLS2outspat=sqrt(7/(2+10))*SdmatrixBspat;
SdLS3outspat=sqrt(7/(2+15))*SdmatrixBspat;

for gmval=1:gm
y1=SdLS1outspat(sitevalue(site),:,gm)';
xi1=MAS(j,4);

u = (xi1-x(r))./(x(r+1)-x(r));
% interp1 (T,SdLS1out(sitevalue(site),:,gm)',RCLS1T(j));
CollectLSspatMAS(site,gmval,j)=0;
else

y2=SdLS2outspat(sitevalue(site),:,gmval)';
xi2=MAS(j,14);

u = (xi2-x(r))./(x(r+1)-x(r));

else
y3=SdLS3outspat(sitevalue(site),:,gmval)';
xi3=MAS(j,15);


u = (xi3-x(r))./(x(r+1)-x(r));

else
End; end ; end; end ; end

SdLS1outmed=sdB;
SdLS2outmed=sqrt(7/(2+10))*sdB;
SdLS3outmed=sqrt(7/(2+15))*sdB;



y1=SdLS1outmed(sitevalue(site),:,1)';
xi1=MAS(j,4);

u = (xi1-x(r))./(x(r+1)-x(r));

CollectLSMASmed(site,1,j)=0;
else
xi2=MAS(j,14);

u = (xi2-x(r))./(x(r+1)-x(r));

else
xi3=MAS(j,15);


u = (xi3-x(r))./(x(r+1)-x(r));

else
end
end
end
end
end

PART OF THE MHAZUS CODING performance points
%Performance point coding
%Bring in the Sd and Sa matrix from the ground motion calculations
SdmatrixURMMSmed=zeros(length(sitevalue),11,1);
SamatrixURMMSmed=zeros(length(sitevalue),11,1);
%Uncorrelated (CollectLS)
%Samatrixb
zeta=0.05;
for site=1:52
for gmval=1:gm
[C1LSd_PP,C1LSa_PP] =
intersections(C1LSdcurve,C1LSacurve,SdmatrixB(site,:,gmval),SamatrixB(site,:,gmval));
if isempty(C1LSd_PP)
eta=0.55;
SdmatrixC1LS(site,:,gmval)=eta*SdmatrixB(site,:,gmval);
SamatrixC1LS(site,:,gmval)=eta*SamatrixB(site,:,gmval);
intersections(C1LSdcurve,C1LSacurve,SdmatrixC1LS(site,:,gmval),SamatrixC1LS(site,:,gmval));
end
CollectC1L(site,gmval,1:4)=0;
CollectC1L(site,gmval,5)=1;
else
m_delta=C1LSd_PP(1)/C1LDpt;
zeta_eq=zeta+0.565*((m_delta-1)/pi*m_delta); %THIS PART IS BEING EDITED AND IS JUST AN ITERATION
FOR areaundercurve.m
eta = sqrt(10/(5+zeta_eq*100));
if eta<0.55
eta = 0.55;
end
SdmatrixC1L(site,:,gmval)=eta*SdmatrixB(site,:,gmval);
SamatrixC1L(site,:,gmval)=eta*SamatrixB(site,:,gmval);
intersections(C1LSdcurve,C1LSacurve,SdmatrixC1L(site,:,gmval),SamatrixC1L(site,:,gmval));

[C1LSd_PP,C1LSa_PP]=areaundercurve(C1LSdcurve,C1LSacurve,SdmatrixC1L(site,:,gmval),SamatrixC1L(site,:,gm
val)),C1LSd_PP,C1LSa_PP;
CollectC1L(site,gmval,1:4)=0;
CollectC1L(site,gmval,5)=1;

else

%then get the performance point - use with the fragility curves for
for i=1:5
if i==1
PC1L(i)=1-cdf('norm',(1/betads)*log(C1LSd_PP(1)/C1LSdbar(i)),0,1);
CollectC1L(site,gmval,i)=PC1L(i);
elseif i==5
CollectC1L(site,gmval,i)=1-PC1L(4);
else
PC1L(i)=1-cdf('norm',(1/betads)*log(C1LSd_PP(1)/C1LSdbar(i)),0,1);
CollectC1L(site,gmval,i)=PC1L(i)-PC1L(i-1);
end
end
end
end
%C1M - Uncorrelated

[C1MSd_PP,C1MSa_PP] =
intersections(C1MSdcurve,C1MSacurve,SdmatrixB(site,:,gmval),SamatrixB(site,:,gmval));
if isempty(C1MSd_PP)
eta=0.55;
SdmatrixC1MS(site,:,gmval)=eta*SdmatrixB(site,:,gmval);
SamatrixC1MS(site,:,gmval)=eta*SamatrixB(site,:,gmval);
intersections(C1MSdcurve,C1MSacurve,SdmatrixC1MS(site,:,gmval),SamatrixC1MS(site,:,gmval));
end
if isempty(C1MSd_PP)
CollectC1M(site,gmval,1:4)=0;
CollectC1M(site,gmval,5)=1;
else
m_delta=C1MSd_PP(1)/C1MDpt;
zeta_eq=zeta+0.565*((m_delta-1)/pi*m_delta);
eta = sqrt(10/(5+zeta_eq*100));
if eta<0.55
eta = 0.55;
end
SdmatrixC1M(site,:,gmval)=eta*SdmatrixB(site,:,gmval);
SamatrixC1M(site,:,gmval)=eta*SamatrixB(site,:,gmval);
intersections(C1MSdcurve,C1MSacurve,SdmatrixC1M(site,:,gmval),SamatrixC1M(site,:,gmval));
[C1MSd_PP,C1MSa_PP] = areaundercurve(C1MSdcurve,C1MSacurve,SdmatrixC1M(site,:,gmval),
SamatrixC1M(site,:,gmval),C1MSd_PP,C1MSa_PP);
if isempty(C1MSd_PP); CollectC1M(site,gmval,1:4)=0; CollectC1M(site,gmval,5)=1;
else

%adapted from SELENA %(dy,ay) First trial yield point
%(dp2,ap2) Intersection of the line with the capacity curve
%(dpi,api) First trial performance point
%Disp,acc Vectors containing the capacity curve data
function [cd1,cd2]=areaundercurve(dy,ay,dp2,ap2,dpi,api);
%First Integrate the two areas (before dp2, ap2 and after dp2,ap2)
Disp=SdmatrixB;Acc=SamatrixB;
findind=find(Disp<=dp2);firstx=Disp(findind); firsty=acc(findind);
areafirst=trapz(firstx,firsty);

%findind=find(Disp>=dp2&Disp<=dpi);
findind=find(Disp<=dpi); asecondx=Disp(findind);
asecondy=acc(findind); areaseconda=trapz(asecondx,asecondy);
areasecond=areaseconda-areafirst;
%Now compute area of geometrical forms.
lado1=ap2; lado2=dpi-dp2; areacuadrado=lado1*lado2; lado3=api-ap2; areatriangulo=lado2*lado3/2;
area2=areasecond-areatriangulo-areacuadrado; lado4=dp2;
arearectangulogrande=lado1*lado4; lado5=dy; lado6=ay;
areatriangulogrande=lado6*lado5/2; lado7=ap2-ay; lado8=dp2-dy;
areatriangulopequeno=lado7*lado8/2;arearectangulopequeno=lado7*lado5;
area1=arearectangulogrande-areatriangulogrande-arearectangulopequeno-areatriangulopequeno-areafirst;
(cd1,cd2)=area2-area1;

%then get the performance point - use with the fragility curves for
for i=1:5
if i==1
PC1M(i)=1-cdf('norm',(1/betads)*log(C1MSd_PP(1)/C1MSdbar(i)),0,1);
CollectC1M(site,gmval,i)=PC1M(i);
elseif i==5
CollectC1M(site,gmval,i)=1-PC1M(4);
else
PC1M(i)=1-cdf('norm',(1/betads)*log(C1MSd_PP(1)/C1MSdbar(i)),0,1);
CollectC1M(site,gmval,i)=PC1M(i)-PC1M(i-1);
end
end
end
end

Fragility Curve example for HAZUS

%Given the following capacity curves
and capacity points.
SRC5aLS1input=sort(totalmatrix(1:numbuild,102));

SRC5aLS1Tinput=sort(totalmatrix(1:numbuild,213));

SRC5aLS1T=median(SRC5aLS1Tinput);
%distributions to gain the distributions
SRC5aLS1med=log(median(SRC5aLS1input));
SRC5aLS1sigma=log(std(SRC5aLS1input));
SRC5aLS1=logncdf(SRC5aLS1input,SRC5aLS1med,SRC5aLS1sigma);

SRC5aLS1Sa=SRC5aLS1med/(SRC5aLS1T/(2*pi))^2;
0 0.02 0.04 0.06 0.08 0.1 0.12
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Displacement, S
d
[m]
A
c
c
e
l
e
r
a
t
i
o
n
,

S
a

[
m
/
s
2
]
C1L
C1M
C1H
URML
URMM

SRC5aLS1Sdcurve = [0 SRC5aLS1med SRC5aLS2med SRC5aLS3med];
SRC5aLS1Sacurve = [0 SRC5aLS1Sa SRC5aLS2Sa SRC5aLS3Sa];
SRC5aLS1Sacurve1 = [0 SRC5aLS1Sa SRC5aLS1Sa SRC5aLS1Sa];

C1MSlinput=C1MSdbar(1)*-4*exp(betads):0.01:C1MSdbar(1)*4*exp(betads);
C1MSlmed=log(C1MSdbar(1));
C1MSlsigma=betads;
C1MSl=logncdf(C1MSlinput,C1MSlmed,C1MSlsigma);
C1MMoinput=C1MSdbar(2)*-
4*exp(betads):0.01:C1MSdbar(2)*4*exp(betads);
C1MMomed=log(C1MSdbar(2));
C1MMosigma=betads;
C1MMo=logncdf(C1MMoinput,C1MMomed,C1MMo
sigma);
C1MExinput=C1MSdbar(3)*-
C1MExmed=log(C1MSdbar(3));
C1MExsigma=betads;
C1MEx=logncdf(C1MExinput,C1MExmed,C1MExsigm
a);
C1MCoinput=C1MSdbar(4)*-
C1MComed=log(C1MSdbar(4));
C1MCosigma=betads;
C1MCo=logncdf(C1MCoinput,C1MComed,C1MCosig
ma);

figure(41);
plot(SRC5aLS1input,SRC5aLS1,'linewidth',2.0); hold on
plot(SRC5aLS3input,Avalue,'linewidth',2.0); hold on
plot(C1MSlinput,C1MSl,'r','linewidth',2.0); hold on
plot(C1MMoinput,C1MMo,'r','linewidth',2.0);
plot(C1MExinput,C1MEx,'r','linewidth',2.0);
plot(C1MCoinput,C1MCo,'r','linewidth',2.0);
xlabel('S_d [m]');
xlim([0 0.7])
ylabel('P(DS>=Sd)');
title('Fragility Curve, DBELA SRC5a LS1,2 and 3, vs.
HAZUS C1M Sl,Mo,Ex,Co');

Using inputs from HAZUS Technical
Manual Chapter 5
%URML,slight-moderate-extensive-complete
for i=1:4
URMLmodh(i)=URML59(2);
URMLdeltaR(i)=URML59(2+i);
URMLSdbar(i)=URMLdeltaR(i)*URMLmodh(i);
URMLSdbar1sd(i)=URMLSdbar(i)*exp(betads);
URMLSdbarmin1sd(i)=URMLSdbar(i)*exp(-betads);
URMLSd(i)=URML59(5+2*i);
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
S
d
[m]
P
(
D
S
>
=
S
d
)
Fragility Curve, DBELA RC5b LS1,2 and 3, vs. HAZUS C1M Sl,Mo,Ex,Co
0 0.02 0.04 0.06 0.08 0.1 0.12
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Displacement [m]
F
r
a
g
i
l
i
t
y

c
u
r
v
e
s

f
o
r

C
1
M
slight
moderate
extensive
complete

PURMLSd(i)=cdf('norm',(1/betads)*log(URMLSd(i)/URMLSdbar(i)),0,1); %check = 0.5
end

%URMM,slight-moderate-extensive-complete
for i=1:4
URMMmodh(i)=URMM59(2);
URMMdeltaR(i)=URMM59(2+i);
URMMSdbar(i)=URMMdeltaR(i)*URMMmodh(i);
URMMSdbar1sd(i)=URMMSdbar(i)*exp(betads);
URMMSdbarmin1sd(i)=URMMSdbar(i)*exp(-betads);
URMMSd(i)=URMM59(5+2*i);
PURMMSd(i)=cdf('norm',(1/betads)*log(URMMSd(i)/URMMSdbar(i)),0,1); %check = 0.5
end

Thus, the damage distributions are produced for MHAZUS and MDBELA in large matrices.
The files used within the SELENA script to help with consultation of the main report
are as follows:- (changing att_sub.m, humanloss.m, and computetool.m)
earthquake.txt
%Earthquake scenarios information
%1st column is the weight for the logic tree scheme, %2nd column is latitude in degrees (e.g.: 50.90)
%3rd column is longitude in degrees (eg.: 10.90), %4th column is focal depth in km (eg.: 20), %5th column is Ms
magnitude (6.0), %6th column is Mw magnitude (6.0), %7th column is Fault orientation in degrees from North
(e.g. 0.0), %8th column is Dip Angle in degrees (e.g: 0.0), %9th column is Fault Mechanism:Strike-
Slip/Normal(1);Reverse(2);All(3), %10th column is the numerical code for the spectral shape as found in
spectral.m function. For example: 1 is UBC 2006 spectral shape.
1.00 28.84 40.9 25.00 7.20 7.20 75.00 90.00 1 1

attenuationboore.txt
%0.50 7 307 1007
1.00-1-301-1001
1.00 2 302 1002
builtarea.txt - built area as produced for the entire 5x50 values
capacity1.txt
capc_C1L-pre.mat 7 0.0010 %C1L
capc_C1M-pre.mat 7 0.0010 %C1M
capc_C1H-pre.mat 7 0.0010 %C1H
capc_URML-pre.mat 7 0.0010 %URML
capc_URMM-pre.mat 7 0.0010 %URMM

ecfiles.txt
1.00 elosssd.txt elossmd.txt elossed.txt elosscd.txt
elosscd1.txt
%NO C1L C1M C1H URML URMM LABEL
1 182.8 208.5 208.5 182.8 182.8 %RES
elossmd1.txt
1 58.0 66.2 66.2 58.0 58.0 %RES
elossed1.txt

1 184.6 210.5 210.5 184.6 184.6 %RES
elossmd1.txt
1 28.1 32.1 32.1 28.1 28.1 %RES

fragility1.txt
%mbt smedian sbeta mmedian mbeta emedian ebeta cmedian cbeta Pre-
CodeSeismicDesignLevel
1 0.0183 0.98 0.0292 0.94 0.0732 0.90 0.1829 0.97 %C1L
2 0.0305 0.73 0.0488 0.77 0.1219 0.83 0.3048 0.98 %C1M
3 0.0439 0.71 0.0701 0.80 0.1755 0.94 0.4389 1.01 %C1H
4 0.0081 1.15 0.0165 1.19 0.0411 1.20 0.0960 1.18 %URML
5 0.0127 0.99 0.0257 0.97 0.0640 0.90 0.1494 0.88 %URMM

HEADER.txt
%GEOUNIT LONGI LATIT SOILT C1LN C1LS C1LM C1LE C1LC C1MN C1MS C1MM C1ME C1MC C1HN C1HS
C1HM C1HE C1HC URMLN URMLS URMLM URMLE URMLC URMMN URMMS URMMM URMME URMMC
NUMB

headerocc.txt
%GEOUNIT RES

USING EQUATIONS MOVING ON FROM SPENCE(2007)
injury1.txt
%Slight Moderate Extensive Complete CompleteCollapse Label (numbers in percentage)
1 0.05 0.20 1.00 10 50 %C1L
2 0.05 0.20 1.00 10 50 %C1M
3 0.05 0.20 1.00 10 50 %C1H
4 0.05 0.20 1.00 10 50 %URML
5 0.05 0.20 1.00 10 50 %URMM
6 0.05 0.20 1.00 10 50 %NONE

injury2.txt
1 0.005 0.020 0.50 8 15 %C1L
2 0.005 0.020 0.50 8 15 %C1M
3 0.005 0.020 0.50 8 15 %C1H
4 0.005 0.020 0.50 8 15 %URML
5 0.005 0.020 0.50 8 15 %URMM
6 0.005 0.020 0.50 8 15 %NONE

injury3.txt
1 0.0 0.0 0.01 4 10 %C1L
2 0.0 0.0 0.01 4 10 %C1M
3 0.0 0.0 0.01 4 10 %C1H
4 0.0 0.0 0.01 4 10 %URML
5 0.0 0.0 0.01 4 10 %URMM
6 0.0 0.0 0.01 4 10 %NONE

injury4.txt
1 0.0 0.0 0.01 4 10 %C1L
2 0.0 0.0 0.01 4 10 %C1M
3 0.0 0.0 0.01 4 10 %C1H
4 0.0 0.0 0.01 4 10 %URML

5 0.0 0.0 0.01 4 10 %URMM
6 0.0 0.0 0.01 4 10 %NONE

numbuild.txt num of buildings as defined at the start

ocupmbt1,ocupmbt2, ocupmbt3, ocupmbt4, ocupmbt5.txt also 50 geocells.

Ocupmbtp.txt
%MBT RES Label
1 0.0343 %C1L
2 0.8306 %C1M
3 0.0853 %C1H
4 0.0344 %URML
5 0.0155 %URMM
6 0.0 %NONE

poptime.txt
%HOUR INDOOR OUTDOOR Label
1 0.99 0.01 % night 02:00 am
2 0.99 0.01 % day 10:00 am
3 0.99 0.01 % commuting 17:00 pm

soilfiles.txt
1.00 soilcenter1.txt

vulnerfiles.txt
1.0 capacity1.txt fragility1.txt

ECONOMIC LOSSES
Economic losses are directly calculated from the matrices of damage distributions large
amount of data so cannot be presented.

The other economic losses not detailed within this report will be available on the attached
DVD for a cell by cell basis for 100GMs taking the average of each cell and the standard
deviation. Shown are a couple of diagrams showing the results. Here is an extract of the
code:-
storeynum=[1 2 3 4 5 6 7 8 9 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 2 3 4 5 6 7 8 9 1 2 3 4];
floorarea=221.58;
cost=[175.8 200.5];
HAZEndCost=zeros(50,gm);
HAZEndCostcorr=zeros(50,gm);
for i=1:37
if storeynum(i)<=4; costtype(i)=storeynum(i)*floorarea*cost(1);
else ; costtype(i)=storeynum(i)*floorarea*cost(2);

end; end
for gmval=1:gm
HAZEndCost(1,gmval)=sum(0.16*costtype(1:37).*HBNs1uncorr(1:37,2,gmval)')+sum(0.33*costtype(1:37).*HBN
s1uncorr(1:37,3,gmval)')+sum(1.05*costtype(1:37).*HBNs1uncorr(1:37,4,gmval)')+sum(1.04*costtype(1:37).*H
BNs1uncorr(1:37,5,gmval)');
End;
After applying all of the damage distribution changing.

For the 100GMs for the spatially correlated GMs, this was applied to the MHAZUS and
MDBELA options in order to produce MDRs. From the 100 damage distributions, the mean
and standard deviation could be extracted (expressed in millions of euros for the standard
deviations).
Site
MDBELA
MDR
MDBELA STDEV of
REPAIR
MHAZUS
MDR
MHAZUS STDEV
(REP.)
2434
0.90 6.2 0.96 9.8
2489
0.89 3.0 0.94 4.7
2490
0.90 20.6 0.94 33.0
2491
0.88 22.4 0.95 36.0
2492
0.80 27.7 0.86 47.4
2493
0.81 31.8 0.86 53.9
2494
0.68 1.0 0.87 1.7
2543
0.69 0.4 0.79 0.7
2544
0.69 0.4 0.80 0.7
2545
0.71 0.3 0.81 0.6
2546
0.71 0.4 0.82 0.7
2547
0.78 20.3 0.83 34.9
2548
0.78 32.1 0.83 55.0
2549
0.78 27.5 0.84 47.3
2550
0.62 50.5 0.67 89.1
2551
0.63 40.0 0.69 69.7
2552
0.67 0.6 0.87 1.2
2601
0.68 0.8 0.80 1.5
2602
0.70 0.8 0.81 1.5
2603
0.54 0.5 0.64 0.9
2604
0.52 0.3 0.64 0.6
2605
0.59 28.9 0.66 50.6
2606
0.59 40.7 0.66 71.6
2607
0.60 19.3 0.67 34.1
2608
0.81 39.4 0.85 66.8
2609
0.65 1.5 0.85 2.9
2657
0.69 1.0 0.79 1.8
2658
0.68 0.8 0.79 1.5
2659
0.69 0.4 0.80 0.8

2660
0.51 0.1 0.63 0.2
2661
0.55 10.0 0.63 17.7
2662
0.56 6.3 0.66 11.4
2663
0.77 35.2 0.83 61.1
2664
0.78 26.2 0.84 44.7
2665
0.63 0.8 0.83 1.5
2666
0.08 0.0 0.80 0.0
2716
0.51 0.4 0.63 0.8
2717
0.50 0.3 0.62 0.4
2718
0.53 3.9 0.64 7.2
2719
0.55 4.4 0.64 7.8
2720
0.74 1.6 0.82 2.9
2721
0.65 0.8 0.82 1.5
2722
0.65 0.8 0.83 1.4
2723
0.35 0.1 0.83 0.2
2773
0.26 0.0 0.77 0.0
2774
0.51 0.4 0.62 0.8
2775
0.53 7.3 0.63 13.0
2776
0.54 10.6 0.64 19.2
2777
0.74 4.6 0.82 8.1
7678
0.28 0.0 0.83 0.1

The SELENA OUTPUT for the median and 1 standard deviation greater case. It can be seen
that the losses incurred will be significantly higher (expressed in Millions of Euros).
SELENA MEDIAN 84%
GEOUNIT
REPAIR
(MEuros)
REPLACE
(MEuros) MDR
REPAIR
(MEuros)
REPLACE
(MEuros) MDR
2434 0.035 0.037 0.95 0.038 0.037 1.04
2489 0.015 0.016 0.94 0.017 0.016 1.04
2490 0.107 0.114 0.94 0.119 0.114 1.04
2491 0.128 0.132 0.97 0.137 0.132 1.04
2492 0.100 0.125 0.80 0.123 0.125 0.99
2493 0.122 0.145 0.84 0.143 0.145 0.98
2494 0.005 0.006 0.86 0.006 0.006 1.00
2543 0.001 0.002 0.71 0.002 0.002 0.94
2544 0.002 0.002 0.89 0.002 0.002 1.02
2545 0.001 0.002 0.88 0.002 0.002 1.01
2546 0.001 0.002 0.60 0.002 0.002 1.01
2547 0.066 0.088 0.75 0.087 0.088 0.99
2548 0.078 0.141 0.55 0.139 0.141 0.99
2549 0.105 0.125 0.84 0.121 0.125 0.97
2550 0.110 0.210 0.52 0.186 0.210 0.88
2551 0.073 0.171 0.42 0.151 0.171 0.88
2552 0.003 0.004 0.69 0.003 0.004 0.92
2601 0.003 0.004 0.76 0.004 0.004 1.02
2602 0.003 0.004 0.74 0.004 0.004 1.02
2603 0.002 0.003 0.63 0.002 0.003 0.84
2604 0.001 0.002 0.79 0.002 0.002 0.93

2605 0.059 0.120 0.49 0.087 0.120 0.73
2606 0.101 0.170 0.60 0.147 0.170 0.87
2607 0.053 0.080 0.66 0.060 0.080 0.75
2608 0.097 0.174 0.56 0.154 0.174 0.88
2609 0.006 0.009 0.72 0.008 0.009 0.94
2657 0.004 0.005 0.87 0.005 0.005 1.01
2658 0.003 0.004 0.67 0.004 0.004 0.93
2659 0.002 0.002 0.88 0.002 0.002 0.99
2660 0.001 0.001 0.75 0.001 0.001 0.83
2661 0.029 0.045 0.66 0.037 0.045 0.82
2662 0.014 0.028 0.48 0.026 0.028 0.92
2663 0.126 0.155 0.81 0.144 0.155 0.93
2664 0.085 0.116 0.73 0.103 0.116 0.90
2665 0.003 0.005 0.63 0.004 0.005 0.94
2666 0.000 0.000 0.81 0.000 0.000 0.99
2716 0.001 0.002 0.46 0.002 0.002 0.95
2717 0.001 0.001 0.65 0.001 0.001 0.91
2718 0.008 0.018 0.43 0.015 0.018 0.86
2719 0.012 0.019 0.61 0.017 0.019 0.89
2720 0.004 0.007 0.59 0.006 0.007 0.83
2721 0.002 0.004 0.58 0.003 0.004 0.82
2722 0.003 0.004 0.72 0.004 0.004 0.84
2723 0.001 0.001 0.84 0.001 0.001 0.88
2773 0.000 0.000 0.82 0.000 0.000 0.96
2774 0.001 0.002 0.65 0.002 0.002 0.83
2775 0.019 0.033 0.59 0.030 0.033 0.92
2776 0.028 0.047 0.58 0.042 0.047 0.87
2777 0.010 0.021 0.48 0.017 0.021 0.82
7678 0.000 0.000 0.87 0.000 0.000 0.99
7679 0.000 0.000 0.000 0.000
7680 0.000 0.000 0.000 0.000
TOTAL 1.63 2.40668 0.68 2.21 2.40668 0.92
An example of a damage state matrix is available upon request, because given the size
of the data it is very difficult to display.

SOCIAL LOSSES
The social losses can be shown on the same GIS format as formulated earlier and is shown
here for MDBELA, as was produced in Daniell et al. (2009) using equations by Spence
(2007).
T1=[0.329 0.3 0.19 0.03 0.002 0.15]; T23=[0.208 0.3 0.23 0.04 0.002 0.22]; T4=[0.097 0.3 0.27 0.05 0.003
0.28]; M1=[0.236 0.5 0.12 0.08 0.004 0.06]; M23=[0.165 0.5 0.15 0.10 0.005 0.08]; M4=[0.094 0.5 0.18
0.12 0.006 0.1];


Inj(1,1:37)=[0.329 0.208 0.208 0.097 0.097 0.097 0.097 0.097 0.097 0.208 0.097 0.097 0.097 0.097 0.097
0.097 ... 0.329 0.208 0.208 0.097 0.097 0.097 0.097 0.097 0.097 0.208 0.208 0.097 0.097 0.097 0.097 0.097
0.097 0.236 0.165 0.165 0.094]; Inj(2,1:37)=[0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3
0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 ... 0.3 0.5 0.5 0.5 0.5];
Inj(3,1:37)=[0.19 0.23 0.23 0.27 0.27 0.27 0.27 0.27 0.27 0.23 0.27 0.27 0.27 0.27 0.27 0.27 0.19 0.23 0.23
0.27 0.27 0.27 0.27 0.27 0.27 ... 0.23 0.23 0.27 0.27 0.27 0.27 0.27 0.27 0.12 0.15 0.15 0.18];
Inj(4,1:37)=[0.03 0.04 0.04 0.05 0.05 0.05 0.05 0.05 0.05 0.04 0.05 0.05 0.05 0.05 0.05 0.05 0.03 0.04 0.04
0.05 0.05 0.05 0.05 0.05 0.05 ... 0.04 0.04 0.05 0.05 0.05 0.05 0.05 0.05 0.08 0.1 0.1 0.12];
Inj(5,1:37)=[0.002 0.002 0.002 0.003 0.003 0.003 0.003 0.003 0.003 0.002 0.003 0.003 0.003 0.003 0.003
0.003 0.002 0.002 0.002 0.003 0.003 0.003 0.003 0.003 0.003 ... 0.002 0.002 0.003 0.003 0.003 0.003 0.003
0.003 0.004 0.005 0.005 0.006];
Inj(6,1:37)=[0.15 0.22 0.22 0.28 0.28 0.28 0.28 0.28 0.28 0.22 0.28 0.28 0.28 0.28 0.28 0.28 0.15 0.22 0.22
0.28 0.28 0.28 0.28 0.28 0.28 0.22 0.22 0.28 0.28 0.28... 0.28 0.28 0.28 0.06 0.08 0.08 0.1];
%WILL BE CONVERTED TO TEXT FILES INPUT
population = dlmread('populationdata.txt',' ',[1 0 52 3]);
storeynum=[1 2 3 4 5 6 7 8 9 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 2 3 4 5 6 7 8 9 1 2 3 4];

daytime=population(1:50,4); nighttime=population(1:50,3);
for i=1:50; for j=1:37
storeysbuild(i,j)=storeynum(j)*buildvalue(i,j);
end; end
for i=1:50
daypopstorey(i)=population(i,4)/sum(storeysbuild(i,1:37));
nightpopstorey(i)=population(i,3)/sum(storeysbuild(i,1:37));
end
for gmval=1:gm; for j=1:6;
DBsocialuncorr(1,gmval,j)=sum((DBNs1uncorr(1:37,4,gmval)').*nightpopstorey(1).*storeynum(1:37).*Inj(j,1:37)
);
End; End;

MHAZUS and MDBELA social loss results are in the main body of the text. Further
modifications to the code in OPAL-GEM2 will be produced in order to make user-friendly
input and user-input for regional social loss functions.

Longitude []
L
a
t
i
t
u
d
e

[
]
DBELA: Map of the total casualties during the day
28.89 28.9 28.91 28.92
40.98
40.985
40.99
40.995
41
41.005
41.01
41.015
41.02
41.025
41.03
0
100
200
300
400
500
600
700
Longitude []
L
a
t
i
t
u
d
e

[
]
DBELA: Map of the total casualties during the night
28.89 28.9 28.91 28.92
40.98
40.985
40.99
40.995
41
41.005
41.01
41.015
41.02
41.025
41.03

Full output tables are included in the main body of the text for MDBELA, MHAZUS and
SELENA. It can be seen that losses in the night are mainly in the south where there is mostly
residential housing. This could have been inferred from remote sensing. The casualties in the
day are reasonably well spread but are most in the geocells of soft soils.

SELENA OUTPUT for median values using the details of Spence (2007)
Population Deaths Night Injuries Day Injuries
GC Night Day Night Day UI I1 I2 I3 I4 UI I1 I2 I3 I4
2434 3646 2917 616 493 224 685 599 115 7 179 548 479 92 5
2489 1373 1099 205 164 73 224 199 37 2 59 179 159 30 2
2490 23854 19085 3279 2624 1178 3618 3181 607 36 943 2895 2545 485 29
2491 11833 9464 1768 1414 759 2217 1761 376 22 607 1773 1409 301 17
2492 13191 10553 1057 846 422 1253 1040 211 12 337 1002 832 169 10
2493 15340 12272 1549 1239 599 1786 1515 299 18 479 1428 1212 239 14
2494 133 133 11 11 7 20 12 4 0 7 20 12 4 0
2543 593 9137 30 469 24 56 33 9 1 367 869 514 145 8
2544 602 9284 59 906 34 86 61 14 1 521 1319 934 219 12
2545 507 7810 49 750 28 70 50 12 1 424 1079 771 179 10
2546 612 9431 22 346 21 47 26 8 0 328 722 394 119 6
2547 7441 5952 491 392 183 557 479 94 6 146 446 383 75 4
2548 8202 6559 257 205 124 345 260 58 3 99 276 208 47 3
2549 11219 8973 1101 881 472 1370 1094 231 13 377 1096 875 185 11
2550 16795 13434 469 375 220 635 474 108 6 176 508 379 86 5
2551 24221 19380 459 367 284 620 468 98 5 227 496 375 78 4
2552 90 90 3 3 3 8 4 1 0 3 8 4 1 0
2601 1061 16358 66 1015 43 104 70 18 1 666 1609 1075 270 15
2602 947 14589 55 854 38 88 59 15 1 585 1362 905 225 12
2603 736 11347 29 446 16 42 30 7 0 244 650 459 109 6
2604 488 7516 34 518 20 50 35 8 0 305 774 537 129 7
2605 10173 8137 256 205 103 303 252 51 3 82 243 201 41 2
2606 9842 7870 360 288 154 451 359 77 4 123 361 287 61 4
2607 4671 3735 233 187 92 274 229 46 3 74 219 183 37 2

2608 20841 16674 686 549 266 788 671 132 8 213 630 537 105 6
2609 219 219 9 9 9 26 11 5 0 9 26 11 5 0
2657 1297 19993 122 1876 62 160 123 26 2 950 2461 1891 407 23
2658 1013 15621 44 677 26 67 46 11 1 406 1032 707 174 10
2659 679 10463 64 987 36 91 66 15 1 550 1407 1012 233 13
2660 354 5453 21 324 12 30 22 5 0 189 462 333 76 4
2661 3650 2919 172 138 85 239 175 40 2 68 191 140 32 2
2662 2281 1825 48 39 30 83 52 15 1 24 66 42 12 1
2663 17780 14226 1479 1184 593 1764 1458 298 17 474 1412 1166 238 14
2664 13219 10576 875 700 356 1046 862 176 10 285 837 690 140 8
2665 122 122 4 4 5 13 5 2 0 5 13 5 2 0
2666 8 8 0 0 1 2 0 0 0 1 2 0 0 0
2716 669 10316 14 214 14 32 16 5 0 221 491 251 82 4
2717 421 6484 17 256 12 30 18 5 0 187 468 279 79 4
2718 1406 1124 25 20 18 47 27 8 0 14 37 22 6 0
2719 1588 1270 63 50 31 89 64 15 1 25 71 51 12 1
2720 648 519 22 18 12 34 23 6 0 10 27 19 5 0
2721 95 95 2 2 3 9 3 2 0 3 9 3 2 0
2722 94 94 4 4 4 11 5 2 0 4 11 5 2 0
2723 31 31 1 1 2 6 2 1 0 2 6 2 1 0
2773 229 3537 12 182 18 37 16 6 0 280 573 240 95 5
2774 669 10316 27 413 17 43 28 7 0 265 666 436 111 6
2775 2715 2171 98 78 51 141 100 24 1 41 112 80 19 1
2776 3869 3095 134 107 70 193 137 33 2 56 155 110 26 1
2777 1696 1356 37 29 19 54 38 9 1 15 43 30 7 0
7678 25 25 1 1 2 6 2 1 0 2 6 2 1 0
7679 0 0 0 0 0 0 0 0 0 0 0 0 0 0
7680 0 0 0 0 0 0 0 0 0 0 0 0 0 0
243188 353657 16441 22862 6876 19950 16259 3352 195 11660 31095 23194 5200 295

From this analysis, comes the idea of adapting MDBELA to a MHAZUS situation.

Idea for the defining LS1 at yield, and LS3 at the end of the capacity curve for HAZUS
method via a DBELA capacity curve.
For one of the reduced number of buildings, its gradation-based value can be applied as a
value between 0 and 4, based on displacement ductility and period, although each point is
calculated based on a different iteration of the reduction factor for a certain limit state and this
causes difficulties in finding the exact ratio. However, this can be calculated by a continuous
type of system between reduction factor and period based on ductility, in order to gain an
exact position of the continuous ductility value. (i.e. using the LS3 reduction factor spectra
and comparing it to LS3, separately from the LS2 value, which will be compared to the LS2
reduction factor spectra (accounting for viscous damping of the equivalent SDOF system).
Appendix D: Multicriteria Analysis and Selection from OPAL-GEM1

Daniell, April 2009 D1

APPENDIX D: MULTICRITERIA ANALYSIS AND
SELECTION FROM OPAL-GEM1
This section involved shows the criteria used in the analysis of the 28 different ELE software
packages. This list of criteria is by no means exhaustive but these are the components
determined to be most important through this thesis, and it is required that users make their
own informed opinions after going through the OPAL procedure. The format for OPAL-
GEM2 has been applied by taking into account these criteria, with respect to the many ELE
software packages reviewed and assessed. Where 6 components are mentioned in the table
below, this refers to the exposure, hazard, vulnerability, damage-loss conversion, GIS and
results. These criteria could be applied to any such project.
FULL CRITERIA
ANALYSIS
Qual/
Num
INDICATOR
Optimal
Solution
Open-Source
Qual/
Num
Fully open source on all levels with
panel administrator to ensure
progress and user support
Fully Open
Global Availability Qual
Internet based and normal software
versions, current
Available
Ease of download Qual
Not just on Internet, downloadable, fast,
useable
Very Easy
Instructions for use Num Manual updated every few months Updated
Documentation Num
Clear methodology, results and updating
with new versions
Clear
Ability to contact
developer
Qual
Email, phone contact, updated website and
open discussion
Able, 24/7
Ease of coding Qual
Extensible, Clear, public domain, separate 6
components
Understandable
Coding language used Qual
Fortran, C++, Web-based, Java, VB, XML,
Excel, Python, Matlab, SOSEWIN, self-
organising systems.
Python, OOP,
GUI
No. of components
open-source
Num
Which components of the system are open
and closed source?
All Open
No GIS License Num
No cost associated with GIS, No trial period.
Fully integrated into system.
Open GIS
Data Access Num Does the user have access to all data? Total
Virtual Communities Qual
Setup of virtual communities like Matlab
Central. User-added code, and help.
Virtual
Community
Enabled
Extensible
Architecture
Qual
Any user, Anywhere for personal use.
Checked for public use. Architecture added
Extensible

to new alarm systems etc.
Software
Qual/
Num
Adaptable, easy to use, free, hybrid,
fast, license-free software
Free, Fast,
Accurate
Integration with
internet
Qual
Online updates available when online. User
community page also current.
Internet-Useable
Open editing (Wiki) Qual
A wiki style interface so that users can
update code, leave ideas, and update bugs
in software
WikiOPAL
enabled
Computation speed Num.
Testing for the various options -
optimisation of coding
Fast
Global Language Qual
English used for the analysis, translatable,
simple architecture/Globally available
software language
English,
Translatable
Hardwiring Qual
Functions are hardwired in many cases into
the coding - SELENA etc.
Not hardwired
Adaptability (Allow
additions)
Qual
Not a wiki update but a direct system to
handle bugs, email-based with a user no.
Adaptable
Recent developments Qual
Able to use new versions of intensity scale,
new response spectra etc.
Recent
(Dynamic)
Update method
(versions)
Qual
Not via DVD update - email user list - and
downloadable
Download or
Email
Hybrid methods Num.
Use of more than 1 method to adapt to diff.
situations (Multiple Level Systems of all 6
components)
Multiple
Reliance on past
methods
Num.
Collaboration, or previous models - inherent
decision-making
Uses
collaboration
Current Application Qual.
Be relevant to current conditions and
population/settlement data
Current
Development Status Num.
Is provided and updated list of bugs fixed
included, longevity of the project
Current
Updated recently Num. How recently updated?
Recent
(Dynamic)
Computing Power
required
Num.
Level of hardware required, GHz, GB and
method
Minimal needed
Licensing Num. Software-registration, free access, etc.
No licenses
required
Global testing under all
platforms
Qual.
Windows, Linux, Unix, Mac, Vista,
Windows95 and various service packs - no
need to change hardware
Global testing OK
Learning and Tutorials Qual.
Learning and Tutorials provided for user
awareness
Included
Optimisation Qual.
Software fully optimised with no slow parts
and user warnings provided
Optimised
Data plotting Qual.
Integration with user-based data plotting -
many options
In-built
Backup Systems Qual.
Backup systems in place for loss of data,
more datasets, etc.
In-place
Previous Knowledge
needed
Qual/
Num
OPAL Procedure imparted to user, or
if known, level-based system
All Users
possible
OPAL Procedure
Not
Eval.
Familiarity of the users with this procedure
Familiar after
OPAL
Ease of Use Qual.
User knowledge check, advanced and
normal settings
Many options
Tutorial Based Loss Not Run through explaining methodology, to the Present

Analysis Eval. final results
Technology
Qual/
Num
Technological Systems in place for
the 6 components and updates
applied and able to be progressively
improved
Latest
Technology
GIS Licensing Num.
GIS- closed or open source. Many open
source ELE use closed GIS.
No GIS license
required
GIS Production Num.
Integrated GIS, or output in GIS format. Use
of GIS for exposure selection?
Uses GIS
Dynamic Improvement Qual.
Technology updates kept up to date and
applied to software package
Dynamic
Remote Sensing
Accuracy
Num.
%accuracy in terms of height, building type,
GIS systems and other methods
Accurate
Remote Sensing
(MIHEA, ground
truthing)
Qual.
What methods used and how well have they
been improved and checked?
Mult. Methods
used.
Rapid Response/Real
Data
Qual.
Updating of Systems in place to apply real
data to software then rapid response
Possible
Exposure
Qual/
Num
Inventory Elements, Typology
checks, Temporal and Spatial
Changes, Global Cover
Current and
Complete
Collection Num.
Method of collection - collaboration -
reliance on technology etc.
All collected
Global cover Num.
% global population covered - building type,
cost, age of building, demographics etc.
All covered
Inventory Elements Num.
What is the use of the elements and their
importance?
Full Inventory
Critical + Lifelines Qual.
Most important elements are able to be
damage modelled and risk analysis.
Identified
Temporal Data Qual.
Temporal changes to the buildings recorded
and satellite images - retrofitting history,
seismic code (if any) adhered to
Present
Spatial Data Num.
Spatial changes able to be recorded and to
be collected
Present
Location Qual.
Accuracy of the exposure calculation at the
location which is being tested
Accurate
Accuracy of Typology Num.
The number of bins used and able to be
used to maintain computational speed
Accurate
Allow addition of
building types
Qual.
Not set and hardwired and new building
types able to be calculated
Allow
Population
Assessment
Num.
Accuracy of the population at any time of
day, demographics
Accurate
Risk Indicators Qual.
Exposure data used to identify risk
indicators
Possible
Data Management Qual.
Ease of data access, time taken to access,
easy storage method
Quick and Easy
Hazard/Demand
Quantification
Qual/
Num
Earthquake Hazards Considered,
Adaptability, History and Real-time,
PSHA/DSHA and Instrumental details
All Options
considered
Spectrum-Based vs.
Intensity
Qual.
Spectrum-based technique used or
intensity-based method - or hybrid
Multiple
User-defined
earthquakes and
events
Num.
Possible to define earthquakes as a user, or
a user-defined GM
Yes
Observed GMs Num.
Past Eqs or real-time GMs used to produce
GMs
Yes

Empirical GMs Num. GMPEs and their global application Yes
Theoretical GMs /
basis
Num.
Produced Theoretical modelling of ground
motions, Level of theoretical modelling of
source, path and site effects
Yes
Ability to update GMs Num.
Is the user able to apply regional GMs and
new GMs?
Yes
PSHA (Reliance) Qual.
Probabilistic techniques used, Poissonian or
Time-Dependent, Multiple options
Option available
DSHA (Reliance) Qual.
Either observed or deterministic - MCE -
and accurate data, possible logic tree
applied to parameters
Option available
Disaggregation Num.
Can the data be disaggregated into the
components (M-D-)?
Option available
Spatial or Temporal
correlation
Num.
Spatial and temporal correlation between
GMs possible?
Option available
Stochastic Catalogues Num.
Stochastic catalogues applied in order for
completeness?
Option available
Optimisation
techniques
Qual.
Demand data and other storage
optimisation?
Yes
Historic EQ Catalogues Qual.
Are historic EQ catalogues used and what
is their extent?
Present and
Current
Active Fault Database Qual.
How extensive is the active fault database
in the software package - history on fault
Present and
Current
Instrumental EQ
catalogue
Qual.
Instrumental data and the ability to use such
applications
Present and
Current
Soil database/Site
Class
Qual.
Extent of the site classification scheme
used + conversion system used to final
ordinates
Complete
Geodetic Data
Standards
Qual.
Quality and accuracy, collection, code types
and data, transfer and exporting, metadata
and accessibility for GPS and GIS
Quality,
Collection,
Transfer, Use is
accurate and
quick
Liquefaction Qual.
Liquefaction taken into account, Intensity
based, detailed or not
Yes
Fault Rupture Qual.
Fault rupture taken into account, GIS
based, detailed or not
Yes
Landslide/SS Qual.
Landslides and SS taken into account,
Empirical or Analytical?, detailed or not
Yes
Tsunami (linking into
other software)
Qual.
Tsunamis, links to other software, basis,
details etc.
Yes
Other - Aftershock +
Volcano + Quake
Lakes
Qual.
Are aftershocks, volcanoes, quake lakes
taken into account, basis and details
Yes
Integration of H/V data
etc.
Qual.
Extent of site velocity and characterisation
methods
Integrated if quick
Vulnerability
Qual/
Num
Empirical and Analytical, Occupancy,
Structural, Variability and Quality of
Vulnerable Stock
All options
considered
Empirical Ability Num. Use of empirical methods in 1 level or more Multiple
Rapid Response Qual.
How easily is vulnerability damage
constrained exactly for rapid response?
Well
Correlation with
Damage
Num. Damage state correlation Yes
Analytical Method
used
Num.
Use of analytical methods - CSM,
displacement based - 1 level +
Multiple

Complexity of
Algorithm
Qual.
This affects user-knowledge and coding
complexity
Complex with
Speed and
Accuracy
Damage states Num.
Which damage states are used and how
well are they defined
Defined
accurately
Uncertainty Qual. How well is uncertainty construed?
Taken into
account
Global Vuln. Fragility Qual.
PAGER-CAT or otherwise, how well are the
fragility functions globalised
Yes
Ability to detect
indicators
Qual.
Methods used to pick up vulnerability
indicators
Yes
Occupancy Qual. Use and Occupancy Rate All wanted
Structural Qual.
Basic Structural features i.e. materials used,
irregularities in plan and elevation, building
height and internal characteristics, complex
failure mechanisms
All wanted
Quality of stock Qual.
Age and no. of age brackets, correlation to
seismic codes, variability
All wanted
Material
variability/Building
differences
Qual.
Probabilistic techniques and distributions to
quantify building variability
Yes
Pictures/GIS Qual.
Pictures and GIS of vulnerable houses via
screening methods etc? Panoramica/Street-
view
Yes
Risk and Damage-
Loss Conversion
Qual/
Num
Economic and Social Loss
Equations, Dynamic and Social
Changes, Visualisation and
Cost/Benefit
Current and
Accurate
Economics (level of
layers)
Num.
Direct, Indirect, analysis, GNP basis, repair,
MDR
All layers
Social (level of layers) Num.
Deaths, levels of injuries, homeless, shelter
needs etc.
All layers
Complexity Qual.
How complex are the calculations from
damage-loss conversion (MAEviz vs.
QLARM)
Complex If
desired
Output Accuracy Num. Calibration with real Eqs Accurate
Dynamic Vulnerability Qual.
How well is the dynamic changes of
vulnerable regions taken into account in
equations?
Dynamic
Social Vulnerability Qual.
Age, demographics, community awareness
programs included in the analysis and
output
Social Vuln.
Included
Age of Eqns Qual.
Age and use of the equations for social and
economic costing. New data, new accuracy.
Current
Uncertainty Qual.
How well is the uncertainty constrained -
are logic tree approaches used?
Taken into
account
Visualisation Qual.
Visualisation via GIS or graphs of
comparative scenarios or for 1EQ
Yes
Decision possibilities Qual.
Logic Tree Approach employed or Expert
Opinion setup
Many options
Single or Portfolio -
Risk
Insurance/Financing
Qual.
Does it take single and portfolio of buildings
into account for insurance purposes
Many options
Disaggregation Qual. Can the economic data be disaggregated Yes
Cost/Benefit Analysis Num. Option possible? Yes
Use for land planning
and zoning
Qual.
GIS layering and scenario overlaying to
help governments
Yes

Forecasting
Qual/
Num
Ability to have speed and accuracy
to define the losses of an EQ before
for a deterministic or probabilistic
scenario
Accurate and
Fast
Speed Num. For a certain large scale calculation - speed Fast
Accuracy Num.
Large scale accuracy - based on
comparison with past earthquakes
Accurate
Online Database of
tested scenarios
Num.
For use in Post EQ - can apply quick
previous estimates.
Extensive and
Online
Government
Involvement
Qual.
Stakeholder meetings and setup of systems
using the program. Smaller, readable
versions of EQ Loss estimation
Readable
Post-EQ
Qual/
Num
Ability to have speed and accuracy
to define the losses of a real-time EQ
+ disaster management components
Fast and
Accurate
ShakeMaps Qual.
Produced or not? Collaboration with people
(eg. USGS 'Did you feel it?')
Produced
Direct Management Qual.
Uses lifeline data, and the forecast to
implement directly into management after
an EQ
Yes
DMT Qual.
Technological tools used before and after
earthquake as part of the system
Yes
Speed of Calc. Num.
Speed from earthquake source to an output
loss estimate
Fast
Accuracy Num. Comparison with previous earthquakes Accurate
Optimisation of
Building Choice
Qual.
Using screening methods, before and after -
disaster management goes to most at-risk
areas
Accurate and
Current
Functionality/Communi
cations
Qual.
How easily is the information transferred to
relevant people (Army, government)
Easily
Topography/GIS Qual.
Output of Topography and GIS data for use
on the ground
Outputted
Comparison with other
ELEs
Qual.
Reads in PAGER2, ELER, WAPMERR,
SAFER and builds comparison table, time
and accuracy in real-time
Compared with
all estimates
Aftershock modelling Qual.
In-tune with the Hazards looked at - warning
system production
Done
Output
Qual/
Num
Ease of understanding output - GIS,
recommendations, summary
Easy and
Extensive
GIS Qual.
Format easily visible within the GIS program
in the ELE package
Easily Produced
One Page Summary Qual.
Similar to the onePAGER but with more
features - economics, visualisation,
uncertainties
Produced
Format Qual.
Data processing time-consuming after.
Need easy application - simple to put in
Open Office Math or already graphed
Easy
Recommendations for
Communities
Qual.
Dependent on the country, city, study
undertaken - retrofitting - could also include
a relevant contact
Produced
Region Specific
Recommendations for
Codes
Qual.
Comparison of the damage through an
assessment with codes i.e. masonry
buildings 1-3 stories high are most likely to
collapse
Produced

Appendix E: Correspondence with other ELE Software Package Producers

Daniell, April 2009 E1
APPENDIX E: CORRESPONDENCE WITH OTHER ELE
SOFTWARE PACKAGE PRODUCERS

Contact Topic and result
Ed Anderson, Frances
Ghesquire, Stuart Gill
CAPRA Details, Documentation and pre-release information
Trevor Dhu, David Robinson,
Duncan Gray
EQRM editing to the code, problems with the code, general
discussion about methods within EQRM
Phil Cummins Remote Sensing details in the Asia-Pacific region, PHIVOLCS
project and other AusAid projects.
Ron Eguchi, Charlie Huyck,
ImageCatInc, Shubharoop
Ghosh
EPEDAT sending information and possibly a copy of EPEDAT
software information was sent in hard form
InLET details and documentation given
Geohazards International,
Fumio Kaneko, RMSI
RADIUS details and software possibility from Geohazards
International however, no comment from OYOCorp
Helen Crowley, Rui Pinho,
Ihsan Bal
DBELA details and methods associated with the coding and use of
DBELA
Max Wyss QLARM and QUAKELOSS2 information and the coding keen for
discussion and correspondence. Also discussion of SAFER.
Friedemann Wenzel etc. EQSIM code wanted and documentation, documentation gained.
However, need to wait for the EQSIM software, it needs to be
revitalized but has much potential and could be used to base the
coding of OPAL-GEM2 on.
Nina Frolova Extremum QUAKELOSS Max Wyss gave details and will
further discuss
Numerous Geoscience
Australia, University of
Adelaide, ANU and
proprietary Australian people
Attempt to gain dataset of building inventory for different locations
in Australia. Gained EQRM Matlab and also Newcastle walk-down
survey

Numerous other people were contacted with no response, as shown in the table of contact
details in Table 4-1.

A
p
p
e
n
d
i
x

E
:

C
o
r
r
e
s
p
o
n
d
e
n
c
e

w
i
t
h

o
t
h
e
r

E
L
E

S
o
f
t
w
a
r
e

P
a
c
k
a
g
e

P
r
o
d
u
c
e
r
s

D
a
n
i
e
l
l
,

A
p
r
i
l

2
0
0
9

E
2

MSc Dissertation 2009 Comparison and Production of Open Source Earthquake Loss Assessment Packages James Daniell

Dissertation2009 Daniell

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dissertation2009 Daniell

Uploaded by

Copyright:

Available Formats

Istituto Universitario

You might also like